DEVELOPING EVIDENCE-BASED DESIGN METRICS AND

METHODS FOR IMPROVING HEALTHCARE SOUNDSCAPES

A Dissertation
Presented to
The Academic Faculty

by

Selen Okcu

In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy in the
School of Architecture

Georgia Institute of Technology

May, 2011
 DEVELOPING EVIDENCE-BASED DESIGN METRICS AND
METHODS FOR IMPROVING HEALTHCARE SOUNDSCAPES

Approved by:

Dr. Craig Zimring, Advisor Dr. Ermal Shpuza

School of Architecture Department of Architecture
Georgia Institute of Technology Southern Polytechnic State University

Dr. Erica Ryherd, Co-advisor Howard Pelton

School of Mechanical Engineering Pelton Associates
Georgia Institute of Technology

Dr. Sonit Bafna

School of Architecture
Georgia Institute of Technology

Date Approved: 3/18/2011

 ACKNOWLEDGEMENTS

I am grateful to my committee for their guidance and assistance in completing this

dissertation work. Dr. Craig Zimring was an open minded supporter at all times and a

guide. Dr. Ryherd introduced me to the interesting world of acoustics. Her willingness

and openness to address multi-disciplinary research was always motivating. Dr. Bafna`s

thoughtful comments provided some of the very interesting and key directions in this

study. I highly appreciate Dr. Shpuza`s collaboration that helped facilitate some of this

work. Our early brainstorming sessions helped me start critically thinking about the

relationships between design and acoustics. He was also very generous to share with me

his computerized floor-plate shape analysis tool. I am also grateful to Mr. Pelton for

providing his expert knowledge. It has been very much mind opening to work with

experts from different disciplines.

I also would like to thank to numerous people outside the Georgia Institute of

Technology who contributed to this study directly and indirectly. I appreciate Dr. Owen

Samuel`s thoughtful comments and suggestions. I am also thankful to nurse educators

Ann Huntley and Mary Still, registered nurses Tim Rice and Anya Freeman and to all

Neuro-ICU and MedSurg ICU nurses, patients and family members for their continuous

help and patience during noise level measurements in the units.

I cannot say enough about my dearest parents (Ayse and Hikmet Okcu) for their role in

getting me where I am. There is no greater comfort than having a supportive and loving

iii
family who believes in you deeply. They were always there during the good and bad

times to help and support me to pursue my goals. I also thank to my brother

(Ahmet Okcu) for taking care of my parents when I was not physically there. I am also

very thankful to my dearest Tim and Mr. and Mrs. Frank for their support and generosity.

I can never thank enough to my dearest Tim for his kindness and warmness, essentially

for being a wonderful human being with a good soul.

iv
 TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS iii

LIST OF TABLES ix

LIST OF FIGURES xi

LIST OF ABBREVIATIONS xvi

SUMMARY xix

CHAPTERS

CHAPTER 1. RESEARCH PROBLEM

1.1 BACKGROUND 1
1.2 PROBLEM STATEMENT 2
1.3 SCOPE AND OBJECTIVES 2
1.4 RESEARCH QUESTIONS 4
1.5 SIGNIFICANCE 8

CHAPTER 2. HOSPITAL SOUNDSCAPES AND STAFF OUTCOMES

2.1 LITERATURE REVIEW

2.1.1 Noise levels in hospitals 11
2.1.2 Hospital noise metrics 11
2.1.3 Hospital noise level measurement methods 12
2.1.4 Hospital noise and its effects on staff members 14
2.1.5 Medical alarms as noise sources and their effects 24
2.2 EMPIRICAL STUDY 1:
CHARACTERIZING SOUND ENVIRONMENTS OF ICUS
2.2.1 Scope 28
2.2.2 Methodology 29
2.2.3 Findings 32
2.2.4 Discussion 46
2.3 EMPIRICAL STUDY 2:
TIME-BASED VARIATIONS OF HOSPITAL NOISE
2.3.1 Scope 48
2.3.2 Methodology 48
2.3.3 Findings 51
2.3.4 Discussion 59

v
 2.4 EMPIRICAL STUDY 3:
EFFECTS OF PARTICULAR HOSPITAL NOISE SOURCES
2.4.1 Scope 61
2.4.2 Methodology 61
2.4.3 Findings 61
2.4.4 Discussion 64

CHAPTER 3. CHARACTERISTICS OF PATIENT AUDITORY

MONITORING

3.1 LITERATURE REVIEW

3.1.1 Patient monitoring in critical care units
3.1.1.1 Overview of patient monitoring 65
3.1.1.2 The profile of caregivers conducting patient
monitoring 67
3.1.1.3 Patient monitoring and other critical nurse tasks 69
3.1.1.4 Types of patient monitoring 69
3.1.1.5 Visual patient monitoring 70
3.1.1.6 Auditory patient monitoring 70
3.1.2 ICU auditory cues 72
3.1.3 Factors impacting auditory monitoring 74
3.1.3.1 Sound detection 75
3.1.3.2 Sound recognition 76
3.1.3.3 Sound localization 78
3.2 EMPIRICAL STUDY:
CHARACTERIZING AUDITORY PATIENT MONITORING
3.2.1 Scope 82
3.2.2 Methodology 83
3.2.3 Findings 85
3.2.4 Discussion 94

CHAPTER 4. ACOUSTICS OF ARCHITECTURAL SPACES

4.1 ROOM ACOUSTICS 98

4.2 ROOM ACOUSTICS PARAMETERS 99
4.3 ACOUSTICS OF PROPORTIONAL SPACES 99
4.3.1 Sound behavior in rooms 99
4.3.2 Floor-plate design and acoustics 100
4.3.3 Statistical analysis of floor-plate design and acoustics 104
4.4 ACOUSTICS OF NON-PROPORTIONAL SPACES 108
4.4.1 Sound behavior in long spaces 108
4.4.1.1 Field measurements 109
4.4.1.2 Simulation studies 111
4.5 ACOUSTICS OF LONG SPACES WITH BRANCHES 117
4.6 CONCLUSION 125

vi
CHAPTER 5. MEASUREMENTS AND SIMULATIONS OF THE
ACOUSTICS OF LONG ENCLOSURES

5.1 EMPIRICAL STUDY 1:

VALIDATION of CATT ACOUSTIC PREDICTIONS 127
5.1.1 L shaped corridors 129
5.1.2 Race track corridor 132
5.1.3 T-shaped corridor 138
5.1.4 Conclusion 143
5.2 EMPIRICAL STUDY 2:
SOUND BEHAVIOR IN SINGLE vs. INTER-CONNECTED
CORRIDORS 143
5.2.1 Methodology 144
5.2.2 Findings 145
5.2.3 Conclusion 147

CHAPTER 6. DESIGN AND ACOUSTICS OF INTERCONNECTED

HOSPITAL CORRIDORS

6.1 BACKGROUND 149

6.1.1 Auditory monitoring in the hospital corridors 149
6.1.2 Effects of reverberation time on auditory monitoring 151
6.1.3 Design characteristics of hospital corridors 153
6.1.3.1 Scope 153
6.1.3.2 Overall corridor design 154
6.1.3.3 Corridor spatial organization 159
6.2 PILOT STUDY:
LINKING ACOUSTICS AND DESIGN VARIABLES OF
ACTUAL EDUCATIONAL CORRIDORS 163
6.2.1 Scope 163
6.2.2 Methodology 163
6.2.3 Findings 166
6.3 EMPIRICAL STUDY 1:
LINKING ACOUSTICS AND DESIGN VARIABLES OF 60
THEORETICAL CORRIDOR MODELS 169
6.3.1 Scope 169
6.3.2 Methodology 170
6.3.3 Findings 180
6.4 EMPIRICAL STUDY 2:
LINKING ACOUSTICS AND DESIGN VARIABLES OF 20
THEORETICAL CORRIDOR MODELS 184
6.4.1 Scope 184
6.4.2 Methodology 184
6.4.3 Findings 188
6.5 DISCUSSION 195

vii
CHAPTER 7. VALIDATION STUDY

7.1 SCOPE 198

7.2 METHODOLOGY 198
7.3 FINDINGS 211
7.4 DISCUSSION 224

CHAPTER 8. CONCLUSION

8.1 SUMMARY OF FINDINGS 226

8.2 CONTRIBUTION 235
8.3 LIMITATION 237
8.4 FUTURE WORK 238

APPENDICES

APPENDIX A: DEFINITION OF TERMS 239

APPENDIX B: DESIGN QUALITIES OF THE TWO UNITS 248
APPENDIX C: NOISE LEVEL FIELD MEASUREMENTS 251
APPENDIX D: NOISE LEVEL MEASUREMENT GUIDE 257
APPENDIX E: HOSPITAL NOISE SURVEY QUESTIONS 261
APPENDIX F: OCCURRENCE RATE 267
APPENDIX G: IMPULSE RESPONSE FIELD MEASUREMENTS 270
APPENDIX H: IMPULSE RESPONSE MEASUREMENT GUIDE 276
APPENDIX I: COMPUTER MODELING 278
APPENDIX J: JUST NOTICEABLE DIFFERENCE (JND) 296

REFERENCES 299

VITA 312

viii
 LIST OF TABLES

Page

Table 2.1: Mean perception ratings of noise-induced nurse outcomes in the two ICUs 38

Table 2.2: Mean perception ratings of loudness and annoyance in the two ICUs 39

Table 2.3: Correlations between objective noise and subjective noise levels 41

Table 2.4: Speech interference levels at different locations in the two ICUs 46

Table 2.5: Annoyance and loudness perceptions for weekdays vs. weekend 51

Table 2.6: Annoyance and loudness perceptions during different times of the day 53

Table 2.7: Perceived medical alarm-induced outcomes 62

Table 2.8 Distribution of medical-alarm induced outcomes 63

Table 3.1 Distribution of averaged RT30 across frequency 94

Table 5.1: Material absorption coefficients of the L-shaped corridor 130

Table 5.2: JND values for the room averaged RT30 131

Table 5.3: JND values for RT30 measured at 150ft away from the source 132

Table 5.4: Material absorption coefficients of the race track design corridor 134

Table 5.5: JND values for room averaged RT30 136

Table 5.6: JND values for RT30 at the selected receiver locations 137

Table 5.7: Material absorption coefficients of the T-shaped corridor 139

Table 5.8: JND values for room averaged predicted and measured RT30 140

Table 5.9: JND values for RT30 at the selected receiver locations 142

Table 6.1: Partial correlation analysis results linking corridor design and RT30 167

Table 6.2: Partial correlation analysis results linking radial distance and RT30 169

Table 6.3: Design characteristics of theoretical model groups 180

Table 6.4: Correlation analysis results (absorption coefficient = 0.3) 182

ix
Table 6.5: Correlation analysis results (absorption coefficient = 0.1) 183

Table 6.6: Partial correlation analysis results linking RT30, and RGD + VF 190

Table 6.7: Regression analysis results 190

Table 6.8: Bivariate correlation analysis results 193

Table 6.9 Summary of the empirical study findings conducted in Chapter 6 196

Table 7.1: Absorption coefficients of materials applied to surfaces of the two wards 211

Table 7.2: JND results of measured RT30 in the two wards 212

Table 7.3: JND results of simulated RT30 at the “T” and “+”shaped intersections 214

Table 7.4: JND results of RT30 at the corridors w/ different number of branches 215

Table 7.5: JND results of RT30 at the corridors with different number of turns 218

Table 7.6: JND results of RT30 at the corridors with different corridor length 220

Table 7.7: JND results of RT30 at the corridors with different total corridor length 222

Table 7.8: JND results of RT30 at the corridors with similar total corridor length 224

x
 LIST OF FIGURES

Page

Figure 1.1: Phases of the research xxii

Figure 2.1: Overall LAeq, LMax, LMin and LPeak noise levels in the two ICU`s 32

Figure 2.2: Occurrence rate of different LMax noise levels in the two ICU`s 33

Figure 2.3: Overall A-weighted equivalent sound pressure levels 35

Figure 2.4: Occurrence rate of LMax noise levels at the nurse stations 36

Figure 2.5: Occurrence rate of LMax noise levels in the MedSurg-ICU 36

Figure 2.6: Occurrence rate of LMax noise levels in the Neuro-ICU 37

Figure 2.7: Average SPL across frequency in the MedSurg-ICU 42

Figure 2.8: Average SPL across frequency in the Neuro-ICU 42

Figure 2.9: Non-weighted noise fluctuation levels, annoyance and loudness levels 44

Figure 2.10: Floor plans of the two ICUs showing the distribution of nurse stations 50

Figure 2.11: Floor-plans of the nurse stations in the two ICUs 50

Figure 2.12: Occurrence rate of Lpeak noise levels during weekdays 55

Figure 2.13: Occurrence rate of Lpeak noise levels at the weekend 55

Figure 2.14: Occurrence rate of Lpeak noise levels during the day times 56

Figure 2.15: Occurrence rate of Lpeak noise levels during the shift changes 57

Figure 2.16: Occurrence rate of Lpeak noise levels during the night times 57

Figure 2.17: Overall Leq, Lmax, Lpeak and Lmin levels in the two ICU`s 58

Figure 2.18: Overall occurrence rates of Lpeak noise levels in the two ICU`s 59

Figure 2.19: Percentage of nurses who agreed that they sometime tune out alarms 63

Figure 3.1: Floor-plans of the two ICUs showing circulation patterns 84

Figure 3.2: Perceived importance of visual vs. auditory monitoring 86

xi
Figure 3.3: Perceived importance of auditory monitoring vs. visual monitoring 87

Figure 3.4: Perceived importance of different medical alarms for patient safety 88

Figure 3.5: Frequency of auditory monitoring at different locations in the two ICU`s 90

Figure 3.6: Perceived sound task performance of nurses in the two ICUs 91

Figure 4.1: Commonly applied concert hall floor plate shapes 101

Figure 4.2: Royal Albert Hall with elliptical floor-plate shape 102

Figure 4.3: Boston Symphony Hall with rectangular floor-plate shape 103

Figure 4.4: Berlin Philharmonie Concert Hall with vineyard floor-plate shape 104

Figure 4.5: Floor-plate shapes of theoretical design models 106

Figure 4.6: Spatial distribution of RT30 in a single corridor 110

Figure 4.7: Spatial distributions of RT30 and EDT in tunnels 111

Figure 4.8: RT30 and EDT due to absorptive qualities of boundaries 113

Figure 4.9: RT30 and EDT due to reflective qualities of boundaries 114

Figure 4.10: Floor-plans of two long enclosures with different length 115

Figure 4.11: RT30 and EDT due to cross-sectional size 116

Figure 4.12: Site-plan of historical underground galleries 118

Figure 4.13: Floor-plan of historical interconnected underground galleries 118

Figure 4.14: Spatial distribution of RT30 along a “+” street junction 120

Figure 4.15: Floor-plans of theoretical long enclosures with a single branch 121

Figure 4.16: Distribution of RT30 in long enclosures with a single branch 122

Figure 4.17: Floor-plans of theoretical long enclosures with multiple branches 123

Figure 4.18: Distribution of RT30 in long enclosures with multiple branches 124

Figure 5.1: 3D-CATT model and floor-plan of an L-shaped corridor 129

Figure 5.2: Room averaged RT30 in a L-shaped corridor 131

Figure 5.3: Measured vs. predicted RT30 at a selected receiver location 132

xii
Figure 5.4: 3D-CATT model and floor-plan of a race track corridor 133

Figure 5.5: Room averaged RT30 in a race track design corridor 135

Figure 5.6: RT30 at the selected receiver locations in a race track design corridor 136

Figure 5.7: 3D-CATT model and floor-plan of a T shape corridor 138

Figure 5.8: Room averaged measured RT30 in a T-shaped corridor 140

Figure 5.9: Measured vs. predicted RT30 at the selected locations (T-shaped) 141

Figure 5.10: Interior of the single corridor setting 144

Figure 5.11: Interior of the T- shaped corridor 144

Figure 5.12: Measured RT30 in two different corridor settings 146

Figure 6.1: Spatial distribution of sound and noise sources in the two ICUs 151

Figure 6.2: A speech sound that is masked differently in two reverberant rooms 152

Figure 6.3: Total length of nursing unit corridors in different hospitals 154

Figure 6.4: Floor plans of nursing units located in different hospitals 155

Figure 6.5: Floor plans of clustered corridors 157

Figure 6.6: Number of clustered corridors in the nursing units 159

Figure 6.7: Nursing unit layout types 160

Figure 6.8: Sound propagation in race track and tree-like design corridors 161

Figure 6.9: Total corridor length of corridor clusters 162

Figure 6.10: Plans of the corridors located in the educational buildings 165

Figure 6.11: Plans of theoretical models 172

Figure 6.12: 3D-CATT models of theoretical tree-like design hallways 174

Figure 6.13: 3D-CATT models of theoretical race track design hallways 178

Figure 6.14: Floor plans of theoretical corridor models 187

Figure 6.15: The relationship between RGD + VF and RT30 at 500Hz 191

Figure 6.16: The relationship between RGD + VF and RT30 at 1kHz 192

xiii
Figure 6.17: Scatter plots showing the relationship b/w RGD, VF and RT30 194

Figure 6.18: Scatter plots showing the relationship b/w RGD, VF and RT30 195

Figure 7.1 Impulse response measurements in the two ICU corridors 200

Figure 7.2: Distribution of source and receiver locations in the Neuro-ICU 200

Figure 7.3: Distribution of source and receiver locations in the MedSurg-ICU 201

Figure 7.4: Source and receiver locations at the “T” and “cross” intersections 202

Figure 7.5: Source and receiver locations along the corridors with multiple branches 203

Figure 7.6: Source and receiver locations along the corridors without any branches 204

Figure 7.7: Distribution of source and receivers (2 turns away from the source) 205

Figure 7.8: Distribution of source and receivers that is 1 turn away from the source 205

Figure 7.9: Distribution of source and receivers along a long corridor 206

Figure 7.10: Distribution of source and receivers along a shorter corridor 207

Figure 7.11: Distribution of source and receivers in the Neuro-ICU corridors 208

Figure 7.12: Distribution of source and receivers in the MedSurg-ICU corridors 208

Figure 7.13: Distribution of source and receivers in the idealized Neuro-ICU 209

Figure 7.14: Distribution of source and receivers in the MedSurg-ICU corridors 210

Figure 7.15: Averaged RT30 in the MedSurg-ICU and Neuro-ICU corridors 212

Figure 7.16: RT30 at the “T” and “+”shaped intersections ( 0.3) 213

Figure 7.17: RT30 at the “T” and “+”shaped intersections ( 0.1) 214

Figure 7.18: RT30 at the corridors with different number of branches ( 0.3) 215

Figure 7.19: RT30 at the corridors with different number of braches ( 0.1) 215

Figure 7.20: RT30 at the corridors 1-2 turns away from the source ( 0.3) 217

Figure 7.21: RT30 at the corridors 1-2 turns away from the source ( 0.1) 217

Figure 7.22: RT30 in the corridors with different length ( 0.3) 219

Figure 7.23: RT30 in the corridors with different length ( 0.1) 219

xiv
Figure 7.24: RT30 in the corridors with different total length ( 0.3) 221

Figure 7.25: RT30 in the corridors with different total length ( 0.1) 221

Figure 7.26: RT30 in the corridors with similar total length ( 0.3) 223

Figure 7.27: RT30 in the corridors with similar total length ( 0.1) 223

xv
 LIST OF ABBREVIATIONS

Acoustics

BEM: Boundary element method

C50/80: clarity

D50/80: definition

dB: Decibel

dBA: A-weighted sound pressure level

dBC: C-weighted sound pressure level

EDT: Early decay time

FEM: Finite element method

G: Sound strength

IACC: Inter-aural cross correlation

ISM: Image source modeling

JND: Just noticeable difference

LEF/LF/LFC: Lateral fraction

Leq: Equivalent continuous sound pressure level

Lmax: Maximum sound level

Lmin: Minimum sound level

Ln centiles: Exceedance level

Lpeak: Peak sound pressure level

ILD: Inter-aural level difference

ITD: Inter-aural time difference

xvi
MLS: Maximum length sequence

NC: Noise criterion

NCB: Balanced noise criterion

PSI: Preferred speech interference

RC: Room criteria

RT: Reverberation time

SI: Speech intelligibility

SIL: Speech interference levels

SNR: Signal to noise ratio

SPL: Sound pressure level

Ts: The centre time

Healthcare and others

AACN: American association of critical care nurses

ASTM: American society for testing of materials

BSI: British standards institute

CEN: European committee for standardization

DDHNS: Disturbance due to hospital noise survey

ICU: Intensive care unit

ISO: International standard organization

OSHA: Occupational safety and health administration

RGD: Relative grid distance

SCCM: Society of critical care medicine

SD: Standard deviation

xvii
WHO: World health organization

VF: Visual fragmentation

xviii
 SUMMARY

Healing and clinical work requires a complex choreography of architectural acoustic

design in healthcare settings that is only beginning to be understood. In most healthcare

settings, medical staff members conduct vital tasks that may have life-and-death

implications. Patients visit hospitals to heal. Their expectations include fast recovery,

restful sleep, and privacy (e.g., speech privacy). However, sound environment qualities of

care settings often fall far from supporting the mission of hospitals. There is strong and

growing evidence showing that soundscapes in healthcare settings impact healing, errors

and stress for patients, families and staff but it is still not clear what measures of the

sound environment best predict key healthcare outcomes and what design strategies best

impact those measures. By using a multi-method approach, this study aims to develop a

toolkit of evidence-based design strategies by statistically defining the relationships

between three types of variables: (1) architectural floor-plate design metrics, (2) acoustic

metrics, and (3) occupant response.

In Chapter 2, the study explores the effects of hospital sound environments on caregiver

health and performance outcomes by comparing the sound environments of a traditional

critical care setting (ICU) with an award winning design ICU. For the objective and

subjective assessment of sound environments of the two ICUs, continuous noise level

(field) measurements and an online survey of caregivers are utilized. According to

statistical analysis results, critical care sound environments with different designs can

vary drastically and impact caregivers` perceived wellbeing and overall task

xix
performance. Differences between healthcare sound environments and nurses` annoyance

and loudness perceptions can also vary during different times of the day and days of the

week. Moreover, particular noise sources such as impulsive noises are likely to dominate

the ICU sound environments and interfere with perceived caregiver health and

performance. Despite their extensive use, traditional noise metrics sometimes may not be

effective in capturing unique the characteristics of healthcare sound environments. This

study validates the effectiveness of a new more detailed noise metric, “occurrence rate”,

in capturing the differences between acoustic characteristics of healthcare sound

environments.

In Chapter 3, the study analyzes the association between healthcare sound environments

and specific type caregiver tasks: “critical sound tasks” (i.e., patient auditory monitoring).

In order to gain detailed information about critical sound tasks conducted by caregivers

and assess caregiver`s ability to conduct these tasks in different care settings, an online

survey is utilized. Interesting differences are found between critical sound task

performances of caregivers working in two wards with different designs. In Chapter 3,

the review of literature clarifies specific acoustic characteristics of sound environments

necessary for auditory monitoring. In particular, reverberant qualities of sound

environments can have significant negative impacts on sound localization performance of

the human auditory system. Analysis of the survey data also provides important baseline

information (e.g., types of critical sounds and listening locations) for the following

phases of the research. For example, caregivers reported that they frequently monitor

sounds in the corridors of their units as well as other locations such as patient rooms and

xx
nurse stations. However, the association between architectural floor-plate design and

acoustics of inter-connected hospital corridors has not been investigated in previous

research. To support the development of sound task supportive healthcare environments,

in the following phases of this research more controlled studies are conducted via

acoustic simulation analysis. In Chapter 4, an overview of room-acoustics parameters, the

sound behavior in proportional and non-proportional spaces and the association between

floor-plate design and acoustics are provided by reviewing literature. In Chapter 5, the

effectiveness of the acoustic simulation tool in estimating the reverberant characteristics

of inter-connected corridors is validated by comparing the predicted and measured

acoustic outcomes. In Chapter 6, the statistical analysis results clarifies the association

between floor-plate design characteristics (e.g., corridor length, number of turns and

branch number) and reverberant qualities of inter-connected nursing unit corridors.

Various theoretical models are generated based on the heuristic design analysis of various

nursing units. Acoustic qualities of these theoretical models are analyzed by the validated

acoustic simulation program. Finally in Chapter 7, acoustic modeling predictions of the

real-world wards and field measurements are utilized in Chapter 6 to verify the proposed

effects of floor-plate design on reverberant qualities of corridors.

xxi
Figure 1.1 Phases of the research (See Appendix A for the definition of terms and
descriptions of different analysis methods included in the graph above)

xxii
 CHAPTER 1

RESEARCH PROBLEM

1.1 BACKGROUND

Hospitals are getting noisier each year (Busch-Vishniac et al., 2005). There is strong and

growing evidence of the negative impacts of a poor hospital soundscape such as staff

stress, loss of productivity, medical errors, and oral miscommunication (Busch-Vishniac

et al., 2005; Flynn et al., 1999; Morrison et al., 2003; Ryherd, Persson Waye, &

Ljungkvist, 2008; Topf, 1988; Topf, & Dillon, 1988). Patients may suffer from sleep

disturbance, anxiety, cardiovascular arousal, and decreased wound healing (Baker et al.,

1993; Freedman et al., 2001; Gabor et al., 2003; Hagerman et al. 2005; Hsu, Ryherd &

Persson Waye, 2009; McCarthy, Quimet, & Daun, 1992; Parthasarathy, & Tobin, 2004;

Wysocki, 1996). In critical care units, poor soundscapes might even have life-and-death

implications as caregivers rely on auditory cues such as help calls, and alarms. To date, a

poor hospital soundscape has primarily been addressed using engineering-oriented

solutions such as sound-absorbing acoustic materials. However, the engineering methods

remain limited because of the complex nature of hospital soundscapes and the limited

availability of acoustic materials meeting stringent hygienic hospital requirements

(Busch-Vishniac et al., 2005; Ryherd, Persson Waye, & Ljungkvist, 2008). Additionally,

while many hospitals are clearly noisy there is limited documentation of their actual

characteristics, particularly their architectural layouts. This is surprising as it is well

known that there are significant impacts of architectural design features on the

soundscapes of music halls, offices, and other spaces (Long, 2006). The few studies that

1
do exist show conflicting results; for example, studies conducted before and after

acoustic renovations in different healthcare settings report inconsistent improvements in

the soundscape (Blomkvist et al., 1996). This limits the ability of architects to effectively

design healing soundscapes.

1.2 PROBLEM STATEMENT

A healing healthcare environment is quiet, orderly, and conducive to patient recovery and

safety, family comfort, and employee health and productivity. Too often, healthcare

facilities are noisy and stressful; the complex auditory environment, or “soundscape,” has

long been a key source of complaints. While there is an urgent need for effective

soundscapes in healthcare settings, there are surprisingly no rigorous efforts to enhance

the architects` contribution in addressing this problem from the early design phase

through building commissioning. This study conducts quasi-experimental and

experimental research to relate specific architectural layout metrics, acoustic metrics, and

hospital occupant outcomes.

1.3 SCOPE AND OBJECTIVES

Primary objectives include:

 Objective and subjective noise level analyses;

o Develop new acoustic metrics and methods effective in capturing unique

characteristics of healthcare sound environments

2
 o Test the effectiveness of traditional and more detailed acoustic metrics in

capturing the differences between sound environments of different

healthcare settings

o Assess whether traditional and more detailed acoustic metrics relate to

perceived qualities of sound environments and nurse outcomes

o Assess the effects of particular noise sources on perceived caregiver

outcomes

 Soliciting feedback from caregivers via online surveys

o Assess the importance of specific caregiver tasks (visual vs. auditory

patient monitoring) for patient safety

o Identify critical listening locations

o Identify critical sounds for patient safety

o Identify which acoustic qualities of sound environments negatively affect

critical sound tasks of caregivers

o Assess auditory monitoring performances of caregivers in different care

settings with different designs

 Just noticeable difference analysis

o Validate CATT acoustic modeling program predictions

 Heuristics design analysis

o Identify floor-plate design characteristics of nursing unit corridors

o Provide basis for theoretical design analysis

 Pilot study: Acoustic analysis (via impulse response) of interconnected corridors

in the educational buildings

3
 o Provide basis for the theoretical design analysis

o Assess the link between acoustics and design of inter-connected corridors

based on field measurements

 Experimental study: Theoretical design analysis (via simulation)

o Statistically link acoustics and floor-plate shape qualities of nursing unit

corridors based on acoustic modeling predictions

o Statistically link acoustics and floor-plate design qualities of particular

hallways based on acoustic modeling predictions

 Validation study: Acoustic analysis (via simulation and impulse response) of real

world hospital wards

o Validate the proposed relationship between design and acoustics of inter-

connected corridors

1.4 RESEARCH QUESTIONS

Chapter 2

Chapter 2 introduces the findings of three empirical studies. The main goal of these

studies is to analyze the impact of sound environments on staff outcomes. Empirical

study 1 assesses the objective and subjective noise levels at different locations in the two

ICUs. Empirical study 2 documents the objective and subjective noise levels during

different times at each unit nurse stations by conducting the occurrence rate analysis and

assessing nurse perceptions. Empirical study 3 compares the level of nurse disturbance

due one of the dominant impulsive sounds, clinical alarms, in the two ICUs. Specific

research questions addressed in Chapter 2 are:

4
 1. Do acoustic qualities (objective and subjective) of healthcare sound environments

with different architectural designs differ from each other?

2. Do objective qualities of healthcare sound environments statistically relate to

subjective qualities of healthcare sound environments?

3. Which noise measures are effective in capturing the differences between critical

care sound environments?

4. Do different types of noise sources impact caregiver outcomes similarly in

different healthcare settings?

Chapter3

Chapter 3 provides a comprehensive overview of factors related to auditory monitoring

and introduces the finding of a case study comparing nurses` auditory monitoring

performance in two ICUs with different architectural designs. Specific research questions

addressed in Chapter 3 are:

1. Which acoustic qualities of sound environments relate to auditory monitoring of

critical sounds?

2. Is auditory patient monitoring as critically important as visual patient

monitoring?

3. What are the critical sounds that caregivers monitor for patient safety?

4. Where do caregivers frequently monitor critical sounds in their units?

5
 5. Do patient auditory monitoring performances of caregivers differ in healthcare

settings with different architectural designs and acoustic qualities?

Chapter 4

Chapter 4 provides an overview of room-acoustics parameters, the sound behavior in

proportional and non-proportional spaces, and the association between floor-plate design

and acoustics by reviewing literature. Specific research questions addressed in Chapter 4

are:

1. Do the characteristics of sound behavior in proportional spaces differ from the

characteristics of sound behavior in the non-proportional spaces?

2. What are the design factors that affect the behavior of sound in proportional and

non-proportional spaces?

Chapter 5

Chapter 5 introduces the findings of various impulse response measurements and acoustic

simulation analysis conducted to assess the validity of acoustic modeling tools in

estimating the acoustic qualities of non-proportional spaces. The specific research

question addressed in Chapter 5 is:

1. Are the differences between predicted (via CATT acoustic modeling program)

and measured (via impulse response measurements) acoustic outcomes

perceptible?

6
Chapter6

Chapter 6 presents the results of a heuristic design analysis and statistically explores the

association between acoustics and design variables of interconnected corridors via three

empirical studies. First, a pilot study is conducted in the inter-connected corridors of

educational settings. To test the findings of the pilot study, a more controlled follow up

study (Empirical Study 1) is conducted by using CATT acoustic simulation program.

Another theoretical design analysis (Empirical Study 2) is also conducted to assess the

effects of overall floor-plate shape characteristics of inter-connected corridor systems on

averaged reverberation time values. Specific research questions addressed in Chapter 6

are:

1. What are the floor-plate design characteristics of nursing unit corridors?

2. Do overall floor-plate shape qualities of inter-connected corridors relate to their

acoustic qualities?

3. Do floor-plate design qualities of particular hallways in an interconnected corridor

setting relate to their acoustic qualities?

Chapter 7

Chapter 7 assesses the validity of the previous study findings (in Chapter 6) by analyzing

the acoustics of real-world hospital ward corridors via impulse response measurements

and acoustic simulation analysis. Specific research questions addressed in Chapter 7 are:

7
 1. Do field measurements and acoustic modeling predictions of the real-world wards

verify the findings of Chapter 6?

a. Do reverberant qualities of hallways with different corridor length differ

from each other?

b. Do reverberant qualities of hallways with different number of branches

differ from each other?

c. Do reverberant qualities of hallways with different number of turns (from

the sound source) differ from each other?

d. Do reverberant qualities of inter-connected corridors with different floor-

plate shape qualities differ from each other?

Chapter 8

Chapter 8 summarizes the findings of each chapter and defines the study contributions,

limitations and future work.

1.5 SIGNIFICANCE

This study addresses the issue of wellbeing and task supportive healthcare soundscapes

by diagnosing problematic aspects of healthcare sound environments and proposing

design strategies. The outcomes of this study:

 Address increasing urgency for the development of effective soundscapes in

healthcare settings. The study assesses the effectiveness of intended design

strategies in improving healthcare soundscapes. The U.S. is starting one of the

largest healthcare construction programs in its history, expected to exceed $76

8
 billion a year by 2011(FMI, 2006; Zimring, & Bosh, 2008). Potentially, the

results will have resonating effects on the health of society, as 4.9 million

Americans work in hospitals and many more are treated in them (U.S. Dept of

Labor, 2008).

 Verify the necessity of using effective acoustic metrics and methods for the

diagnosis of problematic aspects of healthcare soundscapes.

 Provide new acoustic metrics to capture the unique characteristics of healthcare

sound environments and validate their effectiveness in predicting occupant

outcomes.

 Diagnose highly problematic sound sources and highlight the necessity of

collaboration between medical equipment engineers/designers and healthcare

providers.

 Provide detailed information about the characteristics of healthcare soundscapes

to support informed decision making and collaboration between architects,

engineers and healthcare providers.

 Expand principles of statistical acoustics that have been widely used to explore

the relationship between acoustics and design of more well-understood spaces to

the complex hospital settings.

 Establish the relationship between hospital floor-plate design and acoustic

metrics.

 Allow architects to design hospital layouts that are more conductive to occupant

health and productivity. To date, a hospital soundscape has primarily been

9
addressed using engineering-oriented solutions such as sound-absorbing

materials.

10
 CHAPTER 2

HOSPITAL SOUNDSCAPES AND STAFF OUTCOMES

Chapter 2 introduces objective and subjective qualities of hospital sound environments by

reviewing literature and conducting three empirical studies.

2.1 LITERATURE REVIEW

2.1.1 Noise levels in hospitals

Growing research on hospital acoustics indicate that hospitals have noisy and multi-

source sound environments. Background noise levels in hospitals have been consistently

increasing (Bush-Vishniac et al. 2005). The study conducted a timeline analysis of noise

levels reported in thirty-five studies published over the last forty-five years. It was found

that since 1960, day time and night time hospital noise levels have risen from 57dBA to

72dBA and from 42 to 60dBA respectively. According to World Health Organization

(WHO) guidelines, background noise levels particularly in the patient rooms should not

exceed 30dBA for day time and 35dBA for night time (Berglund et al., 1999).

2.1.2 Hospital Noise Metrics

Previous research has used different metrics to measure and describe the sound

environment. Most research to-date has focused on characterizing overall noise levels.

Equivalent (Leq), minimum (Lmin), maximum (Lmax), and peak (Lpeak) sound pressure

levels have been most commonly reported. This may be based on the practicality and

convenience of these measures, and because they are incorporated into various guidelines

11
such as WHO. Leq, Lmin, Lmax and Lpeak sound levels provide a good general

overview of the sound environment, but are limited in usefulness. More detailed acoustic

measures such as the exceedance level (Ln), reverberation time (RT), speech

intelligibility (SI), and frequency analysis or noise criteria indicators of spectral content

have been less commonly reported. In the ICU, there are a diversity of noise sources such

as alarms, ventilation systems, conversation, and medical equipment that contribute to a

complex, varying sound environment. More detailed acoustic measures are important to

more fully understand the spectral character of the sound environment, its behavior over

time and the potential to interfere with speech.

2.1.3 Hospital noise level measurement methods

There is a good deal of variance in the methods applied during the sound sample

collections in the previous literature on ICUs. Compared to other spaces, noise levels in

the ICU patient rooms have been commonly documented. Measurements typically took

place either in a representative patient room (Christensen, 2005; McLaughlin et al., 1996;

Ryherd, Persson Waye, & Ljungvist, 2008) or in more than one patient room with

different features such as distance to the nurse station, occupied-unoccupied, number of

patient beds (Aaron et al., 1996; Aitken, 1982; Blomkvist et al., 2005; Bush-Vishniac et

al., 2005; Hilton, 1985; MacKenzie, & Galbrun, 2007; Meyer et al., 1994; Morrison et

al., 2003; Moore et al., 1998; Williams, Drongelen, & Lasky, 2007). There is general

consistency in the location of the sound meter, with most studies locating the microphone

as close as possible to patient head to capture what the patient hears while avoiding any

interference with nurse work flow. Hanging the microphone from the top of the medical

12
tower in the patient room has been introduced as a practical solution (Ryherd, Waye, &

Ljungvist, 2008).

Different durations have been used in the data collection, ranging from a few minutes to

168h, with 24h recordings most widely used. A few ICU noise studies have also

conducted measurements at the nurse stations using 24h or 168h durations (Blomkvist et

al., 2005; Bush-Vishniac et al., 2005; Moore et al., 1998). Busch-Vishniac et al. (2005)

also described the sound environment of hallways. There are also differences in the sound

level meter response times used, with many using a slow response time (1s) as suggested

by the Occupational Safety and Health Administration (OSHA) for typical occupational

noise measurements (ANSI 1996; Bush-Vishniac et al. 2005; Williams, Drongelen, &

Lasky, 2007). Some studies used fast response time (0.125s) as suggested by WHO to

better capture fluctuations (MacKenzie, & Galbrun, 2007; Ryherd, Waye, & Ljungvist,

2008). Averaging intervals also ranged from 5s to 24h (e.g., 30s, 1min, 5min). Among

ICU-noise studies the use of 1min averaging interval was most common. Measurements

were often analyzed as a function of time, with day time and night time average sound

levels commonly reported. Weekdays were most commonly measured, with limited data

on weekends. Morrison et al. (2003) and Ryherd, Waye, & Ljungvist (2008) considered

day and night time based on twelve hour nurse shifts (day time:7am-7pm; night time:

7pm-7am). MacKenzie, and Galbrun (2007) considered the day and night time periods

based on WHO guidelines (16h day time:7am-11pm; 8h night time:11pm-7am).

13
2.1.4 Hospital noise and its effects on staff members

The acoustic environment in hospitals can affect all occupants, including staff, patients,

and visitors. Related with the focus of this study, the following discussion is mostly

limited to its effects on staff members: stress and annoyance; work performance; health

outcomes and work overload. Further information about how hospital acoustics may

impact patients and visitors can be found in sources such as Busch-Vishniac et al. (2005)

and Ryherd et al. (2008). The staff’s wellbeing, efficiency and effectiveness in delivering

care and performing critical tasks is critical to maximize patient safety, satisfaction, and

care quality in ICUs.

Stress and Annoyance

ICUs are stressful care settings and the stress can be exacerbated by the sensory overload

caused by environmental factors, including the high noise levels. Stress is the individual’s

appraisal of a mismatch between perceived demand and perceived self-capabilities to

cope (Clarke, 1984). Depending on the severity and duration, stress may lead to illness

(e.g., elevated blood pressure, indigestion) and behavioral changes (e.g., sadness,

depression, negative attitudes). Anxiety is a psychological response to environmental

stimuli or activity producing arousal (Gast, & Baker, 1989). Excessive anxiety can be

detrimental and lead to different health effects such as sleep deprivation and confusion.

Like anxiety, annoyance is one of the early psychological responses which reflect the

negative quality of the environment stimuli (Baker, 1984), and it relates to the intrusion

of a stimulus on a mental or physical activity.

14
In one study, higher average sound pressure levels predicted higher nurse heart rate,

perceived stress, and perceived annoyance levels in a Pediatric-ICU (Morrison et al.,

2003). The sound pressure level was measured using the A-weighted equivalent sound

pressure level (LAeq). In this study, the sound levels were averaged every 30min.

Regression analysis was used to correlate the LAeq with continuously recorded heart rate

and saliva samples and stress/annoyance ratings that were also collected in every 30min.

During 3h periods, samples were collected from 11 nurses. Simultaneously noise levels

were also recorded. Subjective staff responses were gathered via the U.S. Army Research

Lab-Specific Rating of Event Scale. A total of 33h of sound measurements showed that

in multiple-bed patient rooms overall average sounds levels (Leq) were approximately

61dBA, and average sound levels during day (7am-7pm) and night (7pm-7am) were

6dBA and 59dBA, respectively. Three other noise measures considered in this study were

Lmin, Lmax, Lpeak. Levels were 43dBA, 93dBA, 93-122dBC, respectively.

In another study, nurses working 8h evening shifts were significantly distressed by noise

in two ICUs surveyed (Topf, 1989). It was also found that less sensitivity to noise and

greater personality hardiness (such as commitment, control, and challenge) were linked

with less noise-induced stress. This research did not collect any objective sound data but

explored noise-induced subjective staff stress via personality hardiness measures and

sensitivity to noise with the use of regression. Their sample size was 100 ICU nurses.

Noise-induced stress data was collected via Disturbance Due to Hospital Noise Scale

(DDHNS). Other results were collected via the Weinstein noise sensitivity scale, locus of

15
control, work scale of alienation test, and security scale of California life goals evaluation

schedule.

There is a very limited number of noise-induced nurse stress studies conducted in the

ICUs. The following stress and annoyance studies are therefore included even though

they are not conducted in the ICUs but in other sections of the hospital. In one study,

enhanced reverberation time (RT) and speech intelligibility (SI) via acoustical

interventions were related to improved perceived psychosocial work environment (such

as low work demand, less strain, less pressure) (Blomkvist et al., 2005). The acoustical

intervention consisted of adding sound absorbing ceiling tiles (renovated condition) in an

8-bed coronary care unit that originally had a sound reflective ceiling (base condition).

Perceived psychosocial work environment data was collected repetitively before and after

every 8h shift from 36 nurses in both the base and renovated conditions. Noise levels

were also recorded during the two study periods, each of which lasted for 1 week. The

staff perception was evaluated with multiple psychological factors (such as strain,

tension, irritation, and anger) using a Demand, Control-Support Model. Analysis of

variance (ANOVA) was used for the comparison of the data collected before and after the

renovation. The SI and RT levels were measured in 3 private patient rooms and at the

nurse station before and after the renovation. The RT levels improved from 0.8 s to

0.4sec at the nurse station and from 0.9s to 0.4s in the patient rooms after the design

intervention. Overall average sound levels (Leq) were 57dBA and 56dBA at the nurse

station and 56dBA and 50dBA in the patient room during the base and renovated

conditions, respectively.

16
In another study, noise-induced annoyance was reported as a problem by the hospital

staff (Bayo et al., 1995). In this study, descriptive statistics were used to describe noise-

induced perceived staff outcomes more than causal relationships between objective sound

levels and outcomes. Five- and 10-point scale surveys were used to measure perceived

staff outcomes and loudness, respectively. Among the 300 staff members surveyed, 70%

reported “very much” as the level of their annoyance due to noise. The other noise-

induced outcomes were: work performance (13%), quality of work (13%), personal

satisfaction (17%), and health (16%). Average perceived loudness was reported as 6.5 out

of 10, with 10 being “very loud”. Leq, Ln percentiles, Lmin, and Lmax at different

locations were measured based on 10min, 5min, and 1min sound measurements at 232

different locations inside the hospital on a regular floor (e.g., non-medical areas,

corridors, clinics, operating theatre, wards, intensive care units) and at 121 locations

outside the hospital (e.g., external premises and on the street). The Leq inside the hospital

ranged between 52-75dBA, and outside the hospital between 52-82dBA. The

measurements took place during morning (9am-1pm) and afternoon (4pm-8pm).

To summarize, there is some evidence that high noise levels in care settings contribute to

staff stress and annoyance. However, the number of the studies specifically exploring the

relationship between ICU noise levels on staff outcomes remains very limited.

17
Work Performance

Hospital sound environments that are supportive of hospital tasks could potentially

improve staff effectiveness in delivering care. Improved nurse work performance in ICUs

can prevent adverse events, improve healthcare quality, and optimize resource utilization.

Studies of noise and hospital staff performance are quite limited, and some conflicting

results are seen. Work performance in hospital noise studies has previously been assessed

both quantitatively (i.e., mental efficacy, memory tests, and motion analysis systems)

(Murthy et al., 1995) and qualitatively (i.e., blind observation) (Moorthy et al., 2004).

Perceived noise-induced outcomes were typically assessed by self-report questionnaires

(Ryherd et al., 2008). Sound environments have been characterized objectively in a

variety of different ways (e.g., Leq, background noise levels, preferred speech

interference levels, noise criteria, sound pattern, Ln percentiles, Lmax, Lmin, Lpeak,

spectral qualities) (Hodge, & Thompson, 1990; Moorthy et al., 2004; Murthy et al., 1995;

Ryherd et al., 2008).

One study found that noise in the workplace was perceived to have a negative impact on

staff work performance (91%) and concentration (43%) (Ryherd et al., 2008). Perceived

noise-induced outcomes were gathered from 47 nurses via 5-point scale questionnaire.

Sound levels were measured continuously in a 2-bed patient room of a neurological-ICU

during 5 week days. Different noise criteria methods (i.e., RC , RC Mark II, NC, NCB)

were used to describe the sound quality of ICU noises. It was found that most ICU noises

were hissy (excessive energy from 1-8 kHz). The study also considered the mean length

of times when specific noise level conditions were met during the day and night times

18
such as Lpeak <75dBA (6min, 8min respectively); Lmax<55dBA (10min, 8min

respectively); and Leq<50dBA-restorative period (9min, 13min respectively). The

overall average sound levels (ranged between Leq 53dBA-58dBA). Statistical level

exceedances were also reported. It was found that 90% of the time the maximum and

peak sound levels exceeded 50dBA and 70dBC, respectively.

A Neonatal-ICU study showed that sound that exceeds 55dBA most of the time can

potentially interfere with work (Thomas, & Martin, 2000). This multidisciplinary

literature review study highlighted that tasks requiring rapid reaction time and vigilance

are sensitive to noise.

Noise-induced work performance research has been more commonly conducted in the

operating theatres. For example, it was found that noise conditions with Leq of 77dBA

were related to short term memory and decreased mental efficiency among twenty

anesthesia residents (Murthy et al., 1995). The Trail Making Test, Digital Symbol Test,

and Benton Visual Retentive Test were used for the assessment of cognitive tasks, mental

efficacy, and short term memory. The performance tests were conducted in an

acoustically treated room in which previously recorded operating room noise was played

back. Overall Leq levels were recorded during 5 different types of operations in the

operating room.

Another operating room study found that unpredictable and uncontrollable noise such as

intermittent, intrusive noise (i.e., alarms >75-85dBA) can negatively affect performance

19
and concentration during complex tasks (Hodge, & Thompson, 1990). This study pointed

out that background conversation and peak sound levels (70-80dBA) can reduce the

reliability of oral communication. In this study, only objective sound measurements were

conducted and these results were interpreted with the literature reviewed. Based on the

noise measurements conducted in the operating theatre, 50% of the time it was louder

than 46dBA. The overall average Leq was 51dBA. Lpeak was 108dBA. Background

noise level was 13dBA. Moreover to evaluate the reliability of the verbal communication,

the Preferred Speech Interference (PSI) level (the arithmetic average of three octave

bands centered at 500Hz, 1kHz and 2kHz) was compared with a commonly accepted

value (noise levels between 47-52dB can be acceptable if the distance between speaker

with normal-voice and talker does not exceed 6ft).

Conversely, Moorthy et al. (2004) found no difference in laparoscopic performance by

surgeons under various noise/music conditions. This was potentially related with

surgeons` ability to effectively “block out” interfering environmental conditions.

However their ability to concentrate in noisy conditions can require too much effort. This

study hypothesized that the complexity of the surgical tasks that require high

concentration and skills moderate the effect of noise on a surgeon’s performance. Three

different acoustical conditions (music, noise and quiet) applied during 12 surgeries

yielded no significant differences in the surgeons` performances. Sound levels were

measured and recorded during 500-700min surgeries in the operating theatre. While

surgeon performance was assessed by 2 blinded observers and validated motion analysis

20
system, the noise recorded in the operating theatre played back continuously. Statistical

significance tests and correlation analysis were used.

To summarize, the impact of noise on staff performance (especially in ICUs) has not

been widely examined and the findings of the some of the previous studies are

conflicting. Caregivers and patients can highly benefit from additional research to

determine how the sound environment effects staff`s work performance in ICUs.

Health

The acoustic environment throughout the hospital may contribute to negative acute or

chronic symptoms in staff. Critical care nursing is a very demanding job and it requires

continuous alertness, vigilance, and well-being to conduct critical tasks effectively. One

study found that of the 47 nurses surveyed in an ICU, 66% reported irritation and fatigue,

while 40% reported headaches due to workplace noise (Ryherd et al., 2008). However,

the impact of noise on surgeons` health in the operating theatres has been the primary

focus of the previous literature examining noise and staff health. Due to the use of power

instruments in operating rooms, medical staff members are commonly exposed to high

noise levels. One review study suggested an association between surgeons hearing loss

and 3 sound qualities of intensity, spectral frequency and pattern (i.e., intermittent,

constant) (Thomas, & Martin, 2000).

Another study further examined the relationship between noise exposure levels and

surgeon’s hearing loss in the operating theatre during five surgeries (Love, 2003). Noise

21
dosimeters were used to measure noise exposure levels during 2 knee replacement and 3

hip replacement surgeries conducted by 3 operating room surgeons. Leq levels varied

between 75-82dBA. High Lpeak and Lmax levels were recorded at 140dB and 108dBA,

respectively.

The association between high noise levels in the operating theatre and surgeons` noise-

induced hearing loss was also suggested by Kracht et al. (2007). They recorded Leq noise

levels as high as 66dBA in the operating theatre. They documented the noise levels

during different surgeries such as neurology, urology, cardiology and gastrointestinal

procedures. During the neuro-surgery, Lpeak exceeded 100dB over 40% of the time and

90dB over 95% of the time.

In addition to noise-induced hearing loss, the general health of staff is also a concern.

However, very limited data in this area exists. One study showed that an increase in

average sound levels was significantly related to an increase in heart rate, in addition to

finding relationships between noise and perceived stress and annoyance as discussed

above (Morrison et al., 2003) .

To summarize, there are agreed findings in the literature suggesting that surgeons’

hearing health can be affected by high noise levels during surgery. Noise-induced hearing

loss has particularly been a concern for surgeons in operating theatres. However, noise-

induced health outcomes (including hearing loss) of ICU nurses` have not yet been the

focus of previous research.

22
Work overload

When noise levels exceed a nurse’s coping abilities it can lead to sensory overload

(Baker, 1984). Sensory overload can cause emotional exhaustion, dissatisfaction, and

decreased sense of personal accomplishment. This in turn can cause feelings of

ineffectiveness, ineptitude, low satisfaction, and perceived lack of success (Barling,

2001). Laschinger and Leiter (2006) found that medication error and other adverse events

necessary for patient safety were associated with emotional exhaustion.

Noise-induced stress due to ICU noise was positively related to nurse emotional

exhaustion and burnout (Topf , &Dillon, 1988). It was found that ICU nurses who rotated

shifts underwent more emotional exhaustion. Moreover during different shifts, nurses

reported different levels of emotional exhaustion. A regression method was used to

analyze the impact of covariates (i.e., sensitivity to noise, age, experience) and noise

levels on nurse burnout, emotional exhaustion, and shift. Nurse responses were gathered

via different surveys: the DDHNS for noise induced disturbance, Jones’s Staff Burnout

Scale Health Profession (SBS-HP) for burnout, Maslach`s Burnout Inventory for

emotional exhaustion, Weinstein’s noise sensitivity, and a life event stress and nursing

stress survey. The study subjects were 100 critical care nurses from two hospitals that

specialized in different areas (i.e., cardiac, medical-surgical, urology, neonatal, pediatric).

Reduced reverberation time and enhanced speech intelligibility via acoustical

interventions in an intensive coronary unit was found to improve perceived staff attitude

(Hagerman et al., 2005). Perceived staff attitude was collected via a 10-point scale

23
questionnaire. After the application of sound absorbent materials, noise levels in the

patient rooms dropped 5-6dB, but not at the nurse station. Reverberation time reduced

from 0.9s to 0.4s in the patient rooms and from 0.8s to 0.4s at the nurse station.

Measurements were conducted during the weekdays in relation to the changes in staffing

during weekends. Speech intelligibility was measured in two different ways: RASTI and

staff self-report. Non-parametric significance tests were used to compare the changes

between different acoustic conditions.

To summarize, minimizing work overload is critically important for quality of care,

patient safety, and the overall well-being of staff. Poor acoustical conditions in

workplaces can aggravate staff attitude and perceived work overload. The limited

existing evidence points to a significant problem that should be investigated further to

determine appropriate acoustic conditions that will minimize negative work overload

effects.

2.1.5 Medical alarms as noise sources and their effects

Caregivers continuously monitor auditory cues and respond to them immediately to keep

patients safe. Clinical alarms are one of the key auditory cues in care settings. They are

intentionally designed to sound highly dangerous, frightening, unpleasant, loud, and tense

(Kuwano et al., 2001) because they have to be clearly distinguishable and alerting.

Regardless their importance for patient safety, caregivers may find these sounds highly

interrupting, annoying and overloading. The following section overviews the previous

literature linking clinical alarms and caregiver outcomes.

24
Work performance

Frequently occurring excessive number of loud alarms can be problematic for nurse work

performance. Currently available monitoring systems are able to monitor most

physiological parameters (Chambrin, 2001). Each medical device emits multiple alarms

with different acoustic qualities. However there is a limit to the amount of information

that human auditory systems can process. Cropp et al. (1994) conducted listening tests in

a critical care setting and found that excessive numbers of alarms confuse nurses

(especially less experienced ones) and potentially affect their auditory monitoring

performance. When nurses were presented previously recorded clinical alarms, they were

able to correctly identify only half of the critical alarms. Wallace et al. (1994) also found

the presence of similar clinical alarm detection difficulties in the operating rooms.

Another study highlighted the waste of valuable staff time and delayed response time to a

highly urgent condition related with difficulties during the detection of alarms -

particularly high pitch tones (Meredith, & Edworthy, 1995). In particular caregivers with

unilateral hearing loss can experience more difficulty in localization and discrimination

of clinical alarms (Newby, 1979). Non-hospital research also indicated the difficulties of

learning and remembering the significance of more than 8 different sounds even in

unstressed conditions (Patterson, & Mikoy, 1980).

Alarms providing false information (a.k.a. false alarms) can also be problematic for nurse

work performance. Nurses continuously asses all audible signals occurring in care

settings as any auditory cue that might be clinically significant. However, false alarms

and/or improper alarm settings cause unnecessary interruptions in the nurse work flow.

25
They also potentially lead to misconceptions and create a “cry wolf” environment where

nurses ignore or inappropriately inactivate alarm signals (ECRI, 2007). A study found

that only 10 of the 33 alarms were critical suggesting a serious problem with either

patient or equipment (Cropp et al., 1994). As a result, false alarms are the indicators of

inefficient patient monitoring systems (ECRI, 2006).

Health

Excessively loud alarms can potentially cause health problems among patients and

caregivers. Kahn (1998) reported that noise sources in critical care settings with peak

levels higher than 80dBA are amenable to behavior modification among staff members.

Another study indicated that loud alarms can aggravate the rest of the staff members

(Hedley-Whyte, 1988). Additionally, it has been found that a considerable percentage of

nurses potentially experience sleep problems after an intensive work day with many

alarms (Ryherd et al., 2008). In addition to behavioral consequences, exposure to high

noise levels can lead to hearing disorders and often caregivers might not be aware of such

significant health problems. For example, a study reported 37% of the physicians

working in critical care settings were unaware of their hearing disorder (Wallace et al.,

1994).

Annoyance, Disturbance and Anxiety

Disturbance by false alarms is commonly reported by nurses. Alarms that are not

signaling a medical emergency can cause staff irritation and annoyance (Meredith, &

Edworthy, 1995). A false alarm indicates a violated parameter of a vital sign that has no

26
clinical consequence/ significance (Kerr, & Hayes, 1983). Lawless (1994) surveyed 2176

alarms and found 68% of them were false. Chambrin (2001) found only 5.9% of 3166

alarms required phone call to the physician.

Not only false alarms but noxious and repetitive signals are perceived as bothersome by

caregivers (Chambrin, 2001; Schmidth, & Baysinger, 1986). Parallel with staff

perceptions, patients also perceived clinical alarms as irritating, disturbing and

frightening (Gast, & Baker, 1989; Kerr, &Hayes, 1983). According to SCCM, less

noxious alarms can be provided by the modulation of critical alarms without reducing the

importance or sense of urgency (SCCM, 1995).

Clinical alarms can also increase the anxiety levels of the occupants. One of the key

responsibilities of the caregivers is to provide immediate response to alerting equipment

that potentially poses a risk to patient safety. Difficulties in determining the location of

the alarms delay caregiver response time to an alarm which in turn leads to increased

anxiety in staff (Samuels, 1986). Parallel to caregiver experiences, patients also

experience increased anxiety levels when they have difficulties in distinguishing between

the alarms of alerting equipment as any alarming signal might pose a risk to their health

condition (Kerr, & Hayes, 1983).

27
2.2 EMPIRICAL STUDY 1

CHARACTERIZING SOUND ENVIRONMENTS OF ICUS

2.2.1 Scope

Intensive care units (ICUs) have important but challenging sound environments. Alarms

and equipment generate high levels of noise and ICUs are typically designed with hard

surfaces that reflect sound. A poor sound environment can add to stress and make

auditory tasks more difficult for clinicians. In particular, the Intensive Care Unit (ICU)

sound environment is aurally demanding. Nurses often execute complex tasks in caring

for critically ill patients and it is important to understand the subjective and objective

qualities of the ICU sound environments from the nurses` point of view. However, few

studies have linked detailed analyses of the ICU acoustical environment to the

corresponding reaction of staff members.

In this study, subjective perception evaluations are coupled with objective sound level

measurements to gain a more thorough grasp of how perceptual and physical acoustic

parameters interact in the ICU setting. The subjective and objective qualities of two ICU

environments with different design features are compared at multiple locations within

each ICU. Specifically, the study aims to determine whether objective noise levels and

subjective perceptions differed:

(a) between various locations within an individual ICU;

(b) when comparing similar locations in the two ICU’s;

(c) when comparing overall (average) levels in the two ICU’s to each other;

28
Additionally, relationship between objective and subjective noise levels was examined,

including noise-induced nurse outcomes.

2.2.2 Methodology

The research is conducted in two intensive care units at Emory University Hospital:

MedSurg-ICU and Neuro-ICU. Detailed information about the physical design qualities

of these two settings can be found in Appendix B.

Both ICUs apply similar staffing models using intensivists and nurse practitioners and

accommodate critical care patients with similar acuity levels. In both units, 10 to 12

registered nurses are typically working during each shift. The Neuro-ICU nurses mostly

work 12h shifts (7am-7pm, 7pm-7am); the MedSurg-ICU nurses work both 12h and 8h

shifts (7am-3pm, 3pm-11pm, and 11pm-7am). In both units, nurses can work weekend,

weekday, or combination shifts, in addition to day, night, or combination shifts.

Objective noise measurements

Objective noise level measurements are conducted at 4 different locations in each unit:

(a) centralized nurse station, (b) unoccupied patient room with the door closed, (c)

occupied patient rooms with and without a respiratory ventilator, with the door

continually opening and closing to accommodate patient care, and (d) multiple data

points in the corridors. A total of 96h of samples are collected at the nurse stations in

each unit from Thursday to Monday. Saturday and Sunday is intentionally included as

these days have not typically been included in previous work.

29
In both units, patients with respiratory failure are connected to respiratory ventilator and

most of those patients are critically ill which restricts the entries into the patient rooms.

Due to this limited access, 45min samples are collected from the occupied patient rooms

with a respiratory ventilator. It is possible to conduct longer measurements in the patient

room without respiratory ventilator. In each unit, 24h continuous stationary sound level

measurements are conducted in the occupied patient rooms without a respiratory

ventilator during a weekday. Additionally, 45min samples are collected in an empty

patient room while patient room doors are closed. At the corridors, multiple 15min sound

samples are collected at randomly selected times during day and night. In total,

approximately 246h of sound data are collected from both units. For the calculation of

overall noise levels in each unit, all sound data collected at different locations are taken

into consideration. Noise levels in the two ICUs are generated by different types of

sources. Medical equipment alarms occurring in the patient rooms, patient monitor

alarms occurring both in the patient rooms and at the nurse stations, sound of the ice

machine, phone ringing, staff conversation, and rolling medical carts in the corridors are

some of the common noises in two units. In the MedSurg-ICU, nurses are paged via

overhead pagers. In the Neuro-ICU 3G-phones or regular phones at the alcoves are used

instead.

At the corridors, the microphone is located at a height of approximately 1.4 m (4.5 ft)

slightly off the room center and stabilized on a tripod. In the patient rooms, the

microphone is hung from the ceiling at a height of approximately 1.8 m (6 ft) above

finished floor. Similarly, the microphone is hung from the ceiling at the nurse stations at

30
a height of approximately 1.8 m (6 ft) above finished floor. In the Neuro-ICU, the sound

level meter is set up at the nurse station in the 14-bed cluster. In the MedSurg-ICU, the

sound level meter is located at slightly off the center of the central nurse work zone in the

center of the unit. All sound data is collected using a fast response time for equivalent,

maximum and minimum levels (0.125 s) and one-minute averaging intervals. Three

Larson Davis type 824 sound level meters and Larson Davis 824 Utility software is used.

The dynamic range is 80 dB unweighted from 38 dB to 118 dB.

Subjective noise measurements

An electronic survey is administered to registered nurses working in the Neuro-ICU and

MedSurg-ICU. The survey consists of 4 sections: nurse profile and working conditions,

perceived sound environment in the workplace, perceived impact of noise levels on nurse

outcomes, general hearing health and noise sensitivity. The survey is completed by 35

Neuro-ICU and 23 MedSurg-ICU nurses, corresponding to response rates of 39% and

35%, respectively. In the Neuro-ICU, 85% of the respondents are full-time and 15% part-

time. In the MedSurg-ICU, 70% were full-time, 26% part–time, and 4% PRN (on-call

nurse). In both units more than 80% of the respondents are female and more than 80% of

were younger than fifty years old.

31
2.2.3 Findings

Objective noise levels

The noise levels measured at 4 different locations in each unit (nurse station, occupied

and unoccupied patient rooms, and corridors) are averaged to find single number overall

levels in each ICU. As shown in Figure 2.1, the overall averaged levels in the two units

range from 57-58 dBA Leq, 97-105 dB LMax, 54-58 dB LMin, and 113-120 dBC LPeak.

Given these values, the difference between overall averaged LAeq levels in the Neuro-

ICU and MedSurg ICU are imperceptible.

120
110
Sound Pressure Level

100
(dB re 20 μPa)

90
80
70
60
50
40
30
20
Leq-dBA Lmax-dB Lmin-dB Lpeak-dBC

MedSurg-ICU Neuro-ICU

Figure 2.1 Overall LAeq, LMax, LMin and LPeak noise levels in the two ICU`s

Further analysis is shown in Figure 2.2, which presents the percentage of time that the

maximum noise levels (LMax) exceeded values ranging from 70 to 90 dB. This type of

analysis is referred as the “occurrence rate” in this study, and has been used successfully

in previous hospital studies by the authors to analyze differences between day and night

(Ryherd, Persson Waye, & Ljungkvist, 2008), in addition to related analyses by others

32
(Kracht, Busch-Vishniac, &West, 2007; MacKenzie, & Galbrun, 2007; Williams,

Drongelen, & Lasky, 2007). In both units, LMax exceeds 70 dB more than 98% of the

time and LPeak exceeds 80 dBC more than 96% of the time. Thus the difference in the

occurrence rate of lower level transient sounds (<70 dB LMax and <80 dBC LPeak) and

very high level transient sounds (>90 dB LMax and >100 dBC LPeak) in the two ICU’s

is negligible. However, a difference between the two wards emerges for mid-level

transient sounds, as Figure 2.2 indicates. LMax exceeds 80 dB more of the time in the

MedSurg-ICU than in the Neuro-ICU. Similarly, LPeak exceeds 90 dBC more of the time

in the MedSurg-ICU than in the Neuro-ICU. This indicates that the MedSurg-ICU is a

more impulsive sounding environment overall.

100%
90%
Percentage (%) of time

80%
70%
60%
50%
40%
30%
20%
10%
0%
LMax>70dB LMax>80dB Lmax>90dB

MedSurg-ICU Neuro-ICU

Figure 2.2 Occurrence rate of different LMax noise levels in the two ICU`s

To summarize, the sound environments of two ICU’s are similar based on traditional

measures of overall Leq, LMax, LMin, and LPeak, but are different based on the

33
occurrence rate of mid-level transient sounds. Note that the terminology “mid-level

transient sounds” does not refer to the perceived loudness of the transient level, but

simply serves to distinguish the level ranges analyzed in this study into lower, mid, and

higher regions.

The average noise levels measured at each of the four different locations (nurse station,

occupied and unoccupied patient rooms, and corridors) are shown in Figure 2.3 for each

ICU. The occupied patient rooms without respiratory ventilator are chosen for this

analysis as there has been better access for longer measurements as compared to the

patient rooms with respiratory ventilators as previously discussed. The overall averaged

levels in the two ICU’s range between 45-60 dBA Leq depending on the location. The

unoccupied patient rooms are the quietest (45-52 dBA) and the nurse station, occupied

patient rooms, and corridors have similar noise levels (56-60 dBA). It is observed that the

differences between overall averaged LAeq levels for similar occupied locations in the

Neuro-ICU and MedSurg ICU are imperceptible (e.g. the nurse stations in both units

have similar Leq levels). However, the overall averaged LAeq is higher in the

unoccupied patient rooms of the MedSurg-ICU compared to the Neuro-ICU. Similar

trends are found for traditional measures of overall LMax, and LPeak in the two ICU’s.

34
 Sound Pressure Level
70

A-weighted Equivalent
Sound Pressure Level
60

(dB re 20 μPa)
50
40
30
20

(dBA
10
0
Nurse Occupied Unoccupied
Empty Corridor
Station Patient Patient
Patient
Room w/out Room
Room
Resp. Vent.

MedSurg-ICU Neuro-ICU

Figure 2.3 Overall A-weighted equivalent sound pressure levels (Leq in dBA) at different
locations in the two ICUs

The occurrence rates of LMax and LPeak at each of the 4 locations in each ICU are also

analyzed. At all 4 locations in both units, LMax exceeds 70dB more than 98% of the time

and LPeak exceeds 80 dBC more than 94% of time. As before, a difference between the 4

spaces emerges for mid-level transient sounds, (Figure 2.4-Figure 2.6). For example,

Figure 2.4 shows the central nurse station results, where LMax exceeds 80 dB more often

in the MedSurg-ICU (43%) than in the Neuro-ICU (15%). Figure 2.5 and Figure 2.6

show similar analysis for all spaces in the two ICU’s. The occupied spaces (nurse station,

occupied patient rooms, and corridors) are more impulsive in the MedSurg-ICU (Figure

2.5) compared to the Neuro-ICU (Figure 2.6). The occurrence rates for the unoccupied

patient rooms are similar in the two ICU’s. LPeak occurrence rate analysis for the four

spaces in both ICU’s shows very similar results to LMax occurrence rate results.

35
 100%
90%

Percentage (%) of time

80%
70%
60%
50%
40%
30%
20%
10%
0%
LMax>70dB LMax>80dB LMax90dB
MedSurg-ICU Nurse Station
Neuro-ICU Nurse Station

Figure 2.4 Occurrence rate of different LMax noise levels at the nurse stations

MedSurg-ICU
100%
90%
Percentage (%) of time

80%
70%
60%
50%
40%
30%
20%
10%
0%
Lmax >70dB Lmax >80dB Lmax >90dB
Nurse Station
Lmax-dB Nurse Station Occupied
Lmax-dB Patient R. P. R.
Occupied
Unoccupied Patient R. P.R.
Lmax-dB Unoccupied Corridor
Lmax-dB Corridor

Figure 2.5 Occurrence rate of different LMax noise levels in the MedSurg-ICU

36
 Neuro-ICU
100%
90%

Percentage (%) of time

80%
70%
60%
50%
40%
30%
20%
10%
0%
Lmax >70dB Lmax >80dB Lmax >90dB
Lmax-dB Nurse Station Lmax-dB Occupied P. R.
Lmax-dB Unoccupied P.R. Lmax-dB Corridor

Figure 2.6 Occurrence rate of different LMax noise levels in the Neuro-ICU

Subjective Noise Perception

Does subjective perception differ when comparing two ICU’s to each other?

Overall, the MedSurg-ICU sound environment is perceived as significantly worse for

nurse wellbeing and work performance as compared to the Neuro-ICU sound

environment. Results for 5 nurse outcomes are shown in Table 2.1: perceived loudness,

annoyance, work performance, general health outcomes, and anxiety. A discrete 5-point

rating scale is used in for all 5 outcomes shown. Workplace noise is perceived as louder,

more annoying, and having a greater negative impact on work performance, health

outcomes, and anxiety by the MedSurg-ICU nurses compared to the Neuro-ICU nurses.

The nurses` sensitivity to noise and tolerance to high noise levels in the workplace does

not differ significantly between ICU’s (p<0.05). Overall, the nurses in both units have

rated themselves as not very sensitive to noise and thought they can somewhat tolerate

37
high noise levels. Perceived hearing ability of two unit nurses also does not differ

significantly, and has been rated as “good” on average in both units.

Table 2.1 (a) Mean perception ratings of noise-induced nurse outcomes in the two ICUs
(b) distribution of noise-induced nurse outcomes in the two ICUs

Neuro-ICU MedSurg-ICU
(1: completely disagree; (1: completely disagree; Nurse outcomes
5: completely agree) 5: completely agree)

3.0 4.3** Loudess My workplace is very loud.

I find the noise very annoying in

3.0 3.9* Annoyance my workplace.

The noise in my workplace negatively

1.9 3.0* Performance
affects my work performance.

The noise in my workplace negatively

1.7 3.0** Health
affects my health.

The noise in my workplace increases

2.0 3.9** Anxiety my anxiety levels.

* The mean perception rating is significantly higher at 0.05 level (p <0.05).

** The mean perception rating is significantly higher at 0.01 level (p <0.01).

Noise- Noise- Noise- Noise-

5-point scale Loudness induced induced induced induced
annoyance work perf. health anxiety
Completely disagree-1 23.0% 20.0% 48.6% 60.0% 51.4%
Somewhat disagree-2 14.2% 11.4% 22.8% 20.0% 20.0%
Neuro-ICU Neither agree nor disagree-3 17.1% 25.7% 17.1% 14.3% 2.9%
Somewhat agree-4 34.3% 34.3% 8.6% 2.8% 22.8%
Completely agree-5 11.4% 8.6% 2.9% 2.9% 2.9%
Completely disagree-1 0.0% 4.3% 20.4% 21.7% 13.0%
Somewhat disagree-2 8.7% 8.7% 26.1% 26.1% 0.0%
MedSurg-ICU Neither agree nor disagree-3 4.3% 13.0% 17.4% 13.0% 17.4%
Somewhat agree-4 34.8% 43.5% 26.1% 4.3% 26.1%
Completely agree-5 52.2% 30.4% 14.0% 34.8% 43.5%

38
Does subjective perception differ when comparing similar locations in the two ICU’s to

each other and within an ICU?

Though not all results are statistically significant, the trend is that the noise is consistently

perceived as louder and more annoying at all 4 locations (nurse station, occupied and

unoccupied patient rooms, and corridors) in the MedSurg-ICU as compared to the Neuro-

ICU. Results are shown in Table 2.2. A discrete 5-point rating scale is used in for

loudness and annoyance perception. Nonparametric Mann-Whitney U test results show

that the differences for the nurse stations and in the empty patient rooms are statistically

significant at (p<0.05 or p<0.01).

Table 2.2 Mean perception ratings of loudness and annoyance at different locations in the
two ICUs

Noise-induced annoyance Loudness

MedSurg- Neuro- MedSurg- Neuro-

ICU ICU ICU ICU

Nurse Station 3.8* 2.9 4.1* 3.2

Unoccupied
2.25* 1.6 2.7** 1.8
Patient Room
Occupied
3.0 2.6 3.4 3.1
Patient Room

Corridor 2.8 2.7 3.3 2.9

* The mean perception rating is significantly higher at 0.05 level (p <0.05).

** The mean perception rating is significantly higher at 0.01 level (p <0.01).

39
Differences in perception between 4 locations within each individual ICU are also found

based on the values in Table 2.2. In the MedSurg-ICU, the nurse station is perceived as

significantly louder than other three locations (p<0.05). In the Neuro-ICU, the empty

patient room is perceived as significantly quieter and less annoying than the other three

locations (p<0.05).

Correlations between Objective and Subjective Measures

Spearman nonparametric correlation tests are used to analyze the relationships between

objective noise levels and subjective loudness and annoyance perception. As shown in

Table 2.3, subjective annoyance and loudness perception is significantly and positively

correlated with Leq (dBA) in both ICU’s. The interpretation is that the nurses perceive

the noise as more annoying and louder as the Leq increases. Additionally, the mid-level

transient sounds (LMax>80 dB and LPeak>90 dBC) are significantly related to

annoyance and loudness perception (Table 2.3). Nurses perceive noise as more annoying

and louder the more often (higher % of time) LMax exceeds 80 dB or LPeak exceeds 90

dBC.

40
Table 2.3 Correlations between objective noise levels (i.e., Leq, LMax and LPeak
occurrence rates) and subjective perception of annoyance and loudness

Noise levels ANNOYANCE LOUDNESS

Leq (dBA) 0.36** 0.41**
Neuro
LMax> 80dB 0.35** 0.39**
-ICU
LPeak> 90dBC 0.35** 0.40**
Leq (dBA) 0.38** 0.42**
MedSurg
LMax> 80dB 0.25* 0.26*
-ICU
LPeak> 90dBC 0.35** 0.41**
Leq (dBA) 0.38** 0.44**
Overall
LMax> 80dB 0.38** 0.43**
(two units)
LPeak> 90dBC 0.38** 0.45**

* Correlation is significant at 0.05 level (p< 0.05)

** Correlation is significant at 0.01 level (p< 0.01)

Spectral content

The frequency distribution of noise levels for the two ICU’s are shown in Figure 2.7 and

Figure 2.8. Similar trends in frequency content are seen for the 4 locations within each

individual ICU, with the unoccupied patient room generally quieter than the occupied

spaces.

41
 MedSurg-ICU
80

Sound Pressure Level

70
60

(dB re 20 μPa)
50
40
30
20
10
0
16

31.5

63

125

250

500

1000

2000

4000

8000
Nurse Station
Occupied Paitent Room w/out resp. vent
Empty Patient
Unoccupied RoomR.
Patient
Corridor

Figure 2.7 Average sound pressure levels across frequency in the MedSurg-ICU

Neuro-ICU
80
Sound Pressure Level

70
(dB re 20 μPa)

60
50
40
30
20
10
0
16

31.5

63

125

250

500

1000

2000

4000

8000

Nurse Station
Occupied Paitent Room w/out resp. vent
Unoccupied
Empty Patient
Patient
RoomR.
Corridor

Figure 2.8 Average sound pressure levels across frequency in the Neuro-ICU

42
When comparing the two ICU’s to each other, the sound pressure levels are generally

somewhat higher in MedSurg-ICU, particularly from 250 Hz-8 kHz. At 8 kHz, clearly

noticeable noise level differences have occurred between two ICU’s nurse stations and

occupied patient rooms (e.g., at 8kHz Neuro-ICU nurse station=41dB; MedSurg-ICU

nurse station=49dB). At 250 Hz and 500 Hz, clearly noticeable noise level differences

have occurred between the empty patient rooms in the two wards (e.g., at 500Hz Neuro-

ICU= 40dBA; MedSurg-ICU=51dB). The differences between ICU’s above 250 Hz are

likely related to increased noise from occupants, alarms, and airflow from ventilation

systems in the MedSurg-ICU. Below 250 Hz, sound levels are generally somewhat higher

in the Neuro-ICU. In the empty and occupied patient room, noise level differences

between the MedSurg-ICU and Neuro-ICU at 16 Hz are clearly noticeable. The

difference between the ICU’s at low frequencies may be related to noise from HVAC

equipment located near the center of the Neuro-ICU. The equipment is located in an

outdoor space that is not accessible by the occupants but included in the design to provide

natural light for some patient rooms.

Noise Fluctuations

The traditional fluctuation metric of (L10 - L90) does not have sufficient range to capture

the noise fluctuation differences in two ICU`s and at different locations in each ICU. At

all measurement locations, L33 results are consistently within 1 dB of non-weighted Leq

results. However, noise fluctuations calculated based on the differences between

maximum and averaged sound pressure levels (LMax minus Leq); and peak and averaged

sound pressure levels (LPeak minus Leq) are effective in capturing the noise fluctuation

43
differences. Overall, noise fluctuations are higher in the MedSurg-ICU compared to the

Neuro-ICU as shown in Figure 2.9. The interpretation is that the MedSurg-ICU nurses

are exposed to more dramatic noise level changes compared to the Neuro-ICU nurses.

This result is consistent with the significant differences between nurse outcomes and

subjective loudness and annoyance levels in the two ICUs. Figure 2.9 also shows the

consistency between the noise fluctuation trends and the trend generated by annoyance

and loudness perception of nurses in the two ICUs.

100 5
90
Sound Pressure Level

4.5

Annoyance - Loudness Levels

80
4
(dB re 20 μPa)

70
60 3.5
50 3
40 2.5
30
2
20
10 1.5
0 Corridor 1
Occupied
P.R.
Corridor

Unoccup.
P.R.
Nurse

NSNurse
Station
Occupied

Station
P.R.

Unoccup.
P.R.
OPR

COR

OPR

COR
Neuro-ICU

EPR

EPR
MedSurg-ICU
NS

Neuro-ICU MedSurg-ICU

LMax-Leq (dB)
Lmax-Leq-dB LPeak-Leq (dB)
Lpeak-Leq-dB
Annoyance Loudness

Figure 2.9 Non-weighted noise fluctuation levels (calculated based on noise level
differences between LMax -Leq and LPeak -Leq), annoyance and loudness levels

44
Speech Interference Levels

The potential for the background noise to interfere with speech communication, or

“speech interference level (SIL)” is also evaluated, as shown in Table 2.4. The SIL is

calculated as the arithmetic average of octave band sound pressure levels from 500Hz to

4kHz, with a higher SIL indicating more potential for interference (Mehta, Johnson, &

Rocafort, 1999). In general, the SIL’s in the MedSurg-ICU are slightly higher at all 4

locations compared to the Neuro-ICU. The SIL is highest at the nurse stations, ranging

from 50-53 dB. At this level, two female nurses will be able to (barely) communicate

with each other in normal voice up to a distance of nearly 0.9m-1.2 m (3ft-4 ft). The

distance would increase to 1.6m-2.3 m (5.5ft-7.5 ft) if the nurses raise their voices

(Mehta, Johnson, & Rocafort, 1999). Lower SIL levels can enable safer

communications from longer distances. Moreover, compared to females, males in general

are able to communicate better at longer distances.

45
Table 2.4 Speech interference levels at different locations in two ICUs (also includes
corresponding distances between speaker and listener for reliable communication during
normal voice and raised voice usage)

MedSurg-ICU Neuro-ICU

Speaker- Speaker-
SIL listener distance (ft) SIL listener distance (ft)
(dB) Normal Raised (dB) Normal Raised
voice voice voice voice
Nurse
53 3.0 5.5 50 4.0 7.5
Station
Occupied
51 3.5 6.5 49 4.5 9.0
Patient Room
Unoccupied
40 12.0 23.5 35 21.5 43.0
Patient Room

Corridor 50 4.0 7.5 49 4.5 9.0

SIL: Speech interference level

2.2.4 Discussion

This study discussed objective and subjective characteristics of hospital sound

environments in detail. Overall, the MedSurg-ICU is perceived as significantly louder,

more annoying, and having a greater negative impact of noise on work performance,

health outcomes, and anxiety as compared to the Neuro-ICU. Additionally, the noise is

consistently perceived as louder and more annoying at the all four locations (nurse

station, occupied and unoccupied patient rooms, and corridors) in the MedSurg-ICU as

compared to the Neuro-ICU. However, the nurses` loudness and annoyance perceptions

significantly differed only at the nurse station and in the unoccupied patient room.

Surprisingly, there is little difference between the sound environments of the two ICU’s

46
based on traditional overall measures of Leq, LMin, and LMax, and LPeak, particularly

for occupied spaces. The objective differences between the occupied sound environments

in the two wards only emerges through a more comprehensive analysis of the occurrence

rate, frequency content, and the speech interference level. Further, perceived annoyance

and loudness levels are significantly and positively correlated to mid-level transient

sound occurrence rates. This indicates that the traditional overall noise measures may be

inadequate to capture the differences in perception between hospital sound environments.

Interesting differences are also found when comparing similar locations (nurse station,

occupied and unoccupied patient rooms and corridors) within each individual ICU.

Specifically, the MedSurg-ICU nurse station is perceived as significantly louder than

other locations within that ICU and the Neuro-ICU unoccupied patient room is perceived

as significantly quieter and less annoying than other locations within that ICU. Nurse

stations are highly transient sound environments where sound sources include medical

alarms, telephone ringing, staff talking and laughing, footfall noise, etc. Unlike nurse

stations, empty patient rooms with the doors closed have more stationary sound

environments where the main noise source is the HVAC system. In highly transient

sound environments such as nurse stations, differences in perception are likely more

related to differences in the occurrence rate of maximum and peak levels. On the other

hand, in more stationary sound environments such as unoccupied patient rooms the

differences in subjective perception are likely more related to differences between overall

equivalent levels (Leq).

47
By comparing the objective noise measures and subjective perception between the two

ICU’s and at four locations within each ICU, an interesting conclusion regarding noise

measures emerges. One must consider the overall equivalent level (Leq) or some related

measure of loudness in addition to the occurrence rate, or some related measure of the

transient nature of the sound. The perception of unoccupied spaces (such as empty patient

rooms) will likely be more related to the overall level, while the perception of occupied

spaces (such as nurse stations) will likely be more related to the transient nature of the

sound. The frequency content and SIL (or related speech measures) should also be

considered.

2.3 EMPIRICAL STUDY 2

TIME-BASED VARIATIONS OF HOSPITAL NOISE

2.3.1 Scope

The study discussed in the previous section compared two critical care sound

environments objectively and subjectively. One of the ICUs is perceived as louder, more

annoying, and having a greater negative impact of noise on work performance, health

outcomes, and anxiety. Surprisingly, there is little difference between two ICU sound

environments based on traditional overall noise measures. The objective differences

between the occupied sound environments in the two units only emerges through a more

comprehensive analysis of the “occurrence rate” of peak and maximum levels, frequency

content, and the speech interference level. To assess the effectiveness of the new acoustic

metric, “occurrence rate”, a follow up empirical study is conducted. The second study is

also important to assess the behavior of sound over time in the two critical care settings.

48
2.3.2 Methodology

Subjective noise levels

Two additional sets of subjective noise level analyses are conducted by using the data

collected in the previous empirical study: a) weekday vs. weekend, and b) day vs. night

vs. shift changes. In addition to the components already discussed, the online survey also

includes questions about noise-induced annoyance and loudness levels during different

times of the day and different days of the week. Details of the methodology are shown in

section 2.2.2.

Objective noise levels

Two additional sets of objective noise level analyses are conducted by using the data

from the previous empirical study: a) weekday vs. weekend, and b) day vs. night vs. shift

changes. As previously described in section 2.2.2, 96 h of sound data has been collected

at the central nurse stations of the two ICUs (see Figure 2.10 and Figure 2.11 for nurse

station locations and layouts). In addition to weekdays, noise level measurements are

intentionally conducted during the weekend as these days have not been typically

included in the previous work. Noise level measurements have taken place at the nurse

stations from Thursday to Monday. Sound data collected during the three weekdays

(Thursday-Friday and Monday), and at the weekend (Saturday and Sunday) are analyzed

separately. For the calculation of overall noise levels in each ward, all sound data

collected from Thursday to Monday are averaged. In the second analysis, noise levels

during the day, night and shift times are analyzed separately. Day and night times are

defined based on 12h nurse shift in the Neuro-ICU (7am-7pm and 7pm-7am) and 8h

49
nurse shift in the MedSurg-ICU (7am-3pm, 3pm-10pm and 10pm-7am). In two wards,

nurse shift changes continue about 45-60min after and start 20-25min before the shift.

Figure 2.10 Floor plans of two ICU`s showing the distribution of nurse stations
(left) Neuro-ICU; (right) MedSurg-ICU

Figure 2.11 Floor-plans of nurse stations in the two ICU`s

(left) Neuro-ICU; (right) MedSurg-ICU

50
2.3.3 Findings

Subjective noise levels

Subjective noise level during the weekdays and at the weekend

The first set of subjective noise level analysis is conducted to assess MedSurg-ICU and

Neuro-ICU nurses` annoyance and loudness perceptions for weekdays vs. weekends.

Results are shown in Table 2.5. A discrete 5-point rating scale was used for noise-

induced annoyance and loudness perception. Nonparametric Mann Whitney U test is used

to analyze the significant differences between the perceptions of nurses in the two units.

Compared to Neuro-ICU, the MedSurg-ICU is perceived as louder and more annoying

both during the weekdays and the weekend. However, statistically significant differences

between loudness and annoyance perceptions of the nurses in the two units have occurred

only during the weekdays. Overall, Neuro-ICU nurses` loudness and annoyance

perceptions do not vary much when comparing weekday to weekends. However, the

trend is that MedSurg-ICU nurses found weekends quieter and less annoying compared to

weekdays.

Table 2.5 (a) Annoyance and loudness perceptions for weekdays vs. weekend
(b) distribution of nurse annoyance and loudness perceptions in the two ICUs

Annoyance Loudness
Weekday Weekend Weekday Weekend

Neuro‐ICU 3.2 3.0 3.2 3.1

MedSurg‐ICU 4.0* 3.2 3.9* 3.3

* The mean perception rating is significantly higher at 0.05 level (p<0.05)

** The mean perception rating is significantly higher at 0.01 level (p<0.01)

51
 Table 2.5 continued

Annoyance Loudness
5-point scale
Weekday Weekend Weekday Weekend
Completely disagree-1 25.3% 8.6% 15.0% 10.3%
Somewhat disagree-2 18.4% 8.6% 20.5% 10.3%
Neuro-ICU Neither agree nor disagree-3 17.1% 15.4% 14.3% 18.0%
Somewhat agree-4 22.0% 21.0% 24.0% 25.0%
Completely agree-5 24.0% 28.0% 25.0% 26.0%
Completely disagree-1 4.0% 17.4% 0.0% 10.0%
Somewhat disagree-2 4.3% 4.3% 4.3% 8.0%
MedSurg-ICU Neither agree nor disagree-3 8.7% 26.1% 8.7% 26.1%
Somewhat agree-4 43.0% 24.0% 39.0% 30.0%
Completely agree-5 39.0% 25.0% 40.0% 22.0%

Subjective noise level during the day and night times and shift change

The second set of subjective noise level analysis is conducted to assess Neuro-ICU and

MedSurg-ICU nurses` noise level perceptions during different times of the day. Two unit

nurses are asked to rate their annoyance and loudness perceptions during the day times,

night times, and shift changes. According to Mann Whitney nonparametric significance

test results, the MedSurg-ICU is consistently perceived to be more annoying and louder

during certain times of the day compared to Neuro-ICU (Table 2.6). Although the trend is

that the MedSurg-ICU is perceived worse during all times, statistically significant

differences occurred only during the day times and shift changes. Additionally, the trend

is that in both units nurses generally has found noise levels slightly more annoying during

the shift changes compared to day and night times. However, their loudness perceptions

during the day times and shift changes are very similar. The trend is that night times were

perceived as less annoying and quieter in both units.

52
Table 2.6 Annoyance and loudness perceptions during different times of the day

Loudness Annoyance
Day Night Shift Day Night Shift

Neuro‐ICU 2.9 2.7 3.2 3.3 2.8 3.4

MedSurg‐ICU 3.9* 3.3 4.2* 4.3* 3.4 4.3*

* The mean perception rating is significantly higher at 0.05 level (p<0.05)

** The mean perception rating is significantly higher at 0.01 level (p<0.01)

Annoyance Loudness
5‐point scale
day time night time shift time day time night time shift time
Completely disagree‐1 18.6% 10.0% 8.6% 8.6% 11.4% 25.0%
Somewhat disagree‐2 17.1% 11.4% 17.1% 12.9% 11.4% 17.6%
Neuro‐ICU Neither agree nor disagree‐3 20.0% 12.9% 14.3% 25.0% 13.0% 25.4%
Somewhat agree‐4 16.0% 20.0% 26.0% 27.0% 21.0% 23.2%
Completely agree‐5 22.0% 24.0% 27.0% 23.0% 25.0% 21.6%
Completely disagree‐1 4.3% 13.0% 4.3% 0.0% 16.0% 0.0%
Somewhat disagree‐2 0.0% 15.0% 4.3% 0.0% 18.0% 4.3%
MedSurg‐ICU Neither agree nor disagree‐3 19.0% 18.0% 13.0% 21.7% 13.0% 17.4%
Somewhat agree‐4 28.0% 25.0% 34.0% 36.0% 25.0% 33.0%
Completely agree‐5 44.0% 27.0% 47.0% 45.0% 30.0% 48.0%

To summarize, subjective noise levels in each unit varied during different times of the

day and different days of the week. The trend is that the MedSurg-ICU is consistently

perceived worse. However statistically significant differences occurred only during the

weekdays. Significant differences between nurse perceptions is also evident only during

the day times and shift changes.

Objective noise levels

Noise levels during the weekdays and at the weekend

The first set of objective noise level analysis is conducted to assess the differences

between the two ICU sound environments and occurring in each unit during the

53
weekdays and the weekend. Related with the scope of this study, noise levels in the two

ICUs are analyzed via “occurrence rate” analysis.

The overall averaged sound pressure levels (Leq) at the nurse stations of two units range

between 57-60dBA and 56-58dBA during the weekdays and at the weekend respectively.

Given these results, the difference between two wards during different days of the week

is imperceptible (Mehta et al., 1997).

As shown in Figure 2.12-Figure 2.16, occurrence rate values present the percentage of

time that the peak noise levels exceeded values ranging from 80dBC to 100dBC. In both

units, Lpeak noise levels exceed 80dBC more than 95% of the time during the weekdays

and at the weekend. The occurrence rate of Lpeak noise levels exceeding 100dBC is

negligible in both units during the weekdays and at the weekend. On the other hand, as

shown in Figure 2.12, substantial differences between the two wards emerge for mid-

level Lpeak noise levels during the weekdays. As noted in earlier sections, the

terminology for “mid-level” for 90dBC does not refer to the perceived loudness of the

transient level, but simply serves to distinguish the level ranges analyzed in this study

into lower, mid and higher regions. In the MedSurg-ICU, Lpeak noise levels exceed

90dBC more of the time than in the Neuro-ICU during the weekdays. The difference

between the two units is much more evident during the weekdays as compared to the

weekend. Similar results are also found for Lmax levels. To summarize, although there is

not a difference in overall levels (LAeq), the occurrence rate analysis indicate that the

54
MedSurg-ICU is a more impulsive sounding environment compared to Neuro-ICU,

particularly during the weekdays.

Figure 2.12 Occurrence rate of Lpeak noise levels during the weekdays in the two units

Figure 2.13 Occurrence rate of different Lpeak noise levels during at the weekend in the
two units

55
Noise levels during the day times, night times and shift change

The second set of objective noise level measurements is conducted to assess noise level

differences during different times of the day. The overall averaged Leq levels at the

Neuro-ICU and MedSurg-ICU nurse stations range between 58-61dBA, 55-58dBA and

59-61dBA during the day, night and shift change time, respectively. Given these results,

the difference between two wards in comparing different times of the day is either

imperceptible or just perceptible to the human ear (Mehta et al. 1997).

However, the occurrence rate of mid-level transient sounds is higher in the MedSurg-ICU

at all times of the day as shown in Figure 2.14, Figure 2.15 and Figure 2.16. The

difference between the two units is much more evident during the day times and shift

changes as compared to night times. In the MedSurg-ICU, more percentage of the time

Lpeak noise levels exceed 90dBC during the day times and shift changes compared to

night times. The occurrence rate of the mid-level transient sounds at the Neuro-ICU nurse

station does not vary much during the different times of the day.

Figure 2.14 Occurrence rate of different Lpeak noise levels during the day times in the
two units

56
Figure 2.15 Occurrence rate of different Lpeak noise levels during the shift changes in
the two units

Figure 2.16 Occurrence rate of different Lpeak noise levels during the night times in the
two units

Overall noise levels

For the third set objective noise level analyses, entire 96h data collected at the nurse

station of each unit are considered. Sound samples collected from Thursday-Monday at

each nurse station are averaged to calculate overall noise levels. Overall averaged noise

57
levels at the nurse stations of two units range from 56-60 dBA Leq, 97-105 dB Lmax, 54-

57dB Lmin, 113-120dBC Lpeak (Figure 2.17). As before, based on Leq the difference

between two unit nurse station sound environments is just perceptible to human ear. The

difference between two unit sound environments emerges for mid-level transient sounds.

Figure 2.18 shows the overall levels at the two unit nurse stations where Lpeak exceed

90dBC more often in the MedSurg-ICU (53%) than in the Neuro-ICU (24%).

Figure 2.17 Overall Leq, Lmax, Lpeak and Lmin levels in the two ICU`s

58
Figure 2.18 Overall occurrence rates of Lpeak noise levels in the two ICUs (Please note
that Figure 2.18 and Figure 2.4 are not the same. Figure 2.4 represents the occurrence
rate of different LMax noise levels and Figure 2.18 represents the occurrence rate of
different LPeak noise levels)

2.3.4 Discussion

Empirical study 2 conducts objective and subjective noise level analyses to assess

changing sound environment characteristics of each unit over time and assesses whether

occurrence frequency of transient sounds track with the nurse loudness and annoyance

perceptions during different times of the day and days of the week.

The first set of occurrence rate analysis indicates that MedSurg-ICU is consistently more

impulsive at all times compared to Neuro-ICU. However, occurrence rate of mid-level

transient sounds in the MedSurg-ICU is substantially higher compared to the occurrence

rate of mid-level transient sounds in the Neuro-ICU only during the day time and shift

change. These results are consistent with the perceptions of nurses in the two units.

Nurses have found MedSurg-ICU more annoying and louder at all times. However

59
significant differences between perceptions of the nurses in the two units have occurred

only during the day time and shift change.

The second set of occurrence rate analysis indicates that MedSurg-ICU is consistently

more impulsive compared to Neuro-ICU during the weekdays and the weekend.

However, occurrence rate of mid-level transient sounds is substantially higher in the

MedSurg-ICU compared to the occurrence rate of mid-level transient sounds in the

Neuro-ICU only during the weekdays. These results are also consistent with the

perceptions of nurses in the two units. Nurses have found MedSurg-ICU more annoying

and louder during the weekdays and the weekend. However significant differences

between perceptions of nurses in the two units only have occurred during the weekdays.

In the third set of occurrence rate analysis, the entire 96h data collected at the nurse

station of each unit is considered. The results indicate that the occurrence rate of mid-

level transient sounds is substantially higher at the nurse station in the MedSurg-ICU

compared to the occurrence rate of mid-level transient sounds at the nurse station in the

Neuro-ICU. This result is also consistent with the perceptions of the nurses in the two

units. As presented in Table 2.2, MedSurg-ICU nurses have found the nurse station in

their unit significantly more annoying and louder compared to the Neuro-ICU nurses.

Overall, occurrence rate analysis findings track with the loudness and annoyance

perceptions of the nurses in the two units. This suggests the potential effectiveness of

occurrence rate in predicting the nurse annoyance and loudness perceptions in ICUs.

60
2.4 EMPIRICAL STUDY 3

EFFECTS OF PARTICULAR HOSPITAL NOISE SOURCES

2.4.1 Scope

Clinical alarms are impulsive and frequently reoccurring sound events in ICUs. The

previous empirical studies (1 and 2) discussed in the last two sections have highlighted

potential negative impacts of impulsive characteristics of hospital sound environments on

staff outcomes. As an extension of earlier empirical studies, a third empirical study is

presented below that assesses the effects of a particular highly impulsive sound event

(i.e., clinical alarms) by comparing the nurse outcomes in the two ICUs.

2.4.2 Methodology

Nurse perceptions

Subjective noise level analysis is conducted by using the data collected in empirical study

1. In addition to the components previously discussed, the survey also includes questions

about alarm-induced nurse outcomes. For methodology details, please refer to section

2.2.2.

2.4.3 Findings

Effects of alarms and overall-noise in the two ICUs

A subjective noise level analysis is conducted to compare nurses` disturbance levels due

to alarms in the two units. Mean perception ratings of the alarm-induced nurse outcomes

in the MedSurg-ICU and Neuro-ICU are shown in Table 2.7. The effects of alarms on the

MedSurg-ICU nurse outcomes are consistently significantly higher than the effects of

61
alarms on the Neuro-ICU nurse outcomes. To restate, the MedSurg-ICU nurses have

perceived alarms as louder, more annoying, and having higher negative impacts on their

health, work performance, and anxiety levels. Moreover, the difference between the two

units is the highest for anxiety compared to the other outcomes. In the MedSurg-ICU, the

trend is that alarms have a greater impact on particular nurse outcomes of anxiety levels

and loudness perceptions as compared to the other outcomes. In the Neuro-ICU, the trend

is that alarms have the highest impact on nurses` loudness perceptions as compared to the

other outcomes. All of the other outcomes in the Neuro-ICU are on average less than 3.0,

which would correspond to “disagreement” ratings on the 5-point scale. This indicates

that on average the Neuro-ICU nurses have not perceived alarms as contributing to

outcomes with perhaps the exception of loudness.

Table 2.7 Perceived medical alarm-induced outcomes (i.e., loudness, annoyance, work
performance, health and anxiety) of the nurses in the two ICUs. Higher numbers
represent more negative perception.

Alarm‐ Alarm‐ Alarm‐ Alarm‐ Alarm‐

induced induced induced induced induced
loudness annoyance work perf. health anxiety

Neuro‐ICU 3.2 2.6 1.8 1.8 2.3

MedSurg‐ICU 4.0* 3.4** 2.8** 3.0** 4.0**

* The mean perception rating is significantly higher at 0.05 level (p<0.05)

** The mean perception rating is significantly higher at 0.01 level (p<0.01)

62
Table 2.8 Distribution of medical-alarm induced outcomes (i.e., loudness, annoyance,
work performance, health and anxiety) of the nurses in the two ICUs

Alarm- Alarm- Alarm- Alarm- Alarm-

5-point scale induced induced induced induced induced
loudness annoyance work perf. health anxiety
Completely disagree-1 6.0% 30.0% 3.0% 60.0% 42.9%
Somewhat disagree-2 12.0% 21.1% 15.7% 17.1% 20.0%
Neuro-ICU Neither agree nor disagree-3 13.0% 20.0% 20.7% 14.3% 8.6%
Somewhat agree-4 34.3% 25.4% 10.0% 2.9% 22.8%
Completely agree-5 23.0% 5.0% 8.6% 5.7% 5.7%
Completely disagree-1 0.0% 9.1% 5.0% 19.8% 8.8%
Somewhat disagree-2 17.4% 18.2% 9.0% 18.4% 4.3%
MedSurg-ICU Neither agree nor disagree-3 8.7% 13.6% 12.4% 13.0% 13.0%
Somewhat agree-4 30.4% 35.4% 28.4% 15.0% 30.4%
Completely agree-5 43.5% 22.7% 22.0% 29.8% 43.5%

The nurses in the two units have been also asked to rate their level of agreement with the

following statement: “I sometimes tune out the alarms.” The majority of the nurses

“somewhat agree” with this statement (Figure 2.19). Only 20% of the nurses completely

disagree with this statement.

Figure 2.19 Percentage of nurses who agreed that they sometime tune out alarms
averaged across the two units

63
2.4.4 Discussion

This study compares the effects of clinical alarms on nurse outcomes in the two ICUs.

Alarms are one of the highly impulsive sound events that occur frequently in the ICUs.

Regardless of their good intent, alarms can be highly disruptive of nurse outcomes.

Furthermore, effects of clinical alarms on nurse outcomes can vary significantly between

different care settings. For example, in the MedSurg-ICU nurses have found clinical

alarms more disruptive of their outcomes (i.e., health and performance outcomes, anxiety,

annoyance, and loudness perceptions) compared to nurses in the Neuro-ICU. This finding

also agrees with occurrence rate analysis findings introduced in Section 2.2.3 and Section

2.3.3. According to two empirical study findings, the occurrence frequency of mid-level

transient sounds is greater in the MedSurg-ICU compared to the occurrence frequency of

mid-level transient sounds in the Neuro-ICU. Furthermore, alarms providing wrong

information known as “false alarms” potentially lead to misconceptions and result in

unwanted situations. Unfortunately, considerable amount of the Neuro-ICU and

MedSurg-ICU nurses have reported that they sometimes tune out clinical alarms.

64
 CHAPTER 3

CHARACTERISTICS OF PATIENT AUDITORY MONITORING

In complex hospital sound environments, caregivers conduct vital tasks including patient

auditory monitoring. Chapter 2 examined the overall relationships between subjective

perception and objective sound level measurements to gain a more thorough grasp of how

perceptual and physical acoustic parameters interact in the ICU setting. Further, the

Chapter 2 research suggested that the Acoustic qualities of hospital sound environments

can have significant impact on nurses` auditory monitoring performance. Chapter 3

focuses specifically on the concept of auditory monitoring by providing a comprehensive

overview of the factors related to auditory monitoring including previous research related

to the topic; and presents the results from a case study comparing nurses` auditory

monitoring performance in two ICUs with different architectural designs.

3.1 LITERATURE REVIEW

3.1.1 Patient monitoring in critical care units

3.1.1.1 Overview of patient monitoring

Today’s ICUs are the outgrowth of respiratory care units established to provide

continuous patient monitoring, and life support treatments for patients suffering from

polio or tetanus in the early 20th century (Berenson, 1984). The establishment of these

units was initiated by the invention of a mechanical ventilator called the “iron lung” in

the late 1920s. In the 1960s, ICUs started providing electronic monitoring. Originally,

65
electronic monitoring was developed for cardiac rhythm monitoring to preserve the

health of relatively stable patients (Russell, 1979).

In today`s ICUs, continuous monitoring is provided for critically ill patients by a wide

range of technological services, continuous observation, and frequent measurement of

vital signs (Safar, & Grenvik, 1971) . Critical care units differ based on their specialty

(e.g., multispecialty, specialty); however, patient monitoring is key for all types. In

multispecialty ICUs (e.g., Medical Surgical-ICU), seriously ill medical surgical patients

with a wide spectrum of illnesses are provided care. In specialty ICUs, patients with

similar problems or specific diseases are provided care such as care of premature or

critically ill newborn (Neonatal-ICU), care of critically ill and injured children (Pediatric-

ICU), adult cardiac disease (Cardiac-ICU), trauma care, care of multiple organ

dysfunction in the surgical ICU (Surgical-ICU), care of neurological and neurosurgical

patients (Neuro-ICU).

Patients in ICUs generally suffer from different health problems including failing bodily

functions, multiple and complex sets of medical problems requiring support for two or

more organ systems (Bennett, & Bion, 1999). Patients are accepted to ICUs mainly

because they need monitoring for potential disturbances to those who are critically ill,

and receive life-supporting treatment or continuous intensive nursing and physician care

(Berenson, 1984). ICUs typically differ based on the level of care provided such as

Level-I, Level-II and Step-down critical care units (Miller, & Swensson, 2002). Typically

Level-I critical care units house patients with multisystem failure, and complicated

66
medical needs requiring continuous availability of sophisticated equipment and

specialized medical staff. Level-II critical care settings typically offer care to patients

with single organ failure. Step-down units or intermediate care patients typically require

close monitoring but not intensive medication or therapy.

There is a growing demand for critical care in the U.S. There are approximately 6,000

ICUs caring for 55,000 critically ill patients each day (SCCM, 2006). About 16% of the

annual admissions to U.S. hospitals are admitted to ICUs, which corresponds to about six

million patients and 2% of the U.S. population (Kersten et al., 2003). ICU patients` health

status requires more attentive monitoring as compared to standard nursing-floor patients.

ICU diseases can develop rapidly in seconds as multiple underlying medical problems

can interact and produce severe unpredictable physiologic complications (Hillman, &

Bishop, 2004). Berenson (1984) indicated that 20% to 40% of the ICU patients died in

the hospital after being transferred from the ICU to the regular medical floor. Providing

an environment that is supportive of patient monitoring is necessary for patient safety,

particularly in ICUs.

3.1.1.2 The profile of caregivers conducting patient monitoring

Critical care nurses provide care to patients who are very sick and critical care nursing

can be a very demanding job. Critical care nurses are educated to provide highly

technical expert care and trained to deal with end-of-life cases (Gross, 2006). They

continuously provide vigilant efforts to maintain patient safety while working under time

pressure.

67
Unfortunately, their workplaces are not always equipped with support work

environments. Demanding work conditions and poorly designed work environments can

lead to high levels of cognitive, physical, and emotional workload, negative health

outcomes, and job dissatisfaction among ICU nurses (Aiken et al., 2002; Carayon et al.,

1999; Ulrich et al., 2009). Findings indicate that emotional stress and working under time

pressure can also lead to increased risk for work-related musculoskeletal disorders,

physiological changes, increase in nurse sensitivity to pain, and perceived physical

workload (Carayon et al., 1999). Moreover, according to American Association of

Critical-Care Nurses (AACN) national survey, a significant portion of the nurses are not

satisfied with their jobs: more than 16% of the nurses indicated that their intention is to

quit their job in the following 12 months and about 27% in the next 3 years (Ulrich et al.,

2009). Similarly, Aiken (2002) found that 1 in 4 bedside nurses is considering leaving

their jobs in a year and half of the nurses reported high burnout range. Given these

statistics, a significant drop in the number of nurses is feasible. A federal agency

estimated that the shortfall of nurses could approach 800,000 by 2020 (DHHS, 2002).

On the other hand, there is some evidence showing positive impacts of improved

environmental conditions such as hospital sound environments on nurse outcomes. The

empirical studies in Chapter 2 compared the sound environments of two different ICUs

and found significant differences between nurse outcomes in the different units due to the

acoustic qualities of these sound environments. Another non-hospital study found that

68
task supportive environmental conditions such as effective sound environments in the

workplaces might help to buffer the effects of job stressors (Leather et al., 2003).

3.1.1.3 Patient monitoring and other critical nurse tasks

ICU nurses conduct different critical and non-critical tasks. Highly critical nurse tasks

mainly relate to the care of the sick and maintenance of patient safety (Diers, 2004).

Some of the tasks related with the care of the sick include skin wound care, managing

pain, providing comfort, teaching patients and families how to manage their care after

hospital discharge, adequate nourishment of patients, infection control intervention, and

prevention of hazards. Nurse tasks that critically relate to patient safety include providing

continuous patient monitoring for early detection of adverse events and complications,

medication errors, and mobilization of resources for timely interventions and rescue

(Aiken, 2005). Nurse time is also allocated with non-critical tasks such as waiting on

other systems (e.g., delivery of medication, lab results), retrieving patient supplies, and

looking for equipment (Hendrich et al., 2008). However, the same study found that

majority of the nurse time (78%) is spent on clinical nursing practice functions including

patient monitoring and patient care activities.

3.1.1.4 Types of patient monitoring

Critical care nurses conduct highly routine patient monitoring tasks. These tasks require

vigilant attendance to multitude of cues and continuously alert minds, vigorous body

states, and prompt-accurate decisions (Carnevale, 2009). ICU nurses continuously

monitor the patients` health status and maintain their normal bodily conditions by

69
assessing visual and auditory cues. To recognize any irregularities, nurses remain

attentive to the patients’ physiologic status and treatment devices, and they are

continually “tuned in” to the immediate recognition of any disruption in the patients`

condition. They must remain tuned in even when performing routine maintenance

activities (e.g., medication preparation, blood procurement). In the event of any abnormal

changes, nurses instantly evaluate the significance of the event to patient safety, and

initiate an appropriate response and effective intervention if required.

3.1.1.5 Visual Patient Monitoring

Since the late 19th century, visual patient monitoring has been a significant element of

nursing practice. Visual cues have been used for the early detection of some health

complications. Examples include changes in the color or texture of the skin (e.g., rashes,

bruising), asymmetric chest movements, abnormal bleeding, proper placement of the

patient on the bed, invasive equipment around the patient bed, and any other condition

that might be related with distress and patient comfort (Downes, 2009). Effective visual

monitoring can be enabled by the absence of any barrier between observer and target. It is

identified as inadequate when nurses are assigned to patients whose rooms are physically

distant from each other (Kalish, 2005). To maximize the patient visibility, the design of

today’s critical care units aim to provide a direct line of vision between every patient

room and nurse work areas by adopting different design strategies such as radial design

layout and distributed nurse stations (SCCM, 1995).

70
3.1.1.6 Auditory Patient Monitoring

Like visual monitoring, assessment of auditory cues has become a significant component

of nursing practice (Downes, 2009). As a part of the auditory monitoring process, nurses

attentively listen to the auditory cues, detect the ones that might present risks to patient

safety, and localize them to provide proper and immediate response to abnormal sounds.

ICU nurses develop auditory skills that enable them to differentiate auditory cues from

each other and immediately recognize the meaning of each while considering the

potential risks to the patient health status over time.

Auditory cues can be generically classified as alarm or non-alarm. An immediate

response to triggered alarms in ICUs includes physically assessing the patient and

resolving the situation (Richardson, 2004). Nurses` response to an “urgent” medical

alarm in a timely manner can be highly critical. During personal interviews with ICU

nurses, appropriate initial response time to a highly urgent alarm was found to be 30

seconds or less. Caregiver response time to a moderate risk alarm is expected to be 1-3

minutes, and to a low risk alarm 3-5minutes for patient safety (Phillips, &Barnsteiner,

2005). Lack of proper monitoring can delay the response time and even disable the early

detection of complications which can result in serious patient injury and death due to

ventilator failure, breathing circuit disconnections, etc. (ECRI, 1986). According to ECRI,

ventilator dependent patients need special attention and they should never be left

unattended. Otherwise, caregivers should limit their visits in the unit to only locations

where they can hear and rapidly respond to alarms. This is mainly because ventilator

dependent patients cannot breathe spontaneously for substantial periods of time and

71
cannot mutually ventilate themselves. Impaired gas exchange (i.e., supply of adequate

oxygen to the body and elimination of carbon dioxide) can lead to poor tissue perfusion

which leads to a reduction in oxygen delivery to cells as well as a retention of CO2 in the

blood stream (Hinchliff, Montague, &Watson, 1988).

Nurse effectiveness in providing immediate response is significantly related with

caregivers vigilance level (Clarke, &Aiken, 2003). Growing evidence from hospital and

non-hospital research also indicate the effect of environmental factors on nurse work

performance utilizing auditory cues. These qualities are discussed details in the following

sections.

3.1.2 ICU auditory cues

With the adoption of computer based patient monitoring systems, nurses started

monitoring the clinical alarms as well as non-alarm sounds. Today, electronic monitors

are extensively used in critical care units and medical alarms are accepted as one of the

primary auditory cues key for patient safety.

A variety of different medical equipment is used in ICUs. Each type of equipment can

produce multiple alarms to communicate about the urgency level of the alarming

condition such as high-level urgency (emergency alarms), medium-level urgency

(cautionary alarms) and low-level urgency (alerting alarms) (CEN, 1995; Chambrin et al.,

1999; Meredith, & Edworthy, 1995). High-level urgency alarms (e.g., 3-star cardiac

alarms) indicate an urgent situation that can lead immediately to a vital problem and

requires immediate attention of the registered nurses. Medium-level urgency alarms (e.g.,

72
2-star cardiac alarms) indicate a dangerous situation and require rapid attention of the

registered nurses. Low-level urgency alarms (e.g., equipment failure that would pose a

minimum adverse effect) indicate an alert situation that still requires the attention of staff

but does not necessarily require registered nurse response.

Most ICU patients are attached to some combination of medical equipment which can be

grouped in three categories: patient monitors, infusion devices, and life-support

equipment (Hirose et al., 2005). Some examples of patient monitoring systems include

electrocardiogram-blood pressure monitors, noninvasive blood pressure monitors, pulse

oximeters, and capnometers (Kerr, &Hayes, 1983). Some of the infusion devices include

IV pumps and feeding pumps. Respiratory ventilators, intra-aortic balloon pumps,

hemodialysis units, and anesthesia machines are some of the commonly used life support

equipment in ICUs. Some of the most commonly used medical equipment in ICUs is

cardiovascular monitors, respiratory ventilators, infusion devices, and dialysis machines.

Given the plethora of medical devices, a high density of critical medical alarms is not

uncommon in modern ICUs. One study reported that there are at least 33 different

medical alarms that nurses continuously monitor in ICUs (Cropp et al., 1994). The same

study also found that during an average hour on the day shift at least 50 audible medical

signals occurred in a critical care unit.

In addition to clinical alarm sounds, effective monitoring of non-alarm sounds can be

very important for the early detection of health complications. These sounds can be

categorized under non-speech and speech sounds. Examples of non-speech sounds

include patient bodily sounds such as gagging (the sound similar to choaking), strider

73
(constriction in breathing pattern), and sleep apnea (irregular breathing patterns). Safety

threatening sounds might include patient falls, the sounds of leaking air from the oxygen

tube, and sounds of falling objects. ICU nurses typically monitor breathing rates,

respiratory efforts, pattern of respiration (i.e., rhythm and depth), presence of audible

wheeze, and dry cough to assess patient respiratory health (Owen, 1982). Examples of

critical speech sounds include help calls by patients and caregivers, announcements from

overhead pagers, medical conversations either conducted on the phone (e.g., between

nurse and pharmacists) or in person (e.g., between physician and nurse).

3.1.3 Factors impacting auditory monitoring

Monitoring of both alarm and non-alarm auditory cues can be significantly affected by

environmental factors. One study indicated that improved hospital sound environments

can have a significant impact on speech intelligibly outcomes (Blomkvist et al., 2005). In

addition to environmental factors, strategic design of auditory cues such as alarms and

drug names can also reduce the risks to patient safety. Similarities in orthographic (i.e.,

spelling) and phonological (i.e., sound) qualities of drug names increase the probability

of making false recognitions (Lambert et al., 2001). According to national statistics, 1 out

of 4 medical errors in the U.S. involves drug name confusion (U.S. Pharmacopeia, 1997).

Furthermore, according to another study the development of new auditory warnings

based on vowel sounds can be less irritating and less susceptible to masking by the

background noise (Stanford, Mclntyre, & Hogan, 1985).

74
There are also studies focusing on three of the individual components of the auditory

monitoring process: sound detection, sound recognition, and sound localization. The

findings of hospital and non-hospital research on these three components are discussed in

the following sections. Overall, sound detection, recognition, and localization can be very

challenging tasks in noisy, multi-source, or reverberant environments. Studies conducted

in controlled lab environments as described below provide some insight to the attributes

that might have impact on nurse auditory monitoring.

3.1.3.1. Sound detection

Sound detection involves the ability to hear an auditory cue, or “target.” In settings like

ICUs, detection of auditory cues in noisy and multi-source environments is inevitable.

The human auditory system is capable of focusing attention on one source/speaker, even

when there are multiple competing sound sources in the background (Cherry, 1953).

This phenomenon is known as the “cocktail party effect”. However, Stifelman (1994)

indicated that listening to the target speech signal while simultaneously exposed to two

other competing background signals may require too much effort. There is some evidence

indicating that such demanding sound tasks can be supported by specific acoustic

qualities of sound targets and competing signals. For example, Treisman (1964) found

that listeners` primary signal monitoring performance improved when competing

background signals had similar acoustic features to each other. This phenomenon is

referred to as “auditory stream segregation” (Bregman, 1990). According to this theory,

two competing sound signals with similar acoustic features (e.g., pitch, spatial location)

form one single auditory stream. This enables the perception of multiple competing

75
signals as one single interfering signal. Bess and Humes (1990) found that human hearing

is most accurate for frequencies between 1kHz-4kHz, the frequency range also

corresponding to primary intelligibility of speech signals. Competing signals that are in

the same frequency range as speech signals (1kHz-4kHz) thus can potentially mask a

speech target. Other research indicated that a frequency difference of 110Hz between

target and competing signals can improve the intelligibility of the target speech signal

(Brokx, & Nooteboom, 1982). Blauert (1997) indicates that intelligibility of target sounds

with and without meaning might have a significant impact on signal detection.

There is also some evidence showing that at certain sound pressure levels speech

detection can be more effective. According to Kobayaski (2007), the minimum listening

difficulty occurred when the target speech level was 50-55dBA. At higher or lower

speech levels, listening difficulty increased. The same study indicated that humans have a

tendency to increase their voices as loudness increases, a well-documented phenomenon

known as the “Lombard Effect.” Speech levels generally stayed constant when the

background level was less than 40dBA. However, when background level was higher

than 40dBA, the talker adjusted his/her speech level to maintain approximately 15dBA of

signal-to-noise ratio.

3.1.3.2 Sound recognition

Sound recognition involves identifying or interpreting an auditory cue. One study

conducted listening tests with nurses to assess their ability to identify critical alarms

occurring in ICUs (Cropp et al., 1994). It was found that only 50% of critical alarms

76
presented to nurses were identified correctly. Moreover, caregivers with more than one

year experience scored higher than those with less than one year. Wallace et al. (1994)

also highlighted difficulties experienced by caregivers in detecting alarms in operating

rooms. These findings might be associated with limited capabilities of human auditory

system. Patterson and Mikoy (1980) suggested that humans are not able to easily learn

and remember the significance of more than 8 different sounds even in unstressed

conditions. Similarly, another group suggested that 6-10 auditory warning signals would

be the optimum (Momtahan, &Tansley, 1989). Lambert et al. (2001) highlighted

recognition difficulties associated with complex and diverse of acoustic qualities of

auditory cues. According to the study findings, orthographic (i.e., spelling), and

phonological (i.e., sound) similarities potentially increase the probability of caregivers`

making recognition memory errors.

Recognizing a specific medical alarm is difficult in part due to the similar acoustic

content of the many different alarms in an ICU. Some studies have documented the

acoustic qualities of medical alarms. Wallace (1994) measured the intensity and

frequency of 26 anesthesia equipment alarms in an unused operating room suite. The

majority of the alarms consisted of mid to high range frequencies ranged between 250Hz

to 8000Hz. Almost all of the alarms were multi-frequency signals. Different from sine

tones (a.k.a. pure tones), a multi-frequency tone consists of multiple frequencies. Nine of

the alarms consisted of 4-5 different frequencies including 500, 1000, 2000, 4000, and/or

8000Hz. Other two alarms also had 250Hz content. Five of the multi-frequency alarms

were limited to frequencies higher than 1000Hz and seven of them were limited to

77
frequencies higher than 2000Hz. Only three of the twenty-six alarms were pure tones and

had a frequency of 4000Hz. Loudness levels of these alarms ranged between 45dBA to

80dBA. A Japanese study documented the acoustic characteristics of 73 pieces of

medical equipment (Hirose et al., 2005). Different from the findings of the previous

study, the study found that 90% of the alarms generated by the equipment included only a

single frequency and the majority had discontinuous patterns. The spectral content of the

medical alarms activated by “biological information monitors” including ECG/BP

monitor alarms, NIBP and pulse oximeter ranged between 500 and 4000Hz. The

frequency level of the life support equipment including infusion pumps and ventilators

ranged between 1000-4000Hz. Most of the time, the maximum (Lmax) and minimum

(Lmin) sound pressure levels generated by the alarms exceeded 70dB and 50dB,

respectively. The repeating module of the warning signal composed of alarm and silence

known as “alarm cycle” was mostly 1s for different alarms.

To summarize, in U.S. hospitals medical alarms are expected to have multiple

frequencies so that they are less likely to be masked by the background noise. For

example, ASTM Standard F29.03.04 for anesthesia and respiratory care medical

equipment alarm signals requires that medical alarms should have a fundamental

frequency of 150-1000Hz and at least four frequency components between 300-4000Hz.

3.1.3.3 Sound Localization

Sound localization relates to auditory system’s ability to make decisions about the

direction and distance of the sound event (Moore, 1997). Auditory information that is

78
critical for sound localization is known as a “spatial” auditory cue. Various studies have

been conducted in controlled lab environments to explore the attributes necessary for the

assessment of spatial auditory cues.

Estimating distance and direction

Developments in binaural hearing research explain how the human auditory system uses

binaural cues obtained by two ears during sound localization. The auditory system

essentially assesses time and sound level differences between the sound waves arriving at

the left and the right ear (Blauert, 2005). Due to the different path lengths to the two ears,

the arrival times of the sound waves emitted from a single sound source are not always

the same at the left and the right ear (Thompson, 1877). This acoustic principle is known

as “inter-aural time difference” (ITD). On the other hand, existence of the head between

the two ears creates an acoustic shadow and causes sound attenuation which leads to an

“inter-aural (sound) level difference” (ILD) between the left and the right ears

(Steinhauser, 1877). Follow-up studies found a frequency dependence of these two

binaural cues (Rayleigh, 1907). According to Rayleigh’s “duplex theory”, ITDs are most

critical for the localization of low-frequency sounds and ILDs are most critical for high

frequencies.

The sound level of sources, a listener`s familiarity with sound events, and the acoustic

quality of sources are the major factors for distance and direction judgments. For

example, in a sound field free of reflecting surfaces, every doubling in distance produces

6dB reduction in the sound level at the ears. Distances longer than 15m (49ft) lead to an

79
additional sound attenuation and slight change in the spectral balance of the sound

reaching the ears (Ingard, 1953; Plack, 2005). This is mainly because air absorbs more

high frequency energy than low frequency in relation to the length of the air path.

Study findings indicate that a listener’s ability to make accurate directional estimations is

associated with their familiarity with signals. Makous and Middlebrooks (1990) indicated

a significant improvement in listeners` directional judgment skills after training. Most of

them made errors of less than 5 degrees both in horizontal and vertical directions. Plenge

and Bruschen (1971) found that subjects` directional judgment performance improved

when short speech signals were pronounced by people that they are familiar with. The

authors also reported listeners` tendency to report the unfamiliar signals as if they were

coming from behind.

The acoustic quality of signals, such as duration and spectral content, also influence

localization estimations. For example, Aschoff (1963) explored the effects of duration on

direction estimations of the auditory system (cross ref. Blauert 1997). According to study

findings, listeners were able to hear the noise circling around their heads when signals

from circular array of loudspeakers were played with a slow switching speed. When

switching speed was increased, the noise was heard to move between left and right sides.

When the switching speed further increased, the auditory event was heard approximately

in the middle of the head. As another example, Blauert (1997) explored the effect of

spectral content on sound localization and conducted an experiment with 140 subjects by

using pulses of white and pink noise. Approximately 90% of the subjects succeeded in

80
their direction judgments. Sandel et al. (1955) found that the localization performance

was worst around 1.5 kHz-3 kHz. According to another study, localization precision was

best for frequencies between 200Hz-1 kHz and the error magnitude peaked around 3 kHz

(Mills 1958) .

Effects of acoustic environment on sound localization

Listeners` ability to localize sounds varies depending on the acoustic environment. For

example, in reverberant sound environments, sound localization can be very difficult.

This is mainly because later-arriving reflections can reduce the impact of direct sound

and early-arriving reflections necessary for sound localization. The human ear is capable

of integrating early reflections (typically up to 50-80ms after the arrival of the primary

signal) with the direct sound signal. However, strong and later-arriving reflections

(typically those greater than 80ms after the arrival of the primary signal) can cause

changes in the perception of the auditory event such as changes in the direction of the

primary sound signal (Blauert, 1997). Hartman (1983) conducted listening tests in a

controlled room with human subjects to assess the impact of signal qualities and room

acoustics on sound localization performance. Subjects localized broadband sounds easier

when design interventions were adopted to reduce reverberation time such as including

absorptive surface and lowering ceilings. Compared to a higher ceiling condition, at a

low ceiling configuration the reverberation time was about half of the high ceiling

condition. The study also concluded that the localization of the sine tones is independent

of the room reverberation time. In other words, when localizing sine tones the human

auditory system is limited to benefit from the useful reflections in the rooms. Different

81
from complex nature of the sounds in hospitals, sine tones are composed of a single

frequency. As with sound detection, studies also show that sound level differences

between target signal and competing signals can affect sound localization. Houtgast and

Plomp (1968) suggested sound localization can be more effective if the level of

competing noise is 15dB or more below than the target signal. Lorenzi et al. (1999)

indicated sound localization accuracy remains unaffected by competing noise until a 0–6

dB signal-to-noise ratio is reached.

3.2 EMPIRICAL STUDY

CHARACTERIZING AUDITORY PATIENT MONITORING

3.2.1 Scope

Intensive Care Unit (ICU) patients are one of the most sensitive patient populations in a

hospital. Mortality rates among ICU patients range between 10% to 20% in most

hospitals (SCCM 2006). To promote patient safety, ICU nurses continuously monitor the

health status of their patients, and strive to maintain appropriate bodily conditions by

assessing visual and auditory cues, and responding immediately to abnormal changes.

The ability to conduct uninterrupted patient monitoring is a challenging task for ICU

nurses, in particular because of other competing tasks and environmental factors.

Effective monitoring of visual cues is feasible in close proximity to the patient as it

requires the absence of a barrier between observer and target. Study findings indicate that

increased visibility of patients in the ICUs through design strategies can help improve

patient safety (Joseph, & Rashid, 2007).

82
Different from visual monitoring, auditory monitoring is an “eyes-free” cue assessment

technique. When caregivers` hands and eyes are busy, auditory monitoring can enable

uninterrupted monitoring of patients from different locations in the care settings.

Therefore, it is one of the early requirements for novice nurses to familiarize themselves

to auditory cues that might pose a risk to patient safety. Nurses` ability to conduct

auditory monitoring more effectively can be enhanced by improving the qualities of

hospital sound environments. Even though the significance of visual monitoring in ICUs

for patient outcomes including patient falls has been well documented, there is limited

information about auditory monitoring. This study aims to provide (1) a comprehensive

overview of previous research related to auditory monitoring; and (2) to describe a case

study examining auditory monitoring in two ICUs.

3.2.2 Methodology

A case study is conducted in the two ICUs described in Chapter 2 in order to assess the

perceived auditory monitoring abilities of nurses. In addition to the components already

discussed, the online survey contains questions about nurse tasks highly important for

patient monitoring, ICU sounds necessary for patient safety, and key listening locations

in the unit. Please refer to section 2.2.2 for additional details on the subjects.

The units are similar in staffing and patient acuity as discussed in section 2.2.2: similar

number of private patient rooms and staffing models (intensivists and nurse

practitioners), acuity levels of patients, nurse working hours. On the other hand, the

designs of these two units are different with the following specific differences that might

83
impact auditory monitoring. As shown in Figure 3.1, different from the design principles

of Neuro-ICU, MedSurg-ICU layout is designed to be more compact to reduce walking

distances that results in smaller volumes and shorter corridors. For example, total length

of the staff corridors in MedSurg-ICU is 73 m (240 ft) while it is 183 m (600 ft) in

Neuro-ICU. Moreover, spaces in the Neuro-ICU are installed with high performance

acoustic ceiling tiles (with higher sound absorption qualities), while it is only regular

ceiling tiles (with less sound absorption qualities) from 1980`s in the MedSurg-ICU.

Figure 3.1 Floor-plans of two ICUs showing circulation patterns and impulse response
measurement locations

As discussed earlier, reverberation can potentially impact auditory monitoring. Therefore,

impulse response measurements are conducted to objectively assess the reverberant

qualities of the corridors in the two wards. The impulse response measurements have

taken place at 6 different receiver locations in two hallways and in an unoccupied patient

84
room in each ward as shown in Figure 3.1. The measurement system consists of a GSR

omni-directional dodechahedral loudspeaker with Outline amplifier, ProSonus EASERA

Gateway data acquisition system (DAQ), laptop equipped with EASERA software v.1.1,

and Larson Davis 824 sound level meter microphones. A maximum length sequence

(MLS) excitation signal is used. The sound source is located slightly off the center of the

rooms in each unit. The receivers are located about 7.6m-10.6m (25ft-35ft) away from

the source in each hallway. In the unoccupied patient rooms of the two wards, the source

is also located about 1.3m (4.5ft) away from the receivers. In both the corridors and

patient rooms of the two wards, receivers are placed at distances outside the critical

distance to avoid significant impact of direct sound on the reverberation time. In the

patient rooms of the two wards, the critical distances range between 0.76m- 0.9m (2.5ft-

3ft). In the corridors the critical distances are 4.5m (15ft) in the Neuro-ICU and 1.8m

(6ft) in the MedSurg-ICU.

3.2.3 Findings

Perceived importance of nurse tasks

The Neuro-ICU and MedSurg-ICU nurses have been asked to rate the perceived

importance of different methods of patient monitoring according to their relevance in

critical care nursing. The distribution of the nurse responses with the distribution normal

curve are shown in Figure 3.2. More than 85% of the Neuro-ICU and MedSurg-ICU

nurses “completely agreed” that visual and auditory patient monitoring are important

tasks in critical care nursing, while the majority of the rest “somewhat” agreed that these

two tasks were important. Mean levels for perceived importance of visual and auditory

85
monitoring are 4.7 and 4.8 out of 5, respectively. Based on significance test results, there

are no statistically significant differences between the perceptions of nurses in the two

units (p>0.05).

“Visual patient monitoring is a crucial task in critical care nursing.” “Overall auditory monitoring (listen, recognize and respond to auditory cues)
is a crucial task in critical care nursing.”

Figure 3.2 Perceived importance of visual vs. auditory monitoring averaged across the
two units

More detailed analyses are shown in Figure 3.3 which represents the perceived

importance of auditory monitoring components in relation to visual monitoring. The

majority of the nurses from two units believe that individual sound tasks conducted

during auditory monitoring (i.e., ability to hear, differentiate and localize auditory cues)

are also highly important nurse tasks. Based on analysis of variance test results, there are

no significant differences between the perceived importance of visual monitoring and

three auditory monitoring tasks (p>0.05).

86
Figure 3.3 Perceived importance of three auditory monitoring tasks vs. visual monitoring
averaged across the two units

Perceived importance of auditory cues

Nurses from both units have been also asked to rate the importance of different medical

alarms for patient safety. Results are shown in Figure 3.4. Almost all nurses believe that

ventilator alarms are “very” critical for patient safety. More than 70% of the nurses

reported hissing sounds of the respiratory ventilators are also very critical for patient

safety. These two auditory cues are monitored to ensure patient’s adequate and proper

breathing. The hissing sound of the ventilator equipment is an indicator of proper

functioning of pressure relief valves. The alarms generated by the medical ventilator

might be associated with different events such as secretion in endotracheal tube (ETT), a

kink in the vent tubing, patient biting on ETT, increased airway pressure, a disconnect in

the vent tubing, displaced ETT, patient anxiety or pain, hypoxia, hypercapnia, and lack of

full delivery of breath.

87
As shown in Figure 3.4, the perceived importance of other non-ventilator alarms (patient

monitor, IV-pump, nurse call and feeding-pump alarms) varied. About 85% of the nurses

believed that patient monitor alarms are “very” critical for patient safety. Patient monitor

alarms inform any abnormal changes in the physiologic parameters such as heart rate,

temperature, ECG, SpO2, blood pressure, and CO2. A specific type of patient monitor

alarm, known as the “code” indicates highly critical occurrences that might have life-and-

death implications such as heart failure, cardiac arrest. The percentage of nurses who

believe that the IV-pump alarms and nurse call alarms are “very” critical for patient

safety was 63% and 55%, respectively. The majority of the rest believe that these tasks

are “somewhat” critical. Not many nurses (20%) have rated the feeding pump alarms as

“very” critical for patient safety, but about 60% of the nurses have reported that they are

“somewhat” critical for patient safety.

Not critical at all Very critical

Figure 3.4 Perceived importance of different medical alarms for patient safety averaged
across the two units

88
In addition, nurses have been also asked to report any other auditory cues including alarm

and non-alarm sounds necessary for patient safety. Some of those reported alarms include

bed rail alarms, bed exit alarm, CRRT alarms (from the dialysis machine), PCA-pump

alarms (from the medication-dispensing unit), V-tach, V-fib, and physiological

parameters (from patient monitors). The latter three as well as the code alarms might lead

to rapid heart failure. Examples of non-alarm auditory cues mentioned include staff help

calls which might indicate the need for additional resources and assistance or a

significant problem with the patient, patient help calls which generally occur when a

patient is not capable of reaching the call light, unusual or distress noise from patient

rooms (e.g. patient getting out of the bed, patient fall), family help calls, and patient

bodily sounds.

Key listening locations in the MedSurg-ICU and Neuro-ICU

Nurses have been also asked how often they use auditory monitoring in various locations

in their ICUs. Specifically, they have been asked about corridors, patient rooms, central

nurse stations, medication prep zones, and supply rooms. Note that both ICUs contained

these various spaces. Additionally, nurses in the Neuro-ICU have been asked about

distributed nurse stations, as this is a design feature unique to this unit and distributed

nurse stations are not contained in the MedSurg-ICU. As Figure 3.5 shows, in both the

Neuro-ICU and MedSurg-ICU, the locations where the most auditory monitoring

occurred are the corridors, patient rooms, and central nurse stations. The Neuro-ICU

nurses additionally have reported that they also frequently monitor auditory cues at

medication preparation zones and at the distributed nurse stations. On the other hand,

89
49% of the MedSurg-ICU nurses have reported that they rarely monitor auditory cues at

the medication preparation zones. In both wards, they rarely monitor auditory cues at the

supply rooms. In the MedSurg-ICU, the medication preparation and supply room spaces

are separated by walls that extend from floor to ceiling. In the Neuro-ICU, the medication

preparation areas are the extensions of corridors.

Figure 3.5 Frequency of auditory monitoring at different locations in the two ICUs

90
Auditory monitoring performance of nurses in MedSurg-ICU and Neuro-ICU

Nurses have been also questioned about their ability to recognize, hear, and localize

auditory cues. As shown in Figure 3.6, the results vary. As shown in Figure 3.6a, the

majority of the MedSurg-ICU and Neuro-ICU nurses “somewhat” or “completely” agree

that they can hear the critical sounds in the unit. This percentage is somewhat higher in

MedSurg-ICU but the difference is not significant.

Figure 3.6 Perceived sound task performance of nurses in the two ICU`s: (a) ability to
hear critical sounds (b) ability to differentiate critical sounds (c) ability to localize critical
sounds

91
 Figure 3.6 continued

Also as shown in Figure 3.6b, the majority of nurses “completely” agree that they can

differentiate the critical sounds in the unit. This percentage is slightly higher in the

Neuro-ICU. As shown in Figure 3.6c, the majority of nurses “somewhat” or “completely”

agree that they can guess the location of the critical sounds in the unit. In the MedSurg-

ICU, sound localization performance of nurses is slightly higher. However the

differences between the perceptions of the nurses in the two units are not significant.

92
To summarize, hearing and sound localization are perceived to be somewhat higher in the

MedSurg-ICU, but differentiation is perceived to be somewhat higher in the Neuro-ICU.

The differences in perception might be related to unit architecture. As summarized in

previous sections, sound localization performance is mainly related with background

noise levels and reverberation time. As discussed earlier, the overall noise levels are the

same in each unit, although the MedSurg-ICU has been found to be more “peaky.”

However, the MedSurg-ICU is a more compact unit and the reverberation time was

measured to be lower. It is possible that the more compact setting and lower

reverberation time of the MedSurg-ICU led to the improved ability of nurses hear and

localize sounds, despite unit being more “peaky”. The reason for the difference in

differentiation perception between the two wards is less clear; one hypothesis is that the

more “peaky” sounds in the MedSurg-ICU cause confusion when differentiating one

alarm (which is a peak type of sound) from another.

Differences between RT30 in MedSurg-ICU and Neuro-ICU

To compare the length of the reverberation time (RT30) in the two wards, impulse

response measurements are conducted. Averaged RT30 in the corridors of the MedSurg-

ICU ranges between 0.44 and 0.56s across different frequencies (Table 3.1). In one of the

corridors, the RT30 is as low as 0.3s. In the patient room of the MedSurg-ICU, averaged

RT30 (250 Hz- 4 kHz) is 0.5s.

93
Table 3.1 Distribution of averaged RT30 across frequency measured in the corridors and
in the unoccupied patient rooms in the two ICUs

250Hz 500Hz 1kHz 2kHz 4kHz

corridor 0.73 0.75 0.87 0.82 0.7
Neuro‐ICU
patient room 1 0.77 0.71 0.77 0.76
corridor 0.56 0.5 0.52 0.54 0.44
MedSurg‐ICU
patient room 0.83 0.34 0.81 0.28 0.25

In the Neuro-ICU corridors and unoccupied patient rooms reverberation times are higher

compared to RT30 measured in the MedSurg-ICU. Averaged RT30 in the corridors of

Neuro-ICU ranges between 0.70 and 0.87s across different frequencies. In one of the

corridors the RT30 is as high as 0.92s and the lowest RT30 measured in different

corridors is 0.65s. In the patient room of the Neuro-ICU, averaged RT30 (250 Hz-4 kHz)

is 0.8s. Given these values, the difference between reverberation times in the Neuro-ICU

and MedSurg-ICU are perceptible based on ISO/DIS 3382-1 standards (Bork, 2000).

3.2.4 Discussion

Factors that impact the effectiveness of auditory monitoring performance of nurses

particularly in ICUs have not been investigated as much as visual patient monitoring.

This chapter provides a comprehensive overview of the factors related to auditory

monitoring, including previous research related to the topic. Additionally, results from a

case study examining nurses` perception of auditory monitoring in two ICUs are

presented.

94
In the case study, ICU nurses strongly agree that auditory monitoring is highly important

for patient safety in addition to visual monitoring. In addition, nurses ability to hear,

differentiate, and localize auditory cues are also perceived to be critically important for

patient safety. However, previous research shows that these sound tasks are hard to

accomplish in complex sound environments. Hospital sound environments are extremely

complex as they are multi-source, noisy, and most times reverberant in nature. Previous

research indicates that particularly in highly reverberant spaces, sound localization can be

very challenging. High background noise levels can also potentially negatively affect the

detection of target sounds. Furthermore, the human auditory system has limited

monitoring capabilities. Pushing the limits of the auditory system to monitor high number

of sound events can require excessive efforts, in particular for novice nurses.

In the case study, interesting differences are found when comparing the importance of

different ICU sounds for patient safety. Some of the alarm sounds are perceived to be

more important than others. Specifically, the majority of nurses believe that medical

ventilator alarms and patient monitor alarms are very important for keeping patients safe.

A considerable percentage of nurses also believed that nurse call, feeding pump, and IV-

pump alarms were critically important for patient safety.

Additionally, it is found in the case study that nurses listen to the auditory cues from

different key locations in the critical care units. Some of those key locations include:

patient rooms, nurse stations, and corridors. It needs to be noted that all these spaces are

all connected through corridors. In the Neuro-ICU where the medication room is also

95
connected to the corridors, it was also reported as a listening location. On the other hand,

since in both units` supply rooms are separated from the corridors with walls from floor

to ceiling, they are not reported as listening locations. Considering these results, the

connectivity of spaces with corridors might be a significant indicator of key listening

locations.

By comparing nurses` ability to conduct sound tasks in the case study MedSurg-ICU and

Neuro-ICU, an interesting conclusion regarding effective environmental factors for

auditory monitoring emerges. Apparently, nurses` perceived hearing and localization

performance is higher in the MedSurg-ICU compared to the Neuro-ICU even though the

Neuro-ICU is installed with high performance absorptive acoustic ceiling tiles and the

MedSurg-ICU was found to be more “peaky”. Parallel to nurse perceptions, RT30 levels

(based on impulse response measurements) in the corridors and in the patient room are

perceptibly higher in the Neuro-ICU compared to MedSurg-ICU. This indicates the

potential significant impact of different design features on task supportive sound

environments such as spatial design features associated with RT30 levels in addition to

absorption qualities of materials. To systematically assess suggested relationships

between different design features and reverberation time, more controlled follow up

studies are also conducted. The study results are introduced in the following sections.

Overall, lack of proper monitoring of auditory cues can potentially delay the nurses`

rescue efforts by disabling the early detection of complications, and could result in

serious patient injury and even death. For effective auditory monitoring in hospitals, both

96
environmental acoustic (e.g., reverberation time, overall noise levels) and sound source

specific (e.g. acoustic qualities of sounds, number) solutions should be researched and

adopted.

97
 CHAPTER 4

ACOUSTICS OF ARCHITECTURAL SPACES: A LITERATURE REVIEW

Chapter 3 demonstrated that acoustic qualities of hospital sound environments can have

significant impact on vital sound tasks (patient auditory monitoring) conducted by

caregivers. Previous research indicates that design features of architectural settings such

as floor-plate shape qualities can impact acoustic qualities of the built environment. This

chapter provides an overview of the previous literature liking design and acoustics,

acoustics parameters used in room-acoustics research and sound behavior in proportional

and non-proportional spaces.

4.1 ROOM ACOUSTICS

The human auditory system is capable of differentiating the acoustic qualities of the

sound produced inside and that produced outside an enclosed space (Mehta et al., 1997).

Sound produced inside a room bounces back and forth from boundary surfaces. On the

other hand, sound generally travels freely away from the source outdoors, unless major

reflecting surfaces are present. Characteristics of an enclosed space such as volume,

spatial proportions, floor-plate shape, and material qualities of boundary surfaces

significantly affect the sound qualities and sound levels. This field of acoustic research

exploring the sound behavior in enclosed spaces with different design implications is

known as “room acoustics”.

98
4.2 ROOM ACOUSTICS PARAMETERS

Acoustic measures used in the analysis of rooms are called room-acoustics parameters.

For this chapter, room-acoustics parameters can be grouped in two categories:

reverberation time (RT) and other room-acoustic metrics including sound strength (G),

clarity (C), definition (D), early decay time (EDT), lateral fraction (LF/LFC) and centre

time (Ts). Reverberation time has been a well-established room-acoustics parameter

widely used in architectural acoustics for the assessment of various types of architectural

spaces including music halls, hospitals, worship places and more recently in long

enclosures. With the developments in perceptual acoustic research (known generally as

“psychoacoustics”), additional room-acoustic parameters were proposed to assess

different subjective attributes of sound quality, particularly in concert halls. The

motivation was to identify and quantify different aspects of the overall acoustic quality of

the music halls that are important for listeners' perception. Detailed information about the

room acoustics parameters relevant to this chapter can be found in the “Definition of

Term” portion of the appendix (Appendix A).

4.3 ACOUSTICS OF PROPORTIONAL SPACES

4.3.1 Sound behavior in rooms

This section reviews sound propagation theories and empirical data presented in the

previous research.

The classic sound propagation theory (a.k.a diffuse field theory) was developed to predict

the behavior of sound in concert halls. According to this theory, direct sound becomes

99
negligible approximately after 10m (19.6ft) (a.k.a. “reverberation radius” or “critical

distance”) (Galindo et al., 2005). At the critical distance, sound energy of the direct and

reverberant sound fields are equal. After this point acoustic qualities are mainly defined

by volume and RT. Additionally, empirical data shows that reverberation time in rooms

does not vary much by distance or the position of the receiver (Abdou, 2003; Lundeby &

Vigran, 1999; Pelorson, Vian, & Polack, 1992).

4.3.2 Floor-plate design and acoustics

Floor-plate design qualities of architectural settings can have a significant impact on

acoustic qualities of sound environments. Previous studies linking floor-plate design and

acoustics took place mainly in the concert halls as discussed in the following section.

Since the 18th century, concert hall floor-plates took a variety of different shapes. Typical

concert hall floor-plate shapes include shoebox (rectangular), fan-shape, vineyard and

horseshoe (Figure 4.1) (Barron, 1993).

100
 Shoe-box Fan

Vineyard Horseshoe

Figure 4.1 Commonly applied concert hall floor-plate shapes (from Barron, 1993)

There are also concert halls with elliptical floor-plate shapes but not many. One of the

reasons is that elliptical floor-plates focuses sound at the centers of the geometry. This

focusing effect leads to non-uniform distribution of sound energy and generates an

unwanted focal effect (Cox & D'Antonio, 2009). A rare successful example of this type is

Royal Albert Hall which required many efforts from acousticians (Figure 4.2) (Zhang,

2005).

101
 Figure 4.2 Royal Albert Hall with elliptical floor-plate shape (from Zhang, 2005)

Concert hall floor-plate shapes can also influence qualities of lateral sound (i.e.,

reflections coming from the side walls). Early lateral reflections contribute to the feeling

of “spaciousness” in concert halls. “Spaciousness” is a desired sound quality specifically

in concert halls. Findings of recent studies indicate that parallel and narrow walls of shoe-

box concert halls with rectangular floor-plate shapes improve the early lateral reflections

(Mehta et al., 1999). This is a primary factor explaining the successful acoustic

performance of many shoe-box shaped concert halls. Beranek and Hann surveyed

acoustic qualities of various concert halls (Hann, & Fricke, 1995). According to their

findings, two thirds of the concert halls rated as “excellent” were in the shape of shoebox.

A well known example is Boston Symphpny Hall (Figure 4.3).

102
Figure 4.3 Boston Symphony Hall with rectangular floor-plate shape (from Hann, &
Fricke, 1995)

Developments in acoustic technology and scientific knowledge enabled the application of

more complex floor-plate shapes such as vineyard since the beginning of the 20th century.

In vineyard floor-plates, a concert hall is subdivided into smaller audience zones

surrounded by side walls. This type of design improves sound quality by providing lateral

reflections. The Berlin Philharmonic is one of the most successful modern concert halls

with vineyard floor-plate shape (Figure 4.4, Mehta et al., 1999).

103
Figure 4.4 Berlin Philharmonie Concert Hall with vineyard floor-plate shape (from Mehta
et al., 1999)

4.3.3 Statistical analysis of floor-pate design and acoustics

The previous section introduced typical floor-plate shapes applied in concert halls.

Several studies statistically and systematically analyzed the impact of floor-plate design

qualities on acoustic outcomes in the concert halls. This section reviews the findings of

these scientific studies.

Early studies indicated the significant impact of hall width in predicting sound qualities

in concert halls (Gade, 1990; Klosak & Gade, 2008; Schroeder, Gottlob, & Siebrasse,

1974). Gade (1990) conducted impulse response measurements in 32 European halls and

statistically analyzed the relationship between room-averaged acoustic data and design

variables. By comparing the expected (based on classic sound propagation theory) and

104
measured results, the study concluded that volume and reverberation time are the two

main factors affecting new room-acoustic parameters such as reverberance and clarity.

This finding supports how well the classic theory predicts acoustic conditions (e.g., EDT,

C, L) of the concert hall as a function of RT and volume. The study also indicated the

significant impact of floor-plate shape on clarity. Based on regression analysis results, the

study found that clarity is positively correlated with hall width and angle between side

walls. In other words, clarity is expected to be higher in wide/ fan-shaped walls compared

to narrow/rectangular halls. The interpretation is that wider rooms tend to minimize the

distance between source and receiver and increase the seating capacity of the halls and

splayed side walls directs the early sound energy to the rear of the room (Mehta et al.,

1999). Gade (1990) also found a fairly high correlation between hall width and LEF

(r=0.73). The regression equation suggested by the study is included below.

Schroeder, Gottlob, and Siebrasse (1974) correlated subjective perception with objective

design and acoustic parameters (i.e., volume, width, time delay, reverberation time,

definition and inter-aural coherence) of 11 European concert halls by conducting

statistical analysis. For the subjective evaluations, a pre-recorded signal was played in

each hall and re-recorded at the receiver locations with the use of an artificial head.

Acoustic parameters were obtained from measured impulse response at the artificial

head`s ears. Subjective evaluation of the recorded signals took place in an anechoic

chamber. According to statistical analysis results, reverberation time (positively), inter-

105
aural coherence (negatively), volume (negatively) and width of the halls (negatively)

were significantly correlated with subjective “consensus preference factor”. Additionally

width was negatively correlated with reverberation time. These findings indicate the

acoustical disadvantages of larger halls.

A recent study used acoustic simulations and generated various theoretical design models

to analyze the association between floor-plate shape and acoustic characteristics of

shoebox-shaped concert halls (Klosak, & Gade, 2008). In total 24 theoretical models

were generated, as shown in Figure 4.5. Only two design variables were changed among

those theoretical models: a) volume and b) length-to-width ratio. The floor-plate shapes

of the theoretical models ranged from square to elongated rectangle as shown in Figure

4.5.

Figure 4.5 Floor-plate shapes of theoretical design models (from Klosak, & Gade, 2008)

106
 Figure 4.5 continued

Acoustic simulation results calculated for 24 theoretical models indicated that uniformity

of the distribution for sound quality metrics such as clarity (C80) and strength (G)

decreases as the shape gets more elongated. This was particularly true in the rooms with

smallest volume (8000m3). Similar results were observed for increasing volume. The

study also highlighted the impact of both width and length in predicting lateral fraction

(LF). Among theoretical models with similar L/W ratio, as the width increased LF values

(i.e., the amount of the useful reflections arriving from the side walls) decreased. The

study suggested the following regression equation showing the relationship between

length-to-width and early lateral energy fraction (LF).

To summarize, various studies analyzed the association between floor-plate design

qualities and acoustic outcomes and found statistically significant relationships. Even

though these studies were particularly conducted in concert halls and correlated the

107
design features and acoustic parameters more specific to concert halls, their findings in

general underline the association between floor-plate design and acoustics.

4.4 ACOUSTICS OF NON-PROPORTIONAL SPACES

4.4.1 Sound behavior of long spaces

Long enclosures are considered as non-proportional spaces related with their extreme

spatial proportions. Examples of long enclosures are corridors and underground train

stations. Acoustics of these non-proportional spaces differ from the acoustics of more

traditional rooms. Principles of classic sound propagation theory and researched

relationships between room design features and acoustics do not necessarily apply to non-

proportional spaces. The following section defines the differences between sound

environments of proportional and non-proportional spaces and introduces the sound

behavior characteristics in long enclosures.

Classic theory assumes the existence of diffuse field and uniform distribution of the

sound rays in all directions (Kang, 2002b). In proportional spaces where a sufficiently

diffuse sound field exist (e.g., concert halls), beyond the reverberation radius, the

contribution of direct sound becomes negligible; therefore, SPL is considered to be

approximately constant. Similarly, reverberation time based on Eyring and Sabine

formulas is considered to be the same at any point in the room beyond the reverberation

radius. On the other hand, sound field in long enclosures is not uniformly diffuse related

with the non-proportional/extreme dimensions. It is one of the key reasons why sound

108
behavior in long enclosures is quite different compared to sound behavior in the

sufficiently diffuse sound fields found in more traditionally-shaped rooms.

Various studies have been conducted to analyze the behavior of sound in long enclosures

with rectangular floor-plate shapes. The main goal of these studies was to explain how

different qualities of sound vary by distance and long enclosure design features (e.g.,

cross section size). The majority of the studies systematically documented acoustic

qualities of long enclosure sound environments (e.g., sound attenuation, early decay time

and reverberation time) by conducting impulse repose measurements at changing

distances from the sound source. In the following sections, the findings of these studies

are introduced.

4.4.1.1 Field measurements

Kang (2002c) conducted field measurements in different long enclosures including a

corridor and multiple underground stations with rectangular floor-plate shapes. These

field studies primarily explored the changing sound behavior with distance. The length,

width and height of the corridor were 42.5m (139ft), 1.56m (5.2ft) and 2.83m (9.3ft)

respectively. The boundaries of the corridor were geometrically reflective and average

absorption was low (approximately 0.1 at 500Hz and 1 kHz). The results of the field

measurements conducted at different distances from the sound source are shown in

Figure 4.6 below. Overall, reverberation time results measured at different receiver

locations increases with increasing source-receiver distance.

109
Figure 4.6 Spatial distribution of reverberation times in a single corridor (from Kang,
2002c)

The underground stations studied by Kang (2002c) were deep tube stations with circular

cross-sectional shape (named Old Street, Warren Street and St John`s Wood). The results

of the field measurements conducted at different distances from the sound source are

shown in Figure 4.7 below. As the source and receiver distance increases, the

reverberation time and early decay time increase along the length until about 40m (131ft)

and then become approximately stable or decrease slightly. In general reverberation time

values were greater than early decay time values, indicating that sound decay curves of

these long enclosures are not linear.

110
Figure 4.7 Spatial distributions of RT30 and EDT levels in tunnels (from Kang ,2002c)

4.4.1.2 Simulation studies

The number of studies exploring the association between design and acoustics of long

enclosures remains limited. Kang conducted a series of detailed computer simulation

studies to systematically assess the impact of different design factors on acoustics of long

enclosures (Kang, 2002c; Kang, 2002d). In these theoretical studies, mainly the design

and acoustics of long enclosures with rectangular floor-plate shapes (e.g., single

corridors) were analyzed. Related with the focus of this thesis, the following section is

limited to studies exploring the impact of long enclosure designs on reverberant qualities.

More detailed information about the relationship between long enclosure design and

sound attenuation can be found in Kang (2002).

111
Design factors affecting reverberant qualities of long enclosure sound environments

Different design factors can affect length of reverberation time in long enclosures such as

corridor length, absorptive qualities of the surface materials, sound source type and cross

section dimensions. Kang conducted various computer simulation studies to

systematically assess the impact of design features on length of the reverberation time in

long enclosures (2002c).

A set of theoretical models were generated to assess how reverberation time varies along

the corridor length in long enclosures with different absorption coefficients (Figure 4.8).

Long enclosures with 6mx4m (20ftx13ft) cross section were assigned two absorption

coefficients: 0.1 and 0.2. The boundaries of the long enclosures were geometrically

reflective. For the less absorptive case ( = 0.1), reverberation time values increased

rapidly until a maximum and then decreased very slightly in the long enclosures. For the

more absorptive case ( = 0.2), reverberation time values increased less rapidly until a

maximum and then decreased with a steeper slope as compared to the ( = 0.1) case.

112
Figure 4.8 RT30 and EDT due to absorptive qualities of boundaries in long enclosures
(from Kang, 2002c)

A second set of theoretical models were generated to assess the impact of diffusely and

geometrically reflective boundaries on reverberation time values. Figure 4.9 shows

interesting differences between behaviors of sound in these two long enclosures. In the

geometrically reflective long enclosure, reverberation time increased slightly and then

decreased slightly. Interestingly, in the diffusely reflective long enclosure, reverberation

time consistently increased. Moreover, reverberation time values were generally 30-60%

longer in the diffusely reflective boundaries case.

113
Figure 4.9 RT30 and EDT due to reflective qualities of boundaries (from Kang, 2002c)

A third set of theoretical models was generated to assess the impact of corridor length on

reverberation time values (Figure 4.10). One of theoretical models was 60m (196ft) long

and the other was 120m (394ft) long. The boundaries of both long enclosures were

diffusely reflective and the end walls were open. Reverberation time values along the

corridor length in the shorter case were about 20-30% less compared to the longer case.

114
Figure 4.10 Floor-plans of two long enclosures with different length (from Kang, 2002c)

A fourth set of theoretical models was generated to assess the shape and size qualities of

long enclosure cross sections on length of reverberation time in long enclosures with

diffusely reflective boundaries. Theoretical models with 5 different cross sections (and

thus different aspect ratios) were generated including 12mx8m (39ftx26ft), 16mx6m

(52ftx20ft), 24mx4m (79ftx13ft), 6mx4m (20ftx13ft) and 8mx8m (26ftx26ft).The length

of all models were 120m (393ft) and all surfaces were assigned an absorption coefficient

of 0.2. As shown in Figure 4.11, for a given cross-sectional area, the reverberation time

values could vary significantly with the aspect ratio. Reverberation time values become

greater as the cross-section tends towards square. In addition, in the long enclosures with

larger cross section, reverberation time values were longer. Similar results were observed

for long enclosures with geometrically reflective boundaries.

115
Figure 4.11 RT30 and EDT due to cross-sectional size (from Kang, 2002c)

A fifth set of theoretical models were generated to assess the distribution of boundary

absorption in long enclosures with diffusely reflective boundaries. Theoretical models

had constant amounts of absorption with five different distributions ranged from evenly

distributed absorption in cross section (D1) and one boundary strongly absorbent and the

rest is reflective (D5). From D1 to D5 the reverberation time values decreased

continuously and the variation was about 30%. Reverberation time was the longest with

absorption that is evenly distributed in cross section and the shortest when one boundary

was highly absorbent.

116
Kang also systematically analyzed the impact of street design on length of the

reverberation time and other acoustic outcomes such as sound attenuation by conducting

computer simulations (2002d). Similar to other long enclosure types, in the urban streets

with rectangular floor-plate shape, reverberation time increased systematically with

increasing distance between source and receiver. Similarly, reverberation time values also

increased significantly with increasing the building height.

4.5 ACOUSTICS OF LONG SPACES WITH BRANCHES

The previous section reviewed the findings of theoretical acoustic studies conducted in

long enclosures with rectangular floor–plate shapes. There are very few studies that have

documented the behavior of sound in more complex long enclosures, such as

interconnected long enclosures with complex floor-plate shapes (e.g., long enclosures

with branches). The following section reviews the findings of the studies conducted in

interconnected long enclosures with complex floor-plate shapes.

Abel et al. (2008) analyzed the association between design and acoustics of historical

underground galleries used for ritual purposes by conducting field measurements.

Underground labyrinthine galleries were arranged in a series of small rectangular alcoves

off narrow corridors (Figure 4.12, Figure 4.13). Some of the other features of the galleries

were rectilinear shapes, narrow width, short height and stone walls.

117
 Figure 4.12 Site-plan of historical underground galleries (from Abel et al., 2008)

Figure 4.13 Floor-plan of historical interconnected underground galleries (from Abel et

al., 2008)

118
Impulse response measurements were conducted at multiple different locations. As

shown in Figure 4.13, location K is a representative receiver location located at one of the

alcoves and D is a representative source location located at another alcove. Overall, in

the underground galleries reverberation time was short (generally less than 0.5s) to

enable the rhythmic sound of shell trumpets. Some other sound qualities include dense

and energetic early reflections and low-inter-aural cross correlation. The study also found

that number of turns (in the path between source and receiver) have significant impact on

reverberation time. Increasing the number of turns between source and receiver led to

higher reverberation time values. Reverberation time was shortest along straight lines

between source and receiver.

Kang (2002d) systematically compared the sound fields in street canyons with different

designs that incorporated a main street / side street configuration. A set of computer

simulations was generated to assess how source location affects reverberation time. As

shown in Figure 4.14, when the sound source moved to different locations in the main

street with two side streets, a slight increase (10%) in average reverberation time levels

was observed. Moreover, related with lack of direct sound, in the side streets

reverberation time was systematically longer compared to the values in the main street.

119
Figure 4.14 Spatial distribution of reverberation time (in second) along a “+” street
junction (from Kang, 2002d)

Liu and Lu (2009a) compared the sound fields of straight long enclosures without any

branch and the sound fields of the long enclosures with one branch (vertical, left inclined

and right inclined) via 1:10 scale physical models (Figure 4.15). The dimensions of the

full scale long enclosure were 94m-length (308ft), 7m-width (23ft) and 5m –height (16ft).

However the characteristics of the branches such as length, width and angle were not the

same.

120
Figure 4.15 Floor-plans of theoretical long enclosures with a single branch (from Liu, and
Lu, 2009a)

Reverberation time varied at different frequencies (Figure 4.16). For example at the low

frequencies, RT30 was longest for the straight long enclosure, second longest for the long

enclosure with a widened left inclined branch, third longest for the long enclosure with a

left inclined branch and the shortest for the long enclosure with an extended left inclined

branch.

121
Figure 4.16 Distribution of RT30 in long enclosures with a single branch
(top): 500Hz; (bottom):1000Hz (from Liu, and Lu, 2009a)

Another study by Liu and Lu (2009b) analyzed the sound fields of long enclosures with

multiple vertical or inclines branches via 1:10 scale physical models. The dimensions of

the full scale long enclosure were 94m –length (308ft), 7m –width (26ft) and 5m-height

(16ft). Sound source and receivers were arranged along the central line of the main

enclosure (Figure 4.17).

122
Figure 4.17 Floor-plans of theoretical long enclosures with multiple branches (from Liu,
and Lu, 2009b)

RT30 values of long enclosures with multiple branches were frequency dependent

(Figure 4.18) and frequency has little impact on RT30 levels in the enclosures with one

branch. Long enclosures with multiple branches had lower RT30 levels compared to long

enclosures with one branch and without any branch. In the low frequency range,

orientation of the branches did not affect the results. However, in the high frequency

range, orientation of the braches had a significant impact on RT30 values. Overall, the

123
study findings indicated that sound field of a long enclosure with multiple branches is

more complex and inhomogeneous compared to long enclosures without branches or with

one branch.

Figure 4.18 Distribution of RT30 in long enclosures with multiple branches

(top): 500 Hz; (bottom):1 kHz (from Liu, and Lu, 2009b)

124
4.6 CONCLUSION

Reverberation time is a universal acoustic metric that has been used to assess the acoustic

characteristics of various spaces including long spaces and rooms. Newer room-acoustic

metrics were developed such as early decay time, clarity, sound strength to particularly

assess the acoustic qualities of music halls that relate to audience experience. Even

though some studies have documented the RT and EDT levels in long enclosures, there

are still discussions whether the newer acoustic metrics are applicable for the acoustic

analysis of long enclosures.

Related with the design qualities of spaces such as spatial proportions, the acoustic

characteristics of their sound environments can vary significantly. Therefore acoustic

theories applicable in one type of space (e.g., concert halls) might not be applicable in

other types of spaces (e.g., corridors). For example, the principle of classic room acoustic

theory that is generally valid in music halls is not necessarily applicable in long

enclosures. Various studies have been conducted in the music halls and statistically

analyzed the relationship between specific design characteristics of floor-plate shapes

such as width, length and acoustic outcomes. However the findings of these studies are

limited to more proportional shaped rooms and do not necessarily apply to long

enclosures.

The number of studies conducted in long enclosures exploring the association between

design and acoustics is growing. The acoustic conditions in long enclosures were studied

via field measurements or computer simulations. Computer simulations enable the

125
systematic analysis of the acoustic conditions in long enclosures. However most of those

studies mainly considered straight long enclosures such as urban street or underground

train stations. A few studies explored the reverberant qualities of long enclosures with

complex floor-plate geometries such as underground stations with branches or staggered

urban streets. However, very limited information exists on the acoustic conditions of

inter-connected corridors located in the buildings. Dimensional qualities of building

corridors differ from other long enclosures (e.g., underground stations, urban streets)

which could potentially affect acoustic outcomes differently.

Overall, findings of long enclosure studies agreed that as the shape of the long enclosures

gets more complicated, sound behavior in these settings also becomes highly

complicated. However, more systematic research is necessary to clarify the impact of

specific design features on acoustics of inter-connected corridors with complex floor-

plate shapes.

126
 CHAPTER 5

MEASUREMENTS AND SIMULATIONS OF THE ACOUSTICS OF

LONG ENCLOSURES

The majority of the previous research has assessed the acoustic qualities of hospital

sound environments by conducting field measurements. The use of acoustic simulation

tools in healthcare acoustics is still not very common. It is probably because healthcare

settings are composed of non-proportional spaces such as interconnected corridors as

well as proportional spaces such as patient rooms. Various studies have tested the

effectiveness of acoustic simulation tools in predicting acoustic qualities of proportional

spaces such as concert halls and long enclosures with simple geometries (Kang 2002; Li,

and Lu, 2004; Li, and Lu, 2005; Yang, and Sheild, 2004). However, the number of

studies assessing the effectiveness of acoustic simulation tools in predicting acoustic

qualities of non-proportional spaces particularly interconnected corridors still remains

limited. This study tested the effectiveness of an acoustic modeling program that uses

hybrid prediction method (CATT-Acoustics V8) in predicting acoustic qualities of

complex long enclosures.

Accurate simulation of sound behavior is highly complex. Two key factors necessary for

accuracy of acoustic predictions are: a) the overall prediction method adopted by the

acoustic modeling developers, and b) modeling settings selected by the user. Appendix I

contains more information on computerized acoustic modeling prediction methods,

modeling settings and the findings of studies assessing the validity of computerized

acoustic modeling tools.

127
5.1 EMPIRICAL STUDY 1

VALIDATION of CATT ACOUSTIC PREDICTIONS

Hospitals are composed of proportional and non-proportional spaces such as inter-

connected spaces. Even though the use of an acoustic simulation program with hybrid

prediction program has been validated for proportional spaces, its use for complex non-

proportional spaces has not been validated. To assess the effectiveness of CATT acoustic

simulation program in predicting the acoustic qualities of complex non-proportional

spaces (i.e., interconnected spaces), empirical study 1 conducts series of validation

studies in the actual inter-connected corridors. Previous study findings agree that ISM

(Li, and Lu, 2005) and ray tracing methods (Yang, and Sheild, 2004) perform well in

predicting the acoustics of long spaces. Hybrid methods incorporate the best features of

ISM and ray tracing methods. However, the use of hybrid methods for acoustic analysis

of long spaces has not been validated yet. In addition, particularly in complex settings,

diffraction is one of the important factors that contribute to sound quality as well as

reflection and absorption. However, not many acoustic simulation programs are capable

of handling diffraction. CATT offers an automatic edge diffusion function to emulate

diffraction from edge diffusion. The effectiveness of this feature and some other useful

CATT features has not been tested yet in this context. CATT also offers an alternative

prediction method other than corrected tail ray tracing method (RTC) to handle the late

reflections effectively in complex geometries. By comparing the predicted (CATT) and

measured (in situ impulse response) results, this study assessed the effectiveness of

CATT-Acoustics in predicting the acoustic qualities of inter-connected corridor spaces.

128
5.1.1 L-shaped corridors

Methodology (L shaped corridor)

The first set of impulse response measurements has taken place in an L-shaped corridor.

This corridor is located in an educational building from the 1960s on the Georgia Institute

of Technology campus. The total volume of the L-shaped corridor is 397 m3 (14,019ft3).

Its dimensional properties are as follows: height: 2.7m (8.8ft); width: 2.4m (7.8ft) and

length: 60m (197ft). In total, the L-shaped corridor includes 31 doors and 18 windows.

Figure 5.1 shows the location of the sound source and the receivers. Related with the

focus of this study, the measurements have taken place only in the non-visual sound

fields of the corridors. This study examines the association between design and acoustics.

However, in the visual sound field, direct sound significantly affects the acoustic

outcomes. To isolate the interfering effects of direct sound, this study only focuses on the

sound behavior in the non-visual sound field.

Figure 5.1 3D-CATT model and floor-plan of an L-shaped corridor

The same space is modeled with the use of CATT-Acoustics simulation program. Similar

materials are assigned to enclosure surfaces and receivers are located at identical

129
locations. Table5.1 shows the absorption qualities of materials assigned to the enclosure

surfaces. Two types of ceiling tiles are used. The one represented as “medium

absorption” in Table5.1 is gathered from Mehta et al. (1997). Idealized absorption

coefficients are also used for the ceiling tiles represented as “low absorption” in Table5.1.

The reason is that majority of the educational settings visited in this pilot study are more

than 40 years old, damaged in some places, and more than likely they are not performing

at the “medium absorption” level.

Table 5.1 Material absorption coefficients of the L-shaped corridor

SurfaceMaterial 250Hz 500Hz 1kHz 2kHz 4kHz SC

Door Wood 0.21 0.10 0.08 0.06 0.06 A.E
CeilingAcoustic ceiling tile 0.38 0.38 0.38 0.38 0.38 20%
(low absorption)
Acoustic ceiling tile- 0.33 0.54 0.69 0.69 0.56 20%
(medium absorption)
Floor Vinyl on concrete 0.03 0.03 0.03 0.03 0.02 10%
Window Glass 0.25 0.18 0.12 0.07 0.04 A.E
Wall Painted concrete 0.05 0.06 0.07 0.09 0.08 30%

Findings (L shaped corridor)

Room-averaged predicted and measured reverberation time results across frequency are

shown in Figure 5.2. When idealized low absorption coefficients are considered, the

difference between measured and predicted results is lower. The “just noticeable

difference”-(JND) for the room averaged reverberation time levels range between 0% and

20% as shown in Tables 5.2 and 5.3. According to ISO 3382 the differences as low as 5%

are perceptible, but other sources show JNDs as high as 39% are perceptible (Meng,

Zhao, & He, 2006). Appendix J contains additional information about JND for

130
reverberation time. Regardless, the study findings are parallel overall to the findings of

previous research. For example, the JND results are quite similar to those reported by

Kang (2002) where he later used the validated acoustic simulation program for several

theoretical studies conducted in long enclosures. Therefore, the findings of this section

suggest the potential effectiveness of hybrid method in predicting the room-averaged

reverberation time outcomes for L-shaped corridors. However, it needs to be noted that

this validation analysis only took into account non-visual sound field.

Figure 5.2 Room averaged measured vs. predicted reverberation time in an L-shaped
corridor

Table 5.2 Just noticeable difference values for room averaged reverberation time (non-
visual sound field)

250Hz 500Hz 1kHz 2kHz 4kHz

JND 0% 20% 14% 17% 12%
(low abs)

131
Figure 5.3 Measured vs. predicted reverberation times at the receiver location 164ft
(50m) away from the sound source

Table 5.3 Just noticeable difference values for reverberation times measured at 164ft
(50m) away from the source

250Hz 500Hz 1kHz 2kHz 4kHz

JND 10% 20% 18% 20% 7%

(low abs)

5.1.2 Race Track Corridor

Methodology (race track design corridor)

The second set of impulse response measurements has taken place in a complex inter-

connected corridor system with a race track layout design. This corridor is located in a

different educational building from the 1960s era on the Georgia Institute of Technology

campus. The total volume of the racetrack design corridor system is 895 m3 (31,606ft3).

Its dimensional properties are as follows: height: 2.5m (8.2ft); width: 1.8-2.5 m (5.9ft-

8.2ft) and total length: 173m (567ft). In total, this race track corridor includes 48 doors

and 26 windows as shown in Figure 5.4.

132
Figure 5.4 3D-CATT model and floor-plan of a race track corridor

The same space is modeled with the use of CATT-Acoustics simulation program. Similar

materials are assigned to enclosure surfaces and receivers are located at identical

locations. Table 5.4 shows the absorption coefficients of the materials assigned to

surfaces of the acoustic model.

133
Table 5.4 Material absorption coefficients of the race track design corridor

Surface Material 250Hz 500Hz 1kHz 2kHz 4kHz SC

Door Wood 0.21 0.1 0.08 0.06 0.06 A.E

Acoustic ceiling tile- 0.34 0.34 0.34 0.34 0.34 20%

Ceiling (low absorption)
Acoustic ceiling tile- 0.33 0.54 0.69 0.69 0.56 20%
(medium absorption)
Floor Vinyl on concrete 0.03 0.03 0.03 0.03 0.02 10%

Window Glass 0.25 0.18 0.12 0.07 0.04 A.E

Wall Tile 0.01 0.01 0.01 0.02 0.02 30%

Findings (Race track design corridor)

The room-averaged predicted and measured reverberation time values across frequencies

are shown in Figure 5.5. Similar to previous validation analysis, only sound behavior in

the non-visual sound field is considered. JND for the room averaged reverberation times

mostly range between 2% and 20% with a few exceptions (Table 5.5). Again based on

the previous study findings, the difference between predicted and measured results is

small enough to suggest the potential effectiveness of hybrid method in predicting

reverberation qualities of race track design corridors.

134
Figure 5.5 Room averaged measured vs. predicted reverberation times in a race track
design corridor

135
Table 5.5 Just noticeable difference values for room averaged reverberation time (non-
visual field)

JND 250Hz 500Hz 1kHz 2kHz 4kHz

(low abs)
L1-7 15% 7% 8% 12% 17%
L8-14 14% 11% 20% 22% 14%
L15-17 2% 25% 5% 10% 9%

Figure 5.6 Measured vs. predicted reverberation times at the selected receiver locations in
a race track design corridor

136
 Figure 5.6 continued

Table 5.6 Just noticeable difference values for reverberation times at the selected receiver
locations (non-visual field)

JND 250Hz 500Hz 1kHz 2kHz 4kHz

(low abs)
L5 22% 5% 3% 10% 16%
L8 5% 21% 22% 3% 20%
L12 16% 11% 6% 14% 24%
L16 7% 25% 11% 20% 3%

137
5.1.3 T-Shaped Corridor

Methodology (T-shaped corridor)

Third set of impulse response measurements took place in another complex inter-

connected corridor system with a T-shaped layout design. This corridor is located in a

different educational building from 1960s era on the Georgia Institute of Technology

campus. The total volume of the corridor system with T-shaped floor-plated shape is

900m3 (31,783ft3). Its dimensional properties are as following: height: 2.4m (7.8ft);

width: 2.3m (7.5ft) and total length: 111m (366ft). In total, the T-shaped corridor

includes 33 doors and 19 small and 3 big windows from as shown in Figure 5.7.

The same space is modeled with the use of CATT-Acoustics simulation program.

Similar materials are assigned to enclosure surfaces and receivers are located at identical

locations. Table 5.7 shows the absorption coefficients of the materials assigned to

surfaces of the acoustic model.

Figure 5.7 3D-CATT model and floor-plan of a T shaped corridor

138
Table 5.7 Material absorption coefficients of the T-shaped corridor

Surface Material 250Hz 500Hz 1kHz 2kHz 4kHz SC

Door Wood 0.21 0.1 0.08 0.06 0.06 A.E

Ceiling Acoustic ceiling tile 0.39 0.39 0.39 0.39 0.39 20%
(low absorption)
Acoustic ceiling tile- 0.33 0.54 0.69 0.69 0.56 20%
(medium absorption)
Floor Vinyl on concrete 0.03 0.03 0.03 0.03 0.02 10%

Window Glass 0.25 0.18 0.12 0.07 0.04 A.E

Wall Painted concrete 0.05 0.06 0.07 0.09 0.08 20%

Findings (T-shaped corridor)

Room averaged predicted and measured acoustic outcomes across frequencies are shown

in the graphs below (Figure 5.8). JND values for room averaged RT30 levels range

between 6% and 23% (Table 5.8). Figure 5.9 shows the measured and predicted RT30

levels at the selected receiver locations. JND values for RT30 levels at the selected

receiver locations ranged between 3% and 24% (Table 5.9). Similar to the previous

analysis, based on the previous study findings, the difference between predicted and

measured results is small enough to suggest the potential effectiveness of hybrid method

in predicting reverberation qualities of T-shaped corridors.

139
Figure 5.8 Room averaged measured vs. predicted reverberation times in a T-shaped
corridor

Table 5.8 Just noticeable difference values for room averaged predicted and measured
reverberation times (non-visual field).

JND 250Hz 500Hz 1kHz 2kHz 4kHz

(low abs)
L1-7 22% 6% 23% 22% 12%
L8-11 21% 20% 15% 15% 6%

140
Figure 5.9 Measured vs. predicted reverberation times at the selected receiver locations in
a T-shaped corridor

141
 Figure 5.9 continued

Table 5.9 Just noticeable difference values for reverberation times at the selected receiver
locations (non-visual field)

JND 250Hz 500Hz 1kHz 2kHz 4kHz

(low abs)
L2 15% 14% 20% 22% 23%
L4 12% 3% 22% 22% 14%
L6 24% 11% 21% 18% 13%
L10 20% 21% 19% 22% 22%

142
5.1.4 Conclusion

This part of the study has tested the effectiveness of CATT in estimating the acoustics of

interconnected corridors. Related with the goals of this study, the measurements only

have taken place in the non-visual sound field. Based on the jnd thresholds suggested by

ISO 3382 standards, the differences between predicted and measured results are

perceptible (i.e., 5% or greater). However, this is not a surprise when the findings of

previous validation studies are considered. Typically the differences between measured

and predicted RT30 results reported by other computer modeling studies are above the

jnd thresholds suggested by ISO 3382. As the findings from this empirical study are

parallel overall to the findings of previous research, this study suggests the potential

effectiveness of hybrid method in predicting the room-averaged reverberation time

outcomes for L-shaped, race track, and T-shaped corridors.

5.2 EMPIRICAL STUDY 2

SOUND BEHAVIOR IN SINGLE vs. INTERCONNECTED CORRIDORS

In addition to assessing the validity of an acoustic simulation tool in predicting acoustic

qualities of long enclosures via Empirical Study 1, the other goal of this chapter is to

understand the behavior of sound in long enclosures. Empirical Study 2 analyzes the

differences between the diffuse sound fields of single and interconnected corridors via

multiple impulse response measurements (in situ measurements) by conducting field

measurements.

143
5.2.1 Methodology

Impulse response measurements have taken place in two different settings: a single

corridor and a T-shaped corridor. Both settings are located in educational buildings on the

Georgia Tech campus and had similar surface materials such as vinyl flooring, ceiling

tiles, wood doors and concrete walls. Figure 5.10 shows the interior of the single corridor

and Figure 5.11 shows the interior of the T-shaped corridor.

Figure 5.10 Interior of the single corridor setting

Figure 5.11 Interior of the T- shaped corridor

144
5.2.2 Findings

Figure 5.12 below shows the sound behavior in three corridor segments located in the

two different corridor settings. The first corridor setting has a rectangular shape floor-

plate. The second setting is an inter-connected corridor with T-shape floor-plate. As

shown in the graphs included in Figure 5.12, reverberation times are not always the same

at different locations of the two long enclosures. In the rectangular corridor, reverberation

time values followed similar trends across frequency, where there is a trend for them to

linearly increase with increasing distance from the sound source. However the sound

behavior is more complex in the inter-connected hallways of the T-shape corridor. Thus,

the behavior of sound in the non-visual-sound field (a.k.a. absence of direct sound) of the

T-shaped corridor is more complex compared to the sound behavior in the visual-sound

field (a.k.a. presence of direct sound).

145
 Measured RT30 Levels along the Corridor Length

Measured RT30 Levels along the Corridor Length (Visual)

Non-visual sound field Measured RT30 Levels along the Corridor Length (Non‐Visual)

Visual sound field

Figure 5.12 Measured reverberation times for the receivers located in the visual and non-
visual fields of two different corridor settings with rectangle and T shaped-floor-plates

146
5.2.3 Conclusion

This study has conducted field measurements and acoustic simulations (1) to test the

effectiveness of CATT in estimating the acoustic of interconnected corridors and (2)

analyze the behavior of the sound in the single and inter-connected corridors.

In Empirical Study 1, usually the CATT- predicted results are within 5-22% accuracy,

with a few exceptions. Even though predicted results are likely perceptibly different from

the measured results, the agreement between them can still be considered acceptable

based on the findings of the previous studies.

Overall, the findings of the validation studies conducted in this part of the study suggest

the potential positive impact of CATT features such as hybrid prediction method,

automatic edge function and diffuse reflections on the accuracy of the predicted RT30

outcomes particularly for the non-visual sound field of the long enclosures. Different

from the visual sound field, in the non-visual sound field there is no direct visual

connection between source and receiver.

In Empirical Study 2, the spatial distribution of reverberation times in the single and

inter-connected corridors is similar to results reported by the previous research. In the

single corridor, the trend of the data set is approximately linear for all frequencies.

Similar to previous study findings, RT30 levels increases with increasing distance. This

indicates the significant impact of distance on reverberation times particularly in the

single corridors. Different from the single corridor case, the trend of the data sets

recorded in the T-shaped corridor is not linear. In particular, the shape of the data set

147
trend lines in the non-visual sound field of the T-shaped corridor is more complex

compared to that in the visual sound field. This can be explained by the dominant effects

of the reflections in the non-visual sound field. To further assess the impact of different

design factors on reverberation time values in addition to distance, this study also

conducts more controlled follow up studies The results are discussed in the following

Chapters.

148
 CHAPTER 6

DESIGN AND ACOUSTICS OF INTERCONNECTED HOSPITAL CORRIDORS

Field study findings conducted in the previous chapters indicate the potential significant

impact of design features on the acoustic qualities of healthcare settings and highlight the

necessity of conducting more controlled studies. Controlled studies can enable the

analysis of the impact of particular design features on the acoustic qualities of healthcare

settings while controlling for other interfering factors. Chapter 6 systematically explores

the association between design and acoustics of interconnected nursing unit corridors

with a focus on their reverberant qualities by conducting acoustic simulation analysis as

well as in-situ impulse response measurements.

6.1 BACKGROUND

6.1.1 Auditory monitoring in the hospital corridors

Auditory patient monitoring is one of the key nurse tasks which might have life-death

consequences. Preliminary study findings discussed in section 3.2.3 indicate that nurses

monitor auditory cues at different locations including patient rooms, corridors, central

and de-central nurse stations in the critical care units. In the patient rooms, assessment of

auditory cues is supported with visual cues. Moreover, the nurses are very close to the

sound source and receive direct sound which is one of the critical factors for effective

auditory monitoring. However when they are mobile walking in the corridors, they most

times do not receive direct sound. Therefore, conducting auditory monitoring in the

corridors can be highly challenging.

149
Figure 6.1 shows the distribution of critical sounds, interfering noise sources, active (i.e.,

physician and nurse) and passive listeners (i.e., patient and visitors) in two ICUs. In these

units one nurse is typically responsible for two patients in the unit. In these sound maps,

lines are used to connect active listeners and the critical sound events that they monitor.

Particularly in the ICUs and nursing units, caregivers are usually mobile in the corridors

either dealing with other tasks or exchanging information with other caregivers.

Moreover, in most hospitals, the local sound environments of nurse stations (central and

de-central) are most times directly connected to the sound environments of corridors.

Therefore, acoustic qualities of corridors also potentially affect the nurses` auditory

monitoring performances at the nurse stations.

150
Red: noise sources
Yellow: Passive listener (i.e. patient and visitors)
Blue: Active listeners (i.e. caregivers)
Green: Critical sounds (i.e. medical alarms)

Figure 6.1 Spatial distribution of sound and noise source in the Neuro-ICU and MedSurg-
ICU

6.1.2 Effects of reverberation time on auditory monitoring

Reverberant qualities of hospital sound environments can significantly affect perception

of auditory cues. As described before, reverberation time represents the rate of the sound

decay. Extended sound decay increases the length of reverberation time and masks the

intelligibility of sounds including speech and non-speech sounds. For example, in a

typical speech stream as the sound of a syllable decays, it tends to mask the sound of the

subsequent syllable. The masking effects of two rooms with different reverberant

qualities are shown in Figure 6.2. In the more reverberant room (RT30=2s) where the

decay is slower, more sound is masked therefore speech intelligibility is lower. A hospital

study also indicated the positive impact of reduced reverberation time values on speech

151
intelligibility levels and improved psychosocial work environment reported by nurses

(Hagerman et al., 2005).

Figure 6.2 A speech sound that is masked differently in two rooms with different
reverberation (from Mehta et al., 1997)

Reverberation time potentially impacts localization of auditory cues in hospitals. As

described in Section 3.1.3, the human auditory system uses different cues when localizing

sounds such as inter-aural differences in arrival time and intensity and spectral cues.

However, late reflections caused by the boundary surfaces of rooms can result in

dramatic physical changes to the sound waves such as sound energy and direction. “Late”

reflections arrive later than the direct sound and early reflections. When late reflections

dominate the direct sound and useful early reflections, and they negatively affect the

sound localization (Litovsky, Colburn, Yost, & Guzman, 1999; Blauert, 1997).

Therefore, the human auditory system mainly benefits from the first arriving energy

when localizing sounds because it contains accurate localization information (Wallach et

al., 1949). In cases where there is no direct sound such as in non-visual sound fields, the

human auditory system depends only on the early arriving energy. Therefore, particularly

152
in the non-visual sound field the impact of late reflections on sound localization can be

highly detrimental.

Overall, high reverberation times can negatively affect sound tasks conducted by

caregivers in the care settings. To assess task supportive design features of care settings,

this research conducted more controlled studies and examined the association between

different design features and reverberation time. The results of these studies are

introduced in the following sections.

6.1.3 Design characteristics of hospital corridors

6.1.3.1 Scope

The architectural program of healthcare settings is composed of various functional spaces

such as corridors, patient rooms, public and private visitor areas, staff work areas and

lounges and medical care support areas. As discussed in Section 2, this study focuses on

the design and acoustic characteristics of corridor settings. Caregivers spend a

considerable amount of their time in the corridors while navigating between spaces. They

also conduct critical sound tasks in the corridors including conducting critical medical

conversations (e.g., patient care management, medication dosage and medical procedure)

and localization and assessment of critical sounds. Poorly designed corridor settings can

potentially aggravate the difficulty that caregivers experience while conducting critical

sound tasks. In this part of the study, design characteristics of corridor settings located in

17 actual nursing units from hospitals located around the United States are analyzed.

153
6.1.3.2 Overall Corridor Design

Corridors in nursing units connect different types of spaces such as patient rooms, nurse

stations and care support areas. As shown in Figure 6.3, the total length of the corridors

connecting different spaces in the entire nursing unit range between 87m (285ft) and

467m (1,532ft). Additionally, a nursing unit can consist of different numbers of

connected hallways forming complex floor-plate shapes as shown in Figure 6.4. The

nursing unit in Paimio Hospital and Bellevue Hospital consist of two intersecting

hallways. In the rest of the nursing units, the number of hallways ranged between 3 and

17.

Figure 6.3 Total length of the nursing unit corridors in different hospitals, expressed as
number of hallways

154
Figure 6.4 Floor plans of nursing units located in different hospitals (Floor-plans of the
hospitals were collected by a group of PhD students including myself as a part of the
Hospital Typology Class led by Dr John Peponis, Dr Sonit Bafna and Dr Craig Zimring)-
All drawings are in the same scale

155
Figure 6.4 (continued)

156
Except for the nursing units of Paimio Hospital and Bellevue Hospital, all nursing units

shown in Figure 6.4 consist of multiple patient-care clusters. Related with care

management and infection control purposes, typically patient-care clusters and their

corridors are segregated from other patient-care clusters with operable doors. The result

is that each patient-care cluster operates as a unique sound environment. This study

conducted another design analysis to assess floor-plate design characteristics of nursing

unit patient-care cluster corridors segregated with doors. A total of 43 different types of

clustered corridors were reviewed (Figure 6.5).

Figure 6.5 Floor plans of clustered corridors

157
 Figure 6.5 (continued)

The number of segregated patient-care corridors with different floor-plate shapes varies

among different nursing units and ranges between 1 and 6 as shown in Figure 6.6.

158
Figure 6.6 Number of clustered corridors in the nursing units

6.1.3.3 Corridor Spatial Organization

Spatial organization of spaces on the nursing unit layouts has been classified in different

categories by James and Tatton-Brown (1986) based on the examination of more than 60

hospital designs (Figure 6.7). These layout design categories are nightingale, corridor or

continental, duplex or Nuffield, racetrack or double corridor, courtyard, cruciform or

cluster and radial type. The logic behind the formation of these layout configurations is

outside the scope of this study. However it needs to be noted that each of these layout

types differently contribute to visual patient monitoring based on different organizations

of nurse stations around clusters of patient rooms. Detailed information about these

layouts can be obtained from the following sources: James and Tatton-Brown (1986);

Rashid (2006).

159
 Figure 6.7 Nursing unit layout types (from James, & Tatton-Brown, 1986)

Based on their influence on sound propagation, the floor-plate shapes of the 44 different

clustered corridors shown in Figure 6.5 can be grouped in two categories: race track

design and tree-like design. In a race track design, corridors surround the centralized care

160
support areas and make a loop around those care support areas. On the other hand a tree-

like design includes straight, segmented or intersecting corridors. As shown in Figure 6.8,

in a race track design corridor setting sound propagates in two major directions away

from a source. On the other hand in a tree-like design corridor, sound propagates in one

major direction.

Figure 6.8 Sound propagation in race track and tree-like design corridors (a) Race track
design hallways (b) tree-like design corridor (Blue dot represents the sound source and
arrows represent the sound rays)

Tree-like design corridors can have a variety of different floor-plate shapes including I,

T, L, cross and different combinations of I, L and T. Race track design corridors can also

have various floor-plate shapes including rectangular, triangular and circular floor-plate

shapes. However, race track design with rectangular floor-plate shapes is the most

common. Moreover, in many cases parallel hallways in these rectangular floor-plates are

connected with intersecting sub-hallways. The distance of these sub-hallways to parallel

main hallways ranges between 10m (33ft) and 24m (79ft).

Overall, the width of the corridors (including both tree-like design and race track design)

varies between 2m (6.5ft) to 4.7m (15.4ft) while usually it is approximately 2.7m (8.8ft).

The total length of the corridors composed of one hallway ranges between 8m (26ft) and

161
34m (111ft) as shown in Figure 6.9. However, usually the corridor clusters are composed

of multiple hallways. The total length of the corridor clusters with multiple hallways

ranges between 30m (98ft) and 210m (689ft). Except Bellevue and Paimio, the length of

each individual hallway composing these corridor clusters mostly ranges between 15m

(49ft) and 40m (131ft).

Figure 6.9 Total corridor length of corridor clusters

162
6.2 PILOT STUDY

LINKING ACOUSTICS AND DESIGN VARIABLES OF ACTUAL

EDUCATIONAL CORRIDORS

6.2.1 Scope

In sufficiently diffuse sound fields, reverberation time is primarily a function of two room

parameters: room volume and absorption. In non-diffuse sound fields such as long

enclosures, there are different design factors that have been found to have impact on

reverberation time values such as size and shape of the cross section, distribution of

absorptive materials on boundary surfaces and reflective qualities of surface materials

(diffuse vs. geometrical). However, the impact of floor-plate design of corridors on

reverberation time has not yet been systematically investigated. To statistically assess the

association between floor-plate design variables and reverberant qualities of corridor

sound environments, a pilot study is conducted in the inter-connected corridors of the

actual educational settings. The other goal of this pilot study is to provide input for the

more controlled acoustic simulation studies that follow. In this pilot study, design

characteristics of particular hallways composing inter-connected corridor systems are

correlated with the reverberant qualities of the sound environments.

6.2.2 Methodology

Field measurements have taken place in 5 corridor settings located in 3 different

educational buildings on the campus of Georgia Institute of Technology. The floor-plate

shapes of these interconnected corridors are shown in Figure 6.10. Each inter-connected

corridor is composed of several hallways and the length of the hallways ranged between

163
15m (49ft) to 45m (148ft). The red dot placed in the corridor drawings is used to

represent sound source locations and lines are used to illustrate the paths where receivers

are systematically distributed 5m (16ft) apart from each other. Color code is applied to

represent the receivers located at different hallways that are 1 and 2 turns away from the

sound source.

Design characteristics of each single hallway represented with an ID number in Figure

6.10 are correlated with mean and standard deviation of reverberation time values

measured in the same hallway. Some of the design variables considered in this analysis

include corridor length, number of branches (number of connected corridors to a single

corridor), number of turns and radial distance from the source. The effects of these design

variables on reverberation time values are assessed while controlling for other design

factors such as material and volume. The acoustic measure considered in the analysis is

mean reverberation time values. Mean values (represented as “RT-Mean” in Table 6.1)

represent the overall/averaged reverberation time values in each hallway. The mid

frequency range (500Hz and 1 kHz) is considered in particular because of its significant

importance for human hearing perception.

164
Figure 6.10 Plans of the corridors located in the educational buildings where impulse
response measurements were conducted

In this study, the design variable “number of turns” represents the order of the visibility

zones from the sound source location. For example the hallway/zone visible from the

sound source location is considered as “0-order visibility zone” and the hallway/zone

perpendicularly connected to the 0-order visibility zone is considered as “1st order

visibility zone”. Similarly, the hallway/zone perpendicularly connected to this 1st order

visibility zone is considered as “2nd order visibility zone”. The zone visible from the

sound source is referred as “visual field” in the following sections of this work. Receivers

located in the hallways where they do not have any visual connection with the sound

source (a.k.a. lack of direct sound energy) referred as the “non-visual sound field” are

165
found to experience longer reverberation times. The “critical distance” in all educational

corridor settings ranges between 5m (16.4ft) and 15m (49ft). Critical distance is the

distance where both the energy of the direct and reflected sound is equal. Conducting

acoustic measurements outside the critical distance is important to avoid the effects of

direct sound on acoustic outcomes. In this research, all impulse response measurements

are conducted outside the critical distance.

6.2.3 Findings

Partial correlation analysis results presented in Table 6.1 indicate that when other

interfering design variables are controlled (total volume, average distance, corridor

length, number of branches and material), the number of turns design variable is

significantly and positively correlated with mean reverberation time values. In other

words, receivers located at the hallways where they have a visual connection with the

sound source (a.k.a. presence of direct sound energy) are found to experience lower

reverberation times. Compared to receivers located in the non-visual zones with lower

visibility order, receivers located in the non-visual zones with higher visibility order are

also found to experience longer reverberation time. It needs to be noted that these

interpretations are limited to 1 and 2 turns because of the characteristics of the hallways

included in the sample group.

166
Table 6.1 Partial correlation analysis results linking corridor design variables, mean of
reverberation times (RT-Mean) and standard deviation of reverberation times (RT-SD)

RT-Mean
Independent corridor Control Frequency Partial
design variable Variables correlation
coefficient

Number of branches
500Hz 0.60**
Number of turns
Corridor length Material
Avg. distance
Volume 1kHz 0.50**

Corridor length
500Hz - 0.35
Number of turns
Number of branches Material
Avg. distance
Volume 1kHz - 0.30

Corridor length
500Hz 0.60**
Number of branches
Number of turns Material
Avg. distance
Volume 1kHz 0.70**

* When controlled for the other variables, the correlation b/w variables is statistically significant at 0.1 level (p<0.1)
** When controlled for the other variables, the correlation b/w variables is statistically significant at 0.05 level (p<0.05)

The design variable number of branches is used to represent the number of hallways

connected to a single hallway. According to partial correlation analysis, when other

interfering design variables are controlled (total volume, average distance, corridor

length, number of turns and material), the number of branches design variable is

negatively but not significantly correlated with mean reverberation time. In other words,

this finding suggests that in the hallways with higher number of branches, reverberation

time values are expected to be lower. The impact of corridor length on reverberation

time values is also investigated with partial correlation analysis. The length of the

corridors is found to be significantly and positively correlated with mean reverberation

167
time values when other interfering design variables were controlled (total volume,

average distance, number of turns, number of branches and material). This finding

suggests that as the length of a particular hallway increases, overall (averaged)

reverberation time value in the particular hallway are also expected to increase. The

results of this analysis also agreed with the earlier sound behavior analysis results. As

shown in Figure 5.12, reverberation time tends to increase with increasing hallway

length.

Another partial correlation analysis is conducted to assess the impact of radial distance on

reverberation time values. This analysis is partially different from the others because the

mean reverberation time values are not used this time. Instead, reverberation time values

measured at each receiver location are directly correlated with radial distance (direct

distance instead of walking distance) between the sound source and the receiver location.

As shown in Table 6.2, when volume and material are controlled for, radial distance is

found to be significantly and positively correlated with reverberation time at discrete

distances. This finding suggests that receivers located at positions with longer radial

distance from the sound source are expected to experience longer reverberation time

values compared to receivers located at positions with shorter radial distance.

168
Table 6.2 Partial correlation analysis results between corridor design variables of radial
distance and reverberation time at discrete distances.

Independent corridor Control Variables Frequency RT at discrete distances

design variable Partial correlation
coefficient

500Hz 0.56**
Radial distance Material
Volume 1kHz 0.50*

* When controlled for the other variables, the correlation b/w variables is statistically significant at 0.1 level (p<0.1)
** When controlled for the other variables, the correlation b/w variables is statistically significant at 0.05 level (p<0.05)

Overall, the findings of the pilot study analysis indicate the potential effects of floor-plate

design features on the acoustic qualities of corridor sound environments. In the following

sections, these findings are followed up with more detailed and controlled simulation

studies.

6.3 EMPIRICAL STUDY 1

LINKING ACOUSTICS AND DESIGN VARIABLES OF 60 THEORETICAL

CORRIDOR MODELS

6.3.1 Scope

The pilot study introduced in previous section has statistically investigated the

relationship between reverberation time and design variables of real life corridors with

complex floor-plate shapes. Some of these design features such as number of turns and

corridor length are found to be significantly associated with mean reverberation time

values. These interpretations are based on field measurements conducted in the actual

settings. In this part of the study, a more controlled follow up study is conducted to

enable the systematic analysis of the relationship between design and acoustic

169
characteristics of hallways located only in the non-visual sound fields of corridors with

complex floor-plate shapes. The motivation for focusing on non-visual sound fields is to

control for the interfering effects of direct sound present in the visual-sound field on

acoustic outcomes.

6.3.2 Methodology

Based on the heuristic analysis findings described in Section 4.2.1, 60 different

theoretical models are generated as shown in Figure 6.11. The total lengths of the

corridors composed of multiple hallways range between 30m (98ft) to 215m (705ft). The

length of each individual hallway included in these theoretical models range between

15m (49ft) and 40m (131ft). Among all theoretical models, corridor width (3m/9.8ft) and

height (3m/9.8ft) are the same.

Several layers of grouping are used to group the 60 theoretical models to control for the

effects of interfering design factors, as shown in Figure 6.11. The theoretical models are

first grouped based on their layout design types: tree-like design and race track design.

For each layout design type, 30 different theoretical models are generated. Theoretical

models in each group are further grouped based on number of turns. In addition to layout

design type, controlling for number of turns is necessary because Phase 1 pilot study

findings have indicated the significant association between this corridor design variable

and reverberation time values. The red dots in Figure 6.11 represent a theoretical sound

source. The hallways of the theoretical models that are 1 turn and 2 turns away from the

sound source are highlighted with blue and red lines respectively. In total, 15 different

170
theoretical models are generated for each number of turns category. Those models are

further grouped under three categories to allow systematic study of different design

variables including branch number, branch distance and corridor length.

Acoustic qualities of these models are analyzed with CATT acoustic simulation program.

The sound source is located 5m (16ft) away from one of the end walls. Multiple receivers

are located 5m (16ft) apart from each other along the hallways highlighted with red or

blue lines. Two different acoustic analyses are conducted to assess the impact of two

levels of absorption on the association between the design and acoustic variables. At first,

all surfaces are assigned a material with 0.1 absorption coefficient (). Second, all

surfaces are assigned a material with 0.3. These particular absorption coefficients are

chosen because they represent room averaged absorption coefficients of hospital settings

installed with lower and higher performance acoustic ceiling tiles. In total, 120 different

acoustic simulation analyses are conducted.

171
Test cases #1: tree-like design | 1 turn

Test cases #2: tree-like design | 2 turns

Figure 6.11 Plans of theoretical models grouped based layout type, number of turns,
number of branches, branch distance and corridor length criteria.

172
 Figure 6.11 (continued)

Test cases #3: race track design | 1 turn

Test cases #4: race track design |2 turns

173
Figure 6.12 3D-CATT models of theoretical tree-like design hallways

174
Figure 6.12 (continued)

175
Figure 6.12 (continued)

176
Figure 6.13 3D-CATT models of theoretical race track design hallways

177
Figure 6.13 (continued)

178
Figure 6.13 (continued)

179
6.3.3 Findings

First, 8 different partial correlation analyses are conducted. For each analysis, 15

theoretical models are considered. Design characteristics of each sample group are

summarized in Table 6.3.

Table 6.3 Design characteristics of theoretical model groups

Number of Number of
Layout design Surface material
turns theoretical models
Tree-like design Absorption coefficient=0.3 1 turn 15 cases
Tree-like design Absorption coefficient=0.3 2 turns 15 cases
Tree-like design Absorption coefficient=0.1 1 turn 15 cases
Tree-like design Absorption coefficient=0.1 2 turns 15 cases
Race track design Absorption coefficient=0.3 1 turn 15 cases
Race track design Absorption coefficient=0.3 2 turns 15 cases
Race track design Absorption coefficient=0.1 1 turn 15 cases
Race track design Absorption coefficient=0.1 2 turns 15 cases

Overall, findings of this follow up study agree with the earlier findings of the pilot study

introduced in Section 6.2. In the first analysis, all theoretical models are applied with

absorption coefficient = 0.1 and in the second analysis all theoretical models are applied

with absorption coefficient = 0.3. As shown in Table 6.4-Table 6.5, corridor length is

positively and significantly correlated with mean (averaged) reverberation time values.

This finding suggests that overall receivers in the longer hallways are likely to experience

longer reverberation time values. Number of branches is negatively and significantly

correlated with the mean reverberation time values. This finding suggest that as the

180
number of hallways (a.k.a. branch) connected to a particular core hallway increase,

overall reverberation time values in that particular core hallway are expected to decrease.

Another analysis is conducted to assess whether the way hallways are connected to a

particular hallway affect reverberation time values though the variable branch distance.

The findings indicate that the distance between the branch and the closest hallway

parallel to it is positively correlated with mean reverberation time values. However these

correlations are not statistically significant. Consistent with the pilot study findings,

number of turns is also positively and significantly correlated with mean reverberation

time values when controlled for interfering design factors. This finding suggests that as

the receivers navigate to corridors with higher number turns from the sound source, they

are likely to experience increased reverberation time.

Based on beta values included in Table 6.4 and Table 6.5, the effect size of the number of

turns variable is the highest compared to other design variables including corridor length,

number of branches and branch distance. The interpretation is that one unit change in

number of turns will potentially lead to more increase/decrease in RT30 compared to one

unit change in corridor length, number of branches and branch distance. The corridor

length variable has the second highest effects size. The interpretation is that one unit

change in corridor length will potentially lead to more increase/decrease in RT30

compared to one unit change in number of branches and branch distance. Number of

branches variable has the third highest effect size on the mean RT30.The interpretation is

181
that one unit change in corridor length will potentially lead to more increase/decrease in

RT30 compared to one unit change in branch distance.

Based on beta values represented in Table 6.4 and Table 6.5, changes in RT30 associated

with one unit change in the number of turns, corridor length, number of branches and

branch distance range between 0.01-0.57s. According to ISO standards, a minimum of

5% increase in RT30 is noticeable. Therefore, design interventions that lead to a

minimum of 5% increase/decrease in the existing RT30 levels will lead to noticeable

changes in RT30.

Table 6.4 Partial correlation analysis results (showing the strength of the relationship
between different corridor design variables and reverberation time values in hallways
with absorption coefficient = 0.3 surface materials. The sample group consisted of 60
theoretical models)

Control Frequency Number of Mean-RT

Absorption variable cases Partial Beta Values
coefficient = 0.3 Correlation
Branch number 0.25*
Length Branch distance 500Hz N=60 (p=0.066) 0.32*
Layout type
Turn 0.24*
Volume 1kHz N=60 (p=0.075) 0.33*
Ave dist.
Length -0.23*
Branch number Branch distance 500Hz N=60 (p=0.099) -0.28*
All Layout type
Turn -0.23*
cases Volume 1kHz N=60 (p=0.090) -0.30*
Ave dist.
Length 0.01
Branch Branch number 500Hz N=60 (p=0.96) 0.01
distance Layout type
Turn 0.08
Volume 1kHz N=60 (p=0.57) 0.10
Ave dist.
Branch number 0.24*
Turn Branch distance 500Hz N=60 (p=0.084) 0.52*
Layout type
Length 0.25*
Volume 1kHz N=60 (p=0.073) 0.57*
Ave dist.
* The correlation between variables is statistically significant at 0.1 level (p<0.1).

182
Table 6.5 Partial correlation analysis results (showing the strength of the relationship
between different design variables and reverberation time values in hallways with
absorption coefficient = 0.1 surface materials. The sample group consisted of 60
theoretical models)

Absorption Control Mean-RT Beta values

coefficient = 0.1 variable Frequency Number of Partial
cases Correlation
Branch number 0.24*
Length Branch distance 500Hz N=60 (p=0.078) 0.27*
Layout type
Turn 0. 23*
Volume 1kHz N=60 (p=0.082) 0.30*
Ave dist.
Length -0.23*
Branch Branch distance 500Hz N=60 (p=0.096) -0.25*
All cases number Layout type
Turn -0.20*
Volume 1kHz N=60 (p=0.094) -0.29*
Ave dist.
Length 0.01
Branch Branch number 500Hz N=60 (p=0.96) 0.01
distance Layout type
Turn 0.03
Volume 1kHz N=60 (p=0.85) 0.03
Ave dist.
Branch number 0.27*
Turn Branch distance 500Hz N=60 (p=0.052) 0.53*
Layout type
Length 0.26*
Volume 1kHz N=60 (p=0.058) 0.57*
Ave dist.

* The correlation between variables is statistically significant at 0.1 level (p<0.1).

183
6.4 EMPIRICAL STUDY 2

LINKING ACOUSTICS AND DESIGN METRICS OF 20 THEORETICAL

FLOOR PLATE MODELS

6.4.1 Scope

Empirical Study 1 analyzed the relationship between design characteristics and

reverberation time of 60 theoretical models developed from heuristic analyses. Empirical

Study 1 focused on the effects of design features of particular hallways located in

complex inter-connected corridor systems such as corridor length, number of branches,

etc. Instead of design qualities of particular hallways, this part of the study focuses on the

effects of overall floor-plate shape qualities of nursing unit corridors by using two new

floor-plate shape metrics. Furthermore, instead of the 60 theoretical corridor models,

empirical study 2 utilizes 20 new theoretical floor plate models.

6.4.2 Methodology

The two new design metrics introduced are used to quantify shape characteristics of

corridor floor-plates: visual fragmentation (VF) and relative grid distance (RGD)

(Shpuza, & Peponis, 2008). Floor-plate design metrics introduced in Chapter 6.3 are used

to quantify design qualities of particular hallways in a corridor system. In this section, the

RGD and VF are used to quantify shape qualities of the entire corridor floor-plates.

Relative grid distance (RGD) values represent the degree to which a floor-plate shape

diverges from square with regard to total distances between pairs of locations in the

shape. Square is assigned the smallest RGD value of 1. Higher RGD values indicate

more elongated floor-plate shapes. Visual fragmentation (VF) values reflect the

184
convexity of a floor-plate shape as the amount of directional changes needed to connect

any two locations on the floor-plate. Visual fragmentation values range between 0 and 1.

VF value of 0 is assigned to any convex shaped floor-plate. Higher visual fragmentation

values indicate more fragmented/jagged floor-plate shapes and those with holes.

Studies exploring the relationship between of floor-plate design and acoustics have

mostly considered simple floor-plate shape metrics like width-to-length ratio. Even

though these metrics have been successfully used for the shape analysis of simple floor-

plate shape geometries, they do not provide a comprehensive picture of the complex

floor-plate shape qualities such as fragmentation and compactness. RGD and VF are

chosen because these two metrics have been successfully used for the quantification of

complex floor-plate shape qualities as they relate to key programmatic considerations

such as the layout of work stations (Shpuza, & Peponis, 2008). Moreover, the underlying

logic of RGD and VF involves computing distance between grid cells and computing the

number of turns. Whereas these were initially developed because of their potential social

and programmatic impacts, there are in fact the same considerations that impact the

sound environment. One of the correspondences is related with similarities between path

distance considerations. For RGD calculations, a floor-plate is represented with a grid

system. RGD calculations take into account every possible distance needed to connect

each cell to all other cells located in the grid system. Interestingly, when sound rays are

emitted by the sound source, they travel different distances till they get to the receiver`s

ear. It is mainly because each sound ray is reflected by a different combination of

boundary surfaces. The other correspondence is related with similarities between

185
directional change considerations. VF calculations take into account all directional

changes needed to connect any two locations on the grid system. It is also well known

that directional changes (a.k.a. turns) leads to some changes in the behavior of sound.

To explore the relationship between RGD, VF and RT30, 20 different theoretical design

models are developed based on the heuristic design analysis of nursing unit corridor

floor-plates (Figure 6.14). As shown in Figure 6.9, the total corridor length of nursing

unit corridor clusters mainly range between 30m to 120m. Mid range total corridor length

value of 66m/216ft is assigned to all theoretical models. The width of the nursing unit

corridors range between 2m to 4.7m but most times it is about 3m. Therefore, the width

of the theoretical corridors is set to 3m.

As shown in Figure 6.5, about 2/3rd of the nursing unit corridor clusters are tree-like

design corridors and the rest of them are race track design corridors. In this analysis, all

of the theoretical models are examples of tree-like corridor design. Theoretical models

with more complex floor-plate shapes are purposefully excluded from the sample group;

particularly because RGD and VF values have not varied sufficiently for corridors with

more complex floor-plate shapes. The complexity of floor-plate shapes of the nursing

unit corridor clusters varied. For the same reason, floor-plate shapes of the theoretical

models are also designed to have different levels of geometric complexity. Ten of them

are composed of a single hallway with different floor-plate shapes. The rest of them are

composed of two main hallways connected with another hallway as shown in Figure

6.14. The number of intersecting corridor parts range from 2 to 6. This number is quite

186
 similar to the number of the intersecting corridor parts of the nursing unit corridor

clusters as shown in Figure 6.5.

RGD and VF shape metrics have been used for the quantification of floor-plate shape

characteristics of office spaces; however, they have not been used for the floor-plate

shape analysis of long enclosures. This study tests the applicability of these two design

metrics for floor-plate shape analysis of long enclosures and analyzes the association

between design and acoustics of corridors by considering these two metrics.

1 2 3 4 5

8 9 10
7
6

11 12 13 14 15

16 17 18 19 20

Figure 6.14 Floor plans of theoretical corridor models with different floor-plate shape
qualities

187
Acoustic characteristics of the theoretical models are analyzed with CATT acoustic

modeling program. A sound source (represented with a red dot) is located at the center of

the connecting corridor as shown in Figure 6.14. In this analysis two considerations are

taken into account to control for the effects of direct sound on reverberation time values.

First, the length of the hallway where the sound source is located is kept the same

(10m/33ft) in all theoretical models. Second, only non-visual sound field characteristics

are reported in the acoustic analysis of these theoretical models. Multiple receivers are

systematically distributed 5m (16.4ft) apart from each other along the highlighted paths

located in the non-visual sound fields of the theoretical models.

6.4.3 Findings

Two partial correlation analyses are conducted to assess the individual effects of VF and

RGD on mean reverberation time values. The reverberation time of each theoretical

model is represented with a single number. This single number is calculated by averaging

the reverberation time values calculated for each receiver located in the non-visual sound

field. The floor plate design variables of VF and RGD of each theoretical model are also

represented with single numbers provided by Qelize (www.morpostudio.net/qelize).

Qelzie is a Java applet designed for calculating the shape measures of RGD and VF.

According to analysis results as shown in Table 6.6, relative grid distance is positively

and significantly correlated with mean reverberation time values. This finding suggested

that as the floor-plate shape gets more elongated, averaged reverberation time values are

expected to increase. Interestingly, this finding is consistent with earlier findings of this

188
study. Elongated corridors potentially have longer corridors and according to previous

study findings reverberation time values are likely to be higher in the corridors with

higher corridor length.

In addition, visual fragmentation is negatively and significantly correlated with mean

reverberation time values. This finding suggests that as the shape gets more fragmented/

jagged, reverberation time values are likely to decrease. This finding can also be expected

based on the earlier study findings. Given the same total corridor length, more

fragmented long enclosures potentially have shorter individual corridors and in corridors

with shorter corridor length, reverberation time values are likely to be lower. Similarly, a

higher number of branches potentially results in more fragmented long enclosures. In the

empirical study 1, the number of branches is found to be significantly and negatively

correlated with reverberation time values.

Linear regression analysis is also conducted to assess the amount of variance caused by

the RGD and VF characteristics of the floor-plate shapes. As shown in Table 6.7, 75% of

the variance in reverberation time values (specifically for 500Hz) is explained with VF

and RGD. The variance accounted for is 65% at 1kHz. For visual assessment of the

correlation between floor-plate metrics and mean reverberation time values, scatter-plots

are also generated as shown in Figure 6.15 and Figure 6.16.

189
Table 6.6 Partial correlation coefficients showing the strength of the relationship between
floor-plate shape characteristics (RGD + VF) and reverberation time.

Control Mean-RT30 (500Hz) Mean-RT30 (1kHz)

variables Partial Correlation Partial Correlation
Visual fragmentation RGD -0.70* -0.60*
Relative grid distance VF 0.80* 0.75*

* The correlation between variables is statistically significant at 0.05 level (p <0.05).

Table 6.7 Regression analysis results

R2 Beta values R2 Beta values

(500Hz) (500Hz) (1kHz) (500Hz)
Visual fragmentation (VF) 75% -0.29 (p<0.05) 65% -0.27 (p<0.05)
Relative grid distance (RGD) 0.06 (p<0.05) 0.06 (p<0.05)

190
Figure 6.15 Scatter plots showing the relationship between RGD + VF and RT30 at
500Hz.

191
Figure 6.16 Scatter plots showing the relationship between RGD + VF and RT30 at 1kHz

192
A follow up analysis is conducted to explore the relationship between visual

fragmentation, relative grid distance and RT30. As shown in Table 6.8, bivariate

correlation analysis results indicate that VF and RGD values are not significantly related.

For visual assessment of the correlation between floor-plate shape metrics and mean

reverberation time values, scatter-plots are also generated as shown in Figure 6.17 and

Figure 6.18. In these scatter-plot graphs, two real world hospitals are also included in

addition to 20 theoretical models. Like theoretical models, acoustic qualities of these two

real world hospitals are analyzed with CATT acoustic modeling program. Mean RT30 is

consistently and significantly higher in the two real world hospitals compared to the

theoretical models. This is probably related with significant differences between the total

corridor lengths of the two real world hospitals and theoretical models.

Table 6.8 Bivariate correlation analysis results

Bivariate correlation Relative grid distance

Visual fragmentation 0.01 (p>0.05)

193
Figure 6.17 Scatter plots showing the relationship between RGD, VF and RT30 at 500Hz.
(MedSurg-ICU (VF=0.8, RGD=2.2, RT30= 1.75); Neuro-ICU (VF=1; RGD=2.2,
RT30=1.80)).

194
Figure 6.18 Scatter plots showing the relationship between RGD, VF and RT30 at 1kHz.
(MedSurg-ICU (VF=0.8, RGD=2.2, RT30=1.80); Neuro-ICU (VF=1; RGD=2.2,
RT30=1.90)).

6.5 DISCUSSION

The experimental studies introduced in Chapter 6 have statistically analyzed the

relationship between design and acoustic qualities of inter-connected corridors by

conducting heuristics design analysis, floor-plate shape analysis, field measurements,

acoustic simulations and statistical analyses. The summary of these empirical study

findings can be found in Table 6.9. More detailed discussion about these findings can be

found in the following paragraphs.

195
 Corridor Design Variable Impact on Reverberation
Increased Number of Turns Increased Mean RT
Increased Corridor Length Increased Mean RT & SD
Increased Number of Branches Decreased Mean RT
Increased Branch Distance Decreased SD
Increased Relative Grid Distance Increased Mean RT
Increased Visual Fragmentation Decreased Mean RT

Table 6.9 Summary of the empirical study findings conducted in Chapter 6

The main motivation of these empirical studies is to assess how different aspects of

corridor floor-plate design of nursing units impact reverberation times. Reverberation

time is one of the critical acoustic measures predicting the effectiveness of critical sound

tasks by caregivers such as critical conversations and auditory monitoring. Particularly, in

critical care units, caregivers continuously assess and localize auditory cues. Previous

study findings indicate that the capability of the human auditory system to monitor and

localize sounds is reduced in sound environments with long reverberation times.

A pilot study is conducted in the inter-connected corridors of several educational settings.

An impulse response measurement method is used to assess the acoustic characterstics of

these corridor settings. The pilot study findings suggest the potential significant impact of

design features of particular hallways (e.g., number of turns, corridor length, and number

of branches) on reverberation time values. For example, according to findings of the

study, in the hallways that are 2 turns away from the sound source, averaged

reverberation time values are likely to be higher compared to hallways 1 turn away from

the sound source. Moreover, in the longer corridors averaged reverberation time values

196
are likely to be higher compared to shorter corridors. On the other hand, in the hallways

with higher number branches, the reverberation time values are expected be lower

compared to hallways with lower number of branches.

To test the findings of the pilot study, Empirical Study 1 is designed to be a more

controlled follow up study using CATT acoustic simulation program. Based on heuristic

analysis of the existing nursing unit corridors, 80 different theoretical models with

different design implementations are generated and in total 140 acoustic simulations are

conducted. Findings of Empirical Study 1 are quite similar to the pilot study findings. For

example, corridor length of the hallways located in the non-visual sound fields is

significantly and positively correlated with averaged reverberation time values. In

constrast, number of branches variable is significantly and negatively correlated with

reverberation time values.

Empirical Study2 is conducted to assess the effects of additional floor-plate shape

characteristics of inter-connected corridor systems on averaged reverberation time values.

According to this analysis, RGD is positively correlated with mean reverberation time

values and VF is negatively correlated with mean reverberation time values. Moreover,

the relationhip between RGD and VF is not statistically significant.

197
 CHAPTER 7

VALIDATION STUDY

7.1 SCOPE

The previous section has discussed the findings of the theoretical studies linking design

and acoustics of inter-connected corridors. To assess whether theoretical study findings in

Chapter 6 track with the study findings based on real-world hospital corridor geometries,

a series of follow up studies is conducted. The acoustics and design of inter-connected

corridors of the two real-world hospital wards is analyzed by conducting field

measurements and acoustic simulation analysis.

7.2 METHODOLOGY

The design features of the two units are reviewed in detail in Section 2.2.2. Shortly, the

Neurological ICU (Neuro-ICU) is a fairly new 20-bed unit. The Neuro-ICU has a tree-

like design corridor system with long corridors. The total length of the hallways is 200m

(656ft).

The Medical Surgical ICU (MedSurg-ICU) is a 1980s era 20-bed unit. The MedSurg-ICU

has a triangular shape race track design corridor system with shorter corridors. The total

length of the corridors is 80m (262ft).

198
In-situ impulse response measurements

Impulse response measurements are conducted to assess the reverberant qualities of the

two ward corridors objectively (Figure 7.1). The impulse response measurements have

taken place at 4 different receiver locations in two hallways in each ward as shown in

Figure 7.2 and Figure 7.3.The preliminary measurements first have taken place in the

non-visual sound field. However, these measurements have to be excluded as an adequate

signal to noise ratio (SNR) could not be achieved. In the other non-hospital settings, the

SNR can be improved by simply increasing the level of the driving sound signal.

However, ICUs are noise sensitive settings and occupied 24/7. This limits the level of the

driving sound signal for the impulse response measurement. In the MedSurg-ICU and

Neuro-ICU, the critical distances are 3m (9.8ft) and 5m (16.4ft), respectively. At all

times, the receivers are located beyond the critical distance. The distance between source

and receivers are 8m (26ft) and 12m (39ft) in both wards. Detailed information about the

impulse response measurement procedure can be found in Appendix G and Appendix H.

During the measurements, all patient room doors are closed in both wards. All staff

members including the directors of the both wards and the visitors are informed about the

procedure.

199
 Figure 7.1 Impulse response measurements in the hospital corridors

Figure7.2 Distribution of source and receiver locations in the Neuro-ICU

200
Figure 7.3 Distribution of source and receiver locations in the MedSurg-ICU

Acoustic simulation analysis

Idealized 3D acoustic models of the two real world ward corridors are also generated.

The models allowed for control of the interfering effects of the differently distributed

materials on the boundary surfaces. Original corridor lengths are assigned to 3D acoustic

models of the two ward corridors. Acoustic simulation analysis of these corridors is

conducted in 6 phases.

In phase 1, the impacts of T and cross shaped corridor intersections on reverberation time

levels are tested (Figure 7.4). The analysis has taken place only in the corridors of the

Neuro-ICU as there is no T-shaped configuration in the MedSurg-ICU. First, source 1

(S1) is activated and reverberation times at the R1 and R2 locations are recorded. Second,

source 2 (S2) is activated and reverberation times at the R3 and R4 locations are

recorded. The distances between source and receivers are kept the same for both test

201
cases. To represent the acoustic conditions associated with the T shaped intersection,

reverberation times predicted for receivers R1-R2 are averaged. Similarly, predicted

results for receivers R3 and R4 are averaged to represent the acoustic conditions

associated with cross shaped intersection.

Figure 7.4 Source and receiver locations considered for the acoustic analysis of “T” and
“cross” shaped intersections in the phase 1 Neuro-ICU simulations

In phase 2, the impact of number of branches on reverberation time is tested. The analysis

has taken place in the corridors of both the Neuro-ICU and MedSurg-ICU (Figure 7.5-

7.6). The receivers R1, R2 and R3 are located in one of the Neuro-ICU corridors with

multiple branches (Figure 7.5); and receivers R3, R4 and R5 are located in one of the

MedSurg-ICU corridors without any branches (Figure 7.6). Predicted RT30 levels for

202
R1-R2-R3 and R4-R5-R6 are averaged to represent the overall acoustic conditions along

the Neuro-ICU corridor and the MedSurg-ICU corridor, respectively.

Figure 7.5 Distribution of source and receivers along the corridor with multiple branches
in the phase 2 Neuro-ICU simulation

203
Figure 7.6 Distribution of source and receivers along the corridor without any branches in
the phase 2 MedSurg-ICU simulation

In phase 3, the impact of number of turns on reverberation time is tested (Figure 7.7-7.8).

In the Neuro-ICU, the receivers R1 and R2 are located along the corridor that was two

turns away from the sound source (Figure 7.7). In the MedSurg-ICU, the receivers R3

and R4 are located along the corridor that is one turn away from the sound source (Figure

7.8). The distances between source and receivers are kept the same in the Neuro-ICU and

MedSurg-ICU. Predicted results for R1-R2 and R3-R4 are averaged to represent the

overall reverberant qualities of the corridors, respectively.

204
Figure 7.7 Distribution of source and receivers that are 2 turns away from the source in
the phase 3 Neuro-ICU simulation

Figure 7.8 Distribution of source and receivers that are 1 turn away from the source in the
phase 3 MedSurg-ICU simulation

205
In phase 4, the impact of corridor length on reverberation time is tested (Figure 7.9-7.10).

In the Neuro-ICU, receivers R1 and R2 are placed along the longest corridor (Figure 7.9).

In MedSurg-ICU, receivers R3 and R4 are placed along a shorter corridor (Figure 7.10).

Receivers R1-R2 and R3-R4 are averaged to represent the overall acoustic conditions

along the corridors in Neuro-ICU and MedSurg-ICU, respectively.

Figure 7.9 Distribution of source and receivers along a long corridor in the phase 4
Neuro-ICU simulation

206
Figure 7.10 Distribution of source and receivers along a shorter corridor in the phase 4
MedSurg-ICU simulation

In phase 5, the impact of overall corridor design on reverberation time is tested by

placing the receivers in different corridors of each unit (Figure 7.11-7.12). Visual

fragmentation values for Neuro-ICU and MedSurg-ICU are 1.0 and 0.8, respectively. The

interpretation is that floor-plate shape of the Neuro-ICU is more fragmented compared to

floor-plate shape of the MedSurg-ICU. Relative grid distance value for Neuro-ICU and

MedSurg-ICU is the same (2.2). In other words, floor-plate shapes of the Neuro-ICU and

MedSurg-ICU is similarly elongated. Predicted reverberation time values for R1-R2-R3-

R4-R5 are averaged to represent the acoustic conditions in the idealized corridors of the

Neuro-ICU. Predicted reverberation time values for R6-R7-R8-R9-R10 are averaged to

represent the acoustic conditions in the MedSurg-ICU corridors.

207
Figure 7.11 Distribution of source and receivers in the phase 5 Neuro-ICU corridor
simulation

Figure 7.12 Distribution of source and receivers in the phase 5 MedSurg-ICU corridor
simulation

Finally in phase 6, another analysis is conducted to test the impact of overall corridor

design on reverberation time (Figure7.13-7.14). In this analysis, an idealized version of

208
the Neuro-ICU floor-plate design is considered instead of the original floor-plate design

used in the previous phase 5 analysis (Figure 7.13). The original design of the MedSurg-

ICU corridors remained unchanged (Figure 7.14). The motivation is to test the impact of

two different corridor designs with similar total corridor length. Visual fragmentation

values for the idealized design of the Neuro-ICU and MedSurg-ICU is 1.0 and 0.8,

respectively. The interpretation is that the idealized floor-plate shape of the Neuro-ICU is

more fragmented compared to the floor-plate shape of the MedSurg-ICU. The relative

grid distance values for the idealized floor-plate shape of the Neuro-ICU and MedSurg-

ICU are 1.9 and 2.2. In other words, the floor-plate shape of the MedSurg-ICU is more

elongated. Predicted reverberation time values for R1-R2-R3-R4-R5 are averaged to

represent the acoustic conditions in the idealized corridors of the Neuro-ICU. Predicted

reverberation time values for R6-R7-R8-R9-R10 are averaged to represent the acoustic

conditions in the MedSurg-ICU corridors.

Figure 7.13 Distribution of source and receivers in the phase 6 idealized Neuro-ICU
corridor simulation

209
Figure 7.14 Distribution of source and receivers in the phase 6 MedSurg-ICU corridor
simulation

In all phases, similar materials are applied to the boundary surfaces of the two wards. The

absorption and scattering coefficients of the surface materials are shown in Table 7.1. In

each phase, the same analysis is repeated for two different types of ceiling tiles:

absorption coefficient = 0.1 and absorption coefficient = 0.3. This is necessary to test

how different material applications affect the impact of floor-plate design on reverberant

qualities of the corridors.

210
 Table 7.1 Absorption coefficients of materials applied to the surfaces of the two wards

Surface Material 250Hz 500Hz 1kHz 2kHz 4kHz SC

Wall Drywall 0.08 0.05 0.04 0.03 0.03 30%

Ceiling Tile 0.3 0.3 0.3 0.3 0.3 20%

 0.3)
Ceiling Tile 0.1 0.1 0.1 0.1 0.1 20%
 0.3)
Floor Vinyl 0.02 0.03 0.03 0.03 0.02 10%

7.3 FINDINGS

Impulse response measurements

The averaged measured reverberation times in different corridors of each unit are shown

in Figure 7.15. In the MedSurg-ICU, averaged reverberation times range between 0.44s

and 0.56s across frequency. In one of the corridors reverberation times are as low as

0.30s and the highest was 0.76 s. In the Neuro-ICU corridors, averaged reverberation

times are consistently higher compared to averaged reverberation times in the MedSurg-

ICU (Figure 7.15), even though the Neuro-ICU is installed with absorptive acoustic

ceiling tiles. The reverberation times in the Neuro-ICU range between 0.70 s and 0.87 s.

The lowest reverberation time recorded in the Neuro-ICU corridors is 0.67 s and the

highest was 0.92 s. The difference between the reverberation times measured in two ward

corridors range between 33% and 68%. This result indicates the potential significant

impact of corridor design in addition to absorptive surface material applications on the

reverberation time. For example, the floor-plate design of the MedSurg-ICU corridor is

more compact with shorter corridors. According to the previous Chapter 6 findings, in the

211
corridor settings with more compact and shorter corridors, reverberation times are likely

to be less.

Figure 7.15 Averaged measured reverberation times in the MedSurg-ICU and Neuro-ICU
corridors

Table 7.2 Just noticeable difference results of measured reverberation times in the two
wards

250Hz 500Hz 1kHz 2kHz 4kHz

JND
(impulse 33% 50% 68% 55% 62%
response
measurement)

212
Acoustic Simulation Analysis

Simulation Phase 1: impact of corridor intersection type on RT30

According to the averaged simulation results, reverberation time is perceptibly higher at

the T-shaped corridor intersection compared to the “+” shaped corridor intersection

particularly at 250Hz, 500Hz, 1kHz and 2kHz (Figure 7.16 and Figure 7.17). In the light

of the Chapter 6 findings, this result is expected because compared to a T-shaped corridor

a cross shaped corridor has more branching corridors. In Chapter 6, increased number of

branches is found to be significantly positively and negatively correlated with the

reverberation time.

Figure 7.16 Averaged simulated reverberation times at the phase 1 “T” and “+”shaped
intersections (ceiling tile  0.3)

213
Figure 7.17 Averaged simulated reverberation times at the phase 1 “T” and “+”shaped
intersections (ceiling tile  0.1)

Table 7.3 JND results of simulated reverberation times at the phase 1 “T” and “+”shaped
intersections

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 12% 13% 10% 10% 8%
 0.3
JND
ceiling tile 14% 15% 9% 8% 9%
 0.1

Simulation phase 2: impact of branch number on RT30

In the Neuro-ICU corridor with multiple branches, averaged reverberation times are

lower compared to the reverberation times in the MedSurg-ICU corridor without any

branches. Similar to the phase 1 previous analysis, the difference is not as significant at

high frequencies compared to the difference at lower frequencies including 250Hz,

214
500Hz and 1kHz (Figure 7.18 and Figure 7.19). This result is expected because in

Chapter 6, increased number of branches is found to be significantly and negatively

correlated with the reverberation time.

Figure 7.18 Averaged simulated reverberation times at the phase 2 corridors with
different number of branches (ceiling tile  0.3)

Figure 7.19 Averaged simulated reverberation times at the corridors with different
number of branches (ceiling tile  0.1)

215
Table 7.4 Just noticeable difference results of simulated reverberation times at the
corridors with different number of branches

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 10% 12% 11% 4% 7%
( 0.3)
JND
ceiling tile 11% 13% 13% 9% 7%
( 0.1)

Simulation phase 3: impact of number of turns on RT30 levels

The averaged reverberation times are higher in the corridor that is 2 turns away from the

sound source (Neuro-ICU) compared to the reverberation times in the corridor that is 1

turn away from the source (MedSurg-ICU). The difference is higher at lower frequencies

including 250Hz, 500Hz and 1kHz (Figure 7.20 and Figure 7.21). This result is expected

because in Chapter 6, increased number of turns is found to be significantly and

positively correlated with reverberation time. However it is also possible that slightly

larger volume of Neuro-ICU might be contributing to the larger RT30.

216
Figure 7.20 Averaged simulated reverberation times at the phase 3corridors that are 1
turn and 2 turns away from the source (ceiling tile  0.3)

Figure 7.21 Averaged simulated reverberation times at the phase 3 corridors that are 1
turn and 2 turns away from the source (ceiling tile  0.1)

217
Table 7.5 Just noticeable difference results of simulated reverberation times at the phase
3 corridors that are 1 and 2 turns away from the sound source

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 11% 13% 10% 6% 7%
 0.3)
JND
ceiling tile 18% 16% 12% 5% 6%
( 0.1)

Simulation phase 4: impact of hallway length on RT30

The averaged reverberation time is longer in the longer corridor of the Neuro-ICU

compared to the averaged reverberation time in the shorter corridor of the MedSurg-ICU.

In particular, the difference is higher at lower frequencies including 250Hz, 500Hz and 1

kHz (Figure 7.22 and Figure 7.23). This result is expected because in Chapter 6,

increased corridor length is found to be significantly and positively correlated with the

reverberation time. However it is also possible that slightly larger volume of Neuro-ICU

might be contributing to the larger RT30.

218
Figure 7.22 Averaged reverberation times in the phase 4 corridors with different corridor
length (ceiling tile  0.3)

Figure 7.23 Averaged reverberation times in the phase 4 corridors with different corridor
length (ceiling tile  0.1)

219
Table 7.6 Just noticeable difference results of simulated reverberation times at the phase
4corridors with different corridor length

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 12% 11% 13% 9% 6%
 0.3
JND
ceiling tile 13% 16% 15% 9% 10%
 0.1

Simulation phase 5: impact of overall floor-plate shape on RT30

Even though, the Neuro-ICU has a more fragmented floor plate shape, the mean of the

reverberation times both measured and predicted in different corridors of the Neuro-ICU

are higher compared to the MedSurg-ICU. The relative grid distance values of the two

unit floor-plate shapes are similar. On the other hand, the total corridor length of the

Neuro-ICU is longer. Given these results, it is plausible to conclude the potential

significant impact of total corridor length on the mean RT30 of the two unit corridors.

Consistent with previous analysis, the difference is higher at lower frequencies including

250Hz, 500Hz and 1 kHz (Figure 7.24 and Figure 7.25).

220
Figure 7.24 Averaged simulated reverberation times in the phase 5 corridors with
different total corridor length (ceiling tile  0.3)

Figure 7.25 Averaged simulated reverberation times in the phase 5 corridors with
different total corridor length (ceiling tile  0.1)

221
Table 7.7 Just noticeable difference results of simulated reverberation times at the phase
5 corridors of two wards with different total corridor length

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 11% 13% 14% 7% 9%
 0.3)
JND
ceiling tile 15% 18% 17% 8% 6%
 0.1)

Simulation phase 6: impact of overall floor-plate shape on RT30 levels

Different from previous analysis, the mean of the predicted reverberation time in the

idealized corridors of the Neuro-ICU is shorter compared to the mean reverberation times

measured in the MedSurg-ICU corridors. In particular the difference is greater at low

frequencies including 250Hz, 500Hz and 1 kHz (Figure 7.26 and Figure 7.27). These

results can be expected based on the statistically significant relationships (found in

Chapter 6) between floor-plate shape qualities and mean RT30. Idealized floor-plate

shape of the Neuro-ICU is more fragmented and less elongated compared to the original

floor-plate shape of the MedSurg-ICU. According to Chapter 6 findings, in more

fragmented and less elongated floor-plates, mean RT30 is expected to be less.

222
Figure 7.26 Averaged simulated reverberation times in the phase 6 corridors with similar
length and different shape (ceiling tile  0.3)

Figure 7.27 Averaged simulated reverberation times in the phase 6 corridors with similar
length and different shape (ceiling tile  0.1)

223
Table 7.8 Just noticeable difference results of simulated reverberation times at the phase
6 corridors of two wards with similar total corridor length

250Hz 500Hz 1kHz 2kHz 4kHz

JND
ceiling tile 12% 14% 10% 9% 8%
 0.3)
JND
ceiling tile 14% 17% 13% 12% 6%
 0.1)

7.4 DISCUSSION

To assess the findings of previous theoretical studies conducted in Chapter 6, acoustic

qualities of two real world hospital ward corridors are analyzed in Chapter 7 by

conducting impulse response measurements (in situ measurements) and simulation

studies.

Interestingly, even though the Neuro-ICU is installed with absorptive ceiling tiles,

averaged RT30 is consistently and perceptibly higher in the Neuro-ICU compared to

MedSurg-ICU. This suggests the potential significant impact of design features on RT30

in addition to absorptive qualities of surface materials. For example the MedSurg-ICU

has a more compact layout with shorter corridors and thus a shorter reverberation time.

Simulation analysis is conducted in 6 phases. In each phase, the impact of particular

corridor design features on RT30 levels is tested including corridor length, number of

224
branches, number of turns and overall design (race track vs. tree-like design) by

comparing predicted reverberation times in two wards.

Overall, the differences between the simulated averaged reverberation times measured in

different corridors of the two wards are higher at lower frequencies including 250Hz,

500Hz and 1 kHz. At all times, the simulation study findings agree with the findings of

Chapter 6. For example, the means of the predicted reverberation times in the corridors

with longer corridor length, lower number of branches and higher number of turns from

the sound source are higher. However, it is possible that slightly larger volume of Neuro-

ICU might also be contributing to the larger RT30. In addition, (given two simulated unit

corridors have similar total corridor length, width and height, and equal absorption

treatment were applied), averaged reverberation times are higher in the more race track

design unit (MedSurg-ICU) compared to the more fragmented and more compact

idealized Neuro-ICU corridors. On the other hand, (when original designs of the two unit

corridors were considered), increased corridor length in the Neuro-ICU led to significant

increase in the reverberation times.

To summarize in the units with shorter, more compact, fragmented corridors with

multiple number of branching hallways, reverberation time levels are likely to be less.

Moreover higher number of turns also potentially increases reverberation time.

225
 CHAPTER 8

CONCLUSION

8.1 SUMMARY of FINDINGS

Chapter 1

In Chapter 1, introductory information about the general structure of the thesis and the

goals of the study are provided.

Chapter 2

By conducting three empirical studies, Chapter 2 documented the objective and

subjective qualities of the hospital sound environments with different architectural

designs, assessed the effectiveness of newer acoustic metrics in capturing caregiver

perceptions, and evaluated the impact of particular noise sources on caregiver outcomes.

In total, three different empirical studies have been conducted. These studies took place

in two 20-bed ICUs: a new Neurological-ICU and a 1980s-era Medical-Surgical-ICU.

Empirical study 1 assessed the objective and subjective noise levels at different locations

in the two ICUs. The MedSurg-ICU is perceived as louder, more annoying, and having a

greater negative impact of noise on work performance, health outcomes, and anxiety as

compared to the Neuro-ICU. Surprisingly, there is little difference between two ICU

sound environments using traditional overall noise measures. The objective differences

between the occupied sound environments in the two units emerged more clearly through

a more comprehensive analysis of the “occurrence rate” of peak and maximum levels,

226
frequency content, and the speech interference level. Furthermore, mid-level transient

sound occurrence rates are significantly and positively correlated to perceived annoyance

and loudness levels.

Empirical study 2 documented the objective and subjective noise levels during different

times at each unit nurse stations by conducting the occurrence rate analysis and assessing

nurse perceptions. Sound environments of the two unit nurse stations are significantly

different based on the occurrence rate analysis. This is consistent with nurse perceptions.

The MedSurg-ICU is consistently perceived as more annoying and louder during

different times of the day and days of the week. Similar to the previous empirical study

findings, these two sound environments are similar based on traditional Leq noise level

analysis.

Empirical study 3 compared the level of nurse disturbance due to overall-noise and

alarms in the two ICUs. MedSurg-ICU nurses have perceived medical alarms as more

detrimental to their work performance and health outcomes, anxiety and annoyance levels

compared to Neuro-ICU nurses. Interestingly, alarms are perceived to be as disturbing as

overall-noise by nurses in both units.

Overall, chapter 2 findings suggest that in particular impulsive characteristics of

healthcare sound environments potentially lead to negative health, wellbeing and

performance outcomes among caregivers. Moreover the use of more detailed acoustic

metrics such as occurrence rate (a newer metric used in this study) can provide a better

227
picture than traditional measures of the aspects of the hospital sound environments as

they impact user experience and well being. Moreover, objective and subjective qualities

of hospital sound environments can vary between settings with different designs.

However conducting more controlled studies are necessary to identify the impact of

particular design features on the acoustic qualities of healthcare settings.

Chapter 3

By reviewing literature and conducting an empirical study, Chapter 3 provided a

comprehensive overview of factors related to auditory monitoring. Additionally, results

from a case study comparing nurses` auditory monitoring performance in two ICUs with

different architectural designs are presented.

The literature review findings indicate that different factors can affect the monitoring of

auditory cues including environmental factors and acoustic qualities of signals. For

example, in highly reverberant and noisy spaces, localization and detection of auditory

cues can be very challenging. In reverberant spaces, late arriving reflections can diminish

the positive effects of early reflections. In spaces with high background noise levels, the

signal-to-noise ratio decreases and therefore the signal becomes less distinguishable.

As a part of the empirical study, nurses are asked to rate the importance of different nurse

tasks and different types of including auditory cues for patient safety. The ICU nurses in

both units have indicated that auditory monitoring is as highly important and similar in

importance to visual monitoring for patient safety. Moreover, a majority of the nurses

228
have found various medical alarms highly important for patient safety such as ventilator

alarms, patient monitor alarms, IV pump alarm, and nurse call as well as ventilator

hissing sound. The nurses also indicated that they monitor these critical sounds at

multiple different locations in their units. These key listening locations include: patient

rooms, nurse stations, corridors in both wards and the medication preparation zone in the

Neuro-ICU.

Nurses in the two units are also asked to rate their ability to conduct sound tasks in their

units. In the MedSurg-ICU, nurse hearing and localization performance are perceived to

be higher compared to the Neuro-ICU even though the Neuro-ICU included a high

performance absorptive acoustic ceiling. Acoustic ceiling tiles are commonly used to

decrease noise levels and reverberation times. Similarly, in MedSurg-ICU averaged

reverberation times are perceptibly shorter compared to Neuro-ICU. This finding

indicates the potential significant impact of other design features such as volume and

corridor length on reverberant qualities of hospital sound environments in addition to

absorption qualities of surface materials.

Overall, chapter 3 findings suggest that like visual monitoring, auditory monitoring is a

highly critical nurse task for patient safety. Nurses monitor a variety of different critical

sounds including alarm and non-alarm sounds at different locations in the nursing units.

Nurses` auditory monitoring performance can vary in healthcare settings with different

reverberant qualities. The design of hospitals is complex and different design features

might contribute to reverberant qualities of hospital sound environments. Therefore,

229
conducting more controlled studies is necessary to identify design features impacting

reverberant qualities of hospital sound environments.

Chapter 4

By reviewing previous literature, Chapter 4 had provided an overview of room-acoustics

parameters, the sound behavior in proportional and non-proportional spaces, and the

association between floor-plate design and acoustics.

Reverberation time has been commonly used for the acoustic analysis of different types

of spaces including long spaces and rooms. Newer room acoustic-parameters including

early decay time, clarity and sound strength have been developed particularly for the

acoustic analysis of concert halls (a proportional space). However, applicability of these

newer acoustic parameters to other spaces has not yet been resolved.

Related with their design qualities such as spatial proportions, the acoustic characteristics

of sound environments can vary significantly. Therefore acoustic theories applicable in

one type of space might not be applicable to another space. Various studies have been

conducted in concert halls and statistically analyzed the relationship between specific

design characteristics of floor-plate shapes such as width, length and acoustic outcomes.

However, the number of studies that systematically and statistically analyzed the

relationship between design and acoustics of long enclosures still remains limited. Some

studies explored the acoustics of long enclosures with complex floor-plate geometries

such as underground stations with branches or staggered urban streets. However, not

230
many studies investigated the acoustic conditions of inter-connected corridors located in

the buildings.

Overall, the study findings indicate that as the shape of the long enclosures gets more

complicated, sound behavior in these settings also becomes highly complicated. This

research conducted more controlled follow-up studies to clarify the impact of specific

design features on the acoustics of inter-connected corridors with complex floor-plate

shapes.

Chapter 5

Chapter 5 has conducted various impulse response measurements and acoustic simulation

analysis to assess the effectiveness of acoustic modeling tools in estimating acoustic

qualities of non-proportional spaces.

Impulse response measurements (in situ measurements) have been commonly used for

the assessment of reverberant qualities of architectural settings including proportional and

non-proportional spaces. Acoustic simulation programs have been particularly used for

estimating the acoustic qualities of proportional spaces (rooms) such as concert halls.

Various validation studies have tested the reliability of these results predicted by the

acoustic modeling programs. In particular, the effectiveness of hybrid prediction methods

has been validated. However, not many studies have tested the effectiveness of prediction

methods in estimating the acoustics of long enclosures. This study has conducted various

impulse response measurements and acoustic simulation analyses in long enclosures with

231
different floor-plate shapes to compare the difference between field measured and

predicted results. Most times predicted results are within 5-22% accuracy with a few

exceptions. This is parallel with the finding of previous studies. For example, in Kang`s

(2002b) study the difference between predicted and measured results range between 10-

25%. He has also used the same simulation program to conduct theoretical studies and

explore the relationship between design and acoustics of long enclosures.

Another analysis has been conducted to compare the behavior of sound in a T-shaped

corridor and a single corridor. Different from single corridor, the trend of the data sets

recorded in the T-shaped corridor is not linear. In particular, the shape of the data set

trend lines in the non-visual sound field of the T-shaped corridor is more complex

compared to that in the visual sound field. This can be explained with the dominant

effects of the reflections in the non-visual sound field.

Overall, the findings of Chapter 5 suggest the potential effectiveness of an acoustic

simulation tool with hybrid prediction (i.e., CATT) in studying the acoustic qualities of

complex hospital sound environments.

Chapter 6

Chapter 6 presents the results of a heuristic design analysis, and statistically explores the

association between acoustics and design variables of interconnected corridors.

232
First, a pilot study is conducted in the inter-connected corridors of educational settings.

The pilot study findings suggest the potential significant impact of design features of

particular hallways (e.g., number of turns, corridor length, and number of branches) on

reverberation time values. For example, according to findings of the study, in the

hallways that are 2 turns away from the sound source, averaged reverberation time values

are likely to be higher compared to hallways 1 turn away from the sound source.

Moreover in the longer corridors, averaged reverberation time values are likely to be

higher compared to shorter corridors. On the other hand, in the hallways with higher

number branches, the reverberation time values are expected be lower compared to

hallways with lower number of branches.

To test the findings of the pilot study, a more controlled follow up study (Empirical Study

1) is conducted by using the CATT acoustic simulation program. Various theoretical

design models are generated based on the findings of the heuristic design analysis.

According to the findings, the corridor length of the hallways located in the non-visual

sound fields is significantly and positively correlated with averaged reverberation time

values. In constrast, the number of branches variable is significantlyand negatively

correlated with reverberation time values.

Another theoretical design analysis (Empirical Study 2) is conducted to assess the effects

of overall floor-plate shape characteristics of inter-connected corridor systems on

averaged reverberation time values. According to this analysis, RGD is positively

correlated with averaged reverberation time values and VF is negatively correlated with

233
mean reverberation time values. Overall these findings agree with the Empirical Study 1

findings as higher RGD values indicate more elongated floor-plates with potentially

longer corridors. On the other hand, more fragmented floor-plates with higher VF values

are likely to have shorter corridors.

Overall, the findings of Chapter 6 suggest that in addition of design features of particular

hallways such as corridor length, number of turns and number of branching hallways,

overall design of nursing unit floor-plate shapes can have significant impact on the

reverberant qualities of hospital sound environments To provide less reverberant sound

environments, designers should consider more compact and more fragmented nursing

unit floor-plate shapes.. All else equal, nursing units with shorter corridors and with more

branching hallways have shorter reverberation times. Also, the more turns that care

providers are from the patient rooms, the poorer their ability to monitor critical sounds

originated in the patient rooms.

Chapter 7

In Chapter 7, the validity of the previous study findings (in Chapter 6) has been assessed

by analyzing the acoustics of real-world hospital ward corridors via impulse response

measurements and acoustic simulation analysis.

The impulse response measurements are conducted in different corridors of the two ICUs

already mentioned. In the Neuro-ICU averaged RT30 levels are consistently and

perceptibly higher compared to averaged RT30 in the MedSurg-ICU corridors. This

234
suggested the significant impact of different design features in addition to absorptive

qualities of the surface materials. For example, the MedSurg-ICU has a more compact

corridor layout with shorter corridors and thus a shorter reverberation time.

Simulation analysis is conducted in 6 phases. In each phase, the impact of a particular

corridor design feature on RT30 is assessed. At all times, the simulation study findings

have agreed with the findings of Chapter 6. For example, the mean of the reverberation

times measured/predicted in the corridors with longer corridor length, lower number of

branches and higher numbers of turns from the sound source are higher. Averaged

reverberation times are also lower in the more fragmented and more compact unit

(Neuro-ICU) compared to less fragmented and less compact (MedSurg-ICU) when both

the simulated units has similar total corridor length, width and height and equal

absorption treatments. On the other hand, increased total corridor length in the Neuro-

ICU led to significant increase in the reverberation times.

Overall, Chapter 7 findings confirm the findings of Chapter 6. In other words, when

designing real world hospital floor-plate geometries, designers can refer to the findings of

theoretical study findings presented in Chapter 6.

8.2 CONTRIBUTION

The research findings contribute to ongoing efforts to improve the currently problematic

hospital sound environments in different ways. This research:

235
 Provides a detailed understanding of the hospital sound environments by studying

noise propagation and architectural acoustics

 Diagnoses problematic aspects of hospital sound environments that particularly

relate to caregiver outcomes including performance and wellbeing

 Proposes and validates a newer acoustic metric effective in capturing unique

qualities of hospital sound environments

 Tests and validates the effectiveness of more detailed acoustic metrics (compared

to traditional metrics) that have not been commonly used in noise propagation

analysis in the hospitals

 Provides a detailed understanding of patient auditory monitoring particularly

conducted in the ICUs

 Provides a multi-disciplinary study framework that links three different types of

variables including design, acoustics and occupant outcomes

 Expands the relationship between design and acoustics by conducting statistical

analysis

 Clarifies design features of long enclosures affecting sound environments and

floor-plate design qualities of nursing unit corridors conductive of critical sound

tasks

 Expands the limited knowledge on acoustics of long enclosures by examining

acoustics of interconnected corridors

 Suggests the use of hybrid acoustic simulation programs for the analysis of

hospital sound environments and long enclosures in particular interconnected

corridors

236
  Provides an opportunity for designers, engineers and care providers to think and

discuss hospital sound environments starting from the early design phases by

linking design and acoustics

 Defines a new area where sound localization research is applicable

8.3 LIMITATIONS

Limitations

External validity

 The studies introduced in Chapter 1 and 2 are conducted in the ICUs and the

study participants are ICU nurses. Related with the sample group characteristics,

generalizability of the Chapter 1 and 2 findings are limited to ICUs.

 In Chapter 6, acoustic simulation study has considered theoretical design models

generated based on the heuristic design analysis of the nursing units. Related with

the limited size of the sample group, the generalizability of the Chapter 6 findings

are limited to nursing units.

 In Chapter 6, the floor-plate design qualities of corridor settings are correlated

with reverberant qualities of hospital sound environments. The floor-plate shape

analysis program and floor-plate shape metrics used in Chapter 6 are only

effective in quantifying the shapes of the tree-like design nursing unit corridors

with limited complexity. Due to the limited size of the sample group, the findings

of this analysis are only generalizable to tree-like design corridors.

237
8.4 FUTURE WORK

Hospitals have complex and problematic sound environments. This research was able to

address particular problems of the hospital sound environments by statistically assessing

the relationship between architectural design features, acoustics and caregiver outcomes.

Future work should also consider patient and visitor outcomes. In addition, this study

focused on self-reported outcomes. While they present challenges in access and

measurement future work might explore other safety, quality and well-being outcomes

such as communication errors, medication errors and physiological stress.

Moreover, this research focused particularly on reverberant and sound level qualities of

hospital sound environments. Even though reverberation time has significant impact on

speech intelligibility levels, this research did not focus on the direct relationship between

speech intelligibility and design features. In fact, speech intelligibility was one of

acoustic metrics that was not reliably predicted by CATT Acoustics V8 for the complex

long enclosures modeled. Therefore, future research should also explore reliable acoustic

measurement/prediction methods for the study of speech intelligibility in hospitals. Also,

the design and acoustics of nursing units was the focus of this study. Future research

should also analyze acoustics and design features of different healthcare settings such as

emergency departments.

238
 APPENDIX A: DEFINITION OF TERMS

The following definitions of the acoustics terms are gathered from the following sources:

Mehta (1997); Long (2006); and Everest, and Pohlmann (2009).

Absorption coefficient: A measure of the sound absorbing property of a surface. More

specifically, absorption coefficient is defined as the fraction of the incident sound energy

absorbed by a surface.

Artificial/dummy head: A binaural acoustic measuring system consisting of two

microphones built into the ears of an artificial human head and torso.

Occurrence rate: Percentage of the time that specified peak and maximum sound levels

exceed certain sound pressure levels.

A-weighting: A means of electronically simulating the unequal sensitivity of the human

ear at various frequencies by filtering.

Binaural hearing: A recording and playback configuration emulating hearing with two

ears.

Centre time (Ts): Center time is a measure used to describe where the sound energy is

concentrated in the echogram (energy distribution map of the sound). Low values of Ts

indicate that the arriving sound is concentrated in the early part while high values indicate

early reflections are weak or decay is slow.

C-weighting: An attempt to electronically reduce the unequal sensitivity of the human ear

at various frequencies by filtering. C-weighting is similar to A-weighting except that C-

weighting is more applicable to loud sounds and low frequency sounds.

239
Clarity (C): Clarity is a measure used to quantify the ability to distinguish individual

sounds (e.g., music, warning, whirring machine) from the general audible stream, or the

degree to which rapidly occurring individual sounds are distinguishable. Clarity values

depend on the sound level difference between early and late arriving reverberant energy.

Direct sound and early reflections are usually considered to be useful sounds that

improve clarity. The cutoff value that represents the transition from early to late sound is

typically considered to be 80ms for music and 50ms for speech.

Cocktail party effect: Humans ability to focus one`s listening attention on a single talker

among a mixture of conversations and background noises.

Critical distance: The distance at which the sound pressure level of the direct and

reverberant sound fields are equal.

Decibel: The decibel is a logarithmic unit that indicates the ratio of a physical quantity

relative to a specified or implied reference level. It is widely known as a measure of

sound pressure level.

Definition (D): Definition can be a measure of speech intelligibility. Definition values

depend on the percentage ratio of the early arriving sound energy to total sound energy.

Direct sound and early reflections are considered to be useful sounds that improve

definition. The cutoff value that represents the transition from early to late sound is

considered to be 50ms for speech.

Diffraction: A change in the direction of propagation of sound as a result of bending

caused by a barrier in the path of the sound wave.

240
Diffuse field: A sound field in which the sound comes in equal intensity from all

directions. For example, a diffuse sound field can be obtained in a large room with highly

reflective surfaces.

Diffuse reflection: Reflection of sound from a rough surface in such a manner that the

sound rays will diffuse in many different directions.

Direct sound: The sound that arrives at a receiver along a direct line from the source

without reflection from any surface.

Early decay time (EDT): Early decay time is another acoustic measure used to quantify

sound decay. EDT is the time required for the initial part the sound to decay 10dB. This

value is multiplied by 6 to extrapolate the results to RT60. Main difference between two

measures is that EDT is composed of a few early reflections while RT consists of

multitude reflections. This is mainly because EDT is calculated from the initial slope of

the reverberation curve.

Exceedance level: Level of sound exceeded during specified percent of the time. For

example, L33 = 50 would mean that 33%of the run time, the decibel level was greater than

or equal to 50dB.

Equivalent sound pressure level: The sound pressure level of a steady sound which, in a

specific time period, has the same energy as the time varying sound.

Frequency: The number of cycles per second measured. The unit of frequency is cycles

per second which is called Hertz (Hz).

Heuristic design analysis: Obtaining information through systematic evaluation.

Hybrid prediction method: A prediction method that optimizes the best features of two

different acoustic prediction models: image source model (ISM) and ray tracing (RT).

241
Image source modeling: A prediction method that determines the images of the source to

calculate the direction of the reflected sound rays off the enclosure surfaces.

Impulse response: Impulse response represents an acoustic “fingerprint” of a setting

from which various key acoustic metrics can be derived.

Impulsive noise: A noise of short duration particularly of high intensity.

Inter-aural cross correlation: Correlating acoustic qualities of the signals received by

the right and left ear.

Inter-aural level difference: Sound level difference between the sound signals reaching

the right and left ear.

Inter-aural time difference: Difference between the arrival times of the sound signals

reaching the right and left ear.

Intermittent sound: A sound which is discontinuous or fluctuates.

Just noticeable difference: Average minimal perceptible change in objective parameters.

Lateral energy fraction (LF/LFC): Lateral fraction is used to assess the relative

contribution of strong early lateral reflections arriving at the listener from either side at

the angles of 20-90degree relative to the front of the listener. Statistically, lateral fraction

is the linear ratio of lateral energy or the energy arriving from the sides within (5ms-

80ms) to total energy within (0ms-80ms).

Lateral fraction: Ratio of sound energy arriving laterally (from the side walls) over

sound energy arriving from all directions.

Long enclosures: Enclosed settings with higher L/H; L/W ratio compared to normal

rooms. (L: length; H: height; W: width).

242
Loudness: An auditory sensation. Loudness depends on factors such as sound pressure

level, frequency of sound and duration.

Maximum sound pressure level: Maximum sound level is the highest sound pressure

level reading of a conventional sound level meter in a specified time interval.

Minimum sound pressure level: Minimum sound level is the lowest sound pressure level

reading of a conventional sound level meter in a specified time interval.

Moderator effect: A situation in which the direction or magnitude of the relationship

between two variables depends on the value of one or more other variables.

Monaural hearing: Hearing auditory signals with one receptor (ears/microphone).

Noise: An unwanted sound.

Noise fluctuations: Time variant nature of the noise levels.

Noise-induced outcomes: Occupant outcomes associated with noise characteristics such

as level.

Non-visual sound field: Sound field/zone that is not visible from the sound source.

Objective noise level: Noise levels quantified with the use of a sound level meter.

Octave band: Frequency interval between two sounds whose frequencies are related to

each other in the ratio of 1:2.

Omni-directional source: A source that radiates sound in all directions equally.

One-third octave band: An octave band divided into three one-third octave bands. The

upper frequency limit of a one-third octave band is 21/3 times its lower frequency limit.

Peak sound pressure level: Peak sound pressure level is different from maximum sound

pressure level. It represents the true peak of sound pressure wave. Therefore the signal

does not pass through a pressure circuit or calculator. For a pure tone the Peak is 3dB

243
above the Maximum sound level. However for varying signals this difference can be

much higher.

Pink noise: A noise whose level decreases with increasing frequency in such a way that

there is a constant energy in every octave.

Precedence effect: Early reflections arriving between 2ms and 50 ms after the arrival of

the direct sound are typically considered to be useful reflections because they can be

combined with the direct sound. The human auditory system localizes sound sources by

considering early/first arriving sounds and this phenomenon is known as precedence

effect.

Ray tracing: Prediction of numerical acoustic measures by tracing acoustic rays based on

the rules of the geometric acoustics.

Relative grid distance: Relative grid distance (RGD) values represent the degree to

which a floor-plate shape diverges from square.

Reverberant sound field: A sound field created by repeated reflections of sound from the

boundaries in an enclosed space.

Reverberation: The continuation of sound in an enclosed space after the initial source has

terminated.

Reverberation time (RT60, RT30, RT15): Reverberation time is a measure of sound

decay. In scientific terms, reverberation time (RT60) is the time it takes sound to decay

60dB or to one millionth of its initial energy. When the background noise in a space is

loud enough that a full 60dB decrease cannot be feasibly measured, reverberation time is

also measured over a range of 30dB (RT30) and 15dB (RT15). In more general terms,

reverberation time refers to the time it takes for sound to fade away after a source has

244
stopped. Strong reflections provided by the enclosure extend the reverberation time.

Excessive values of reverberation time can interfere with the activities conducted in the

settings such as speech intelligibility and sound localization.

Room acoustics: Field of acoustics that defines the sound decay qualities in enclosed

settings.

Room acoustics parameters: Measures used to define acoustic qualities of enclosed

settings such as concert halls and lecture rooms.

Room gain or Sound strength (G): Sound strength is used to assess sound level gained

by the room conditions. It represents “the total energy of an impulse response at a

location in a room relative to the total energy of the same source measured in an anechoic

chamber at a distance of 10m” (Chiang, 1994). Sound strength approximates the

subjective sense of loudness.

Scattering: An irregular diffraction of sound in many directions.

Scattering coefficient: A parameter used in acoustic simulation programs to quantify

diffusely reflective qualities of the boundary surfaces.

Signal to noise ratio (SNR): A measure of signal strength relative to background noise.

Soundscape: An environment of sound (sonic environment) with emphasis on the way it

is perceived and understood by the occupants.

Sound lateralization: A type of sound localization but where the subjects localize the

sounds in their heads as signals are introduced via head phones.

Sound level meter: An instrument for measuring sound pressure levels and other acoustic

parameters.

245
Sound localization: A hearing system`s ability to make accurate judgments about the

distance and direction of the sound source.

Sound pressure level: Sound pressure level is a logarithmic measure of the effective

sound pressure of a sound relative to a reference value. The commonly used reference

sound pressure in air is pref = 20 µPa (rms), which is usually considered as the threshold

of human hearing.

Sound strength: The sound level gained by the room conditions.

Spectral content (frequency spectrum): Distribution of sound energy across different

frequencies.

Specular reflection: Reflection of a sound from a smooth surface in such a manner that

the angle of the incidence is equal to the angle of the reflection.

Speech intelligibility: Intelligibility of speech that is usually measured in the presence of

noise and sometimes reverberation. Speech intelligibility is quantified in different ways

such as the percentage of speech units understood correctly by a listener.

Speech interference levels: Arithmetic average of the sound pressure levels at 500Hz,

1kHz and 2kHz center frequencies and is a measure of the degree to which background

noise interferes with speech.

Subjective noise level: Loudness levels reported by the occupants.

The center time (Ts): A measure used to describe where the sound energy is concentrated

in the echogram.

Visual fragmentation: Visual fragmentation (VF) values reflect the convexity of a floor-

plate shape defined as the amount of directional changes needed to connect any two

locations on the floor-plate.

246
Visual sound field: Sound field/zone that is visible from the sound source.

White noise: A noise whose energy is uniform over a wide range of frequencies.

247
 APPENDIX B: DESIGN QUALITIES

OF THE NEURO-ICU AND MEDSURG-ICU

The Neurological ICU (Neuro-ICU) is a recently opened 20-bed unit (A.B1). This unit

received the “ICU Design Citation” award in 2008, co-sponsored by the Society of

Critical Care Medicine (SCCM), the American Association of Critical Care Nurses

(AACN), and the American Institute of Architects Academy on Architecture for Health

(AIA). The award was for the Neuro-ICU’s design intent to enhance the critical care

environment for patients, families and clinicians. Some unit design features include large

private patient rooms with family studios and distributed nurse work areas and care

support areas. High performance absorptive acoustic ceiling tiles and drop ceiling

applications reside mainly along the two parallel sides of the corridors and at the nurse

stations, painted dry wall, acoustic rubber flooring, and 1.8 m (6 ft) wide (two-wing)

glass patient room doors are some of the surface applications in the unit. The patient care

core of the Neuro-ICU sits on approximately 1,765 m2 (19,000 sqft).

248
Figure A.B1 Floor-plans of two critical care units (Key=a: patient room with family
studio, b: family area around the periphery of the unit, c: central nurse station, d: open
space, e: MED. area (i.e., medication preparation, pixes machine), f: other service/support
areas; grey highlighted zones: de-central nurse stations in the Neuro-ICU or patient
observation cores in the MedSurg-ICU).

The Neuro-ICU has a cluster-type architectural layout, composed of a 6-bed and 14-bed

clusters. Each cluster has a central nurse station with its own care support areas (e.g.,

medication room, supply room), and computerized patient monitoring system. In total,

the unit has two central nurse stations and seventeen distributed nurse work areas. Each

36 m2 (390 sqft) patient room is segregated into two portions: a patient care area

(approximately two thirds of the total area) and a family lounge area. The two areas are

separated by a semi-opaque glass wall. Approximately 1/3rd of the patient care core floor

area is occupied by the corridors, which are 183 m (600 ft) total length. The Neuro-ICU

includes additional spaces such as public family areas, CT scan lab, and a healing garden.

The total area of the Neuro-ICU is approximately is 2,229 m2 (24,000 sqft).

249
The Medical Surgical ICU (MedSurg-ICU) is a 1980s era 20-bed unit (Figure A.B1).

Compared to the Neuro-ICU, the MedSurg-ICU has more traditional surface applications

including standard ceiling tiles, vinyl flooring, approximately 1.5 m (5 ft) wide (two-

wing) glass patient doors, and painted dry wall. The patient care core of this unit is

approximately 817.5 m2 (8,800 sqft).

The MedSurg-ICU has a triangular shape race track layout design. In this layout, medical

and support areas are located in the center and patient rooms are located on the perimeter,

with a corridor separating the two space types. Twenty private patient rooms are

organized around one large triangular-shaped service core that contains a centralized

nurse station and other care support areas such as medication room, equipment and

supply room. The centralized nurse station contains two patient monitoring cores (each

serves up to ten patients) at the corners and has a separate computerized patient

monitoring system. Patient rooms in this unit are approximately 18 m2 (190 sqft). This

particular layout type requires the use of segregated corridors for staff and family

members. Approximately 1/4th of the patient care core floor area is occupied by the staff

corridor, with a total length of 73 m (240 ft). The total MedSurg-ICU sits on

approximately 1161 m2 (12,500 sqft).

Overall, floor-plate shape of the Neuro-ICU is more fragmented compared to MedSurg-

ICU (VF=1, VF=0.8, respectively). On the other hand, floor-plate shapes of the two units

are similarly elongated (RGD=2.2).

250
 APPENDIX C: NOISE LEVEL FIELD MEASUREMENTS

Monaural recordings

Although the human auditory system gathers information with two ears (a.k.a. binaural

hearing) some auditory information such as loudness can be obtained by listening through

one ear (a.k.a. monaural hearing). Based on their practicality and convenience, usually

sound level meters (with a single channel microphone such as a condenser microphone)

are used for the assessment of background noise levels via monaural recordings. The

condenser microphone samples the sounds as shown in Figure A.C1. The preamplifier

prepares an electronic signal for further amplifications and processing. It also keeps the

microphone away from the body of the instrument to avoid strong reflections. The body

of the instrument includes electronic circuits to process the sound detected by the

microphone. The digital signal processor filters the sound in various ways (e.g., A, C and

Flat broadband filters; Real-time FFT and 1/1 and 1/3 octave band filters; Slow, Fast

RMS detectors; Impulse and Peak detectors) and prepares the readouts to be displayed on

the digital LCD display of the instrument or exported to a computer.

251
 Figure A.C1 Larson Davis sound level meter

The use of additional equipment such as lockable “outdoor measurement case” is

necessary if the sound level meter is going to be left unattended in the field (Figure

A.C2). The body of the instrument is located in the outdoor measurement case. Through

the holes located on the sides of the case, the microphone extension cable can be run

from the case to the target location of the microphone.

252
Figure A.C2 Set of equipment necessary for continuous noise level measurement

Placement of sound level meter is critical to capture accurate sound samples. When

appropriate, a sound level meter can be place on a tripod as shown in Figure A.C3.

253
Figure A.C3 Placement of sound level meter in an unoccupied patient room

In the clustered occupied patient rooms, the use of tripod might not be practical. A

practical solution can be hanging the microphone from the ceiling while leaving the

instrument in the case (Figure A.C4 and Figure A.C5).

254
Figure A.C4 Placement of a single channel microphone in an occupied patient room

When applicable the microphone can also be placed on one of the medical equipment

surrounding the patient such as medical boom (Figure A.C5).

255
 Figure A.C5 Placement of a single channel microphone in a ward

Figure A.C6 Placement of a single channel microphone at a nurse station

256
APPENDIX D: NOISE LEVEL MEASUREMENT GUIDE

257
258
259
260
APPENDIX E: HOSPITAL NOISE SURVEY QUESTIONS

261
262
263
264
265
266
 APPENDIX F: OCCURRENCE RATE

This appendix aims to provide more information about occurrence rate analysis.

Occurrence rate represents the % of the time during which background noise levels

exceeded specified peak and maximum sound pressure levels. This study considered

maximum sound pressure values of 70dB, 80dB and 90dB and peak sound pressure levels

of 80dBC, 90dBC and 100dBC as the threshold values. It was mainly because the

occurrence rates of maximum sound pressure levels less than 70dB and higher than 90dB,

and the occurrence rates of peak sound pressure levels less than 80dBC and more and

100dBC did not differ much in the two hospital settings.

Figure A.F1represents the sound pressure levels associated with different sound events to

enable cross comparisons. However it needs to be noted that not only sound level but also

other qualities of sound events such as frequency content and duration can also

significantly affect annoyance and loudness perceptions of the human auditory system.

267
Figure A.F1. Sound pressure levels associated with different sound events

There are not many detailed guidelines providing allowable exposure times to maximum

and peak sound levels. According to OSHA, exposure to an impulsive noise event should

not exceed 140dB peak sound pressure level (OSHA, Occupational Safety and Health

Standards). According to WHO, Lmax sound pressure levels should not exceed 40dB in

the hospital settings particularly in the patient rooms (Berglund et al., 1999).

Calculation of occurrence rate:

To calculate occurrence rates, peak and maximum sound pressure levels are documented

every minute (a.k.a. 1-minute interval) during the study period. Then peak and maximum

sound pressure levels are compared with the specified levels as shown in Figure A.F2.

268
Figure A. F2. Example acoustic data set used for the calculation of occurrence rate of
maximum and peak sound pressure levels.

Different from occurrence rate analysis, traditional metrics used to assess impulsive

characteristics of sound environments only consider the highest LMax and LPeak values

that occurred during the study period. In other words, they represent the characteristics of

a single highly impulsive noise event. Overall, compared to traditional metrics,

occurrence rate analysis of LMax and LPeak sound pressure levels provide a better

picture of the impulsive characteristics of sound events. Therefore, in addition to

traditional metrics, the use occurrence rate analysis can be highly important.

269
 APPENDIX G: IMPULSE RESPONSE FIELD MEASUREMENTS

This appendix introduces the details of gathering impulse response of a room by

conducting field measurements. The use of different equipment is necessary for this type

of field measurement (Figure A.G1). The microphone is located at the receiver position.

It captures the room response to a sound burst produced by the sound source located at

another location. In particular, the ISO 3382-1 standard suggests the use of an omni-

directional sound source and an omni-directional microphone to measure the

reverberation time (ISO 3382-1). Some other components of the system that were used in

this study included an Outline amplifier, ProSonus EASERA Gateway data acquisition

system (DAQ) and lap-top equipped with EASERA software v1.1 (Figure A.G2). The

DAQ system enables the data transfer between different system components including

amplifier, lap-top, and microphone.

DAQ (data acquisition system)

Figure A.G1 Diagram showing the set up of impulse response equipment

270
 Speakon cable Connector

Amplifier
Audio cable

ProSonus EASERA Gateway Box Omni directional loudspkear

Single-channel microphone EASERA software

Figure A.G2 Individual components of the impulse response measurement

271
Maximum Length Sequence (MLS) Method vs. Others

Different sound signals can be used during for impulse response measurements. A

commonly used signal is the “sine sweep” which consists of a swept sine wave with a

frequency sweeping over the desired frequency range (Figure A.G3). This is not a

random signal and the extracted impulse response is free from noise contributed from the

excitation signal. It is considered as a weighted sum of delayed impulses with defined

weighting factors (Acoustics Engineering, 2007). By deconvolving the sweep signal with

source signal, room impulse response is obtained.

Figure A.G3 Shape of the swept sine signal

In this study, a maximum length sequence (MLS) signal was used. It is a periodic,

pseudorandom white noise signal. Its frequency spectrum over one period is flat (Figure

A.G4). It is considered as the weighted sum of delayed impulses with weighting factors

+1 and -1 (Acoustics Engineering, 2007). By calculating the cross-correlation between

the MLS and the signal at the receiver location, the room impulse response is obtained.

Due to cross-correlation, non-correlated noise including the competing background noise

is rejected. Therefore, the extracted impulse response is therefore not “polluted” by any

noise due to the excitation signal. It is one of the reasons why this method enables more

reliable impulse response results in noisy environments such as HVAC background noise.

272
For reliable results, ISO 140 Standard recommends 6dB as the minimum acceptable

value for Signal-to-Noise ratio (ISO 140 Standard).

Figure A.G4 Shape of the MLS signal

EASERA

EASERA is a software package used for the estimation of room-acoustics parameters

such as reverberation time from room impulse response. EASERA`s post processing

engine calculates all acoustic functions and measures according to ISO Standard 3382

based on octave or 1/3 octave filters (EASERA Tutorial). The impulse response of a

room can be captured via EASERA by using a variety of excitation signals such as sine

sweep, white and pink noise and MLS signal as shown in Figure A.G5.

273
 Figure A.G5 EASERA interface

Critical Distance

The energy of the sound source decreases by distance. The distance where both energy

densities of the direct and reflected sound are equal is called “critical distance” (Kutruff,

2000). The total energy of the sound is the sum of direct sound and reflected sound.

Inside the reverberation radius (a.k.a. critical distance) direct sound is dominant, however

beyond the reverberation radius the effect of direct sound is negligible as shown in Figure

A.G6.

274
 Figure A.G6 Critical distance graph (from Kutruff, 2000)

Conducting impulse response measurements outside the reverberation radius is important

as most times the main interest of acousticians is the assessment of reflections caused by

the enclosure boundaries. The following formula is used for the calculation of the critical

distance. The critical distance (rc) is in meter. “A” is the total absorption area and equal to

Sxa. “S” is the total surface area and “a” is the average absorption coefficient of surfaces

in the room.

275
APPENDIX H: IMPULSE RESPONSE MEASUREMENT GUIDE

276
277
 APPENDIX I: COMPUTER MODELING

Different theories have been developed to describe sound behavior in enclosed spaces

such as wave-based acoustics and geometric acoustics. Wave based acoustics considers

sound as pressure waves. Some of the computer simulation methods developed based on

wave-based acoustics principles include Finite Element Method (FEM) or the Boundary

Element Method (BEM). However, these methods are often not practical for architectural

acoustics because the number of modes in a room increases rapidly as frequency

increases. Therefore, the use of these methods is typically limited to predict the acoustics

of small rooms and low frequencies (Smith, 2004). Geometric acoustics consider sound

as rays, similar to light rays used in the field of optics that propagate in straight lines.

This consideration assumes that the dimensions of a rooms are large compared to the

wavelength of the sound. In small rooms, this assumption does not always hold for the

low frequency sounds because wavelengths of low frequency sound is larger compared to

wavelengths of high frequency sounds. Related with this, the use of geometric acoustics

is typically limited to frequencies of 250Hz and above in small rooms. Computerized

acoustic modeling programs have been developed based on geometric acoustics theory

that are capable of simulating different behaviors of sound in the enclosed spaces such as

specular reflection, diffuse reflection and absorption. These programs most times do not

take into account diffraction. However, recent advances in computer simulation programs

enabled the development of algorithms that emulate diffraction from edges. This topic is

discussed in more details below.

278
Some of the computer simulation methods developed based on geometric acoustics

principles include ray tracing, image source method, and hybrid method. In the following

section, the qualities of each method are introduced based on the information gathered

from following sources: Smith 2004, Long 2006, CATT manual, and Odeon Manual.

Some information about the Radiosity method is also included below. Kang (2002a) used

this method for the theoretical studies introduced in Chapter 4.

Ray Tracing

In acoustic modeling applications that use a ray tracing method, a source emits a finite

number of rays representing the sound waves in either an omni-directional pattern (i.e.,

spread of sound rays in every direction) or in a specific pattern based on the directivity of

the sound source (Figure A.I1). The ray tracing method predicts numerical acoustic

measures by tracing acoustic rays based on the rules of the geometric acoustics (e.g.,

generating specular and/or diffuse reflections) and using statistical methods to calculate

energy loss via absorption. The sound rays are traced until they reach a virtual listener

represented with a sphere. Detection of the sound ray is significantly influenced by the

size of the detector. For example, a large spherical detector will record a large number of

hits from the rays compared to a receiver represented with a smaller sphere. By recoding

the energy, direction and arrival time of sound rays, the program generates an echogram

displaying the strength of the reflections at the receiver location over time. One of the

main advantages of the ray tracing method is the shorter computation time because the

sound source emits sound rays randomly. The computational time is proportional to the

number of rays and the order of the reflections used. Another advantage of this method is

279
that multiple receivers at different locations can be investigated because sound is emitted

in every direction.

Figure A.I1 Distribution of sound rays in an enclosed space (only rays crossing the circle
contribute to the impulse response calculations)

Image Source Modeling (ISM)

ISM method determines specular reflections between source and receiver. It generates the

images of the sound source known as virtual sources or image of the source (Figure

A.I2).

Figure A.I2 Image of a sound source (from Cheenne, 2002)

280
These virtual sources replace the boundaries of the room as shown in Figure A.I3

(Cheenne, 2002). ISM first calculates the first order image sources of the main sources in

all reflecting planes. From each of these first order image sources, second order sources

are created by calculating new images sources in all reflecting planes. This procedure is

repeated until the order of reflection and arrival time defined by the user. Once all image

sources are found, the program calculates arrival times by calculating the length of each

image source to the receiver. The amplitude of each reflection is calculated from the

output power and directivity of the source and is adjusted by absorption and diffusion

properties of the planes as well as air absorption.

This method concerns only with the sound reflections reaching the receiver and ignores

reflections that do not reach the receiver. It collects the amplitude, arrival time and the

direction of all reflections. The echogram is provided by summing the contribution of

each image source with corresponding delays and attenuation (Tsakostas, 2004). It needs

to be noted that, each virtual source contributes only a pure impulse of known strength

and delay in the time domain.

In particular, this method has been mainly preferred because of its effectiveness in

handling the early arriving energy or low-order reflections (i.e., direct sound, first and

second order reflections) (Monks, Oh, & Dorsey, 1996).

281
 Figure A.I3 Distribution of sound source images (from Cheenne, 2002)

Hybrid Method

Hybrid algorithms have been developed to optimize the best features of two different

acoustic prediction models: image source model (ISM) and ray tracing (RT). The hybrid

method is the most up-to-date prediction method. Hybrid algorithms uses ray tracing

method to determine the valid image sources. Each ray detected by a receiver is

associated with an image source. These image sources are found by tracing it back from

the receiver location. To keep track of the valid image source, “image tree” is created.

Hybrid methods handle early and late reflections differently by incorporating the best

features of ISM and ray tracing methods. Often the image source method is used to

predict the early part of the impulse response and mostly ray tracing methods is used to

predict the late part of the impulse response.

282
Acoustic modeling programs such as CATT and ODEON use a hybrid algorithm. These

hybrid methods may vary based on the type of receiver used such as cone, beam and

pyramid tracing vs. ray tracing. In this study CATT Acoustics V8 is used. In CATT V8,

Early Part Detailed ISM prediction method uses the image source method to calculate the

early part of the echogram which is an approximation to the squared impulse response of

a room (Smith, 2004; CATT manual). The Full Detailed Calculation prediction method

uses Randomized Tail-Corrected Cone tracing (RTC) or Late-Part Ray Tracing. The use

of latter option is suggested for coupled rooms, and rooms with complex floor-plate

shapes.

Radiosity Method

The radiosity method has been predominantly used to calculate light energy. It is

considered to be applicable in acoustics when used for high frequency band (Kang,

2002a). The radiosity method divides boundaries into a number of patches and distributes

the sound energy of an impulse source to patches (Figure A.I4). The patches are then

determined as sound sources which redistribute the sound energy to other patches. In

summary, this method calculates the energy response at the receiver locations by

calculating the energy exchange between patches.

Figure A.I4 Patches used in the Radiosity method (from Kang, 2002a)

283
Acoustic Modeling User Settings

Behaviors of sound commonly predicted by the computer simulation programs include

diffuse reflection, geometric reflection and absorption. With recent developments,

diffraction can also be predicted by some programs. For the accuracy of the predicted

acoustic outcomes and simulation of sound behaviors in enclosed spaces, user selected

parameters are highly necessary.

Diffuse vs. geometric (specular) reflection

In accordance with “law of reflection”, a light/sound ray is reflected from the boundary

surfaces with a predictable manner. Reflection off smooth surfaces leads to a reflection

known as “specular reflection”. As a rule of thumb, the angle of the incidence is equal to

the angle of the reflection for specular reflections (Figure A.I5). Specularly reflected

sound rays remain concentrated in a bundle upon leaving the surface. Reflection off

rough surfaces leads to a type of reflection known as “diffuse reflection”. Diffusely

reflected sound rays diffuse in many different directions. Diffusive qualities of boundary

surfaces are significant for acoustic outcomes such as reverberation time (RPG Diffuser

Systems). For example, when the surfaces of a room are diffusely reflective, there is

more chance that the sound rays will hit various surfaces including walls and ceiling

(where most time absorbers are located). Diffusely reflective surfaces also provide more

uniform distribution of acoustic qualities in a room.

284
Figure A.I5 Sound reflections off diffusely and geometrically reflective surfaces (from
Cheenne, 2002)

Scattering coefficient

Scattering coefficient is a parameter used in acoustic simulation programs to quantify

diffusely reflective qualities of the boundary surfaces. It is a measure of the total amount

of sound scattered randomly from the boundary surfaces in relation to total reflected

sound energy as shown in the equation below.

The energy that is not scattered is absorbed or specularly reflected by the boundary

surfaces. In more detail, if the incident energy is normalized to1, the total reflected sound

energy (that is not absorbed by the surface) will be (1-α) where α is the absorption

coefficient. This energy is used for specular and diffuse reflections. The component of

the sound energy that is specularly reflected is considered as (1- α)(1-δ) and the

component that is non-specularly reflected (scattered) is considered as (1- α) δ (Figure

A.I6).

285
Figure A.I6 Diffusely and specularly reflected sound rays off rough surfaces (from
Cheenne, 2002)

Rough surfaces are assigned higher values of scattering coefficients to indicate more

scattering of the sound compared to smooth surfaces. The CATT-Acoustics manual

provides some guidelines for approximating the scattering coefficients of the surfaces. A

minimum of 20% scattering is recommended for an average size, flat, smooth surfaces

and 10% for larger flat and smooth surfaces. Some of the other surfaces that CATT

specifies scattering coefficients include audience areas (suggested scattering coefficients

range between 30% and 70%) and rough surfaces with 0.3m surface roughness

(suggested scattering coefficient is as high as 80%). In conditions of doubt, assigning

higher values of scattering coefficient rather than assigning lower values is suggested.

Absorption coefficient

All surface materials absorb sound to some degree and the rest is reflected or transmitted

(Figure A.I7). Materials that are specifically employed for the purpose of absorbing

sound are called “sound absorbing materials”. A standard method of rating the

effectiveness of a sound absorbing material is by its absorption coefficient. The

286
absorption coefficient varies with the frequency of sound. In other words, a surface

material does not absorb the sound similarly across frequencies. To enable

comprehensive analysis of absorption of different frequencies sound, absorptive qualities

of materials are reported across frequencies. Absorption coefficients of a material can

vary greatly across frequencies. However, it is likely that a reflective material has low

values of absorption coefficients compared to absorptive material. In architectural spaces,

values of absorption coefficient in six octaves ranging from 125Hz to 8kHz is typically

considered, though the range can vary in some instances.

Figure A.I7 Reflected, absorbed and transmitted sound (Mehta et al., 1999)

Automatic edge diffusion

Sound is capable of traveling around the corners. In wave-based acoustics, sound is

described as pressure waves. Bending of pressure waves around obstacles such as corner

of buildings and walls of interior spaces is known as “diffraction”. Reflection alone does

not account for all the indirect/reflected sounds at the receiver location. Diffraction also

contributes to qualities of the sound at the receiver location, particularly at the receiver

287
locations where the direct sound is not present (a.k.a. non-visual sound field). For

example, outdoors a person around the corner of a building can hear us even if there is no

reflection and direct sound. When sound diffracted in air at standard conditions, its speed,

frequency and wavelength generally remain unchanged, since the wave remains in the

same medium.

Recent developments in computer simulation research also suggest algorithms that can

predict diffraction caused by the object edges such as doors and windows or boundary

surfaces. For example, CATT Acoustics V8 offers an “automatic edge diffusion”

function which takes into account the impact of diffraction on sound quality at the

receiver locations. To emulate the diffusing effects of diffraction, a plane is assigned

automatic frequency dependent edge diffusion. Briefly, reflections falling within a

quarter of a wavelength from an edge are not specularly reflected but diffusely reflected

as shown in Figure A.I8.

Figure A.I8 Reflective surfaces considered by frequency dependent “automatic diffusion

function” of CATT (from CATT manual)

288
Number of rays

A user defined “number of rays” option is necessary to describe the amount of rays to be

randomly emitted by the sound source. It is one of the significant factors that affect

accuracy of predicted results. However, there are not commonly agreed solutions for

calculating number of rays. CATT Acoustics V8 has an option where the program

calculates the minimum number of rays needed for each particular acoustic model called

“auto number” function.

Truncation time

Truncation time represents the length of time that the sound rays will be traced. There is

not an agreed solution for calculating truncation time but it is suggested that truncation

time should be set to at least 2/3 of the reverberation time. Similar to the “number of

rays” option, CATT Acoustics V8 has an option where the program calculates the

minimum truncation time needed (when selected).

Geometry modeling

Geometry files (also known as “GEO” files) include necessary information to describe

the hall geometry in a text format. It also includes information about x, y, z coordinates

of the planes corners, surface absorption and diffusion properties. The hall geometry can

have any shape and can be composed of various planes (a max of 99,999 according to

CATT). Entry of the geometry data can be defined either by using a text editor (Figure

A.I9) or exporting the drawing from the AutoCAD interface (Figure A.I10). Once a set of

Auto-LISP procedures are introduced to AutoCAD, the entire acoustic model (including

289
all surface planes, source, receiver positions, plane names and absorption names) can be

generated in the Auto-CAD environment. The geometry information is stored in a

separate layer that can be exported as a GEO file compatible with CATT. This method

has been applied widely as well as the text editor method. Martin and Arana (2006)

indicated that they did not find any problem with this procedure.

Figure A.I9 CATT text editor interface

Figure A.I10 Auto-LISP commands used in Auto-CAD

290
It also needs to be noted that high levels of detail in the models do not necessarily

improve the accuracy of the results. In fact, Bork (2000) argues that very detailed

geometries could reduce the accuracy of the predicted results. Parallel with Bork`s (2002)

argument, Naylor and Rindel (1992) suggest that an acoustic model should replicate the

general forms of the room while avoiding unnecessary small surfaces and details.

However, according to Bradley and Wang (2002) the level of model detail on predicted

results does not affect the accuracy of predictions. In this study, related with these

findings, instead of detailed models, simplified models were used.

Validity of Acoustic Modeling Programs

Acoustic modeling programs have been widely used by acoustic consultants and

researchers, particularly for the acoustic analysis of rooms such as concert halls. The

validity of these acoustic modeling programs, in particular the ones using the hybrid

prediction method, in estimating the acoustic qualities of architectural spaces especially

concert halls has been widely studied.

Martin and Arana (2006) compared the results of the impulse response measurements

conducted in the new Symphony Hall in Spain with the predicted results by two acoustic

modeling programs (P1 and P2) that use hybrid prediction method (Figure A.I11). In

general the results provided by the two programs were similar but some differences

occurred between predicted reverberation time and Ts levels.

291
Figure A.I11 Three dimensional geometry of the Symphony Hall in Spain showing
source and receiver locations (from Martin, & Arana, 2006)

In general across different frequencies, measured reverberation time results were higher

compared to predicted results (Figure A.I12). Differences between predicted and

measured results were generally perceptible for reverberation time and other measures

such as Ts, C80, D50 and EDT based on the jnd thresholds suggested by ISO 3382.

Compared to P2, the reverberation time results provided by P1 were generally higher and

more similar to the measured results. Among the two programs, results provided by P1

followed a similar trend to the trend of the measured results. The key reason for such

differences is potentially related with different algorithms used in the programs such as

differently handled statistical corrections for late reflections and diffuse reflections. It

was not explicitly mentioned in the study but based on the description of the program

features, P1 was potentially CATT-Acoustics and the P2 was potentially ODEON. As

mentioned earlier CATT and ODEON are two of the major acoustic modeling programs

that use hybrid prediction method.

292
Overall, the study concluded that commercial acoustic modeling programs largely

coincide with field measured results despite perceptible differences. This conclusion is

only valid when we consider the similar trends followed by predicted and measured

results. A similar trend line analysis is also used in this study to compare the predicted

and measured results visually.

Figure 5.A.I12 Predicted vs. measured RT30 results at 500Hz and 1kHz (from Marin and
Arana 2006)

293
Various Round Robin studies were also conducted to assess the effectiveness of different

programs for room acoustics computer simulations. Vorlender (1995) compared the

predicted results and measured results for a speech auditorium. In total, predicted results

provided by 14 different acoustic modeling programs were considered. Each line in

Figure A.I13 represents the predicted results by different modeling programs and the

thick line represents the field measured results. The simulations were carried out by

different participants independently and most of them were software developers. The

results provided by only three programs using hybrid prediction methods were considered

as reliable. The differences between measured and predicted results provided by these

three programs were still perceptible based on ISO 3382 standards. Another important

finding of the study was the significant impact of absorption coefficients on accuracy of

the results.

Figure A.I13 Reverberation time levels predicted for the same hall by different
participants (from Vorlender, 1995)

294
Acoustic modeling of long enclosures

Various studies analyzed the effectiveness of acoustic simulation programs in predicting

the acoustics of long enclosures. These studies tested the applicability of different

prediction methods such as ray tracing (Yang, and Sheild, 2001) and ISM (Li & Lu,

2004; Li & Lu, 2005). The applicability of different prediction methods including ISM

and ray tracing methods has been generally validated for the acoustic analysis of long

enclosures particularly with rectangular floor-plate shapes. Interestingly, the applicability

of hybrid prediction programs that combines the best features of ISM and ray tracing

method has not been verified for the acoustic analysis of long enclosures. Additionally,

there are very few studies that document the behavior of sound in more complex long

enclosures, such as interconnected long enclosures with complex floor-plate shapes (e.g.,

long enclosures with branches). This study tested the effectiveness of a modeling

program that uses hybrid prediction method (CATT-Acoustics V8) in acoustic analysis of

both simple and complex long enclosures.

295
 APPENDIX J: JUST NOTICEABLE DIFFERENCE (JND)

Various studies have assessed the reliability of the predicted results via acoustic modeling

programs by comparing them with field measured results. The lower the difference

between predicted and field measure results, the better the accuracy/precision of the

predicted result is. This difference is desired to be as low as possible so that the human

auditory system will not perceive the difference. Various studies have been conducted to

assess the human auditory system`s ability to perceive the difference between different

levels of room-acoustics parameters. The unit of difference is defined as “subjective

limen” or “just noticeable difference (jnd)”. If the difference between two levels is less

than one subjective limen then the two levels are not perceptibly different. When

comparing the measured and predicted results, the most desirable condition is that this

difference is less than 1 subjective limen. According to draft international ISO/DIS 3382-

1 standard, subjective limen for reverberation time is 5% as shown in Table A.J1 (Bork,

2000).

296
Table A.J1 Just noticeable difference thresholds for different room acoustic parameters
(from ISO 3382).

However this allowable difference is very small therefore many studies could not obtain

results at this precision (Martin, Arana 2005; Shiokawa, & Rindel, 2007). Similarly,

program developers also concluded that it is highly difficult or even impossible to obtain

results that are within 1 limen range of the measured results (Odeon Tutorial).

The allowable error for reverberation time found by recent studies varied between 4.5%

and 10% (Karjalainen, 2001; Niaounakis, 2002). Even higher results were found by a

more recent study by Meng, Zhao and He (2006). Measured reverberation JND for

different subjects (e.g., audio technicians, common students and audio engineering

students) ranged between 21% and 39% (Table A.J2). However, this study used music

motifs to approximate the experience of the audiences in the concert halls. The

applicability of the results to hospitals is not clear.

297
Table A.J2 Just noticeable difference thresholds for different room acoustic parameters
(from Meng et al., 2006).

Overall, there is not a consensus for the reverberation JND. However, the majority of the

validation studies used jnd values suggested by ISO 3382-1 when assessing the difference

between measured and predicted results.

298
 REFERENCES

Aaron, J. N., Carlisle, M. A., Carskadon, C. C., et al. (1996). Environmental noise as a
cause of sleep disruption in an intermediate respiratory care unit. Sleep, 19, 707-
710.

Abdou, A. A. (2003). Measurement of acoustical characteristics of mosque in Saudi

Arabia. Journal of Acoustical Society of America, 113(3), 1505-1517.

Abel, J.S., Rick, J.W., Huang, P.P., Kolar, M.A., et al. (2008). On the acoustics of the
underground galleries of the ancient Chavin de Huantar, Peru. Presented at the
Acoustics` 08 Conference, Paris.

Acoustics Engineering. (2007). Measuring impulse response using dirac. Technical Note
(TN001). Available at www.acoustics-engineering.com (Accessed November
2010)

Aitken, R. J. (1982). Quantitative noise analysis in a modern hospital. Achieves of

Environmental Health, 37, 361-364.

Aiken, L. H., Clarke, S. P., Sloane, D. M., et al. (2002). Hospital nurse staffing and
patient mortality, nurse burnout, and job dissatisfaction. Journal of the American
Medical Association, 288, 1987-1993.

Aiken, L. H. (2005). Improving quality through nursing. In D. Mechanic, L. Rogut, D.

Colby, et al. (Ed.), Policy Challenges in Modern Health Care. New Brunswick,
NJ: Rutgers University Press.

American National Standards Institute. (1996). Measurement of occupational noise

exposure, ANSI S 12.12, 1996.

Aschoff, V. (1963). On spatial hearing. Arbeitsgem. f. Forschung Nordrhein-Westfalen.

138, 7-38.

Baker, C. F. (1984). Sensory overload and noise in the ICU: sources of environmental
stress. Critical Care Nursing Quarterly, 6, 66-80.

Baker, C. , Garvin, B., Kennedy, C., et al. (1993). The effect of environmental sound and
communication on CCU patients’ heart rate and blood pressure. Research in
Nursing Health, 16, 415–421.

Balogh, D., Kittinger, E., Benzer, et al. (1993). Noise in the ICU. Intensive Care
Medicine, 19 (1), 343-346.

299
Barling, J. (2001). Drowning not waving: Burnout and mental health nursing.
Contemporary Nurse, 11, 247-259.

Barron, M. (1993). Auditorium acoustics and architectural design: E & FN Spon.

Bennett, D., & Bion, J. (1999). Abc of intensive care: organisation of intensive care.
British Medical Journal, 318, 1468-1470.

Berenson, R. A. (1984). Intensive Care Units (ICUs): Clinical outcomes, costs and
decision making (Health Technology Case Study 28). Office of Technology
Assesment, U.S. Congress OTA-HCS-28.

Berglund, B., Lindvall, T. and , Schwela,D (ed). 1999. Guidelines for community noise.
Tech. Report, World Health Organization.

Bess, F. H., & Humes, L. E. (1990). Audiology: The Fundamentals. Baltimore: Williams
& Wilkins.

Bindl, L., Dresbach, K., & Lentze, M. J. (2005). Incidence of acute respiratory distress
syndrome in German children and adolescents: a population-based study. Critical
Care Medicine, 33, 209-212.

Blauert, J. (1997). Spatial Hearing. MIT Press.

Blauert, J. (2005). Analysis and synthesis of auditory scenes. In J. Blauert (Ed.).,

Communication Acoustics. Berlin: Springer-Verlag.

Blomkvist, V., Eriksen, C. A., Theorell, T., et al. (2005). Acoustics and psychosocial
environment in intensive coronary care. Journal of Occupational Environmental
Medicine, 62 (3), 1-8.

Blows, W. T. (2004). The Biological Basis of Nursing: Clinical Observations. London:

Routeledge.

Bork, I. (2000). A comparison of room simulation software- The 2nd Round robin on
room acoustical computer simulation. Acustica, 86, 943-956.

Braasch, J. (2005). Modeling of binaural hearing. In J. Blauert (Ed.), Communication

Acoustics. Berlin: Springer-Verlag.

Bradley, J. S., Sato, H., & Picard, M. (2003). The importance of early reflections for
speech intelligibility in rooms. Journal of Acoustical Society of Japan,113, 3233-
3244.

Bradley, D., & Wang, L. (2002). Effect of model detail level on room acoustic computer
simulations, Journal of Acoustical Society of America, 111, 2389(A).

300
Bregman, A. S. (1990). Auditory Scene Analysis: The Perceptual Organization of Sound.
Cambridge: The MIT Press.

Brokx, J. P. L., & Nooteboom, S. G. (1982). Intonation and the perceptual separation of
simultaneous voices. Journal of Phonetics, 10, 23-36.

Busch-Vishniac, I. J., West, J. E., Barnhill, C., et al. (2005). Noise levels in Johns
Hopkins Hospital. Journal of Acoustical Society of America, 118, 3629-3645.

Carayon, P., Smith, M. J., & Haims, M. C. (1999). Work organization, jobstress and work
related musculoskeletal disorders. Hum Factors, 41, 644-663.

Carnevale, F. A. (2009). Nursing care in the pediatric intensive care unit. In H. R.,
Wheeler, and T. P. Shanley, Science and Practice of Pediatric Critical Care
Medicine. London: Springer-Verlag.

CATT Tutorial.

Chambrin, M. C., Ravaux, I., Calvelo-Aros, D., et al. (1999). Multicentric study of
monitoring alarms in the adult intensive care unit: a descriptive analysis. Intensive
Care Medicine, 25, 1360-1366.

Chambrin, M. C. (2001). Alarm in the intensive care unit: how can the number of false
alarms be reduced? Critical Care, 5, 184-188.

Cheene, D. J. (2002). Modeling and auralization. In G. Ballou (Ed.), Handbook for Sound
Engineers. New York:Elsevier.

Cherry, E. C. (1953). Some experiments on the recognition of speech with one and with
two ears. Journal of Acoustical Society of America, 25, 975-979.

Chiang, W. (1994). Effects of Various architectural parameters on six room acoustical

measures in auditoria. University of Florida.

Christensen, M. (2005). What knowledge do ICU nurses have with regard to the effects
of noise exposure in the Intensive Care Unit? Intensive Critical Care Nursing,
21, 199-207.

Clarke, M. (1984).The constructs stress and coping as a rationale for nursing activities.
Journal of Advance Nursing, 9, 267-275.

Clarke, S. P., & Aiken, L. H. (2003). Failure to rescue. AJN 103:

Comité Européen de Normalisation (CEN)-European Standard. (1995). Electrically

generated alarm signals.

301
Cox, T. J., & D'Antonio, P. (2009). Acoustic absorbers and diffusers: theory, design and
application. New York: Spon Press.

Cropp, A., Woods, L., Raney, D., et al. (1994). Name that tone: The proliferation of
alarms in the intensive care unit. Chest, 105, 1217-1220.

Department of Health and Human Services (DHHS), National Center for Health
Workforce Analysis, Health Resources and Services Administration. (2002).
Supply, Demand, and Shortages of Registered Nurses, 2000-2020.

Diers, D. (2004). Speaking of Nursing: Narratives of Practice, Research, Policy, and the
Profession. Sudbury, MA: Jones and Bartlett.

Downes, J. J. (2009). Development of pediatric critical care medicine: how we get here
and why? In D. S. Wheeler, H. R. Wong, & T. P. Shanley (Ed), Science and
Practice of Pediatric Critical Care Medicine. London: Springer-Verlag.

EASERA Tutorial (2008).

ECRI Institute. (1986). Leaving ventilator dependent patients unattended. Health

Devices, 15(4), 102-103.

ECRI Institute. (2006). Patient safety reviews of medical technology: intensive care
environments. Available at
https://www.ecri.org/Products/Pages/Patient_Safety_Reviews.aspx (Accessed
November 2009).

ECRI Institute. 2007. Too many alarms can be hazardous to patients. Available at
https://www.ecri.org/Press/Pages/AlarmOverload.aspx (accessed November
2009).

Everest, F. A., & Pohlman, K. C. (2009). Master of handbook of acoustics: McGraw-Hill.

Fife, D., & Rappaport, E. (1976). Noise and hospital stay. Am. J. Public Health 66: 680–
681.

Gade, A. C. (1990). The influence of architectural design on the acoustics of concert

halls. Applied Acoustics, 31, 207-214.

Galindo, M., Zamarreno, T., and Giron, S. (2005). Acoustic analysis in Mudejar-Gothic
churches: experimental results. Journal of Acoustical Society of America, 117(5),
2873-2888.

Gardner, D. (1985). Presence. In G. Bulechek, & J. McCloskey (Ed.), Nursing

Interventions: Treatments for Nursing Diagnoses. Philadelphia: W.B. Saunders.

302
Gardner, M. B., & Gardner, R. S. (1973). Problem of localziation in the median plane:
effect of pinnae cavity occlusion. Journal of Acoustical Society of America, 45,
47-53.

Gardner, M.D. (1969). Distance estimation of 0-degree or apperant 0-degree oriented

speech signals in anechoic space. Journal of Acoustical Society of America , 53,
400-408.

Gast, P, & Baker, C. (1989). The CCU patient: Anxiety and annoyance to noise. Critical
Care Nursing Quarterly, 12, 39–54.

Glucksberg, S., & Cowen, J. G. N. (1970). Memory for non-attended auditory material.
Cognitive Psychology , 1, 149-156.

Gross, A. G. (2006). End-of-life care obstacles and facilitators in the critical care units of
a community hospital. Journal of Hospice and Palliative Nursing, 8, 92-1102.

Haan, C. H. & Fricke, F. R. (1995). Musician and music critic responses to concert hall
acoustics, Vol. Trondheim, Norway.

Hagerman, J., Rasmanis, G., Blomkvist, V., et al. (2005). Influence of intensive coronary
care acoustics on the quality of care and physiological state of patients.
International Journal of Cardiology, 98, 267–270.

Hartmann, W. M. (1983). Localization of sound in rooms. Journal of Acoustical Society

of America, 74, 1380-1391.

Hedley-Whyte, I. (1988). Monitoring and alarms philosophy and practice. Clinical

Anaesthesiology, 2, 379-389.

Hendrich, A., Chow, M. P., Skierczynski, B. A., et al. (2008). A 36-hospital time and
motion study: how do medical-surgical nurses spend their time?. The Permanente
Journal, 12, 23-34.

Hillman, K., & Bishop, G. (2004). Routine care of seriously ill. In Clinical Intensive
Care and Acute Medicine. New York: Cambridge University Press.

Hilton, B. A. (1985). Noise in acute patient care areas. Research in Nursing & Health, 8,
283-291.

Hinchliff, S. M., Montague, S. E., & Watson, R. (1988). Physiology for Nursing Practice.
London: Bailliere Tindall.

Hirose, M., Sato, E., Taguchi, M., et al. (2005). Characteristics of auditory alarms for
medical equipment and future issues. Journal of Clinical Engineering, 30(4), 208-
213.

303
Hodge, B., & Thompson, J.F. (1990). Noise pollution in the operating theatre. Lancet,
335 (8694), 891-894.

Houtgast, T., & Plomp, R. (1968). Lateralization threshold of a signal in noise. Journal of
Acoustical Society of America, 44, 807-812.

Imaizumi, H., Kunimatsu, S., & Isei, T. (2000). Sound propogation and speech
transmiision in a branching underground tunnel. Journal of Acoustical Society of
America, 108 (2), 632-42.

Ingard, U. (1953). A review of the influence of meteorological conditions on sound

propogations. Journal of Acoustical Society of America, 25, 405-411.

ISO/DIS 3382-1: Acoustics - Measurement of room acoustic parameters – Part 1:

Performance rooms, International Organization for Standardization (2006).

ISO 140. Acoustics – Measurement of sound insulation in buildings and of building

elements, all parts, International Organization for Standardization.

James, W.P., & Tatton-Brown, W. (1986). Hospitals: Design and Development. London:
The Architectural Press.

Joint Commission on the accrediation of healthcare (JCAHO). (2003). JCAHO national

patient safety goals: practical strategies and helpful solutions for meeting these
goals. Joint Commission Perspectives on Patient Safety 3(1).

Joseph, A., & Rashid, M. (2007). The architecture of safety: hospital design. Current
Opinion on Critical Care, 13,714-719.

Kahn, D. M., Cook, T. E., Carlisle, C. C., et al. (1998). Identification and modification of
environmental noise in an ICU setting. Chest, 114, 535-540.

Kalish, B. J. (2005). Missed nursing care: a qualitative study. Journal of Nursing Care
Quality, 21, 306-313.

Kang, J. (2002a). Basic acoustic theories and modeling techniques. In Acoustics of Long
Enclosures. Thomas Telford:London.

Kang, J. (2002b). Acoustic theories of long spaces. In Acoustics of Long Enclosures.

Thomas Telford:London.

Kang, J. (2002c). Parametric study: design guidance afor long enclosures. In Acoustics of
Long Enclosures. Thomas Telford:London.

Kang, J. (2002d). Parametric study: design guidance for urban streets. In Acoustics of
Long Enclosures. Thomas Telford:London.

304
Karjalainen, M., & Jarvelainen, H. (2001). More about this reverberation science:
perceptually good late reverberation. 111th AES Conference, New York.

Kerr, J. H., & Hayes, B. (1983). An alarming situation in the intensive therapy unit.
Intensive Care Medicine, 9, 103-104.

Kersten, A., Milbrandt, E. B., Rahim, M. T., et al. (2003). How big is critical care in the
US? . Critical Care Medicine, 31, A8.

Klosak, A. K., & Gade, A. C. (2008). Relationship between room shape and acoustics of
rectangular concert halls. Paper presented at the Acoustics`08, Paris.

Kracht, J. M., Busch-Vishniac, I. J. & West, J. E. (2007). Noise in the operating rooms
of Johns Hopkins hospital. Journal of Acoustical Society of America, 121, 2673-
2680.

Kuhl, W. (1969). Differences between hearing in a room and electroacoustic

transmission. Rundfunktech. Mitt, 13, 205-208.

Kutruff, H. (2000). Room Acoustics. New York: Elsevier.

Kuwano, S., Namba, S., & Mizunami, T. (2001). Desirable properties for identification of
auditory warning signals. Acoustical Science and Technology, 22, 311-314.

Lambert, B., Chang, K., & Lin, S. (2001). Effect of orthographic and phonological
similarity on false recognition of drug names. Social Science & Medicine, 52(12),
1843–1857.

Laschinger, H. K., & Leiter, M. P. (2006). The impact of nursing work environments on
patient safety outcomes: The mediating role of burnout/ engagement. Journal of
Nursing Administration, 36, 259-267.

Laws, P. (1972). On the problem of distance hearing and the localization of auditory
events inside the head. Technishe Hochschule, Aachen.

Lawless, S. T. (1994). Crying Wolf: False alarms in pediatric intensive care unit. Critical
Care Medicine, 22, 981-985.

Lerche, E., & Plath, P. (1961). On the localization of sound sources with headphones.
Pflugers Archiv, 274, 91.

Li, K.M., & Lu K.K. (2004). Propagation of sound in long enclosures. Journal of
Acoustical Society of America,116 (5), 2759-2770.

305
Li, K.M., & Lu K.K. (2005). Prediction of reverberation time and speech transmission
index in long enclosures. Journal of Acoustical Society of America, 117 (6), 3716-
3726.

Litovsky, R.Y., Colburn, H.S., Yost, W.A., & Guzman, S.J. (1999). The precedence
effect. Journal of American Society of America. 106, 1633-1654.

Long, M. (2006). Architectural Acoustics. Burlington, MA: Elsevier Academic Press.

Lorenzi, C., Gatehouse, S., & Lever, C. (1999). Sound localization in noise in normal-
hearing listeners. Journal of Acoustical Society of America, 105, 1810-1820.

Love, H. (2003). Noise exposure in the orthopaedic operating theatre: a significant health
Hazard. ANZ Journal of Surgery, 73 (10), 836-838.

Lui, J.C., & Lu P. (2009a). Study on sound propogation in long enclosures with a vertical
or inclined branch of different dimensions. Applied Acoustics, 70, 522-29.

Lui, J.C., & Lu P. (2009b). The effect of multiple branches on sound propogation in long
enclosures. Applied Acoustics, 70, 894-901.

Lundeby, A., & Vigran, T. E. (1999). Uncertainties of Measurements in Room Acoustics.

Acustica, 81, 344-355.

Martin, R.S., & Arana, M. (2006). Predicted and experimental results of acoustic
parameters in the new Symphony Hall Pamplona, Spain. Applied Acoustics, 67, 1-
14.

McCance, K. L., & Huether, S. E. (2002). Pathophysiology: the biological basis for
disease in adults and chidlren. St Louis: Mosby.

Makous, J. C., & Middlebrooks, J. C. (1990). Two-dimensional sound localization by

human listeners. Journal of Acoustical Society of America, 87, 2188-2200.

MacKenzie, D. J., & Galbrun, L. (2007). Noise levels and noise sources in acute care
hospital wards. Building Services Engineering Research and Technology, 28, 117-
131.

McLaughlin, A., McLaughlin, B., Elliott, J., et al. (1996). Noise levels in a cardiac
surgical intensive care unit: A preliminary study conducted in secret. Intensive
Critical Care Nursing, 12, 226-230.

Mehta, M., Johnson, J., & Rocafort, J. (1999). Architectural Acoustics. New Jersey:
Prentice-Hall, Inc.

306
Meng, Z., Fengjie, Z., & He M. (2006). The just noticeable difference of noise length and
reverberation perception. IEEE: T2A-3.

Meredith, C., & Edworthy, J. (1995). Are there too many alarms in the intensive care
unit?: An overview of problems. Journal of Advanced Nursing, 21, 15-20.

Meyer, T. J., Eveloff , S. E., Bauer, M. S., et al. (1994). Adverse environmental
conditions in the respiratory and medical ICU settings. Chest, 105, 1211-1216.

Middlebrooks, J. C., & Green, D. M. (1991). Sound localization by human listeners.

Annual Review of Psychology, 42, 135-159.

Miller, R. L., & Swensson, E. S. (2002). Hospital and Healthcare Facility Design. New
York, NY: W.W Norton.

Mills, A. W. (1958). On the minimum audible angle. Journal of Acoustical Society of

America, 30, 237-246.

Minkley, B. (1986). A study of noise and its relationship to patient discomfort in the
recovery room. Nursing Research, 17, 247–250.

Monks, M., Oh, M., & Dorsey, J. (1996). Acoustic Simulation and Visualization using a
New Unified Beam Tracing and Image Source Approach. Paper presented at the
Proceedings of the Convention of the Audio Engineering Society

Moore, B. C. J. (1997). An Introduction to the Psychology of Hearing. London: Elsevier

Academic Press.

Moore, M. M., Nguyen, D., Nolan, S. P., et al. (1998). Interventions to reduce decibel
levels on patient care units. The American Journal of Surgery, 64, 894-899.

Moray, N. (1970). Attention: Selective Processes in Vision and Hearing. New York:
Academic Press.

Moorthy, K.. Munz, Y., Undre, S., et al. (2004). Objective evaluation of the effect of
noise on the performance of a complex laparoscopic task. Surgery, 136, 25-30.

Morrison, W., Haas, E., Shaffner, D., et al. (2003). Noise stress and annoyance in a
pediatric intensive care unit. Critical Care Medicine, 31, 113–119.

Murthy, V. S., Malhotra, S. K., Bala, I., & Raghunathan, M. (1995). Auditory functions
in anaesthesia residents during exposure to operating room noise. Indian Journal
of Medical Research, 101, 213-216.

Newby, H. A. (1979). The handicap of hearing impairment. In H. A. Newby (Ed.),

Audiology. Englewood Cliffs: Prentice-Hall.

307
National League for Nursing Accrediting Commission (NLNAC). (1933). A concept of
nursing. American Journal of Nursing, 33, 6.

Naylor, G., & Rindel, J. (1992). Predicting room acoustical behavior with the ODEON
computer model. Journal of Acoustical Society of America, 92, 2346(A).

Niaounakis, T.I., & Davies, W.J. (2002). Reverberation time perception in small listening
rooms. Journal of the Audio Engineering Society. 50, 343-50.

ODEON Tutorial.

Occupational Safety and Health Administration (OSHA). Occupational Safety and Health
Standards. Available at
http://www.osha.gov/pls/oshaweb/owasrch.search_form?p_doc_type=STANDAR
DS&p_toc_level=1&p_keyvalue=1910. (Accessed March 2011).

Owen, J. (1982). Patient observation. Nursing Mirror, 154(11): vii-x.

Paolo, L., Athanase, B., & Stevo, J. (2006). The impact of increased heart rate on clinical
outcomes in hypertension: implications for antihypertensive drug therapy. Drugs,
66, 133-144.

Park, S., Song, H., Han, J., et al. (1994). Effect of noise on the detection of rib fractures
by residents. Investigative Radiology, 29, 54-58.

Parthasarathy, S., & Tobin, M. (2004). Sleep in the intensive care unit. Intensive Care
Medicine, 30, 197–206.

Patterson, R. D., & Mikoy, R. (1980). Auditory warnings on civil aircraft the leaming
and retention of warnings. Prepared for Civil Aviation Authonty 7D/S/0142.

Pelorson, X., Vian, J.-P., & Polack, J.-D. (1992). On the variability of room acoustical
parameters: reproducibility and statistical validity. Applied Acoustics, 37, 175-
198.

Phillips, J. A., & Barnsteiner, J. H. (2005). Clinical alarms: improving efficiency and
effectiveness. Critical Care Nursing Quarterly, 28, 317-323.

Plack, C. J. (2005). The Sense of Hearing. London: Lawrence Erlbaum.

Plenge, G., & Brunschen, G. (1971). A priori knowledge of the signal when determining
the direction of speech in the median plane. Presented at the 7th Int Congr on
Acoustics, Budapest.

Rashid, M. (2006). A decade of adult intensive care unit design. Critical Care Nursing
Quarterly, 29 (4) , 282-311.

308
Rayleigh, L. (1907). On our perception of sound direction. Philosophical Magazine, 13,
214-232.

Richardson, W. (2004). Critical alarm system testing. Association of periOperative

Registered Nurses, 80, 280-294.

Rindel, J., Shiokawa, H., Christensen, C., et al. (1999). Comparisons between computer
simulations of room acoustical parameters and those measured in concert halls.
Presented at the Joint Meeting of the Acoustical Society of America and the
European Acoustical Association.

RPG Diffuser Systems. Available at http://www.rpginc.com/research/reverb01.htm

(Accessed Nowember 2010)

Ryherd, E. E., Persson Waye, K., & Ljungkvist, L. (2008). Characterizing noise and
perceived work environment in a neurological intensive care unit. Journal of
Acoustical Society of America, 123, 747-756.

Russell, L. B. (1979). Intensive care. In Technology in Hospitals: Medical Advances and

Their Diffusion. Washington, DC: The Brookings Institution.

Safar, P., & Grenvik, A. (1971). Critical care medicine: organizing and staffing intensive
care units. Chest , 59, 535.

Samuels, S. I. (1986). An alarming problem. Anaesthesiology, 62, 128.

Sandel, T. T., Teas, D. C., Feddersen, W. E., et al. . (1955). Localization of sound from
single and paired sources. Journal of Acoustical Society of America, 27, 842-852.

Schmidth, S. I., & Baysinger, C. L. (1986). Alarms: help or hindrance?. Anesthesiology,

64, 654.

Schroeder, M. R., Gottlob, D., & Siebrasse, K. F. (1974). Comparative study of European
concert halls: correlation of subjective preference with geometric and acoustic
parameters. Journal of Acoustical Society of America, 56(4), 1195-1201.

Shpuza, E., Peponis, J. (2008). The effect of floor plate shape upon office layout
integration. Environment and Planning B: Planning and Design, 35, 318-336.

Society of Critical Care Medicine (SCCM). 2006. Critical care statistics in the United
States. Available at http://sccmcms.sccm.org/NR/rdonlyres/1FCD55C6-71B0-
48CA-B652-31EBB61DD202/504/WebStatisticsPamphletFinalJune07.pdf
(accessed November 2009).

Smith, H. (2004). Geometric acoustic modeling of the LDS Conference Center. Master
Thesis. Department of Physics and Astronomy, Brigham Young University

309
Soni, N., & Chamberlain, S. K. (1999). Monitoring. Best Practice & Research Clinical
Anaesthesiology, 13, 307-319.

Spieth, W., Curtis, J. F., & Webster, J. C. (1954). Responding to one of two simultaneous
messages. Journal of Acoustical Society of America, 26, 391-396.

Stanford, L. M., Mclntyre, J. W. R., & Hogan, J. T. (1985). Audible alarm signals for
anaesthesia monitoring equipment. International Joumal of Clinical Monitoring
and Computing 1: 251-256.

Steinhauser, A. (1877). The theory of binaural edition. Philosophical Magazine, 7, 181-

197 & 261-274.

Stifelman, L. J. (1994). The cocktail party effect in auditory interfaces: a study of

simultaneous presentation. Technical report prepared for MIT Media Laboratory.

Tsakostas, C. (2004). Image Receiver Model: An efficient variation of the 'Image Source
Model' for the case of multiple sound sources and a single receiver. Paper
presented at the Conference of Hellenic Institute Of Acoustics in Thessaloniki.

Thomas, K.A., & Martin. P.A. (2000). NICU sound environment and the potential
problems for caregivers. Journal of Perinatal Medicine, 20(8), 94-99.

Treisman, A. M. (1964). The effect of irrelevant material on the efficiency of selective

listening. The American Journal of Psychology, LXXVII, 533-546.

Topf, M. (1988). Noise-induced occupational stress and health in critical care nurses.
Hospital Topics, 66, 30–34.

Topf, M. (1989). Sensitivity to noise, personality hardiness, and noise-induced stress in

critical care nurses. Environment and Behavior, 21, 717-733.

Topf, M., & Dillon, E. (1988). Noise induced stress as a predictor of burnout in critical
care nurses. Heart and Lung, 17, 567-574.

Ulrich, B. T., Lavandero, R., Hart, K. A., et al. (2009). Critical care nurses’ work
environments. Critical Care Nurse, 29, 93-102.

U.S.Pharmacopeia. (1997). Similar names. USP DI Update.Vol I & II:4-7.

Vorlander, M. (1995). International round robin on room acoustical computer

simulations. Paper presented at the 15th International congress on acoustics,
Trondheim, Norway.

310
Yang, L.N., & Shield, B.M. (2001). The prediction of speech intelligibility in
underground stations of rectangular cross section. Journal ofAmerican Society of
America,109, 266-73.

Wallace, M. S., Ashman, M. N., & Matjasko, M. J. (1994). Hearing acuity of

anesthesiologists and alarm detection. Anesthesiology, 81, 13-28.

Wallach, H. (1949). On sound localization. Journal of Acoustical Society of America, 10,

270-274.

Webster, J. C., & Thompson, P. O. (1954). Responding to both of two overlapping

messages. Journal of Acoustical Society of America, 26, 396-402.

Williams, A. L., Drongelen, W. & Lasky, R. E. (2007). Noise in contemporary neonatal

intensive care. Journal of Acoustical Society of America, 121, 2681-2690.

World Health Organization. (1999). Guidelines for community noise. B. Berglund and T.
Lindvall (Ed.), Technical Report. Geneva, Switzerland.

Zhang, Y. (2005). A method to predict reverberation time in concert hall preliminary

design stage. Unpublished PhD Thesis, Georgia Institute of Technology, Atlanta.

311
 VITA

Selen Okcu was born in Turkey. She received both her Master’s degree in Design

Computing and Bachelor`s degree in Architectural Design from Istanbul Technical

University, College of Architecture. In her professional life she has been deeply involved

in multi-disciplinary research and design teaching. She is currently an active member of

different research groups at Georgia Institute of Technology including Evidence Based

Design Research Group, Healthcare Acoustics Research Group and Health Systems

Institute. As a researcher, she collaborated with researchers from different disciplines

including medicine, psychology, engineering and architecture. She participated in various

sponsored research projects. Her research experience ranges from qualitative and

quantitative assessment of physical environment qualities and their impact on user

outcomes such as assessment of influences of innovative workplace design solutions on

employee outcomes, to combined effects of job stress and workplace noise on

employee`s job satisfaction, job commitment and health outcomes. In 2009, she was

awarded a Newman Medal from the Acoustical Society of America for excellence in the

study of architectural acoustics. She has received more awards from professional

organizations for her acoustics research including the Martin Hirshorn IAC Prize,

ASHRAE Graduate Student Award. Recently, her research that relates layout design to

acoustic outcomes has been awarded by Boston Society of Architects. She has been also

awarded for her research on high performance school buildings and user outcomes by

Emory University. She is also a member of different professional groups including

Acoustical Society of America, Institute of Noise Control Engineering, American Society

312
of Heating, Refrigerating and Ventilation Engineering, Society of Women Engineers, and

US Green Building Council.

313