Age and Ageing 2009; 38: 693–697
C The Author 2009.

2009. Published by Oxford University Press [on behalf of the British Geriatrics Society].

doi: 10.1093/ageing/afp164 All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
Published electronically 3 September 2009

C-reactive protein, severity of pneumonia and

mortality in elderly, hospitalised patients with
community-acquired pneumonia
ULRICH THIEM1,2 , DAVID NIKLAUS1 , BETTINA SEHLHOFF1 , CHRISTOPH STÜCKLE3 ,
HANS JÜRGEN HEPPNER4 , HEINZ GERD ENDRES2 , LUDGER PIENTKA1
1
Department of Geriatrics, Marienhospital Herne, University of Bochum, Widumer Str. 8, D-44627 Herne, Germany
2
Department of Medical Informatics, Statistics and Epidemiology, University of Bochum, Overbergstr. 17, D-44801 Bochum,

Downloaded from https://academic.oup.com/ageing/article/38/6/693/40913 by guest on 26 March 2021

Germany
3
Institute for Diagnostic and Interventional Radiology and Nuclear Medicine, Marienhospital Herne, University of Bochum,
Widumer Str. 8, D-44627 Herne, Germany
4
2nd Department of Internal Medicine/Geriatrics, Medical Center Nuremberg, University Erlangen/Nuremberg,
Prof.-Ernst-Nathan-Str. 1, D-90419 Nuremberg, Germany
Address correspondence to: U. Thiem. Tel: (+49)-2323-499-5912; Fax: (+49)-2323 499 2621. Email: ulrich.thiem@rub.de

Abstract
Background: increasingly, markers of systemic inflammation like C-reactive protein (CRP) levels and white blood count
(WBC) are being used for assessing the prognosis of patients with community-acquired pneumonia (CAP). However, their
predictive value has not been validated in populations of elderly patients.
Objective: to evaluate the prognostic value of CRP and WBC in comparison with the CURB score and the pneumonia severity
index (PSI) in elderly, hospitalised patients with CAP.
Methods: the charts of all patients, aged 65 years and older, who were consecutively admitted to the Department of Geriatrics,
Marienhospital Herne, Germany, for treatment of CAP between January 2001 and September 2005, were reviewed. CRP,
WBC, CURB and PSI were analysed in relation to 30-day mortality.
Results: in a total of 391 patients, median age 80 years, no association was found between CRP or WBC and mortality. In
contrast, the CURB score and PSI were significantly associated with mortality and treatment in the intensive care unit (ICU).
Conclusion: in elderly, hospitalised patients with CAP, admission CRP and WBC are not predictors of the prognosis.

Keywords: community-acquired pneumonia, CURB score, C-reactive protein, mortality, elderly

Introduction the value of these markers remains unclear. A recently pub-

Several risk scores are available for evaluating the prognosis lished study [14] reported a considerable association between
of patients with community-acquired pneumonia (CAP). CRP and mortality, describing CRP as an independent risk
The pneumonia severity index (PSI) described by Fine et al. factor for complications and 30-day mortality. In a receiver
[1] in 1997 is widely used in the United States [2]. In Europe, operator characteristic curve (ROC) analysis, CRP produced
CURB65—a score covering the variables acute confusion, a moderate performance with regard to mortality, while it
serum urea, respiratory rate, blood pressure and age—is used outperformed CURB65 and PSI with regard to complicated
to predict prognosis [2, 3]. Both tools have been validated pneumonia [14]. Another recent paper [15] confirmed a sig-
in various populations, and, in part, in patients aged 65 years nificant association between prognosis and CRP and WBC.
and older [4–10]. Current guidelines for the management In this study, however, the power of CRP and WBC to predict
of patients with CAP recommend the use of these scores mortality was significantly lower than CURB65 and procalci-
[11–13]. tonin. Moreover, CRP and WBC lost their independence as
Biochemical markers of inflammation have also been dis- predictors of prognosis in a multivariate analysis that adjusted
cussed as potentially important prognostic variables. These for pneumonia severity [15].
include, among others, the readily available C-reactive protein Muller et al. [16] described a relationship between WBC,
(CRP) level and the white blood cell count (WBC). However, but not CRP, and mortality. Other authors report divergent

693
U. Thiem et al.

results on the prognostic value of CRP and WBC with regard mation about previous illnesses was collected as previously
to mortality, while a large number of the studies were pri- described [1]. The clinical symptoms and findings collected
marily designed as diagnostic studies [17–25]. Only one pub- were pulse, systolic and diastolic blood pressure, respiratory
lication reported CRP and WBC in elderly patients with CAP rate, temperature and new onset of an acute confusional state
[10]. However, this report focused on the validation of the as previously described [1]. The following laboratory values
CURB score and assessed CRP in only 75 patients, a sample were collected: CRP, WBC, haematocrit, blood glucose, urea
size too small to permit a definitive conclusion. and sodium level on admission as well as pO2 levels and
The aim of the present study was to compare the prognos- pH values from an arterial or capillary blood gas analysis.
tic value of CRP and WBC with that of the established risk The chest x-ray findings were used to document any proof
scores CURB and PSI in elderly, hospitalised patients with of infiltrate and any pleural effusion. Since 2001, all labo-
CAP. Like Myint et al. [10], we used the abbreviated CURB ratory values, written x-ray findings and medical reports on
score that omits the age criterion in comparison to CURB65, patients have been available through a central server system.
as our sample only includes patients aged ≥ 65 years. Cases with missing values were excluded if the other existing
parameters did not permit unequivocal categorisation of the

Downloaded from https://academic.oup.com/ageing/article/38/6/693/40913 by guest on 26 March 2021

CURB score. CURB and PSI were calculated as previously
Patients and methods described [1, 3].
Setting
Sample size calculation and statistics
The Marienhospital is a 530-bed tertiary medical centre in
Herne, Germany. It operates departments specializing in The sample size was calculated using estimates of Myint
geriatrics, pneumology and haematology/oncology, among et al. [10] for the overall mortality of patients ≥65 years with
others. The Department of Geriatrics mainly treats elderly CAP and of Chalmers et al. [14] regarding mortality risk as
and geriatric patients requiring acute medical care. During a function of admission CRP levels (CRP < 100 mg/L and
the study period, the management of pneumonia including ≥ 100 mg/L, respectively). Assuming a significance level of
antibiotic therapy and admission to the intensive care unit 5% (α = 0.05), a power of 80% (1 − β = 0.80), a 30-day
(ICU) was guided by criteria adopted from the 2001 British mortality of 20% overall and 4% in the low-risk group (group
Thoracic Society Guideline on CAP [11]. with low CRP levels), a sample size of 49 patients for each
group was required (Fisher’s exact test). Assuming a higher
Inclusion and exclusion criteria mortality risk of 8% for the group with a low CRP, a sample
size of 148 patients per group or a total of ∼300 patients
During the period between 1 January 2001 and 15 September
would be needed under preconditions that were otherwise
2005, the charts of all patients ≥ 65 years with an ICD-10
the same.
code ‘pneumonia’ as the admission diagnosis were reviewed.
Categorical variables were stated in absolute values and in
For inclusion, the radiological verification of a newly mani-
percent, and constant variables as median, mean and standard
fest infiltrate was required. Patients were enrolled as present-
deviation (SD). The statistical tests on categorical variables
ing with clinically typical pneumonia, if at least two clinical
were performed using the χ 2 test, on constant variables with
symptoms suggestive for respiratory tract infection (i.e. dys-
Student’s t-test. Two-tailed tests with a significance level of
pnoea, cough, new or purulent sputum, fever >38.0◦ C) were
5% (α = 0.05) were used for all analyses. One-way analysis
present. Atypical cases presenting for example with mobil-
of variance (ANOVA) was used to detect differences in both
ity impairment, falls, confusion, incontinence or other signs
CRP and WBC as a function of the CURB score or PSI. We
of clinical deterioration were included when a newly man-
used log-transformation of WBC and CRP, respectively, to
ifest infiltrate was verified and no other cause sufficiently
achieve a symmetrical distribution of values. Logistic regres-
explaining the clinical condition of the patient was detected.
sion was used to analyse the relationship between 30-day
Cases were excluded if they had stenosis-induced or aspi-
mortality or treatment on the ICU as dependent variables
ration pneumonia, pulmonary embolism or were receiving
and CRP or WBC with the CURB score and other factors as
chemotherapy or irradiation within 4 weeks before hospital-
independent variables. Adjusted odds ratios (OR) with 95%
isation. Patients were additionally excluded if they had been
confidence intervals (CI) were estimated for the independent
hospitalised within the previous 4 weeks, lacked radiologi-
variables. The value of individual variables was compared by
cal confirmation of an infiltrate within the first 24 h after
ROC analysis. The sample size was calculated using StatsDi-
admission or were receiving palliative treatment modalities.
rect statistical software, version 2.6.2 (StatsDirect Ltd, Altrin-
Nursing home residents were not excluded.
cham, Cheshire, UK); all statistical analyses were performed
Definition of variables and index calculation with SPSS 14.0 (SPSS Inc., Chicago, IL, USA).
Data were collected on age, sex, nursing home status and
duration of inpatient treatment. The patient’s vital status Results
on day 30 was ascertained by inquiry with the responsible
address registration office whenever these data could not A total of 438 patients ≥ 65 years with an ICD-10 code
be gathered from additional hospital resources. The infor- ‘pneumonia’ on admission were reviewed. Thirty patients

694
 CRP, severity of pneumonia and mortality in elderly, hospitalised patients with CAP

Table 1. Characteristics of the study cohort (n = 391) Table 2. CRP, WBC and pneumonia severity
Characteristic n % CRPa WBCb
................................................................ median 25th 75th median 25th 75th
Age ≥ 80 years 204 52.2 (mg/L) percentile percentile (µL−1 ) percentile percentile
Female sex 200 51.2 ................................................................
Nursing home residency 99 25.3 CURBc
30-day mortality 76 19.4 0–1 points 83.5 29.6 143.0 11,500 8,100 15,900
ICU treatment 72 18.4 2 points 73.9 26.0 151.2 11,900 9,000 14,900
Neoplastic disease 56 14.3 3–4 points 105.7 43.8 169.7 11,850 9,050 16,125
Cardiac disease 261 66.8 PSId
Cerebrovascular disease 152 38.9 Class II/III 63.4 38.8 146.3 12,400 9,150 16,150
Kidney insufficiency 119 30.4 Class IV 73.1 24.7 131.8 11,200 8,100 15,300
Respiratory rate > 30/min 209 53.5 Class V 85.1 28.7 163.1 11,900 8,500 15,300
Blood pressure < 90/60 mmHg 30 7.7
Temperature < 35.0◦ or > 40.0◦ C 11 2.8 a CRP: C-reactive protein.
Pulse > 125/min 27 6.9 b WBC: white blood cell count.

Downloaded from https://academic.oup.com/ageing/article/38/6/693/40913 by guest on 26 March 2021

Haematocrit < 30% 36 9.2 c for CRP P = 0.68, for WBC P = 0.83.

Blood glucose > 250 mg/dL 35 9.0 d for CRP P = 0.63, for WBC P = 0.74.

Urea > 7 mmol/L 251 64.2

Sodium < 130 mmol/L 21 5.4 Table 3. CURB- and PSI-scored pneumonia severity and
Pleural effusion 125 32.0
30-day-mortality
Survivors Non-survivors Total

were excluded, as they did not fulfil the study criteria, with n % n % n %
................................................................
ambiguous x-ray findings and palliative care being the most
CURB a
common reasons. Seventeen cases were excluded due to miss- Points
ing data. The characteristics of the 391 included patients are 0 49 94.2 3 5.8 52 100.0
summarised in Table 1. 1 133 81.6 30 18.4 163 100.0
The median age of the patients was 80 years (mean ± SD: 2 118 77.6 34 22.4 152 100.0
80.0 ± 8.0 years). A 30-day mortality was 19.4% (76/391 3 15 68.2 7 31.8 22 100.0
4 0 0.0 2 100.0 2 100.0
patients). A high proportion of nursing home patients 5 – – – – – –
(25.3%) and a generally high morbidity were striking. In terms PSI b
of pneumonia severity, the majority of patients were assigned Risk class
to a medium- or high-risk group. Scored by the CURB score, I – – – – – –
only 13.3% (52 patients) had 0 points and 80.6% scored 1 or II 1 100.0 0 0.0 1 100.0
III 11 91.7 1 8.3 12 100.0
2 points. According to the PSI, only 1 patient had risk class II, IV 129 86.6 20 13.4 149 100.0
whereas 229 patients (58.6%) were assigned to risk class V. V 174 76.0 55 24.0 229 100.0
CRP showed a median of 79.9 mg/L (mean 99.4 ± Total 315 80.6 76 19.4 391 100.0
87.4 mg/L), WBC a median of 11,600/µL (mean 13,037 ±
9,354/µL). In the 76 patients who died during hospital stay,
aP = 0.001.
bP = 0.053.
the admission median CRP was 76.1 mg/L (mean 100.8 ±
87.3 mg/L), while survivors had a median CRP of 82.1 mg/L
> 2 points. There was also no meaningful increase in CRP
(mean 99.0 ± 87.5 mg/L, P = 0.33). The median WBC was
between different PSI risk classes. No relationship between
11,500/µL (mean 12,553 ± 6,129/µL) for deceased patients
CRP or WBC and CURB or PSI, respectively, was statistically
and 11,600/µL (mean 12,717 ± 6,318/µL) for survivors
significant (Table 2).
(P = 0.54). The median CRP and WBC on admission were
The risk scores CURB and PSI were significantly associ-
not significantly different between patients requiring or not
ated with mortality (Table 3). The OR for mortality adjusted
requiring ICU treatment.
for age and sex was 1.69 (95% CI 1.21; 2.35) for the CURB
Logistic regression analysis on mortality as the dependent
score and 1.83 (95% CI 1.06; 3.14) for the PSI. The area under
variable produced an OR of 1.00 (95% CI 0.97; 1.03) for
the curve (AUC) values from the ROC analysis for CURB
CRP and 1.02 (95% CI 0.99; 1.04) for WBC; when treatment
and PSI with 30-day mortality as the target variable were 0.64
on an ICU was used as the dependent variable, the resulting
(95% CI 0.57; 0.71) (P < 0.001) for CURB and 0.63 (95% CI
OR were 1.02 (95% CI 0.99; 1.05) and 1.00 (95% CI 0.96;
0.56; 0.70) (P = 0.001) for PSI.
1.03), respectively. No relevant changes were revealed after
adjusting for CURB-scored pneumonia severity, age, sex and
further variables (data not shown). Discussion
A clear relationship between CRP and WBC and CURB
score could not be assessed (Table 2). When scored by CURB, Our data did not reveal any association between CRP and
the median for CRP did not show any notable increase until WBC and mortality. Nor did we find any relationship between

695
U. Thiem et al.

pneumonia severity and CRP or WBC. These results appear high-risk elderly patients. As we could reaffirm, established
to contradict those of recent studies. risk scores such as the CURB score and PSI are available that
In one recent publication, Chalmers et al. describe a clear have already shown prognostic impact even in elderly high-
association between CRP during hospital stays and 30-day risk patients. In view of the divergent results, the existing
mortality and other additional clinical endpoints; the rela- literature does not make recommendations for using CRP
tionship remained after adjusting for pneumonia severity [14]. and WBC as prognostic parameters in CAP [14, 15, 22].
Although Krüger et al. confirmed a relationship between CRP Instead, Kruger et al. recommend procalcitonin, not CRP or
and also WBC and mortality in their univariate analysis, this WBC, as a promising marker [15].
relationship was not maintained after adjusting for additional One limitation of our study is its retrospective design.
factors [15]. Muller et al. described an association between Nevertheless, we believe that this does not detract from the
WBC and prognosis, but not CRP, again by univariate anal- validity of our data, seeing as the decisive clinical endpoint of
ysis only [16]. mortality and the most important variables of interest, CRP
The major differences between our cohort and those of and WBC, were easily and reliably determined by retrieval
the studies cited are age and risk profile of the patients. Given from readily available administrative and electronic records.

Downloaded from https://academic.oup.com/ageing/article/38/6/693/40913 by guest on 26 March 2021

the inclusion criterion age ≥65 years, our cohort showed a Furthermore, the retrospective collection of data does not
median age of 80 years which is markedly older than the inherently lead to substantial disagreement with prospec-
populations in the studies cited [14, 15, 22]. The pneumonia tively collected data, as Aujesky et al. [29] have shown for the
severity also showed strong divergence. Based on the CURB PSI. This should likewise apply to CURB, a score essentially
score, 45% of our patients would be ranked as having a high made up of considerably less comparable components. Our
mortality risk, compared with 34% [14] and 5% [15] in the reliance on routine data, however, did not allow us to exam-
other studies. Based on the PSI, just ∼4% of our patients ine more recent inflammatory markers, such as procalcitonin,
would be categorised into the favourable risk classes II or that are used clinically, but not routinely [15, 16, 19, 21, 22].
III, which contrasts with the 40% cited by Mueller et al. [22]. That such markers should be tested in high-risk patients is
Furthermore, our cohort includes the highest proportion of already a claim that Kruger et al. [15] have made in their
nursing home residents. A much greater majority of patients comprehensive validation study.
had cardiac and cerebrovascular co-morbidities and a rele- In conclusion, we were not able to establish a relationship
vant impairment of renal function on admission compared between CRP and WBC and 30-day mortality in a high-risk
to the other studies [14, 15, 22]. The aforementioned factors group of elderly, hospitalised patients with CAP. In such
result in a markedly higher 30-day mortality rate of ∼20% patients, clinicians should continue to base their prognostic
over comparable studies [14, 15, 22]. The aspects discussed assessment on clinical criteria and established risk scores such
above show that our study describes a high-risk cohort with as CURB and PSI. In the future, the prognostic evaluation
consistently poor prognosis, which might be suggested as an of CAP might be potentially improved by investigating more
explanation for the lack of any connection between CRP, novel inflammatory markers, like procalcitonin, in special risk
WBC and prognosis. groups.
A comparable phenomenon has been previously identi-
fied for the prognostic value of risk scores: CURB65 and PSI Key points
are comprehensively validated tools [1, 3–10, 26] that show
overall good to very good performance in predicting mor- • Recent studies support the use of CRP on hospital admis-
tality [1, 6–9, 26]. Validation studies on elderly patients have sion as a prognostic marker in patients with CAP.
reported limited performance of the scores, which can essen- • A new report on some 390 hospitalised patients aged
tially be explained by a clustering of patients in the higher risk 65 years and older questions this recommendation, as no
groups [4, 5, 10]. This had previously led to various proposals relationship between CRP level on admission and 30-day
for changing or supplementing the existing scores for evalu- mortality could be found.
ating the prognosis of elderly patients with a high mortality • In contrast, the validity of two existing prognostic tools,
risk [4, 27, 28]. The performance of both scores determined the CURB score and PSI, could be confirmed.
in our study indeed ranks among the lowest published thus • Clinicians should not rely on CRP, but on their clinical
far, but is actually very consistent with comparable studies [4, impression and validated risk scores to assess the prog-
5, 10]. According to Chalmers et al. [14] and Kruger et al. [15], nosis of elderly people with CAP.
a weaker association exists between CRP, WBC and 30-day
mortality than between CURB65 and mortality. If the same
applies to our cohort, the non-significant result for CRP and Acknowledgements
WBC comes as no surprise given the weak performance of
the CURB score and PSI. The authors’ sincere thanks go to Ms C. Offenberg and
Thereby, we are not arguing against the pathophysiologi- Ms I. Salih, Central Laboratory, Marienhospital Herne, for
cally conclusive relationship between pneumonia severity and their support in researching the laboratory data and to the
the extent of the pretreatment inflammatory reaction. Rather, registration office of the City of Herne for their coopera-
we question the clinical importance of CRP and WBC in tion. The study was performed at the Marienhospital Herne,

696
 CRP, severity of pneumonia and mortality in elderly, hospitalised patients with CAP

University of Bochum, Widumer Str. 8, D-44627 Herne, 13. Mandell LA, Wunderink R, Anzueto A et al. Infectious Dis-
Germany. eases Society of America/American Thoracic Society con-
sensus guidelines on the management of community-acquired
Conflicts of interest pneumonia in adults. Clin Infect Dis 2007; 44: S27–72.
14. Chalmers JD, Singanayagam A, Hill AT. C-reactive protein is
The authors have no conflicts of interest to disclose. an independent predictor of severity in community-acquired
pneumonia. Am J Med 2008; 121: 219–25.
15. Kruger S, Ewig S, Marre R et al. Procalcitonin predicts patients
Funding at low risk of death from community-acquired pneumonia
across all CRB-65 classes. Eur Respir J 2008; 31: 349–55.
The study was funded in part by an unrestricted 16. Muller B, Suess E, Schuetz P et al. Circulating levels of pro-atrial
research grant from Hoffmann-LaRoche, Grenzach-Wyhlen, natriuretic peptide in lower respiratory tract infections. J Intern
Germany. The sponsor played no role in the design, execu- Med 2006; 260: 568–76.
tion, analysis and interpretation of the data or the writing of 17. Schaaf B, Kruse J, Rupp J et al. Sepsis severity predicts outcome
the manuscript. UT was additionally supported by a research in community-acquired pneumococcal pneumonia. Eur Respir

Downloaded from https://academic.oup.com/ageing/article/38/6/693/40913 by guest on 26 March 2021

grant from the Marienhospital Herne, University of Bochum. J 2007; 30: 517–24.
18. Almirall J, Bolibar I, Toran P et al. Contribution of C-
reactive protein to the diagnosis and assessment of severity
References of community-acquired pneumonia. Chest 2004; 125: 1335–
42.
1. Fine MJ, Auble TE, Yealy DM et al. A prediction rule to identify 19. Brunkhorst FM, Al-Nawas B, Krummenauer F et al. Procalci-
low-risk patients with community-acquired pneumonia. N Engl tonin, C-reactive protein and APACHE II score for risk evalu-
J Med 1997; 336: 243–50. ation in patients with severe pneumonia. Clin Microbiol Infect
2. Ewig S, Torres A, Woodhead M. Assessment of pneumonia 2002; 8: 93–100.
severity: a European perspective. Eur Respir J 2006; 27: 6–8. 20. Hansson LO, Hedlund JU, Ortqvist AB. Sequential changes
3. Lim WS, Van Der Eerden MM, Laing R et al. Defining commu- of inflammatory and nutritional markers in patients with
nity acquired pneumonia severity on presentation to hospital: community-acquired pneumonia. Scand J Clin Lab Invest 1997;
an international derivation and validation study. Thorax 2003; 57: 111–8.
58: 377–82. 21. Hedlund J, Hansson LO. Procalcitonin and C-reactive pro-
4. Naito T, Suda T, Yasuda K et al. A validation and potential tein levels in community-acquired pneumonia: correlation with
modification of the pneumonia severity index in elderly patients etiology and prognosis. Infection 2000; 28: 68–73.
with community-acquired pneumonia. J Am Geriatr Soc 2006; 22. Muller B, Harbarth S, Stolz D et al. Diagnostic and prognostic
54: 1212–9. accuracy of clinical and laboratory parameters in community-
5. Mody L, Sun R, Bradley S. Community-acquired pneumonia acquired pneumonia. BMC Infect Dis 2007; 7: 10.
in older veterans: does the pneumonia prognosis index help? 23. Ortqvist A, Hedlund J, Wretlind B et al. Diagnostic and
J Am Geriatr Soc 2002; 50: 434–8. prognostic value of interleukin-6 and C-reactive protein in
6. Buising KL, Thursky KA, Black JF et al. A prospective com- community-acquired pneumonia. Scand J Infect Dis 1995; 27:
parison of severity scores for identifying patients with severe 457–62.
community acquired pneumonia: reconsidering what is meant 24. Smith RP, Lipworth BJ. C-reactive protein in simple
by severe pneumonia. Thorax 2006; 61: 419–24. community-acquired pneumonia. Chest 1995; 107: 1028–31.
7. Capelastegui A, Espana PP, Quintana JM et al. Validation of 25. Smith RP, Lipworth BJ, Cree IA et al. C-reactive protein. A
a predictive rule for the management of community-acquired clinical marker in community-acquired pneumonia. Chest 1995;
pneumonia. Eur Respir J 2006; 27: 151–7. 108: 1288–91.
8. Ewig S, de Roux A, Bauer T et al. Validation of predictive rules 26. Barlow G, Nathwani D, Davey P. The CURB65 pneumonia
and indices of severity for community acquired pneumonia. severity score outperforms generic sepsis and early warning
Thorax 2004; 59: 421–7. scores in predicting mortality in community-acquired pneumo-
9. Aujesky D, Auble TE, Yealy DM et al. Prospective comparison nia. Thorax 2007; 62: 253–9.
of three validated prediction rules for prognosis in community- 27. Kamath AV, Myint PK, Vowler SL et al. Is it time to rethink
acquired pneumonia. Am J Med 2005; 118: 384–92. the urea criterion in CURB-65? Eur Respir J 2006; 27: 1321–2.
10. Myint PK, Kamath AV, Vowler SL et al. The CURB (confusion, 28. Myint PK, Kamath AV, Vowler SL et al. Severity assessment
urea, respiratory rate and blood pressure) criteria in community- criteria recommended by the British Thoracic Society (BTS)
acquired pneumonia (CAP) in hospitalised elderly patients aged for community-acquired pneumonia (CAP) and older patients.
65 years and over: a prospective observational cohort study. Should SOAR (systolic blood pressure, oxygenation, age and
Age Ageing 2005; 34: 75–7. respiratory rate) criteria be used in older people? A compilation
11. Committee BTSPG. BTS Guideline for the Management of study of two prospective cohorts. Age Ageing 2006; 35: 286–
Community Acquired Pneumonia in Adults—Update 2004. 91.
http://www.brit-thoracic.org.uk 2004; (29 July 2008 date last 29. Aujesky D, Stone RA, Obrosky DS et al. Using randomized
accessed). controlled trial data, the agreement between retrospectively and
12. Woodhead M, Blasi F, Ewig S et al. Guidelines for the prospectively collected data comprising the pneumonia severity
management of adult lower respiratory tract infections. Eur index was substantial. J Clin Epidemiol 2005; 58: 357–63.
Respir J 2005; 26: 1138–80.
Received 25 November 2008; accepted in revised form 30 July 2009

697