Original Article

Validity of Pneumonia Severity Index and CURB-65 Severity

Scoring Systems in Community Acquired Pneumonia in an
Indian Setting
Bashir Ahmed Shah1, Wasim Ahmed1, Ghulam Nabi Dhobi1, Naveed Nazir Shah 2, Syed Quibtiya
Khursheed2 and Inaamul Haq3

Department of General Medicine, Sher-i-Kashmir Institute of Medical Sciences1, Department of Chest Medicine,
Government Medical College2, Srinagar, Jammu and Kashmir and Department of Community Medicine 3,
Mamata Medical College, Khammam, Andhra Pradesh, India

ABSTRACT

Background. Little information is available from India regarding prognostic factors in patients with community acquired
pneumonia (CAP).
Methods. Hospital-based prospective study to test the validity of pneumonia severity index (PSI) and the confusion, urea,
respiratory rate, blood pressure, age over 65 years (CURB-65) risk scoring systems in patients with CAP (n=150).
Results. Although both CURB-65 class ≥III and PSI class ≥IV were 100% sensitive in predicting death, CURB-65 class ≥III
had a higher specificity (74.6%) than PSI class ≥IV (52.2%) when used to predict death. In both PSI and CURB-65 risk scoring
systems, mortality rate, need for intensive care unit (ICU) admission, prolonged need for intravenous (I.V.) antibiotics,
prolonged duration of hospital stay and need for admission to ICU increased progressively with increasing scores. The PSI
class ≥IV was more sensitive in predicting ICU admission than CURB-65. The duration of hospital stay was found to have
a weak but significant correlation with PSI and CURB-65 criteria. Defervescence time also had a very weak but significant
correlation with PSI and CURB-65 criteria. Duration of I.V. antibiotics had a moderately strong correlation with CURB-65
criteria but a weak correlation with PSI criteria.
Conclusions. Both PSI and CURB-65 were found to have equal sensitivity to predict death from CAP. Specificity of CURB-
65 was higher than that of PSI. However, PSI was more sensitive in predicting ICU admission than CURB-65.
[Indian J Chest Dis Allied Sci 2010;52:9-17]

Key words: Community acquired pneumonia, CURB-65, Pneumonia severity index.

INTRODUCTION than 45% to 70% of patients.4,5 Streptococcus pneumoniae

is the most commonly isolated pathogen responsible for
Community acquired pneumonia (CAP) is a common 35% to 60% of cases.6,7 Studies reported during the last
disorder with an incidence of about 20% to 30% in two decades from India have also reported a higher
developing countries compared to an incidence of 3% to prevalence of Klebsiella pneumoniae among culture-
4 % in developed countries.1-3 The incidence varies positive pneumonias.8-10 In two Indian studies from
markedly with age, being much higher in the very New Delhi, the prevalence of Mycoplasma pneumoniae
young and the elderly. It is estimated that India together has been reported to be 35% in adults 11 and 27.4% in
with Bangladesh, Indonesia and Nepal account for 40% children.12
of global acute respiratory infection; 90% of mortality is The reported mortality of adults admitted to
due to pneumonia, mostly bacterial in origin.2 hospital with CAP has varied widely (4%–21%).13-15
The cause of CAP is often difficult to establish and While the British Thoracic Society (BTS) multi-centric
despite the recent progress it takes a few days to study recorded a surprisingly low mortality of 5.7%, 16 a
identify the causative micro-organism in the blood or higher mortality (ranging from 21%-25%) has been
sputum samples. The aetiology of CAP remains reported in other studies.17,18 Though definite statistics
uncertain in many patients. Even with the use of are lacking CAP remains a leading cause of death in
extensive laboratory testing and invasive procedures; India too.7 The mortality in a study of CAP reported by
aetiological confirmation being achieved in no more Bansal et al6 was 11 percent. In another Indian study,19 a
[Received: November 27, 2008; accepted after revision: June 23, 2009]
Correspondence and reprint requests: Dr Naveed Nazir Shah, Post Box No 1178, General Post Office, Srinagar-190001,
Jammu and Kashmir, India; Phone: 91-9419016438; Fax: 91-194-2422383; E-mail: [email protected]
10 Scoring Systems in Community Acquired Pneumonia B.A. Shah et al

significantly higher mortality was noticed in patients other immunosuppressive agents, neutropenic patients
aged 50 years or above and in those with underlying co- with absolute neutrophil count <1000/mm 3 ); (iii)
morbid conditions. The mortality of patients with patients hospitalised within previous 14 days; and (iv)
severe CAP requiring admission to an intensive care patients with an alternate diagnosis during follow-up.
unit (ICU) is high. This is likely to be particularly At the time of initial evaluation, the selected
evident in health services where ICU beds are at a patients underwent a complete clinical history and
premium such that only critically ill patients in need of examination; chest radiograph (postero-anterior and
assisted ventilation can be admitted. In the UK, ICU lateral views) at presentation and repeated after 48
based studies report mortality rates of over 50 hours; electrocardiogram; arterial blood gas analysis
percent.14,15,20,21 and serum electrolyte measurement; sputum for gram
It is hoped that the knowledge of relevant staining and culture; blood cultures (in selected
prognostic factors might be useful for early patients); complete blood counts, blood urea nitrogen
identification of patients at high risk requiring intensive and serum creatinine; fasting blood glucose, serum
care treatment. Prognostic scoring systems for CAP bilirubin, aspartate aminotransferase (AST), alanine
have been developed to address these issues. The two aminotransferase (ALT), alkaline phosphatase (ALP),
prominent tools for this purpose are the pneumonia total proteins, serum albumin; lactate dehydrogenase
severity index (PSI), developed in the USA after (LDH) levels, PSI scoring and CURB-65 scoring on the
pneumonia outcome research trial (PORT), and the BTS basis of the points (Tables 1 and 2).
rule, which has recently been modified to the CURB-65 Other investigations like pleural fluid analysis,
rule “confusion, elevated blood urea nitrogen, elevated computed tomography (CT) of the chest, broncho-
respiratory rate, low systolic or diastolic blood pressure
(BP), and age over 65 years (CURB-65)” rule.22,23 The two Table 1. Pneumonia severity index (PSI) scoring
scoring approaches are viewed as being complementary, Patient Characteristics Points
as each has different strengths and weaknesses.
Even though most of the burden in terms of Demographics
Age(years): Male: age —
mortality and morbidity occurs in the developing world,
Female: age —
little has been done to study the factors associated with Nursing home resident +10
an adverse prognosis in CAP in this region. Further, the Co-morbidities
scoring systems currently employed in the western world Neoplastic disease +30
Liver disease +20
have not been validated in developing countries where
Congestive heart failure +10
population demographics and health-care delivery Cerebrovascular disease +10
systems are totally different from the developed world. Renal disease +10
The aim of our study was to test the validity of PSI and Examination findings
CURB-65 severity scoring systems in CAP in an Indian Altered mental status +20
Respiratory rate ³30/minute +20
setting. Systolic blood pressure <90 mmHg +20
Temperature <35oC or ³40oC +15
Pulse ³125/minute +10
MATERIAL AND METHODS Laboratory findings
pH <7.35 (do ABG only if hypoxic +30
In this hospital-based prospective study, 150 patients or COPD)
with CAP attending the out-patient as well as in-patient BUN >10.7 mmol/ L +20
Sodium <130 mEq/L +20
departments of Sher-i-Kashmir Institute of Medical
Glucose ³13.9 mmol/L +10
Sciences (SKIMS), a tertiary care Institute in an urban Hematocrit <0.30 +10
area of Kashmir, India, were studied. Patients PaO2 <60mmHg or oxygen saturation <90% +10
presenting with any opacity on chest radiograph Pleural effusion +30
consistent with the diagnosis of acute pneumonia,
associated with respiratory symptoms, infectious
syndrome and lack of an alternate diagnosis, were
diagnosed to have CAP.24 Risk Class Score
The following patients were excluded from the Low I <51
study: (i) patients known to be positive for human Low II 51 - 70
Low III 71 - 90
immunodeficiency virus (HIV); (ii) chronically
Medium IV 90 - 130
immunosuppressed patients (defined as High V >130
immunosuppression for solid organ transplantation,
post-splenectomy, receiving >10mg/day of pre- dnisone
Hospitalisation is recommended for class IV and V. Class III
or the equivalent for more than 30 days, treatment with is based on clinical judgement
2010;Vol.52 The Indian Journal of Chest Diseases & Allied Sciences 11

Table 2. CURB-65 criteria scoring Table 3. Comparison of various clinical variables in pa-
tients who survived and died
Confusion
Blood urea >7 mmol/L at the time of admission. Variable Survived Died
Respiratory Rate of ≥ 30/minute No. (%) No. (%)
Systolic BP ≤ 90 mmHg or diastolic BP ≤ 60mmHg
Age ≥ 65 years Patients 134 (89.3) 16 (10.7)
Mean age (±SD), years 54.3±16.6 67.2±4.5
A score of 1 is given for presence of each of the variables Male sex 76 (56.7) 13 (81.3)
BP=Blood pressure Smokers 73 (54.5) 16 (100)
Pre-hospitalisation antibiotics 79 (59) 16 (100)
alveolar lavage (BAL) were done depending on the Cough 119 (88.8) 16 (100)
Purulent sputum 101 (75.4) 15 (93.8)
clinical scenario of the patient. Haemoptysis 19 (14.2) 0 (0)
At the clinical end points (hospital discharge or Chest pain (pleuritic) 61 (45.5) 0 (0)
death) the following parameters were recorded: (i) Confusion 33 (31.3) 14 (87.5)
duration of antibiotics; (ii) time taken for defervescence; Pleural effusion* 43 (24.6) 3* (18.8)
Temperature >100 oF 91 (67.9) 3 (18.8)
(iii) need for mechanical ventilation; (iv) need of Pulse >100/min 73 (54.5) 5 (31.2)
admission to ICU; and (v) condition at 30 days after Systolic BP £90mmHg 24 (18) 3 (18.75)
discharge from the hospital. Diastolic BP £60mmHg 30 (22.4) 6 (37.5)
In the present study, in-hospital death or death Respiratory rate ³30/min 77 (57.5) 16 (100)
D-Dimer positive 77 (57.5) 16 (100)
within 30 days of discharge was the main outcome
Hypertension 46 (34.3) 8 (50.0)
studied. Requirement for ICU admission was studied as Diabetes mellitus 22 (16.4) 2 (12.5)
a marker for “severe pneumonia”. Factors associated COPD 4 (3.0) 5 (31.3)
with prolonged duration of antibiotic therapy (defined *=Two of these patients had transudative effusion because of
as need for antibiotics for more than five days), congestive heart failure and only one had synpneumonic
effusion
prolonged hospital stay (defined as hospital stay for
COPD=Chronic obstructive pulmonary disease; BP=Blood pres-
more than seven days) and prolonged time to sure
defervescence (more than three days) were also studied.
Defervescence was defined as resolution of fever, chest The number of patients in different PSI risk classes
pain; respiratory rate ≤ 24 per minute; arterial oxygen is given in table 4A. All the16 patients (100%) who died
saturation (SaO2) of ≥90% while breathing room air; and were in PSI class ≥IV. Mortality in PSI class I to III was
ability to perform basic daily activities without support. 0%; in class IV, 14.1% and Class V, 34.8 percent. Table 4B
represents the sensitivity, specificity, NPV and PPV of
Statistical Analysis different levels of PSI classes for predicting death as an
outcome. Sensitivity and specificity for PSI risk class ≥IV
Sensitivity, specificity, positive predictive value (PPV) to predict death was 100% and 52.2% and PPV and NPV
and negative predictive value (NPV) were calculated for were 20% and 100%, respectively. Mortality in risk class
different PSI and CURB-65 grades with qualitative 0 to II was 0%, in risk class III it was 9.5%, 47.8% in class
variables (death, ICU admissions) as an outcome. The IV and 50% in class V. The sensitivity and specificity
relationship of quantitative variables with PSI and were most favourable for a PSI class ≥IV. Though the
CURB-65 classes was assessed by Spearman’s specificity increased to 88.8% when PSI class V is chosen
correlation co-efficient. as the cut-off, there was an unfavourable drop in the
sensitivity which decreases to 50 percent. The receiver-
RESULTS operating characteristic (ROC) curve for different PSI
classes is shown in figure 1.
In our study (n=150), 89 (59.3%) were males. The mean Table 4A. Mortality in different PSI classes
age (±SD) of males [60.8 (±13.6) years] was higher than Variable PSI Class Total
that of females [48.3 (±17.0) years]. Eighty-nine patients I II III IV V
(59.3%) were smokers of which 74 (83.2%) were males. Number
Clinical characteristics of patients who survived and of patients 25 27 18 57 23 150
died are given in table 3. Eighty-nine patients had one (16.7%) (18%) (12%) (38%) (15.3%) (100%)
or more co-morbidities. The most common co- Deaths 0 0 0 8 8 16
morbidity was hypertension, followed by diabetes (0%) (0%) (0%) (50%) (50%) (100%)
mellitus and chronic obstructive pulmonary disease PSI=Pneumonia severity index
(COPD).
The number and percentage of patients in different
Sixteen patients (10.7%) died. [12 (8%) patients who
risk classes of CURB-65 scoring is given in table 5A.
died in-hospital and four (2.7%) within 30 days after
Table 5B represents the sensitivity, specificity, NPV and
discharge].
12 Scoring Systems in Community Acquired Pneumonia B.A. Shah et al

Table 4B. Sensitivity, specificity, negative and positive predic- Table 5B. Sensitivity, specificity, and negative and positive pre-
tive values for different PSI classes for predicting death dictive values for different CURB-65 classes for predicting
death
PSI Sensitivity Specificity Negative Positive
Class (%) (%) Predictive Predictive CURB-65 Sensitivity Specificity Negative Positive
Value Value class (%) (%) Predictive Predictive
(%) (%) Value (%) Value (%)
³II 100 18.7 100 12.8 ³I 100 20.1 100 13
³III 100 38.8 100 16.3 ³II 100 43.3 100 17.4
³IV 100 52.2 100 20 ³III 100 74.6 100 32
V 50 88.8 93.7 34.8 ³IV 87.5 88.8 98.3 48.3
V 18.8 97.8 91 50
PSI=Pneumonia severity index
CURB-65=Confusion, urea, respiratory rate, blood pressure, age
over 65 years

Figure 1. Receiver-operating characteristic curve for PSI with

death as outcome. PSI=Pneumonia severity index. Figure 2. Receiver-operating characteristic curve for CURB-65
with death as outcome. CURB-65=Confusion, urea, respiratory
PPV of choosing different levels of CURB-65 classes for rate, blood pressure, age over 65 years score.
predicting death as an outcome. The sensitivity and The sensitivity and specificity were most favourable for
specificity of CURB-65 risk class ≥III to predict death a PSI class ≥IV. Though the specificity increases to 94.8%
was 100% and 74.6% and PPV and NPV were 32% and when PSI class V is chosen as the cut-off, there is an
100%, respectively. The sensitivity and specificity were
unfavourable drop in the sensitivity which decreases to
most favourable for a CURB-65 class ≥III. Though the
48.6 percent. The ROC curve for different PSI classes is
specificity increased to 88.8% when CURB-65 class ≥IV
shown in figure 3.
was chosen as the cut-off but there was an unfavourable Tables 7A and 7B represent the sensitivity,
drop in the sensitivity to 87.5 percent. The ROC curve specificity, NPV and PPV of choosing different levels of
for different CURB-65 classes is shown in figure 2. CURB-65 classes for predicting ICU admission as an
Although both CURB-65 class ≥III and PSI class ≥IV
were 100% sensitive in predicting death, CURB-65 class
≥III had a higher specificity (74.6%) than PSI class ≥IV
Table 5A. Mortality in different CURB-65 risk classes

Variable CURB-65 Class Total

0 I II III IV V
Number of 27 31 42 21 23 6 150
patients (18%) (20.7%) (28%) (14%) (15.3%) (4%) (100%)
Deaths 0 0 0 2 11 3 16
(0%) (0%) (0%) (12.5%) (68.75%) (18.8%) (100%)

CURB-65=Confusion, urea, respiratory rate, blood pressure, age over 65

years
(52.2%) when used to predict death.
Tables 6A and 6B represent the sensitivity,
Figure 3. Receiver-operating characteristic curve for PSI with
specificity, NPV and PPV of choosing different levels of
ICU admission as outcome. PSI=Pneumonia severity index,
PSI classes for predicting ICU admission as an outcome. ICU=Intensive care unit.
2010;Vol.52 The Indian Journal of Chest Diseases & Allied Sciences 13

Table 6A. Number of ICU admission in different PSI classes Table 7B. Sensitivity, specificity, negative and positive predic-
tive values for different CURB-65 classes for predicting ICU ad-
Variable PSI Class Total mission
I II III IV V
CURB-65 Sensitivity Specificity Negative Positive
Number of 25 27 18 57 23 150 Class (%) (%) Predictive Predictive
patients (16.7%) (18%) (12%) (38%) (15.3%) (100%) Value (%) Value (%)
ICU 0 0 0 18 17 35
admissions (0%) (0%) (0%) (51.4%) (48.6%) (100%) ≥I 100 23.5 100 28.5
≥ II 100 50.4 100 38.0
ICU=Intensive care unit; PSI=Pneumonia severity index ≥ III 91.4 84.4 97 64.0
≥ IV 74.3 97.4 92.6 89.7
Table 6B. Sensitivity, specificity, negative and positive predic- V 17.1 100 79.9 100
tive values for different PSI classes in predicting ICU admis-
sion ICU=Intensive care unit; CURB-65=Confusion, urea, respi-
ratory rate, blood pressure, age over 65 years
PSI Sensitivity Specificity Negative Positive
Class (%) (%) Predictive Predictive
Value (%) Value (%) ≥IV (60.9%).
The PSI risk class was also significantly associated
³II 100 21.7 100 28
³III 100 45.2 100 35.7 with the admission to ICU (p<0.001), prolonged
³IV 100 60.9 100 43.8 duration of antibiotics (p<0.001), time to defervescence
V 48.6 94.8 85.8 73.9 (p=0.007) and prolonged duration of hospital stay
PSI=Pneumonia severity index (p<0.001) (Table 8). Similarly, CURB-65 score was also
significantly associated with the need for ICU
outcome. The sensitivity and specificity were most admission (p<0.001), prolonged need for antibiotics
favourable for a CURB-65 class ≥III. Though the (p<0.001) and prolonged duration of hospital stay
specificity increased to 97.3% when CURB-65 class ≥IV (p<0.001) (Table 8).
was chosen as the cut-off, there was an unfavourable The duration of hospital stay was found to have a
drop in the sensitivity which decreases to 74.3 percent. weak but significant correlation with PSI and CURB-65
The ROC curve for different CURB-65 classes is shown criteria. Defervescence time also had a very weak but
in figure 4. significant correlation with PSI and CURB-65 criteria.
The PSI class ≥IV is more sensitive in predicting Duration of IV antibiotics had a moderately strong
ICU admission than CURB-65 class ≥III; as CURB-65 correlation with CURB-65 criteria but a weak correlation
class ≥III has a higher specificity (84.4%) than PSI class with PSI criteria (Table 8).

Table 8. Correlation between some outcome parameters and PSI

and CURB-65 criteria in 150 patients with community acquired
pneumonia
Spearman’s rho PSI CURB-65
Duration of Correlation 0.401 0.487
hospital stay coefficient
p-value <0.001 <0.001
Defervescence Correlation 0.218 0.243
time coefficient
p-value 0.007 0.003
Duration of IV Correlation 0.467 0.634
antibiotics coefficient
p-value <0.001 <0.001
Figure 4. Receiver-operating characteristic curve for CURB-65 PSI=Pneumonia severity index; CURB-65=Confusion, urea, res-
with ICU admission as outcome. CURB-65=Confusion, urea, piratory rate, blood pressure, age over 65 years
respiratory rate, blood pressure, age over 65 years score,
ICU=Intensive care unit.
DISCUSSION
Table 7A. Number of ICU admissions in different CURB-65 risk classes

Variable CURB-65 Class Total In the initial management of patients with suspected
0 I II III IV V CAP the clinician is faced with diagnostic and
No. of 27 31 42 21 23 6 150 prognostic challenges, each challenge corresponding to
patients (18%) (20.7%) (28%) (14%) (15.3%) (4%) (100%)
a specific management decision. This emphasises the
ICU 0 0 3 6 20 6 35
admissions(0%) (0%) (8.6%) (17.1%) (57.1%) (17.1%) (100%) importance of prompt, accurate diagnosis and severity
of illness which corresponds to decisions regarding the
ICU=Intensive care unit; CURB-65=Confusion, urea, respiratory
rate, blood pressure, age over 65 years intensity of management. The decision regarding the
14 Scoring Systems in Community Acquired Pneumonia B.A. Shah et al

most appropriate site of care, including whether where an age ≥65 years was given additional score of 1,
admission to hospital is warranted, is the first and single making a total score of 5.
most important decision in the overall management of A major breakthrough was achieved only after the
CAP. It has consequences both for the level of treatment transformation of these rules into a risk score, which
received by the patient as well as the overall costs of resulted from adding one point for each of these
treatment.25 parameters (CURB or for patients aged >65 years
An unchanged mortality of 4% to 21% 13,17 in- CURB-65) by Lim and co-workers. 23,31 The scoring
hospital treated CAP has renewed the interest in system consists of a six-point score determined at the
studying prognostic factors associated with fatal time of initial presentation. In the original study,
outcome. mortality risk in the six separate groups was as follows:
The first landmark study to prognosticate patients group 0, 0.7%; group 1, 3.2%; group 2, 3%; group 3, 17%;
of CAP was conducted by the Research Committee of group 4, 42%; and group 5, 57 percent. These scores
the BTS in 1982. 26 In this study 26 comprising of 453 allowed for predictions very similar to those made by
adults in 25 British hospitals, patients had a 21-fold the PSI. In a subsequent study,32 the absence of any
increased risk of death if they had two of the following CURB criterion was associated with a 30-day mortality
at admission: respiratory rate ≤30/min, diastolic BP of one percent, the presence of one or two with 8%, and
≤60mmHg, urea >7mmol/L. On the basis of these the presence of three or four with 30% mortality.
findings, BTS1 rule was constructed by selecting three In 1997, Fine et al22 introduced the pneumonia
factors, which were highly associated with death at severity index (PSI), a product of the Pneumonia PORT
admission, namely, respiratory rate ≥30/min at study of ambulatory and hospitalised patients with
admission; diastolic blood pressure ≤60mmHg, and CAP. The rule stratifies patients into five classes of risk
blood urea level >7mmol/L. for death within 30 days of presentation. The lowest risk
This rule yielded the highest value among any of class (risk class I) comprises patients who are younger
the rules tested in the Youden index, a statistic than 50 years of age, have none of the five important co-
combining sensitivity and specificity for selection of an existing illnesses and have normal mental status and
optimal rule, assuming equal importance of sensitivity normal or only mildly abnormal vital signs at
and specificity.27 When the first rule was modified to use presentation. Assignment to the remaining risk classes
only three most predictive features (‘confusion’ depends on the presence or absence of a set of medical
replacing ‘urea>7mmol/L), immediate application was history, physical examination, and laboratory findings.
possible with this second rule referred to as BTS2 rule. Total point scores of 70 or less correspond to class II, 71
This modified rule had the highest overall accuracy to 90 to class III, 91 to 130 to class IV, and more than 130
(93%) and the highest specificity (94%) of any rule to class V. Mortality rates in risk classes I, II, and III are
tested, but correctly identified only 39% of the patients low (0.1% to 0.4% in class I and 0.9% to 2.8% in class III),
who died; a positive rule was associated with a relative with correspondingly higher mortality rates in risk
risk of death of 10.2. These two rules were compared classes IV and V. The cumulative mortality rate of
with a more complicated one suggested by patients in risk classes I to III is less than one percent.
Macfarlane, 28 which required at least three of the The variables in PORT study were derived from
following factors: (i) confusion on examination, (ii) and validated in more than 50,000 patients, the largest
white blood cell count ≥10x10 9/L or lymphocytes database ever studied in the history of CAP research.
≥1x109/L; (iii) arterial oxygen tension (PaO2) ≤6.6KPa; The original role of the PSI was to identify those patients
and (iv) blood urea level ≥7mmol/L. It showed an at a low risk of mortality who, therefore, could safely be
overall accuracy of 87%, but identified only 50% of the treated as out-patients. The PSI was subsequently
patients who died, and was associated with a relative confirmed to make valid predictions of mortality by
risk of death of 6.4. several authors, although in some reports mortality
Neill et al29 derived a modified BTS rule (mBTSr) in rates were somewhat lower in the highest risk group. 32-
34
which severe CAP was suggested by the presence of Finally, the PSI was also shown to predict long-term
two or more of: (i) confusion, (ii) respiratory rate ≥30/ outcomes of CAP.35 A major limitation of the PSI is the
minute, (iii) diastolic BP ≤60mmHg; and (iv) blood urea unbalanced impact of age on the score, resulting in a
≥7mmol/L at the time of admission. Those who satisfied potential underestimation of severe pneumonia,
mBTSr had a 36.5-fold greater risk of dying compared particularly in younger otherwise healthy individuals.32
with 22 and 9.9 with BTS1 and BTS2, respectively.29 Nevertheless, the PSI is currently recommended as a
Subsequently, CURB criteria (confusion, urea, tool of severity assessment in the Infectious Diseases
respiratory rate and blood pressure) were developed Society of America (IDSA) guidelines.36, 37
which were similar to mBTSr, but systolic BP <90mmHg Capelastegui et al 38 presented a comparative
was added (either systolic BP <90mmHg or diastolic BP validation of the CURB-65, CRB-65 (which omits the
<60mmHg scores 1). Authors30 also suggested CURB-65 blood urea measurement) and PSI scores in a
2010;Vol.52 The Indian Journal of Chest Diseases & Allied Sciences 15

population of 1,776 patients including 676 outpatients. score to circumvent the use of confusion and blood
The 30-day mortality increased with increasing score, urea. From their analyses, low systolic BP (S) and poor
and predictions of 30-day mortality were equivalent for oxygenation (PaO2: FIO2) (O), advancing age (A), high
all scores as assessed by ROC analysis. This is in respiratory rate (R) were found to be significantly
contrast to the study by Aujesky et al39 comprising 3,181 associated with death from CAP, and derived a new
patients and including 1,094 outpatients, showing a index (SOAR) using these criteria and assessed its
minor but significant advantage for the PSI score in usefulness. They defined severe pneumonia as the
predicting 30-day mortality using area under the curve presence of ≥2 criteria out of four. A score of 1 was given
(AUC) analysis. However, this population for presence of each of the following (dichotomised
predominantly included less severely ill patients (only variables): (i) systolic BP <90 mmHg; (ii) PaO2:FIO2 <250;
6% PS IV as compared with 18% in the present study), (iii) age ≥65 years; and (iv) respiratory rate ≥30 per
thereby limiting the comparability of both populations minute.
studied. In another prospective study from Hong Kong
The CURB-65 score has a major advantage in its (n=1016), Man et al 37 compared the ability of three
simplicity. However, with blood urea nitrogen, it validated prediction rules for CAP to predict mortality:
includes a variable that is not readily available in the 20 variable PSI, the 6-point CURB-65 scale adopted
general practice and not even in some hospitals. by the BTS and the simpler CRB-65. The patients were
Therefore, one of the most remarkable findings of the classified into three risk groups (low, intermediate and
study by Capelastegui et al 38 is the equivalence of high) according to each rule and the ability of the three
predictions made by the CURB and the CRB-65 score, rules to predict 30-day mortality was compared. The
the latter simply replacing blood urea nitrogen by the overall mortality and ICU admission rates were 8.6%
presence of age >65 years. This fits well into findings and 4.0%, respectively. The PSI, CURB-65 and CRB-65
from the data generated by the German Competence performed similarly and the areas under the ROC curve
Network for the study of community-acquired were 0.736 (95% confidence interval (CI) 0.687 to 0.736),
pneumonia (CAPNETZ; unpublished data, T.T. Bauer, 0.733 (95% CI 0.679 to 0.787) and 0.694 (95% CI 0.634 to
Medizinische Klinik III, Bergmannsheil Klinikum der Ruhr- 0.753), respectively. All three rules had high negative
Universität, Bochum, Germany). In a population of 1,312 predictive values but relatively low positive predictive
patients, which included 205 out-patients, CURB and values at all cut-off points. Larger proportions of
CRB-65 had an equivalent predictive power for 14-day patients were identified as low risk by PSI (47.2%) and
mortality. Taken together, there is growing evidence that CURB-65 (43.3%) than by CRB-65 (12.6%). The study
CURB, CURB-65 and CRB-65 all allow for similar concluded that all three predictive rules have a similar
predictions of death from CAP as compared to the PSI, performance in predicting the severity of CAP, but
with the CRB-65 representing the only score that is also CURB-65 was more suitable than the other two for use
easily applicable in out-patients. in the emergency department because of its simplicity of
Overall, the CRB-65 and CURB-65 scores are an application and ability to identify low-risk patients.
impressive example of the value of a simple clinical In a study by Loh et al43 conducted in Malaysia, BTS
approach not requiring sophisticated biochemical, criteria fared poorly in predicting mortality compared
immunological or genetic data in the risk stratification with clinical assessment by attending clinicians (36-fold
of patients with an acute potentially life-threatening increased risk of death by ‘clinical assessment’ vs two-
condition. three-fold by ‘BTS criteria’). These results have
Capelastegui et al 38 have also identified several demonstrated the need for testing the validity of such
additional factors associated with the need for scoring systems in Asian countries and other
hospitalisation not necessarily related to mortality but developing parts of the world that have different
requiring special attention, which should be assessed in demographic characteristics as well as healthcare
all but the lowest risk classes, thereby extending delivery systems than those where such prognostic
previous experiences. 34 These factors comprise co- scoring systems were developed and validated.
morbidities, severe hypoxaemia or hypercapnia, the The comparison between mortality rates in
extent of radiographic infiltrates, and pleural effusions. different risk classes in our study and that of the
Both CURB and CURB-65 include confusion and previous studies22,32,36,38 showed that in all the studies
raised urea (>7mmol/L) in their severity criteria, which mortality rates progressively increase with increasing
may be less useful in the elderly as both conditions are risk scores in both PSI and CURB-65 risk classes.
common in acutely unwell older people.40,41 In this Though in our study mortality rates in PSI risk class I to
regard, Myint et al42 examined the value of ventilation III were lower compared to other two studies by Fine et
perfusion mis-match using the ratio of PaO 2 and al22 and Buising et al36, mortality rates in classes IV to V
fraction of inspired oxygen (FIO 2) for predicting were higher. The latter effect is because Buising et al36
mortality from CAP and to derive alternative severity studied only in-hospital mortality while as in our study,
16 Scoring Systems in Community Acquired Pneumonia B.A. Shah et al

death group included patients who died within 30 days Brandstetter RD, Karetzky M, Cunha BA, editors. The
after discharge. Further, four patients who were Pneumonias. New York: Springer-Verlag; 1993:pp 25-48.
2. Regional situation on health statistics reporting. Health
severely sick and required admission to ICU left the Situation in the South-East Asia Region 1994-1997. New
hospital against medical advice on request and Delhi: EHI/WHO-SEARO. September 2007.
subsequently died at home within few days of leaving 3. Garibaldi RA. Epidemiology of community acquired
the hospital. Comparison of mortality rates in different respiratory tract infections in adults: incidence, etiology and
impact. Am J Med 1985;78:32-7.
CURB-65 risk classes in our study and that of 4. Ishida T, Hashimoto T, Arita M, Ito I, Osawa M. Etiology of
Capelastegui et al38 and Ewig et al32 showed comparable community-acquired pneumonia in hospitalized patients: a-
results. 3-year prospective study in Japan. Chest 1998; 114:588-93.
The comparison of PSI and CURB-65 with respect 5. Lieberman D, Schalaeffer F, Boldur I, Liebermam D,
Horowitz, Friedman MG, et al. Multiple pathogens in adult
to sensitivity, specificity and predictive values have
patients admitted with community acquired pneumonia: a
good sensitivity and NPV but specificity and PPV are one year prospective study of 346 consecutive patients.
less impressive. These results are comparable to those Thorax 1996;51:179-84.
obtained by Man et al.37 Specificity of CURB-65 was 6. Bansal S, Kashyap S, Pal LS, Goel A. Clinical and bacteriolo-
found to be better than PSI probably because a major gical profile of community acquired pneumonia in Shimla,
Himachal Pradesh. Indian J Chest Dis Allied Sci 2004;46:17-22.
limitation of the PSI is the unbalanced impact of age on 7. Kulpati DDS, Khastgir T. Reappraisal of pneumonias. J Assoc
the score, resulting in a potential underestimation of Physicians India 1988;36:660-4.
severe CAP particularly in younger otherwise healthy 8. Madhu SV, Gupta U, Guleria JS, Talwar V. Clinical and
individuals.32 bacteriological profile of hospitalized community acquired
pneumonias: a preliminary study. Indian J Chest Dis Allied Sci
The two scoring CURB-65 and PSI approaches are
1990;32:95-100.
viewed as being complementary, as each has different 9. Kulpati DDS, Kumar A. Flexible fiberoptic bronchoscopy in
strengths and weaknesses. The PSI seems to have been lower respiratory tract infections. Indian J Chest Dis Allied Sci
developed, and best validated, as a way to identify low 1980;22:39-46.
10. Sharma TN, Jain NK, Nanavati V, Mangal HN, Sarkar SK,
mortality risk patients, but the scoring system can
Singh V. Transbronchofiberoscopic bronchial aspiration in
occasionally underestimate severity of illness, especially lower respiratory tract infections. Indian J Chest Dis Allied Sci
in young patients without comorbid illness.33,36 This is 1981;23:73-80.
primarily because the PSI heavily weighs age and co- 11. Dey AB, Chaudhary R, Kumar P, Nisar N, Nagarkar KM.
Mycoplasma pneumoniae and community acquired
morbidity, and does not directly measure CAP-specific
pneumonia. Natl Med J India 2000;13:66-70.
disease severity. In contrast, the CURB-65 approach may 12. Chaudhry R, Nazima N, Dhawan B, Kabra SK. Prevalence of
be ideal for identifying high mortality risk patients with Mycoplasma pneumonia and Chalamydia pneumoniae in
severe illness due to CAP who might otherwise be children with community acquired pneumonia. Indian J
overlooked without formal assessment of subtle Pediatr 1998;65:717-21.
13. Fang GD, Fine M, Orloff J, Arisumi D, Yu VL, Kapoor W, et
aberrations in key vital signs. 34 However, one clear al. New and emerging etiologies for community-acquired
deficiency of the CURB-65 approach is that it does not pneumonia with implications for therapy: a prospective
generally account for comorbid illness, and thus may multicenter study of 359 cases. Medicine 1990;69:307-16.
not be easily applied in older patients who may still 14. Woodhead MA, Macfarlane JT, Rodgers FG, Laverick A,
Pilkington R, Macrae AD. Aetiology and outcome of severe
have substantial mortality risk, even if a mild form of community-acquired pneumonia. J Infect 1985;10: 204-10.
CAP destabilises a chronic, but compensated, disease 15. Hirani NA, Macfarlane JT. Impact of management guidelines
process. Thus, both tools offer a valuable assessment of on the outcome of severe community acquired pneumonia.
patient illness, but from different perspectives, and each Thorax 1997;52:17-21.
16. Mac Farlance J. Community acquired pneumonia. Br J Dis
is best at identifying patients at opposite ends of the
Chest 1987;81:116-27.
disease severity spectrum. 17. Ortqvist A, Hedlund J, Grillner L, Jalonen E, Kallings I,
In conclusion, both PSI and CURB-65 have equal Leinonen M, et al. Aetiology outcome and prognostic factors
sensitivity to predict death from CAP, however in community acquired pneumonia requiring
hospitalization. Eur Respir J 1990;3:1105-13.
specificity of CURB-65 is higher than that of PSI.
18. Pachon J, Prados MD, Capote F, Cuello JA, Garnacho J,
Mortality rates progressively increase with increasing Verano A. Severe community acquired pneumonia: etiology,
risk class in both severity scoring systems. By using the prognosis and treatment. Am Rev Respir Dis 1990; 142:369-73.
knowledge of these criteria, patients of CAP can be 19. Dey AB, Nagarkar KM, Kumar V. Clinical presentation and
better prognosticated as regards severity of their illness predictors of outcome in adult patients with community-
acquired pneumonia. Natl Med J India 1997;10:169-72.
with consequently better triaging of patients, utilisation 20. British Thoracic Society Research Committee and Public
of resources and appropriate treatment to improve the Health Laboratory Service. The aetiology, management and
outcome in this disease. outcome of severe community-acquired pneumonia on the
intensive care unit. Respir Med 1992;86:7-13.
21. Alkhayer M, Jenkins PF, Harrison BDW. The outcome of
REFERENCES community acquired pneumonia treated on the intensive
care unit. Respir Med 1990;84:13-6.
1. Karetzky M. Community-acquired pneumonia. In:
2010;Vol.52 The Indian Journal of Chest Diseases & Allied Sciences 17

22. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Gudiol F. Etiology, reasons for hospitalization, risk classes,
Singer DE, et al. A prediction rule to identify low-risk and outcomes of community-acquired pneumonia in
patients with community-acquired pneumonia. N Engl J Med patients hospitalized on the basis of conventional admission
1997;336:243-50. criteria. Clin Infect Dis 2001;33:158-65.
23. Lim WS, van der Erden MM, Laing R, Boersma WG, Karalus 35. Mortensen EM, Kapoor WN, Chang CCH, Fine MJ.
N, Town GI, et al. Defining community acquired pneumonia Assessment of mortality after long-term follow-up of
severity on presentation to hospital: an international patients with community-acquired pneumonia. Clin Infect
derivation and validation study. Thorax 2003;58:377-82. Dis 2003;37:1617-24.
24. Querol-Ribelles JM, Tenias JM, Grau E, Querol-Borras JM, 36. Buising KL, Thursky KA, Black JF, MacGregor L, Street AC,
Climent JL, Gomez E, et al. Plasma d-Dimer levels correlate Kennedy MP, et al. A prospective comparison of severity
with outcomes in patients with community-acquired scores for identifying patients with severe community
pneumonia. Chest 2004;126:1087-92. acquired pneumonia: reconsidering what is meant by severe
25. Guest JF, Morris A. Community-acquired pneumonia: the pneumonia. Thorax 2006;61:419-24.
annual cost to the National Health Service in the United 37. Man SY, Lee N, Ip M, Antonio GE, Chau SS, Mak P, et al.
Kingdom. Eur Respir J 1997;10:1530-4. Prospective comparison of three predictive rules for
26. Anonymous. Community-acquired pneumonia in adults in assessing severity of community-acquired pneumonia in
British hospitals in 1982-1983: a survey of aetiology, Hong Kong. Thorax 2007;62:348-53.
mortality, prognostic factors and outcome. The British 38. Capelastegui A, Espana PP, Quintana JM, Areitio I, Gorordo
Thoracic Society and the Public Health Laboratory Service. I, Egurrola M, et al. Validation of a predictive rule for the
Q J Med 1987;62:195-220. managment of community-acquired pneumonia. Eur Respir
27. Armitage P. Statistical methods in medical research. Oxford:
J 2006;27:151-7.
Blackwell Scientific Publications, 1971.
39. Aujesky D, Auble TE, Yealy DM, Stone RA, Obrosky DS,
28. Macfarlane JT. Adverse prognostic factors in pneumonia.
Meehan TP, et al. Prospective validation of three validated
Thorax 1983;38:231.
prediction rules for prognosis in community-acquired
29. Neill AM, Martin IR, Weir R, Anderson R, Chereshsky A,
pneumonia. Am J Med 2005;118:384-92.
Epton MJ, et al. Community-acquired pneumonia: aetiology
40. Hodkinson HM. Common Symptoms of Disease in the Elderly.
and usefulness of severity criteria on admission. Thorax
Oxford: Blackwell Scientific Publications; 1976:p.24.
1996;51:1010-16.
41. Warren JL, Bacon WE, Harris T, McBean AM, Foley DJ,
30. British Thoracic Society Standard of Care Committee. BTS
guidelines for the management of community acquired Phillips C. The burden and outcome associated with
pneumonia in adults. Thorax 2001;56 (Suppl. 4):1-64. dehydration among United States elderly. Am J Publ Health
31. Lim WS, Lewis S, Macfarlane JT. Severity prediction rules in 1991;84:1265-9.
community-acquired pneumonia: a validation study. Thorax 42. Myint PK, Kamath AV, Vowler SL, Maisey DN, Harrison BD.
2000;55:219-23. Severity assessment criteria recommended by the British
32. Ewig S, de Roux A, Garcia E, Mensa J, Niederman M, Torres Thoracic Society (BTS) for community-acquired pneumonia
A. Validation of predictive rules and indices of severity for (CAP) and older patients. Should SOAR (systolic blood
community-acquired pneumonia. Thorax 2004;59:421-7. pressure, oxygenation, age and respiratory rate) criteria be
33. Ewig S, Kleinfeld T, Bauer T, Seifert K, Schäfer H, Göke N. used in older people? A compilation study of two
Comparative validation of prognostic rules for community- prospective cohorts. Age Ageing 2006;35:286-91.
acquired pneumonia in an elderly population. Eur Respir J 43. Loh LC, Khoo SK, Quah SY, Visvalingam V, Radhakrishnan
1999;14:370-5. A, Vijayasingham P, et al. Adult community-acquired
34. Roson B, Carratala J, Dorca J, Casanova A, Manresa F, pneumonia in Malaysia: prediction of mortality from
severity assessment in admission. Respirology 2004;9:379-86.