Open access Original research

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
Helicopter and ground emergency medical services
transportation to hospital after major trauma in
England: a comparative cohort study
Oliver Beaumont ‍ ‍ ,1,2 Fiona Lecky,3,4 Omar Bouamra,3 Dhushy Surendra Kumar,5
Tim Coats,6 David Lockey,7 Keith Willett8

1
Clinical Academic Graduate ABSTRACT
School, Oxford University, Background The utilization of helicopter emergency What is already known on this subject
Oxford, UK
2
Department of Trauma and medical services (HEMS) in modern trauma systems has
been a source of debate for many years. This study set to ►► UK prehospital trauma care has evolved rapidly
Orthopaedics, Bristol Royal
Infirmary, Bristol, UK establish the true impact of HEMS in England on survival over the last decade, including the extensive
3
Trauma Audit Research for patients with major trauma. utilization of helicopter emergency medical
Network, University of services (HEMS) services across the major
Methods A comparative cohort design using
Manchester, Manchester, UK trauma networks.
4
Care for Urgent and Emergency prospectively recorded data from the UK Trauma Audit
Care Research (CURE), Health and Research Network registry. 279 107 patients were ►► Due to the nature of prehospital care, there
Services Research Section, identified between January 2012 and March 2017. The are very few randomized trials comparing
School of Health and Related primary outcome measure was risk adjusted in-­hospital HEMS with ground emergency medical services
Research (ScHARR), University (GEMS).
of Sheffield, Sheffield, UK mortality within propensity score matched cohorts
using logistic regression analysis. Subset analyses were ►► A Cochrane review of a number of
5
Department of Critical Care,
Anaesthesia and Pre-­hospital performed for subjects with prehospital Glasgow Coma observational studies concluded that due to
Emergency Medicine, University Scale <8, respiratory rate <10 or >29 and systolic blood methodological weaknesses, the benefit of
Hospital Coventry, Coventry, UK HEMS could not be determined.
pressure <90.
6
Emergency Medicine Academic
Group, University of Leicester, Results The analysis was based on 61 733 adult
Leicester, UK patients directly admitted to major trauma centers:
7
Department of Trauma Sciences, 54 185 ground emergency medical services (GEMS) What this study adds
Blizard Institute, Queen Mary and 7548 HEMS. HEMS patients were more likely
University of London, London,
UK male, younger, more severely injured, more likely to be
►► Rigorous statistical evaluation of prospectively
8
Kadoorie Research Centre, victims of road traffic collisions and intubated at scene.
recorded data from 279 107 patients using
Nuffield Department of Crude mortality was higher for HEMS patients. Logistic
Orthopaedics, Rheumatology propensity score, multiple logistic regression
regression demonstrated a 15% reduction in the risk
and Musculoskeletal Science, and sensitivity analysis.
adjusted odds of death (OR=0.846; 95% CI 0.684 to
University of Oxford, Oxford, UK ►► Our analysis demonstrates a 15% risk-­adjusted
1.046) in favor of HEMS. When analyzed for patients
mortality reduction (OR=0.846; 95% CI 0.684
Correspondence to previously noted to benefit most from HEMS, the odds
to 1.046) for a comparable cohort of severely
Mr Oliver Beaumont, Clinical of death were reduced further but remained statistically
injured patients transported by HEMS versus
Academic Graduate School, consistent with no effect. Sensitivity analysis on 5685
Oxford University, Oxford, UK; ​ GEMS; however, this did not reach statistical
patients attended by a doctor on scene but transported
oliver.​beaumont@​doctors.​org.​uk significance.
by GEMS demonstrated a protective effect on mortality
►► Subset sensitivity analysis demonstrated a
Received 11 May 2020 versus the standard GEMS response (OR 0.77; 95% CI
protective effect of doctors on scene when
Revised 14 June 2020 0.62 to 0.95).
patients were transferred to hospital by GEMS
Accepted 15 June 2020 Discussion This prospective, level 3 cohort analysis
(OR 0.77; 95% CI 0.62 to 0.95, p=0.015).
demonstrates a non-­significant survival advantage for
patients transported by HEMS versus GEMS. Despite the
large size of the cohort, the intrinsic mismatch in patient
trauma in England is £300–£400 million.2 In 2012,
demographics limits the ability to statistically assess
trauma services in England, including prehospital
HEMS true benefit. It does, however, demonstrate an
care, underwent reconfiguration into regional
improved survival for patients attended by doctors on
trauma networks (RTNs) to address well-­reported
scene in addition to the GEMS response. Improvements
deficiencies in trauma care.3–6 Thirty major trauma
in prehospital data and increased trauma unit reporting
centers (MTCs) are at the hubs of these networks
© Author(s) (or their are required to accurately assess HEMS clinical and cost-
providing 24-­hour, multidisciplinary consultant-­led
employer(s)) 2020. Re-­use effectiveness.
permitted under CC BY-­NC. No care for severely injured patients. One hundred and
commercial re-­use. See rights twenty other hospitals are designated as trauma
and permissions. Published units (TUs) with the facility to assess, resuscitate
by BMJ. INTRODUCTION and stabilize critically injured patients prior to
To cite: Beaumont O, Trauma is a leading cause of premature death and transfer to an MTC if needed.
Lecky F, Bouamra O, et al. disability in England, with over 16 000 deaths annu- All UK regions have air ambulance provision.
Trauma Surg Acute Care Open ally and an average of 36 life-­years lost per trauma These are increasingly staffed with a doctor-­
2020;5:e000508. death.1 The estimated annual cost of treating major paramedic crew. They provide a number of critical
Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508 1
 Open access

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
care interventions for time critical conditions (eg, airway compro- HEMS services may use either incident characteristics, initial
mise, ventilatory failure and bleeding) and allow rapid accompa- reports of casualty’s condition or logistic information (such as
nied transfer to appropriate hospitals.7–9 Transfer from scene to distribution of alternative response assets and distance to an
hospital may be by air or in a ground ambulance. Ground emer- MTC) to try to predict cases in which better care might be given
gency medical services (GEMS) do not routinely have doctors as using helicopter transport to provide either additional specialist
crew or possess the skillset to perform the same level of critical clinical skills at the incident scene or more rapid transport direct
interventions such as rapid sequence induction (RSI). to a specialist care center.
The utilization of helicopter emergency medical services This study aims to compare survival outcomes for patients
(HEMS) in modern trauma systems has been a source of debate sustaining major trauma who were delivered to hospital by
for a number of years, notably due to the increased costs HEMS or GEMS following the implementation of RTNs within
compared with GEMS and conflicting reports on outcomes.9–15 England. Our null hypothesis is that risk-­adjusted survival does
Accurate estimation of the clinical benefit of HEMS compared not differ significantly between major trauma patients trans-
with GEMS is limited by the low quality of the available evidence ported to hospital by HEMS versus GEMS in England.
and heterogeneity of study methodologies in the literature.16 17
HEMS dispatch is likely of most benefit to those with more
severe injuries. Patients with abnormal prehospital variables such METHODS
as respiratory rate (RR) and Glasgow Coma Scale (GCS) have The study was conducted by using patient data from the Trauma
been shown to have significant survival advantage when attended Audit and Research Network (TARN)—the largest trauma
by HEMS services.18 However, a high proportion of HEMS call-­ registry in Europe, holding data from all trauma receiving
outs are for non-­life threatening, less serious injuries.19 hospitals in England. An online electronic data collection and
Although a doctor-­ paramedic crew is the most common reporting system has been in use since 2005. Validation proce-
configuration in the UK, significant variation exists in the dures check for accuracy in date/time sequencing, physiological
configuration and staffing of HEMS services, limiting the accu- measurements and investigations. Coders are trained in injury
rate identification of capabilities.20 There is also no single set of coding, and their work is subjected to validation and internal
dispatch criteria for a HEMS response. In the dispatch process, quality checks on a weekly basis. This study did not directly

Table 1 Characteristics of the directly admitted adult population (MTCs only) by mode of arrival: January 2012–March 2017
Mode of arrival
GEMS HEMS Total Unknown final outcome
Total 54 185 7548 61 733 5748
Male, n (%) 31 487 (58.1) 5619 (74.4) 37 106 (60.1) 3891 (67.7)
Age, median (IQR) 61.9 (40.7–80.9) 45.5 (29–60.3) 59.3 (38.5–79.4) 53 (34.5–71.6)
Prehospital GCS, median (IQR)* 15 (14–15) 15 (10–15) 15 (14–15) 15 (11–15)
Prehospital SBP, median (IQR)* 137 (120–155) 129 (110–146) 136 (119–154) 132 (114–152)
ISS 10 (9–20) 20 (10–29) 10 (9–21) 21 (10–29)
Prehospital O2sat, median (IQR)* 97 (95–99) 98 (95–100) 97 (95–99) 97 (95–99)
Prehospital pulse rate, median (IQR)* 84 (72–98) 88 (72–104) 84 (72–99) 86 (73–102)
Prehospital resp rate, median (IQR)* 18 (16–20) 20 (16–24) 18 (16–22) 18 (16–22)
Time from leaving scene to arrival to hospital (mins), median (IQR) 24 (17–34) 22 (16–30) 24 (17–34) 25 (17–36)
Penetrating injury, n (%) 2459 (4.5) 363 (4.8) 2822 (4.6) 162 (2.8)
MOI, n (%)  
 Vehicle incident/collision 13 584 (25.1) 4560 (60.4) 18 144 (29.4) 2494 (43.4)
 Fall from heights 8527 (15.7) 1387 (18.4) 9914 (16.1) 1368 (23.8)
 Low fall 26 303 (48.5) 644 (8.5) 26 947 (43.7) 1286 (22.4)
 Stabbing/shooting 2195 (4.1) 294 (3.9) 2489 (4) 130 (2.3)
 Other 3576 (6.6) 663 (8.8) 4239 (6.9) 470 (8.2)
Prehospital intubation, n (%) 1106 (2) 2115 (28) 3221 (5.2) 744 (12.9)
LOS in days, median (IQR) 10 (5–19) 10 (5–20) 10 (5–19) 12 (4–24)
Neuro center on site, n (%)† 49 153 (90.7) 7357 (97.5) 56 510 (91.5) 4528 (78.8)
Head injury AIS 3+, n (%) 14 481 (26.7) 2703 (35.8) 17 184 (27.8) 2487 (43.3)
Attendance at scene, n (%)
 Consultant 0 (0) 476 (6.3) 476 (0.8) 115 (2)
 Doctor 0 (0) 3438 (45.5) 3438 (5.6) 749 (13)
 Paramedic only 48 962 (90.4) 2834 (37.5) 51 796 (83.9) 4232 (73.6)
 Mortality 4559 (8.4) 1181 (15.6) 5740 (9.3) N/A
*Prehospital characteristics are taken from the earliest recorded data.
†Neurosurgery/neurological care facility.
AIS, Abbreviated Injury Scale; GCS, Glasgow Coma Scale; GEMS, ground emergency medical services; HEMS, helicopter emergency medical services; ISS, Injury Severity Score; LOS,
length of stay; MOI, mechanism of injury; MTCs, major trauma centers; O2sat, oxygen saturation; SBP, systolic blood pressure.

2 Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508

 Open access

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
years old) trauma patients directly admitted to an MTC for the
period between January 2012 and March 2017.
Patients were grouped according to the vehicle mode of arrival
to an MTC—HEMS by helicopter and GEMS by ground ambu-
lance. The GEMS patients with a presence of a doctor on scene
were excluded as this is not a standard GEMS response as the
doctor might have been conveyed by helicopter to the scene.
This cohort were analyzed in a separate sensitivity analysis.
Some air ambulances attend patients with paramedic only crews.
Although these crews do not provide the same level of clinical
intervention as a doctor-­paramedic crew, they were included in
the HEMS group as they include the helicopter asset and these
paramedics often have additional training and additional skills.

Missing data
Missing data were present in all prehospital vital signs such
as systolic blood pressure (SBP), RR, heart rate (HR), oxygen
saturation (O2sat) and GCS. To overcome the bias created by
missing data, an imputation procedure was carried out assuming
that the mechanism of missingness is at random and that is
Figure 1 Flow chart of the study population. EMS, emergency medical the missing value depends on measured variables. The rate of
services; GEMS, ground emergency medical services; HEMS, helicopter missing values varied from 6% for GCS to 12% for O2sat.
emergency medical services; MTC, major trauma center; PS, propensity The imputation was carried out using the Stata procedure ‘mi
score; TARN, Trauma Audit and Research Network. impute’ (StataCorp V.14, 2015), which creates five imputed sets.
The procedure requires multivariate normality of the variables
used in the imputation model, because all of the imputed vari-
include Patient and Public Involvement (PPI), but the database ables were not normally distributed, they were all transformed
used was developed with regular PPI. into normal scores for the imputation, then transformed back to
Patients of all ages are included on the TARN database if they their original scale.21
sustain injury resulting in any of: admission to hospital for 3
days or longer, intensive or high dependency care, interhospital Propensity score
transfer for further care or death in hospital. Patients aged over In observational studies, a direct comparison of the effect of heli-
65 years with an isolated fracture of the femoral neck or pubic copter and ground emergency services (HEMS and GEMS) on
ramus and those with isolated closed limb injuries (excepting the outcome would not be appropriate even with case-­mix adjust-
femoral shaft/condyles) are excluded. ment. This is because the exposed subjects (HEMS) are system-
Prospectively recorded data were used in the current cohort atically different from the unexposed subjects (GEMS) in both
study that includes eligible patients of all ages presenting with measured and unmeasured baseline characteristics. The estima-
blunt or penetrating trauma, submitted to TARN and injured tion of the propensity score for HEMS transportation is carried
between 1 January 2012 and 31 March 2017 and not further out using the ‘psmatch’ procedure in Stata with the following
limited by severity. independent variables: prehospital vital signs (SBP, RR, HR and
The data completeness, which is the ratio of the number of GCS), admission to MTC, intubation at scene, age, gender, most
submitted cases to TARN to the expected number of cases from severe injury in body region and entrapment at scene. These are
the hospital episode statistics (28) database, is 95.4% for MTCs related to likelihood of HEMS dispatch and trauma outcome. To
and 62.3% for TUs. The study population consists of adult (≥16 balance the covariates between the two groups, 1-­to-1 propen-
sity score matching technique was used that consists of pairing
subjects without replacement within a specified distance (caliper)
Table 2 Missing data with an absolute value of 0.05 based on the difference of their
Missing Observed % missing
respective propensity score. The propensity scores were aver-
aged over the imputed sets to create the matched pairs. Though
Prehospital
there is uncertainty on the best methods to perform propensity
 GCS 3696 58 037 6.0
matching, this methodology reflects that performed in previous
 SBP 6472 55 261 10.5 studies and has been proven to effectively reduce bias.22 Matched
 Resp rate 6444 55 289 10.4 pairs of subjects were obtained, and the balance of their covari-
 Pulse rate 4752 56 981 7.7 ates was assessed using standardized differences within 0.1.
 Oxygen saturation 7670 54 063 12.4 Standardized differences are preferred to p values as the latter
are more sensitive to sample size. Standardized differences are
Emergency department
defined as the difference of the mean values of the covariates
 GCS 5074 56 659 8.2
divided by their pooled variances.
 SBP 2902 58 831 4.7
 Resp rate 7818 53 915 12.7
Statistical analysis
 Pulse rate 2531 59 202 4.1 The outcome of interest was in-­hospital mortality, and because
 Oxygen saturation 4042 57 691 6.5 of the matched pair design, a random effect logistic regression
GCS, Glasgow Coma Scale; SBP, systolic blood pressure. was used to evaluate the effect of HEMS and GEMS on outcome.
Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508 3
 Open access

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
Table 3 Characteristics of the propensity score matched directly admitted adult population (MTCs only) by mode of arrival: January 2012–March
2017
Mode of arrival
GEMS HEMS Total
Total 4636 4636 9272 P value
Male, n (%) 3442 (74.3) 3350 (72.3) 6792 (73.37) 0.031
Age, median (IQR) 47.2 (30–62.8) 46.8 (30.7–62.7) 47.2 (29.6–62.8) 0.516
Prehospital GCS, median (IQR) 15 (14–15) 15 (15–15) 15 (14–15) 0.999
Prehospital SBP, median (IQR) 130 (115–147) 130 (113–148) 130 (114–147) 0.999
ISS 16 (9–25) 17 (9–26) 16 (9–25) 0.0001
Prehospital O2sat, median (IQR) 97 (95–99) 98 (96–100) 97 (95–99) 0.0001
Prehospital pulse rate, median (IQR) 86 (73–101) 85 (72–100) 85 (72–100) 0.004
Prehospital resp rate, median (IQR) 18 (16–22) 19 (16–24) 18 (16–23) 0.073
Time to arrival to hospital (min), median (IQR) 23 (15–33) 23 (16–31) 25 (15–33) 0.999
Penetrating injury, n (%) 273 (5.9) 261 (5.6) 534 (5.8) 0.593
MOI, n (%)
 Vehicle incident/collision 2396 (51.7) 2410 (52.0) 4806 (51.8) 0.006
 Fall from heights 1034 (22.3) 963 (20.8) 1997 (21.5)
 Low fall 487 (10.5) 592 (12.8) 1079 (11.6)
 Stabbing/shooting 244 (5.3) 226 (4.9) 470 (5.1)
 Other 475 (10.3) 445 (9.6) 920 (9.9)
Prehospital intubation, n (%) 232 (5.0) 1043 (22.5) 1275 (13.8) <0.0001
LOS in days, median (IQR) 9 (5–18) 9 (5–18) 9 (5–18) 0.999
Neuro center on site, n (%) 4471 (96.4) 4475 (96.5) 8946 (96.5) 0.822
Head injury AIS 3+, n (%) 1465 (31.6) 1452 (31.3) 2917 (31.5) 0.771
Attendance at scene, n (%)
 Consultant, n (%) 0 282 (6.1) 22 (3.0) <0.0001
 Doctor, n (%) 0 2068 (44.6) 2068 (22.3)
 Paramedic only, n (%) 4104 (88.5) 1813 (39.1) 5917 (63.8)
Mortality, n (%) 448 (9.7) 619 (13.4) 1067 (11.5) <0.0001
GCS, Glasgow Coma Scale; GEMS, ground emergency medical services; HEMS, helicopter emergency medical services; ISS, Injury Severity Score; MOI, mechanism of injury; MTCs,
major trauma centers; SBP, systolic blood pressure.

The model used age, gender and their interaction, Injury Severity RR <10 or RR >29. We also conducted a sensitivity analysis to
Score (ISS), Charlson comorbidity index, GCS on arrival and allow inclusion of GEMS patients attended by a doctor at scene.
amount of blood given within 6 hours. The analysis was repeated
on a subset of subjects with GCS <8 and a subset of subjects with
RESULTS
A total of 61 733 eligible adult patients were directly admitted to
MOI:Fall<2m
an MTC: 54 185 for GEMS and 7548 for HEMS. Table 1 shows
Intubation/Ventilation
most severe: Thorax that HEMS patients were in the majority male, younger, more
Trapped: No
Blood given within 6 hours
Comorbidity index 6 - 10
severely injured, more likely to be victims of road traffic colli-
Comorbidity index >10
most severe:Abdo sion, more likely intubated at scene and attended by a doctor.
Prehosp SBP
Attended Neuro centre
most severe: Head
The crude mortality is higher for HEMS patients. The observed
most severe: Spine
most severe: Face imbalance in the patient’s baseline covariates would not be satis-
Prehosp GCS 6 - 8
Prehosp GCS 4 - 5
Preshosp GCS 3 factory to assess the effect of HEMS against GEMS on outcome.
most severe: Pelvis
Trapped at scene: Yes
Prehosp GCS 9 - 12
The study population is described in the flow chart (figure 1).
Prehosp Pulse rate
Prehosp GCS:intubated The amount of missing data is shown table 2.
most severe: Other body region
Comorbidity index:Not recorded
most severe: Limb
The propensity score matching produced two groups of size
Prehosp Resp rate
MOI: Other 4636 with balanced baseline covariates as displayed in table 3.
MOI:Stabbing/Shooting
MOI: Fall>2m
Prehosp O2sat
There are statistically significant differences in some of the
Prehosp GCS 13 - 14
covariates but not clinically significant. The only significant
-1 -.5 0 .5 1
Standardardized difference differences are with the intubation at scene, the presence of
doctor at scene and unadjusted mortality.
Before Adjustment After Adjustment
The balance in the covariates was also assessed by the absolute
standardized differences between HEMS and GEMS groups, and
Figure 2 Standardized difference before and after propensity it showed that the absolute differences after matching was less
matching. GCS, Glasgow Coma Scale; MOI, mechanism of injury; O2sat, than 0.1 (figure 2). Characteristics of those attending the scene
oxygen saturation; SBP, systolic blood pressure. forms part of the intervention and is expected to be imbalanced.
4 Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508
 Open access

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
this through a sensitivity analysis demonstrating that having a
Table 4 Odds of death for HEMS versus GEMS transport to hospital
doctor at scene does have a significant survival impact (OR 0.77;
for major trauma
95% CI 0.62 to 0.95, p=0.015). Further interaction modeling
Criteria OR of death 95% CI P value suggests this does not impact on the effect seen of HEMS versus
Entire matched cohort 0.846 0.684 to 1.046 0.122 GEMS on outcome.
(9272) Due to the nature of prehospital care, internationally, there
Prehospital GCS ≤8 are very few randomized trials comparing HEMS with GEMS.
 Present (1417) 0.733 0.508 to 1.059 0.098 A number of observational studies have been published, but
 Absent (7855) 1.150 0.862 to 1.534 0.340 conclusions attributing improved survival to prehospital care
should be interpreted with caution because of study heteroge-
Prehospital respiratory rate <10 or >29
neity, selection bias and the contribution of the treating hospital
 Present (1007) 0.760 0.498 to 1.161 0.204
centers.16 17 A Cochrane Database Systematic Review (2015)
 Absent (8265) 0.894 0.692 to 1.156 0.393 concluded that due to methodological weaknesses in the multi-
Prehospital hemorrhagic shock (SBP <90) variate regression studies, neither the benefits of HEMS, nor
 Present (596) 0.330 0.078 to 1.391 0.165 its component elements, could be determined. It also pointed
 Absent (8676) 0.944 0.753 to 1.183 0.615 to a need to examine cost and safety.17 The TARN case-­mix
GCS, Glasgow Coma Scale; GEMS, ground emergency medical services; HEMS, adjustment model used in this study also addresses a number
helicopter emergency medical services; SBP, systolic blood pressure. of the limitations of previous Trauma and Injury Severity Score
based analyses.23 A study in the Netherlands, a country with
relatively comparable healthcare and geography to England,
A logistic regression was run on the matched pairs, and a concluded that HEMS saved 5.33 additional lives per 100
reduction on the odds of death in favor of HEMS was observed, dispatches compared with GEMS.9 However, there may be an
but this did not reach significance (OR=0.846; 95% CI 0.684 overestimate of the treatment effect as that study was based on
to 1.046, p value 0.122). The same analyses were repeated on only one regional center, and roughly half of the HEMS cohort
three subsets of subjects with extremes of prehospital variables was excluded due to missing data. There is some limited trial
and improved on the odds of death further in favor of HEMS, evidence, including a single-­center randomized controlled trial
though with lower participant numbers, again this did not reach (RCT), for a reduced mortality in blunt injuries when physicians
significance (table 4). are involved in prehospital care,24 but equally another RCT of
The sensitivity analysis allowed inclusion of patients attended prehospital versus in-­hospital RSI for GCS <9 in blunt major
by a doctor at scene, but where patients were subsequently trauma concluded no survival benefit but an improved 6-­month
transferred to hospital by ground ambulance (5685 patients in disability.11
figure 1), with the revised propensity analysis matching 5015 The analysis presented here could well provide the most exten-
patients in each cohort. This identified a significant protective sive evaluation for the role of HEMS in major trauma in England
effect for doctor present at scene on mortality (OR 0.77; 95% to date and imply that HEMS may be a beneficial resource, but
CI 0.62 to 0.95, p=0.015) but no significant impact of HEMS limitations in methodology require that the non-­ statistically
(OR 1.14; 95% CI 0.94 to 1.38, p=0.195). Further modeling to significant results must also be interpreted with caution.
assess the combined effect of a doctor at scene and HEMS did There may be ‘survival bias’ due to more intensive prehos-
not show any significant interaction (OR 0.84; 95% CI 0.56 to pital resuscitation, in HEMS groups, meaning patients arrive at
1.26, p=0.4). hospital alive whereas equivalent patients in the GEMS group
did not (TARN does not include patients certified dead in the
DISCUSSION prehospital phase). However, HEMS doctors may perhaps be
This study explores the contribution of HEMS to patient more likely to declare a death prehospital, which could make
survival after major trauma since the introduction of RTNs survival appear worse in the GEMS group. Furthermore, HEMS
across England in 2012. HEMS patients directly admitted are unable to transfer patients to TUs without helipads. This may
to MTCs are more severely injured than those conveyed by therefore encourage transfers to MTCs for less severely injured
GEMS, with greater requirements for critical care and opera- patients who could have been suitably treated at the local TU.
tive intervention. We demonstrated a 15% reduction in the risk However, potential for bias here is addressed by the propensity
adjusted odds of death (OR 0.846; 95% CI 0.684 to 1.046) in score analysis.
the HEMS with respect to GEMS. However, this was not statis- The methodology used has accounted for prehospital time
tically significant. The 15% risk adjusted mortality reduction in propensity score analysis, but this is not a perfect substitute
demonstrated is however clinically significant and comparable for distance; a key benefit of HEMS is its ability to cover large
with results reported in the USA.18 When repeated for subsets distances in shorter times and avoid traffic.
of patients previously noted to benefit most from HEMS (GCS The use of propensity score to balance the variation in
<8, abnormal RR or SBP <90),18 the odds of death was reduced case-­mix between the two patient groups improves on previous
further but still demonstrated a non-­significant survival advan- study methodologies. However, patients transported to TUs
tage. This seems to suggest that despite the size of the cohort were excluded, removing from the analysis perhaps one of the
directly admitted to MTCs, due to the intrinsic mismatch in most intrinsic benefits of HEMS in delivering patients from
patient profiles, the ability to statistically test for a mortality long distance directly to MTCs. Incomplete records of mode of
impact from HEMS in England is limited. arrival and final outcome reduced patient numbers from 80 532
An increasingly popular mode of operation for HEMS, partic- to 61 733 introducing again the potential for selection bias.
ularly in urban areas, is to deploy a specialist EMS team by heli- HEMS cost £404–£1689 per mission in the UK,16 while GEMS
copter with the patient subsequently conveyed by road to hospital. cost £144–£216.23 A recent US study suggested that HEMS must
A number of patients therefore had a helicopter-­supported inter- save 3.7 lives per 100 seriously injured to be cost-­effective at $50
vention but not attributed to HEMS. We attempted to address 000 per quality-­adjusted life year.14 Poor dispatch reduces the
Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508 5
 Open access

Trauma Surg Acute Care Open: first published as 10.1136/tsaco-2020-000508 on 16 July 2020. Downloaded from http://tsaco.bmj.com/ on June 26, 2023 by guest. Protected by copyright.
cost-­effectiveness of HEMS; however, the variable costs such as REFERENCES
fuel and servicing are relatively cheap compared with fixed costs 1 Office for National Statistics. Causes of death over 100 years. 2018. https://www.​ons.​
of staffing and depreciation of assets. Current call-­out criteria gov.​uk/​peop​lepo​pulat​ ion​andc​ommunity/​birt​hsde​athsa​ ndma​ rriages/​deaths/​articles/​
caus​esof​deat​hove​r100​years/​2017-​09-​18.
offer limited specificity for patients who would benefit from 2 Hughes G. Trauma care in England. Emerg Med J 2010;27:338.
HEMS. 3 Chiara O, Cimbanassi S. Organized trauma care: does volume matter and do trauma
HEMS teams still vary nationwide in their dispatch criteria, centers save lives? Curr Opin Crit Care 2003;9:510–4.
operating times, airframe capability, operating environment, 4 Royal College of Surgeons of England and the British Orthopaedic Association.
Better Care for the Severely Injured. A joint report from the Royal College of
intention of dispatch (retrieval or expert intervention) and
Surgeons of England and the British Orthopaedic Association. 2000. https://www.​
funding, with many crew configurations changing on a shift-­by-­ rcseng.​ac.​uk/​library-​and-​publications/​rcs-​publications/​docs/​better-​care-​for-​the-​
shift basis, giving heterogeneity of what is meant by a ‘HEMS severely-​injured/.
response’. This heterogeneity needs to be captured to inform 5 National Confidential Enquiry into Patient Outcomes and Death. Trauma: who cares?
future studies. The incorporation of the ‘real world’ varia- 2007. http://www.​ncepod.​org.​uk/​2007report2/​Downloads/S​ IP_​report.​pdf.
6 Moran CG, Lecky F, Bouamra O, Lawrence T, Edwards A, Woodford M, Willett K, Coats
tion in dispatch criteria in this paper strengthens the general- TJ. Changing the System - Major Trauma Patients and Their Outcomes in the NHS
izability of our conclusions. We did not subgroup patients by (England) 2008-17. EClinicalMedicine 2018;2-3:13–21.
the criteria that triggered a HEMS response, as this information 7 Iirola TT, Laaksonen MI, Vahlberg TJ, Pälve HK. Effect of physician-­staffed helicopter
is not well recorded (and in fact the dispatch decision is often emergency medical service on blunt trauma patient survival and prehospital care. Eur
multifactorial). J Emerg Med 2006;13:335–9.
8 Hesselfeldt R, Steinmetz J, Jans H, Jacobsson M-­LB, Andersen DL, Buggeskov K,
As a randomized trial of HEMS versus GEMS will be very Kowalski M, Praest M, Øllgaard L, Höiby P, et al. Impact of a physician-­staffed
challenging, improved understanding of the effects of both the helicopter on a regional trauma system: a prospective, controlled, observational study.
clinical and transport assets within HEMS will require better and Acta Anaesthesiol Scand 2013;57:660–8.
unified prehospital reporting methods with linkage to national 9 Den Hartog D, Romeo J, Ringburg AN, Verhofstad MHJ, Van Lieshout EMM. Survival
benefit of physician-­staffed helicopter emergency medical services (HemS) assistance
trauma registries and outcome data. A clear definition of patients
for severely injured patients. Injury 2015;46:1281–6.
who receive advanced doctor-­paramedic levels of care rather 10 Davenport RA, Tai N, West A, Bouamra O, Aylwin C, Woodford M, McGinley A, Lecky F,
than ‘standard’ paramedic care is essential in understanding the Walsh MS, Brohi K, et al. A major trauma centre is a specialty Hospital not a hospital
benefits of different EMS configurations. of specialties. Br J Surg 2010;97:109–17.
The analysis presented here provides a non-­statistically signif- 11 Bernard SA, Nguyen V, Cameron P, Masci K, Fitzgerald M, Cooper DJ, Walker T, Std BP,
Myles P, Murray L, et al. Prehospital rapid sequence intubation improves functional
icant mortality benefit from HEMS, although a 15% adjusted outcome for patients with severe traumatic brain injury: a randomized controlled trial.
odds of mortality reduction is clinically significant. Further anal- Ann Surg 2010;252:959–65.
ysis demonstrates a protective effect on mortality from doctors 12 National Institute for health and Care Excellence. Major trauma: assessment and
on scene subsequently transported by GEMS. An investment in initial management. 2016. https://www.​nice.​org.​uk/​guidance/​ng39/c​ hapter/​
recommendations#​airway-​management-​in-​prehospital-​and-​hospital-​settings.
prehospital reporting and TU data should be made in order to
13 Leaman AM, Nutbeam T. Trauma systems and medical helicopters in the UK. Injury
conduct further studies of HEMS clinical and cost-­effectiveness 2011;42:12–14.
that are less prone to type 2 error. 14 Delgado MK, Staudenmayer KL, Wang NE, Spain DA, Weir S, Owens DK, Goldhaber-­
Fiebert JD. Cost-­Effectiveness of helicopter versus ground emergency medical services
Acknowledgements Many thanks to Tom Jenks and Dr Carl McQueen for their for trauma scene transport in the United States. Ann Emerg Med 2013;62:351–64.
assistance with data preparation. 15 Galvagno SM, Haut ER, Zafar SN, Millin MG, Efron DT, Koenig GJ, Baker SP, Bowman
SM, Pronovost PJ, Haider AH, et al. Association between helicopter vs ground
Contributors OlB contributed to the literature review, study question and emergency medical services and survival for adults with major trauma. JAMA
design, data analysis and interpretation; prepared the first draft of the report; and 2012;307:1602–10.
contributed to subsequent versions. OmB, FL and KW contributed to study design, 16 Butler DP, Anwar I, Willett K. Is it the H or the EMS in HemS that has an impact
data analysis and interpretation, and drafting of the final report. TC, DSK and DL on trauma patient mortality? A systematic review of the evidence. Emerg Med J
contributed to the data analysis and interpretation, and drafting of the final report. 2010;27:692–701.
All authors approved the final version. OlB, KW and FL are the guarantors. 17 Galvagno SM, Sikorski R, Hirshon JM, Floccare D, Stephens C, Beecher D, Thomas
Funding The authors have not declared a specific grant for this research from any S. Helicopter emergency medical services for adults with major trauma. Cochrane
funding agency in the public, commercial or not-­for-­profit sectors. Database Syst Rev 2015:CD009228.
18 Chen X, Gestring ML, Rosengart MR, Billiar TR, Peitzman AB, Sperry JL, Brown
Competing interests None declared.
JB. Speed is not everything: identifying patients who may benefit from helicopter
Patient consent for publication Not required. transport despite faster ground transport. J Trauma Acute Care Surg 2018;84:549–57.
Provenance and peer review Not commissioned; internally peer reviewed. 19 Bledsoe BE, Wesley AK, Eckstein M, Dunn TM, O’Keefe MF. Helicopter scene transport
of trauma patients with nonlife-­threatening injuries: a meta-­analysis. J Trauma
Data availability statement Data are available on reasonable request. Data 2006;60:1257–66. discussion 65-6.
may be obtained from a third party and are not publicly available. Further detail 20 Department Of Health. Transforming NHS ambulance services: National audit Office,
of analyses performed are available at request from the corresponding author. 2011.
Additional and raw data are stored securely through the TARN database who can be 21 Lunt M. A Guide to Imputing Missing Data with Stata Revision: 1.4. 2011. https://
contacted via ​support@​tarn.​ac.​uk. personalpagesmanchesteracuk/staff/marklunt/mi_guidepdf.
Open access This is an open access article distributed in accordance with the 22 Mitra R, Reiter JP. A comparison of two methods of estimating propensity scores after
Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which multiple imputation. Stat Methods Med Res 2016;25:188–204.
permits others to distribute, remix, adapt, build upon this work non-­commercially, 23 Bouamra O, Wrotchford A, Hollis S, Vail A, Woodford M, Lecky F. A new approach
and license their derivative works on different terms, provided the original work is to outcome prediction in trauma: a comparison with the TRISS model. J Trauma
properly cited, appropriate credit is given, any changes made indicated, and the use 2006;61:701–10.
is non-­commercial. See: http://​creativecommons.​org/​licenses/​by-​nc/​4.​0/. 24 Garner AA, Mann KP, Fearnside M, Poynter E, Gebski V. The head injury retrieval
trial (Hirt): a single-­centre randomised controlled trial of physician prehospital
ORCID iD management of severe blunt head injury compared with management by paramedics
Oliver Beaumont http://​orcid.​org/​0000-​0002-​1396-​9262 only. Emerg Med J 2015;32:869–75.

6 Beaumont O, et al. Trauma Surg Acute Care Open 2020;5:e000508. doi:10.1136/tsaco-2020-000508