Short-term mortality of adult inpatients withcommunity-acquired pneumonia: external validationof a modified CURB-65 scoreMarc Andre Pflug,1 Timothy Tiutan,2 Thomas Wesemann,3 Harald Nüllmann,3

Hans Jürgen Heppner,4 Ludger Pientka,3 Ulrich Thiem3,5

For numbered affiliations seeend of article.

Correspondence toDr Ulrich Thiem, Department ofGeriatrics, MarienhospitalHerne, University of Bochum,Widumer Str 8, Herne D-44627, Germany;ulrich.thiem@rub.de

Received 2 June 2014Revised 6 January 2015Accepted 8 January 2015Published Online First24 January 2015

To cite: Pflug MA, Tiutan T,Wesemann T, et al. PostgradMed J 2015;91:77–82.

ABSTRACTObjective The management of community-acquiredpneumonia (CAP) continues to be a challenge, especiallyin older people. To enable better risk stratification, avariation of the severity scores CRB-65 and CURB-65,called CURB-age, has been suggested. We compared theassociation between risk groups as defined by the scoresand 30-day mortality for a cohort of mainly olderinpatients with CAP.Methods We retrospectively analysed data from theCAP database from the years 2005 to 2009 of a singlecentre in Herne, Germany. Patient characteristics, criteriavalues within the severity scores CURB-65, CRB-65 andCURB-age, and 30-day mortality were assessed. Wecompared the association between score points andscore-defined risk groups and mortality. Sensitivity andspecificity with corresponding 95% CIs were calculated,and receiver operating characteristic (ROC) curve analysiswas performed.Results Data from 559 patients were analysed (meanage 74.1 years, 55.3% male). Mortality at day 30 was10.9%. CURB-age included more patients in the low-riskcategory than CRB-65 (195 vs 89), and the patientgroup had a lower mortality (2.6% vs 3.4%). Whencompared with CURB-65, CURB-age included slightlyfewer patients (195 vs 214) with lower mortality (2.6%vs 4.2%). CURB-age sorted the most patients who diedwithin 30 days into the high-risk CAP group (CURB-age,32; CURB-65, 28; CRB-65, 9), which had the highestmortality (CURB-age, 26.4%; CURB-65, 19.4%; CRB-65, 21.4%). Advantages of CURB-age categories weredepicted through ROC curve analysis (area under thecurve 0.73 (95% CI 0.67 to 0.79) for CURB-agecategories, 0.67 (95% CI 0.60 to 0.74) for CURB-65categories, and 0.59 (95% CI 0.52 to 0.66) for CRB-65categories).Conclusions In comparison with CRB-65 and CURB-65, risk stratification as defined by CURB-age showedthe closest association with 30-day mortality in oursample. Further prospective studies are needed to assessthe potential of CURB-age for better risk prediction,especially in older patients with CAP.

INTRODUCTIONManagement of pneumonia is still a major problemin developed countries. Along with influenza, it isone of the ten leading causes of death in both theUSA (2.1% of all deaths or 17.2 per 100 000 citi-zens annually) and Germany (2.2% of all deaths oroverall 18 889 citizens annually). It is also theleading cause of death due to infectious disease inboth countries. In patients >85 years old in

Germany, annual mortality rates due to pneumoniaand influenza reach 3.0%.1 2 This shows the urgentneed for proper management of pneumonia, whichwould play an important part in providingadequate healthcare and reducing overall morbidityand mortality, especially in older populations.Pneumonia and influenza also have a major eco-nomic impact. In 2005, their direct cost amountedto US$34.2 billion and indirect cost accounted forUS$6 billion in the USA, which have both probablyincreased over the last decade.3

Previous studies have shown that physicians whouse a risk-stratified care management approachalong with their clinical expertise are likely toimprove clinical decision-making when encounter-ing patients with community acquired pneumonia(CAP). Risk stratification can be used to determinethe need for hospital and intensive care unit admis-sion as well as to evaluate the use of antibiotictherapy schemes. The Pneumonia Severity Index(PSI), developed in the USA, has been shown toprovide good guidance for decision-making in theclinical management of CAP.4 However, it is difficultto use in daily clinical practice because of its exten-sive list of 20 criteria. A recent study by Serisieret al5 found that many Australian physicians, despiterecommendations to do so, do not use the PSI, nordo they apply it correctly when choosing to use it. Asimpler score was developed by Lim et al,6 whoaltered the modified British Thoracic Society ruleand established the five-point CURB-65 score(Confusion; Urea >7 mmol/L; Respiratory rate ≥30/min; low systolic (<90 mm Hg) or diastolic(≤60 mm Hg) Blood pressure; age ≥65 years) foruse in daily clinical routine. An even simplervariant, the CRB-65 score, can be used in ambula-tory settings because of its exclusion of the urea cri-terion.6 Many studies have confirmed the validity ofthese scores. However, several studies have shownthat CURB-65 and CRB-65 perform poorly whenused for older patients. Therefore, efforts have beenmade to improve these severity scores.7–14

Myint et al15 proposed a modification toCURB-65 when used with older patients, extendingthe urea criterion to >11 mmol/L and the age cri-terion to ≥85 years, with each scoring one add-itional point. This score variant was namedCURB-age. The cut-off for the urea criterion hasrecently been confirmed to be significant in deter-mining prognosis.16 There has only been one valid-ation study for the CURB-age score, which wasperformed by the research group that devised it.8

Although CURB-age was theoretically thought to

Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802 77

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

be advantageous, its performance was not considered to besuperior to CURB-65.

The objective of our study was to apply the prognostic toolsCRB-65, CURB-65 and CURB-age in a cohort of predominantlyolder inpatients, in which more patients are likely to meet theextended urea and age criteria, and to compare the associationbetween risk groups as defined by the scores and 30-day mortality.

METHODSWe performed a retrospective cohort study using data from ahospital database of inpatients with CAP (box 1) admitted tothe Marienhospital Herne, Germany (box 2) between 2005 and2009.

Inclusion and exclusion criteriaThe database includes patients ≥18 years old who are admittedfor inpatient treatment of CAP. To ensure that all patients metthe inclusion criteria and no exclusion criteria were present, allpatient records retrieved from the CAP database were checkedby two independent reviewers (MAP and TW).Patients were considered to have pneumonia when presenting▸ with at least one classic symptom (new onset of cough, puru-

lent sputum, dyspnoea and/or fever >38.0°C),▸ without an alternative source of infection and▸ with either signs of inflammation (elevated C-reactive

protein or leucocytosis)▸ or typical radiological findings within 48 h of admission.We excluded patients▸ with infectious exacerbation of chronic obstructive pulmon-

ary disease,▸ undergoing chemotherapy and/or radiation,▸ with primary or secondary immunodeficiency (including

HIV) or▸ with other types of pneumonia such as aspiration pneumonia.

Further data collectionComorbidity was assessed with the Charlson Comorbidity Index(CCI)17 18 by linking the database to the hospital informationsystem, which provided International Classification of Diseases10th revision (ICD-10)-coded concomitant diagnoses and dis-eases. Urea levels on admission, as well as further laboratoryvalues, were gathered from the central laboratory server, again bytwo reviewers (MAP and TW). The central server stores allvalues assessed in the laboratory since 2001. Information on sur-vival status or date of death was provided by the city’s Office of

Statistics and Election in Herne, Germany, on 6 September 2012.Microbiological data were not obtained. They are not part of theCAP database and have been recommended by national guide-lines and local policy for only a minority of patients.

Sample size calculationTo calculate the sample size, available data on inpatients withCAP who were admitted to our institution between 2001 and2005 were used. For risk groups as defined by CURB-65, wefound a 30-day mortality of 5% for the low-risk group and>12% mortality for the combined group of intermediate- andhigh-risk patients. We assumed that these two proportionswould also apply to our current sample. For detection of at leastthis difference, a sample size of 248 patients per group is neces-sary, or approximately 500 patients in total (χ2 test, significancelevel p=0.05, power 0.80).

Data analysisBaseline characteristics of patients are reported with absoluteand relative frequencies for categorical variables and with mean,median and range for continuous variables. To assess the associ-ation between baseline characteristics and 30-day mortality, wecalculated ORs and their corresponding 95% CIs. From the rawdata, we calculated the risk scores CRB-65, CURB-65 andCURB-age as well as score-defined risk categories. Thirty-daymortality was compared across the scores. We tested the associ-ation between the scores and mortality using Pearson’s χ2 test.In addition, sensitivity, specificity and the corresponding 95%CI were calculated for each point and risk category of the sever-ity scores. To evaluate the ability to predict the primary endpoint—death due to any cause 30 days after admission—wecompared the different severity scores in a receiver operatingcharacteristic (ROC) curve analysis. For the area under thecurve (AUC), a 95% CI is provided. For all analyses, a two-sidedp value of <0.05 was considered to be significant.

Data analysis was performed using SPSS for Windows V.21.The software CI analysis, V.2.2.0 (T. Bryant, 2011), was used tocalculate CIs for sensitivity and specificity.

EthicsThis study was performed in accordance with the Declaration ofHelsinki of the World Medical Association.19 The study wasapproved by the institutional review board of the University ofBochum, Germany, on 6 August 2012.

RESULTSPatient characteristicsFrom the CAP database, 587 patients were considered eligible,and 559 patients met all the inclusion criteria. Twenty-eightpatients were excluded because of one of the following exclu-sion criteria: advanced healthcare directive such as in palliative

Box 1 Community-acquired pneumonia (CAP) database

In 2005, a nationwide quality assurance programme was mademandatory for all German hospitals providing acute medicalcare for CAP inpatients. The programme is organised by theNational Institute of Quality in Healthcare which is responsiblefor quality assurance for inpatients in Germany. By means of apredefined report sheet, the hospitals have to assess and reportthe following variables for each CAP inpatient: age, sex, referralmode (from private home, nursing home or another hospital),respiratory rate, systolic and diastolic blood pressure, presenceof confusion, need of invasive or non-invasive ventilatorysupport on admission, clinical condition at discharge andoutcome (death or survival). The data allow calculation of theCRB-65 severity score. An analysis of the data for the years2005 and 2006 has been published.8

Box 2 Setting

The Marienhospital Herne is a 575-bed hospital of the University ofBochum, Germany. It operates departments specialising incardiology, gastroenterology, nephrology, oncology and geriatrics,among others. The hospital serves 21 000 inpatients andapproximately 50 000 outpatients annually. Despite being auniversity hospital, it provides free, unrestricted and immediatehealthcare to the population of Herne.

78 Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

situations (18 patients), invasive ventilation on admission (6patients), or referral from another hospital (4 patients). Inalmost 90% of patients included, the CAP diagnosis was radio-logically confirmed. Restricting the analysis to patients withradiologically confirmed CAP did not substantially alter theresults. Therefore, we only present results for the whole sampleof 559 patients.

Patients included in the study had a mean (median; range) ageof 74.1 years (78.2; 18–104). Female patients (250, 44.7%) hada mean (median; range) age of 77.8 years (81.7; 18–104),whereas male patients (309, 55.3%) had a mean (median;range) age of 71.2 years (76.0; 23–95). Baseline characteristicsare shown in table 1.

The 30-day mortality with CRB-65, CURB-65 and CURB-ageThe 30-day mortality for this group of patients was 10.9%. Theassociation between mortality and the risk assessment scores isdepicted in tables 2 and 3, which are stratified by score pointsor risk categories, respectively. In general, mortality increasedalong with rising score points as well as higher risk categoriesfor all three scores. Sensitivity for severe pneumonia (cut-offpoints ≥3 points for CRB-65 and CURB-65, and ≥4 points forCURB-age) was highest for CURB-age (52.5%). Specificity washighest for CRB-65 (93.4%). The ROC curve analysis, providedin figure 1 and table 4, revealed that CURB-age defined risk cat-egories better than the other scores.

Differentiation of patients with severe/non-severe andlow-risk/non-low-risk CAP in CRB-65, CURB-65 andCURB-ageCURB-age (CURB-65/CRB-65) predicted 195 (214/89) patientsto be at low risk (CURB-age ≤1 points, CURB-65 ≤1 points,CRB-65 0 points), where 190 (205/86) patients or 97.4%(95.8%/96.6%) survived 30 days of admission.

CURB-age (CURB-65/CRB-65) predicted 121 (144/42)patients to have severe CAP (CURB-age ≥4 points, CURB-65≥3 points, CRB-65 ≥3 points), of which 32 (28/9) patients or26.4% (19.4%/21.4%) died within 30 days of admission.

Thirty-day mortality and association with patientcharacteristicsThe 30-day mortality was significantly higher in the subgroupof patients who were admitted from nursing homes, had

dementia, or were 65 years or older. It also increased with CCIcategories (8.5%, 10.7% and 14.7%).

DISCUSSIONThe main finding of our study is that the new variant of theCURB-65 score called CURB-age showed a closer associationwith 30-day mortality in adult inpatients with CAP than thecommonly used CRB-65 and CURB-65 scores. In comparisonwith CRB-65, the low-risk groups defined by CURB-age wastwice as large (195 vs 89 patients), but showed lower mortality(2.6% vs 3.4%). The low-risk group defined by CURB-age wasonly slightly smaller than that defined by CURB-65 (195 vs 214patients), but again with lower mortality (2.6% vs 4.2%). InROC curve analysis, the largest AUC —that is, the best relationbetween sensitivity and specificity—corresponded to CURB-age.Improvements in the identification of low-risk patients wouldhave the most clinical impact. Current guidelines recommendconsidering ambulatory management when providing care forlow-risk patients with CAP.20 21 Doing so could reduce health-care costs, save valuable hospital resources, and lower risks asso-ciated with hospital stays. Therefore, better identification oflow-risk patients would be beneficial to both patients and thehealthcare system.

Our findings differ markedly from the validation study ofMyint et al.8 In that study’s 190 inpatient sample, the sensitivityof both CURB-age and CURB-65 was considerably lower thanthat found in our study, namely 50% for CURB-age and 59%

Table 1 Characteristics of the study cohort

Risk factorDeath(n=61)

Survival(n=498) OR (95% CI)

Female sex 31 (50.8) 219 (44.0) 1.32 (0.77 to 2.24)Age ≥65 years 55 (90.2) 379 (76.1) 2.88 (1.21 to 6.85)Nursing home residency 35 (57.4) 111 (22.3) 4.69 (2.71 to 8.13)Congestive heart failure 17 (27.9) 105 (21.1) 1.44 (0.79 to 2.63)Cerebrovascular disease 16 (26.2) 98 (19.7) 1.45 (0.79 to 2.68)Dementia 32 (52.5) 140 (28.1) 2.82 (1.65 to 4.84)Chronic lung disease 14 (23.0) 146 (29.3) 0.72 (0.38 to 1.35)Diabetes mellitus 14 (23.0) 144 (28.9) 0.73 (0.39 to 1.37)Renal disease 20 (32.8) 128 (25.7) 1.41 (0.80 to 2.50)Charlson Comorbidity Index0–1 points 18 (29.5) 213 (42.8) 1.00 (reference)2–3 points 21 (34.4) 196 (39.4) 1.27 (0.66 to 2.45)>3 points 22 (36.1) 128 (25.7) 2.03 (1.05 to 3.94)

Values are number (%).

Table 2 Mortality stratified by score points of three severity scores

Scorepoints

All(n=559),n

30-daymortality(n=61), n(%)

Sensitivity, %(95% CI)

Specificity, %(95% CI)

CRB-650 89 3 (3.4) 100 01 272 24 (8.8) 95.1 (86.5 to 98.3) 17.3 (14.2 to 20.8)2 156 25 (16.0) 55.7 (43.3 to 67.5) 67.1 (62.8 to 71.1)3 39 9 (23.1) 14.8 (8.0 to 25.7) 93.4 (90.8 to 95.2)4 3 0 (0) 0 (0 to 5.9) 99.4 (98.2 to 99.8)

CURB-650 74 2 (2.7) 100 01 140 7 (5.0) 96.7 (88.8 to 99.1) 14.5 (11.6 to 17.8)2 201 24 (11.9) 85.2 (74.3 to 92.0) 41.2 (36.9 to 45.5)3 117 20 (17.1) 45.9 (34.0 to 58.3) 76.7 (72.8 to 80.2)4 26 8 (30.8) 13.1 (6.8 to 23.8) 96.2 (94.1 to 97.5)5 1 0 (0) 0 (0 to 5.9) 99.8 (98.9 to 99.9)

CURB-age0 74 2 (2.7) 100 0

1 121 3 (2.5) 96.7 (88.8 to 99.1) 14.5 (11.6 to 17.8)2 114 12 (10.5) 91.8 (82.2 to 96.5) 38.2 (34.0 to 42.5)3 129 12 (9.3) 72.1 (59.8 to 81.8) 58.6 (54.3 to 62.9)4 73 12 (16.4) 52.5 (40.2 to 64.5) 82.1 (78.5 to 85.2)5 42 18 (42.9) 32.8 (22.3 to 45.3) 94.4 (92.0 to 96.1)6 6 2 (33.3) 3.3 (0.9 to 11.2) 99.2 (98.0 to 99.7)7 0 0 (0) 0 (0 to 0.1) 100 (99.1 to 100)

CRB-65: 1 point for each characteristic at admission: confusion, respiratory rate ≥30/min, low blood pressure systolic (<90 mm Hg) or diastolic (≤60 mm Hg), age≥65 years. CURB-65: 1 point for each characteristic at admission: confusion, urea>7 mmol/L, respiratory rate ≥30/min, low blood pressure systolic (<90 mm Hg) ordiastolic (≤60 mm Hg), age ≥65 years. CURB-age: characteristic at admission:confusion=1, urea >7 mmol/L=1, urea >11 mmol/L=2, respiratory rate ≥30/min=1,low blood pressure systolic (<90 mm Hg) or diastolic (≤60 mm Hg)=1, age≥65 years=1, age ≥85 years=2.

Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802 79

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

for CURB-65. In several other studies that investigatedCURB-65, sensitivity and specificity were reported to varybetween 80–95% and 50–60%, respectively,6 7 22 23 which arewell in line with our data. An important finding in the Myintstudy was the high mortality.8 In total, 54 out of 190 inpatients(28.4%) died during follow-up. This is much higher than themortality of ∼11% in our study, which agrees well with that ofall adult inpatients in Germany.24 There was no significant dif-ference in age between the Myint study (median age 76 years)and our study sample (median age 78 years), thus the highermortality in the former study may suggest that the patients hadmore comorbidities. Comorbidities were also present in ourpatient sample, with more than 25% suffering from chroniclung disease, kidney disease or diabetes. Furthermore, functionalimpairment was prevalent, with nearly a quarter of our patientsresiding in nursing homes and about a third having dementia.Advanced comorbidity was present in almost two-thirds of ourinpatients, as indicated by the CCI. However, these

characteristics do not imply that our sample is truly comparableto that of Myint et al.8

Two prognostically relevant variables, age and urea level onadmission, are potentially undervalued in the establishedCURB-65 score, justifying the need for the new CURB-agescore.8 15 In the past, age and urea have only been included inseverity scores as dichotomous variables, scoring 1 point each ifage was ≥65 years and urea ≥7 mmol/L. Two new scoring cat-egories have since been proposed, extending the age and ureacriteria to ≥85 years and ≥11 mmol/L, respectively, and witheach scoring two points. Both cut-off points are considered tocorrelate with mortality.9 16 25 Extending the already adoptedurea and age criteria allows a simple transition towards the useof the CURB-age score. In addition, the use of both variableshas already been proposed by several guidelines.20 21 The imple-mentation of the extended criteria would not require much add-itional work, and doing so has the potential to improve severityassessment and possibly CAP management.

There are several limitations of our study that should beacknowledged. First, the data obtained from the patient registrywere initially acquired for the purposes of external quality assur-ance. In our perspective, this probably does not affect the valid-ity of the results obtained. In fact, the database contains most ofthe variables required for analysis, including all criteria neces-sary for the calculation of CRB-65. Urea level on admission wasnot included, but was available from a central laboratory serverthat has stored blood variables obtained for all patient admis-sions since 2001. Urea values are collected routinely on hospitaladmission. Retrieving these values means that our patiententries do not have any missing values. Finally, vital status toassess 30-day mortality was obtained from the official localregistration office, which is reliable even when acquiredretrospectively.

Table 3 Mortality stratified by risk categories defined by three severity scores

Risk category All (n=559), n 30-day mortality (n=61), n (%) Sensitivity, % (95% CI) Specificity, % (95% CI)

CRB-65Low 89 3 (3.4) 100 0Intermediate 428 49 (11.4) 95.1 (86.5 to 98.3) 17.3 (14.2 to 20.8)High 42 9 (21.4) 14.8 (8.0 to 25.7) 93.4 (90.8 to 95.2)

CURB-65Low 214 9 (4.2) 100 0Intermediate 201 24 (11.9) 85.2 (74.3 to 92.0) 41.2 (36.9 to 45.5)High 144 28 (19.4) 45.9 (34.0 to 58.3) 76.7 (72.8 to 80.2)

CURB-ageLow 195 5 (2.6) 100 0Intermediate 243 24 (9.9) 91.8 (82.2 to 96.4) 38.2 (34.0 to 42.5)High 121 32 (26.4) 52.5 (40.2 to 64.5) 82.1 (78.5 to 85.2)

Risk categories: CRB-65: low risk=0 point, intermediate risk=1–2 points, high risk=3–4 points; CURB-65: low risk=0–1 points, intermediate risk=2 points, high risk=3–5 points;CURB-age: low risk=0–1 points, intermediate risk=2–3 points, high risk=4–7 points.

Figure 1 Receiver operating characteristic (ROC) curves for riskcategories of three severity scores. For definition of risk categories, seetables 2 and 3.

Table 4 Receiver operating characteristic curve analysis for riskcategories of three severity scores

Risk category Area under the curve 95% CI

CRB-65 0.591 (0.518 to 0.664)CURB-65 0.668 (0.600 to 0.736)CURB-age 0.730 (0.666 to 0.794)

For definition of risk categories, see tables 2 and 3.

80 Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

Furthermore, data were retrieved solely from a single centre.As the hospital is a primary care centre that provides acutemedical care to the population, patient selection as expectedwith a tertiary referral centre is unlikely. However, the character-istics of both the hospital and the population served are likely todiffer in other settings. Therefore, the findings of this studycannot be transferred to other settings without reservation. Inaddition, only inpatients were evaluated. We cannot claim thatlow-risk patients as defined in our sample can be treated safelyas outpatients. This conclusion can only be made with rando-mised controlled trials that adequately investigate this question.

Finally, our findings certainly do not prove superiority ofCURB-age over CURB-65 or CRB-65. Although the sample sizein our study was almost three times larger than that in the studyof Myint et al,8 we still could not perform inference tests—forexample, to detect differences in AUC values from ROC curveanalysis. Therefore, further larger studies assessing the potentialof CURB-age for better identification of low-risk patients areneeded.

Main messages

▸ As common severity scores do not perform optimally inolder inpatients with community-acquired pneumonia (CAP),a score variant of the standard CURB-65, named CURB-age,has been proposed.

▸ In a sample of more than 500 inpatients from a singlecentre in Germany, CURB-age showed a closer associationwith 30-day mortality than standard scores.

▸ The use of CURB-age may improve risk stratification andhence management of inpatients with CAP.

Current research questions

▸ Can low-risk patients with community-acquired pneumonia(CAP) be treated safely as outpatients?

▸ Do different risk-based treatment strategies improve theoutcome of high-risk patients?

▸ Are different risk-based treatment strategies able to reduceside effects in low-risk patients?

Key references

▸ Lim WS, van der Eerden MM, Laing R, et al. Definingcommunity acquired pneumonia severity on presentation tohospital: an international derivation and validation study.Thorax 2003;58:377–82.

▸ Myint PK, Sankaran P, Musonda P, et al. Performance ofCURB-65 and CURB-age in community-acquired pneumonia.Int J Clin Pract 2009;63:1345–50.

▸ Loke YK, Kwok CS, Niruban A, et al. Value of severity scales inpredicting mortality from community-acquired pneumonia:systematic review and meta-analysis. Thorax 2010;65:884–90.

▸ Ewig S, Bauer T, Richter K, et al. Prediction of in-hospitaldeath from community-acquired pneumonia by varying CRB-age groups. Eur Respir J 2013;41:917–22.

Author affiliations1School of Medicine, University of Bochum, Bochum, Germany2College of Medicine, University of Arizona, Tucson, USA3Department of Geriatrics, Marienhospital Herne, University of Bochum, Herne,Germany4Department of Geriatrics, HELIOS Klinikum Schwelm, University of Witten/Herdecke,Schwelm, Germany5Department of Medical Informatics, Biometry and Epidemiology, University ofBochum, Bochum, Germany

Acknowledgements We sincerely thank Patricia Vierhaus, Department of MedicalInformatics, Biometry and Epidemiology, University of Bochum, Germany, for datahandling and technical assistance. We also thank the staff of the Office of Statisticsand Elections, Herne, Germany, for their cooperation.

Contributors MAP collected, analysed and interpreted the data, performed theliterature review, and drafted the manuscript. TT contributed to data analysis,interpretation, and drafting of the manuscript. TW contributed to the design of thestudy, collected the data, and supported data analysis and interpretation. HNsupported data analysis and interpretation. HJH and LP designed the study andinterpreted the data. UT designed the study, contributed to the collection of data,performed analysis and interpretation of the data, and assisted in drafting themanuscript. All authors read and approved the final version of the manuscript.All authors fulfil the criteria for authorship, as mentioned in the ContributorshipStatement.

Competing interests None.

Ethics approval University of Bochum, Germany (approval given 6 August 2012).

Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES1 Hoyert DL, Xu J. Deaths: preliminary data for 2011. Natl Vital Stat Rep

2012;51:1–51.2 Anonymous. Causes of death in Germany 2012 [Todesursachen in Deutschland].

Fed Stat Office Germany 2013;12.3 American Lung Association. Lung Disease Data: 2008. 2008.4 Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients

with community-acquired pneumonia. N Engl J Med 1997;336:243–50.5 Serisier DJ, Williams S, Bowler SD. Australasian respiratory and emergency

physicians do not use the pneumonia severity index in community-acquiredpneumonia. Respirology 2013;18:291–6.

6 Lim WS, van der Eerden MM, Laing R, et al. Defining community acquiredpneumonia severity on presentation to hospital: an international derivation andvalidation study. Thorax 2003;58:377–82.

7 Capelastegui A, Espana PP, Quintana JM, et al. Validation of a predictive rule forthe management of community-acquired pneumonia. Eur Respir J 2006;27:151–7.

8 Myint PK, Sankaran P, Musonda P, et al. Performance of CURB-65 and CURB-agein community-acquired pneumonia. Int J Clin Pract 2009;63:1345–50.

9 Parsonage M, Nathwani D, Davey P, et al. Evaluation of the performance ofCURB-65 with increasing age. Clin Microbiol Infect 2009;15:858–64.

10 McNally M, Curtain J, O’Brien KK, et al. Validity of British Thoracic Society guidance(the CRB-65 rule) for predicting the severity of pneumonia in general practice:systematic review and meta-analysis. Br J Gen Pract 2010;60:e423–33.

11 Chalmers JD, Singanayagam A, Akram AR, et al. Severity assessment tools forpredicting mortality in hospitalised patients with community-acquired pneumonia.Systematic review and meta-analysis. Thorax 2010;65:878–83.

12 Loke YK, Kwok CS, Niruban A, et al. Value of severity scales in predicting mortalityfrom community-acquired pneumonia: systematic review and meta-analysis. Thorax2010;65:884–90.

13 Ochoa-Gondar O, Vila-Corcoles A, Rodriguez-Blanco T, et al. Comparison of threepredictive rules for assessing severity in elderly patients with CAP. Int J Clin Pract2011;65:1165–72.

14 Myint PK, Kamath AV, Vowler SL, et al. The CURB (confusion, urea, respiratory rateand blood pressure) criteria in community-acquired pneumonia (CAP) in hospitalisedelderly patients aged 65 years and over: a prospective observational cohort study.Age Ageing 2005;34:75–7.

15 Myint PK, Kamath AV, Vowler SL, et al. Simple modification of CURB-65 betteridentifies patients including the elderly with severe CAP. Thorax 2007;62:1015–16;author reply 1016.

16 Metersky ML, Waterer G, Nsa W, et al. Predictors of in-hospital vs postdischargemortality in pneumonia. Chest 2012;142:476–81.

17 Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognosticcomorbidity in longitudinal studies: development and validation. J Chronic Dis1987;40:373–83.

18 Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for definingcomorbidities in ICD-9-CM and ICD-10 administrative data. Med Care2005;43:1130–9.

Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802 81

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

19 World Medical Association. WMA Declaration of Helsinki, Ethical Principles forMedical Research Involving Human Subjects. 55th WMA General Assembly, Tokyo,Japan 2004.

20 Hoffken G, Lorenz J, Kern W, et al. [Epidemiology, diagnosis, antimicrobial therapyand management of community-acquired pneumonia and lower respiratory tractinfections in adults. Guidelines of the Paul-Ehrlich-Society for Chemotherapy, theGerman Respiratory Society, the German Society for Infectiology and theCompetence Network CAPNETZ Germany]. Pneumologie 2009;63:e1–68.

21 Woodhead M, Blasi F, Ewig S, et al. Guidelines for the management of adult lowerrespiratory tract infections—full version. Clin Microbiol Infect 2011;17(Suppl 6):E1–59.

22 Menendez R, Martinez R, Reyes S, et al. Biomarkers improve mortality prediction byprognostic scales in community-acquired pneumonia. Thorax 2009;64:587–91.

23 Myint PK, Kamath AV, Vowler SL, et al. Severity assessment criteria recommendedby the British Thoracic Society (BTS) for community-acquired pneumonia (CAP) andolder patients. Should SOAR (systolic blood pressure, oxygenation, age andrespiratory rate) criteria be used in older people? A compilation study of twoprospective cohorts. Age Ageing 2006;35:286–91.

24 Ewig S, Birkner N, Strauss R, et al. New perspectives on community-acquiredpneumonia in 388 406 patients. Results from a nationwide mandatoryperformance measurement programme in healthcare quality. Thorax2009;64:1062–9.

25 Ewig S, Bauer T, Richter K, et al. Prediction of in-hospital death fromcommunity-acquired pneumonia by varying CRB-age groups. Eur Respir J2013;41:917–22.

Warwick university courses

13th – 16th July 2015: Warwick University Short Course. ‘Techniques and Applications ofMolecular Biology: A Course for Medical Practitioners’.A four day course for those in the medical profession wishing to improve their under-standing of the principles and applications of genetic engineering techniques. Optionalaccreditation leads to a masters level Postgraduate Award. Details: Dr Charlotte Moonan,School of Life Sciences, University of Warwick, Coventry, CV4 7AL (Tel: 024 7652 3540;email Charlotte.Moonan@warwick.ac.uk;website http://www2.warwick.ac.uk/fac/sci/lifesci/study/shortcourses/molecularbiology).

29th June – 10th July 2015: Warwick University Short Course. ‘Laboratory Skills’.A ten day course for those wishing to gain hands-on practical experience of vital and fun-damental laboratory techniques. Details: Dr Charlotte Moonan, School of Life Sciences,University of Warwick, Coventry, CV4 7AL (Tel: 024 7652 3540;email Charlotte.Moonan@warwick.ac.uk;website http://www2.warwick.ac.uk/fac/sci/lifesci/study/shortcourses/labskills/ ).

82 Pflug MA, et al. Postgrad Med J 2015;91:77–82. doi:10.1136/postgradmedj-2014-132802

Original article

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from

validation of a modified CURB-65 scorecommunity-acquired pneumonia: external Short-term mortality of adult inpatients with

Hans Jürgen Heppner, Ludger Pientka and Ulrich ThiemMarc Andre Pflug, Timothy Tiutan, Thomas Wesemann, Harald Nüllmann,

doi: 10.1136/postgradmedj-2014-1328022015

2015 91: 77-82 originally published online January 24,Postgrad Med J 

http://pmj.bmj.com/content/91/1072/77Updated information and services can be found at:

These include:

References #BIBLhttp://pmj.bmj.com/content/91/1072/77

This article cites 22 articles, 11 of which you can access for free at:

serviceEmail alerting

box at the top right corner of the online article. Receive free email alerts when new articles cite this article. Sign up in the

CollectionsTopic Articles on similar topics can be found in the following collections

(20)Pneumonia (respiratory medicine) (332)Epidemiology

Notes

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://group.bmj.com/subscribe/To subscribe to BMJ go to:

group.bmj.com on February 13, 2015 - Published by http://pmj.bmj.com/Downloaded from