[Type text]
ONLINE DATA SUPPLEMENT FOR:
Severity assessment tools for predicting mortality in hospitalised patients with Community-Acquired Pneumonia: Systematic review and Meta-Analysis.
James D Chalmers1, Aran Singanayagam2, Ahsan R Akram2, Pallavi Mandal2, Philip M Short3, Gourab Chowdhury2, Victoria Wood1, Adam T Hill2
1. University of Edinburgh, Edinburgh, UK.
2. Department of Respiratory Medicine, Royal Infirmary of Edinburgh, Edinburgh, UK
3. University of Dundee, Tayside, UK.
Corresponding author
Dr James Chalmers
Department of Respiratory Medicine
Royal Infirmary of Edinburgh
51 Little France Crescent
Old Dalkeith Road
Edinburgh
EH16 4SA
E-mail:
Keywords: Pneumonia, Severity assessment, mortality, meta-analysis
ADDITIONAL METHODS
Definition of Severity scores
Pneumonia Severity Index[1]
The Pneumonia Severity Index is a well-validated prediction scale for 30-day mortality in community-acquired pneumonia is composed of the following twenty characteristics: demographics including age, sex and nursing home residence; co-morbid illness including neoplastic disease, cerebrovascular disease, congestive cardiac failure, chronic renal disease and chronic liver disease; physical examination findings including altered mental status, respiratory rate 30/min, systolic blood pressure <90mmHg, temperature <35oC or >40oC and pulse >125/min; laboratory findings including pH <7.35, blood urea >10.7 mmol/L, sodium <130 mEq/L, glucose >13.9 mmol/L, haematocrit <30% and PaO2 <60mmHg; radiographic findings including pleural effusion.
Using these data, patients are classified into 5 risk classes. In the original Pneumonia Patient Outcome Research Team cohort study, 30-day mortality ranged from 0.1% for patients with a class 1 rating to 27% for patients with a class 5 rating. Typically, patients with PSI class IV/V are regarded as severe pneumonia and recommended for hospitalization. For the purposes of this meta-analysis, 2x2 tables were constructed for mortality in both PSI class IV/V and class V alone.
CURB65/CRB65[2]
CURB65 is a validated method of predicting inpatient mortality associated with CAP that is recommended by the British Thoracic Society. It consists of new onset mental confusion, urea >7 mmol/L, respiratory rate 30 breaths/min, systolic blood pressure <90mmHg or diastolic blood pressure 60 mmHg and age 65 years.
British Thoracic Society guidelines suggest that patients with a CURB65 score of 0–1 be considered for outpatient treatment; that patients with a CURB65 score of 2 be considered for short inpatient hospital stay; and that patients with a CURB65 score 3 have severe pneumonia that requires inpatient management, and intensive care or high dependency environment care should be considered, particularly for patients with a CURB65 score 4. For the purposes of this meta-analysis both 3 and 4 were used to construct 2 x 2 tables of the relationship between CURB65 and mortality.
CRB65 is a simplified version of the CURB65 score that does not include the urea criterion. It stratifies patients into low risk (CRB65= 0 or 1), intermediate risk (CRB65= 2) or high risk groups (CRB65 = 3 or 4). CRB65 is recommended for outpatient use in the British Thoracic Society guidelines.[3]
HAYDENS CRITERIA FOR QUALITY ASSESSMENT
The following table (Table E1) describe Haydens criteria.[4] These quality assessment criteria were modified to apply to observational studies of community acquired pneumonia severity scores.
Haydens criteria / Components / As applied to CAP studies1. Study sample represents the population of interest, design appropriate to limit potential bias / a) Source population clearly defined
b) Study population described
c) Study population represents population of interest / i) Population limited to CAP and excludes other diagnoses
ii) Requires Chest x-ray confirmation and uses recognised definition
iii) Enrolls consecutive, unselected patients
iv) Demographics are representative of CAP cohorts internationally.
2. Loss to follow-up, study data adequately represent the sample / d) Completeness of follow-up described
e) Completeness of follow-up adequate / i) Appropriate follow up to determine mortality
ii) Limited number of patients lost to follow-up.
3. Prognostic factor of interest is adequately measured in study participants / f) Prognostic factors defined
g) Prognostic factors appropriately measured / i)CURB65/CRB65/PSI score calculated according to standard definition
II) Measurement made on admission and recorded prospectively
III) Missing values minimised and appropriately dealt with
4. The outcomes of interest are adequately measured in study participants / h) Outcome defined
i) Outcome measured appropriately / j) Mortality
k) 30-day mortality or alternative outcome determined appropriately.
5. Important confounders are accounted for. / l) Confounders defined and measured
m) Confounders accounted for / Not applicable as prognostic scores are used independently.
6. Appropriate statistical analysis / n) Analysis described
o) Analysis appropriate
p) Analysis provides sufficient presentation of data / l) Uses and reports PPV/NPV/sensitivity/specificity and/or receiver operator characteristic curve for pneumonia severity scores.
II) displays data for different cut-points of each pneumonia severity score
TABLE E1- Haydens Criteria for quality assessment, modified to apply to studies of community acquired pneumonia.
ADDITIONAL RESULTS
The Table (E2) provides details of each of the studies included in the meta-analysis.
First author name / Study Population / Score(s) assessed / Setting and design / N / Age(years) / Study Outcome / Mortality rate / Study Objective/
Conclusion
Ananda-Rajah[24] / Retrospective chart review / PSI, CURB65 / Melbourne, Australia,
2002 / 408 / 72 +/- 16 / 30-day mortality / 15.4% / Comparison of PSI and CURB65
Aujesky, D[15] / Guideline implementation trial / PSI, CURB65 / 32 hospitals in Pennsylvania, Connecticut USA
2001 / 3181 / 63 / 30-day mortality / 4.6% / Randomised controlled trial comparing effect of intensity of guideline implementation (PSI)
Barlow, G[25] / Prospective, before and after quality improvement study / CURB65, CRB65 / 2 Hospitals, Tayside, UK. 2001-2003. / 419 / Median 74 / 30-day mortality / 18.9% / Study to increase proportion of patients receiving antibiotics within 4 hours of admission
Bauer, TT (CAPNETZ)[37] / Prospective cohort study / CRB65 / 10 inpatient and outpatient centres Germany,
2003-2004 / 1967 / 66 +/- 18 / 30-day mortality / 4.3% / Validation of CRB65
Buising, KL[26] / Prospective cohort study / CURB65, PSI / Melbourne, Australia,
2003-2006 / 722 / Median 74 / In-hospital mortality / 9.8% / Derivation of a new severity score (CORB)
Capelastegui , A[27] / Prospective cohort study / PSI, CURB65, CRB65 / Galdakao Hospital, Basque region, Spain
2000-2004 / 1776 / 61.8 +/- 20.5 / 30-day mortality / 6.7% / Validation of severity scores
Challen, K[28] / Retrospective case note review / CURB65 / Manchester, UK 2005. / 186 / NR / In-hospital mortality / 22.6% / Comparison of CURB65 with modified early warning score
Chalmers, JD[20] / Prospective cohort study / PSI, CURB65, CRB65 / 2 hospitals, Edinburgh, UK 2005-2008 / 1007 / 66 (50-78) / 30-day mortality / 9.6% / Modification of the CURB65 score.
Charles, PG[29] / Prospective cohort study / PSI, CURB65 / Multicentre, Australia,
2004-2006. / 882 / 65.1 +/- 19.9 / 30-day mortality / 5.7% / Derivation of new severity score (SMART-COP)
Chen CZ[41] / Prospective cohort study / PSI / Taiwan,
2005-2006 / 250 / NR / In hospital mortality / 12% / Evaluation of repeated measurements of PSI
Davydov L[42] / Prospective cohort study / PSI / Multicentre,
USA
1998-1999 / 875 / 66.5 +/- 17.7 / In hospital mortality / 2.7% / Audit treatment of patients according to PSI class.
Dedier, J [43] / Retrospective database / PSI / 38 US Academic hospitals / 1062 / Median 64 (range 18-98) / Hospital mortality / 6% / Study the effect of processes of care on outcome.
Ewig, S[44] / Prospective cohort study / PSI / Single centre, unknown location, 1998-2001 / 489 / 67.8 +/- 17.1 / In-hospital mortality within 30-days / 6.8% / Comparison of severity scores.
Ewig S[38] / Retrospective administrative database / CRB65 / Nationwide database, Germany
2005-2006 / 388,406 / Median 76 / In-hospital mortality / 14.1% / National epidemiological survey
Feagan, B[45] / Retrospective chart review / PSI / 20 hospitals across Canada
1996-1997 / 858 / 69.4 +/- 17.7 / 30-day in hospital mortality / 14.1% / Evaluate the treatment and outcome of CAP in Canada
Fine, MJ- Medisgroup[7] / Administrative database / PSI / MEDISGROUP 1
78 hospitals in USA,
1989
MEDISGROUP 2
193 Hospitals in Pensylvania
1991 / 52,238 / NR / 30-day mortality / 10.5% / Derivation and internal validation of PSI
Fine MJ- PORT[7] / Prospective cohort study / PSI / 5 Hospitals, USA and Canada, 1991-1994 / 2287 / NR / 30-day mortality / 4.9% / Validation of PSI
Flanders, WD[46] / Retrospective chart review / PSI / 22 Hospitals, Atlanta, USA
1994-1995 / 1,024 / NR / Not reported. Assumed in-hospital mortality / 4.8% / Recalibration of PSI
Garau J[47] / Retrospective care note review / PSI / Multicentre, Spain 2001-2002 / 3233 / 66.6 +/- 18.5 / In-hospital mortality / 8.7% / Investigating factors affecting length of stay and mortality.
Garcia-Vazquez E[48] / Retrospective study / PSI / Murcia, Spain, 2003. / 211 / 63 (range 13-100) / In hospital mortality / 7.1% / Derivation of simplified score.
Goss CH[49] / Prospective cohort study / PSI / Seattle, WA, USA 1994-1996 / 522 / 46 (range 18-100) / In hospital mortality / 3.3% / To evaluate resource utilisation among patients in low risk PSI groups.
Huang, DT[30] / Prospective cohort study / PSI, CURB65 / Multicentre, USA
2001-2003 / 1651 / 65 +/- 18.5 / 30-day mortality, 90 day mortality / 6.4%
9.8% / Assessment of procalcitonin as a severity marker (GenIMS)
Johnstone, J[39] / Prospective, population based cohort study / PSI / 6 hospitals, Alberta, Canada 2000-2002 / 2,906 / 68.9 +/- 17.9 / 1 year mortality (30-day mortality as secondary end-point) / 13% / Long term mortality and morbidity of CAP patients
Lim, WS[10] / Combination of 3 prospective cohort studies / CURB65, CRB65 / Nottingham, UK 1998-2000. Christchurch and Waikato, New Zealand
1999-2000. Alkmaar, Netherlands, 1998-2000. / 1068 / Mean 64.1 / 30-day mortality / 8.3% / Derivation of CURB65, CRB65
Man, SY[18] / Prospective cohort study / PSI, CURB65, CRB65 / Hong Kong
2004 / 1016 / 72 +/- 7.2 / 30-day mortality / 8.6% / Validation of severity scores for 30-day mortality
Menendez, R[31] / Prospective cohort study / PSI, CURB65, CRB65 / 2 Hospitals, Spain
2003-2004 / 453 / 67.3 +/- 17.1 / Treatment failure
(includes mortality) / 6.8% / Investigation of CRP, cytokines and procalcitonin as markers of treatment failure
Migliorati, PL [50] / Retrospective chart review / PSI / Single Centre, Italy / 148 / 70.3 +/- 17.3 / 30-day mortality / 12.2% / Validation of PSI in Italy.
Ortega L[40] / Prospective cohort study / PSI / Barcelona, Spain, Dates not stated / 128 / 64 +/- 8 / In-hospital mortality / 3.1% / Evaluation of the pneumonia severity index
Phua, J[32] / Retrospective cohort study / PSI, CURB65 / Singapore,
2004-2007 / 1242 / 65.7 +/- 20.1 / In hospital mortaity / 14.7% / Validation of IDSA/ATS criteria for severe CAP.
Renaud, B[8]
Pneumocom 1 / Randomised controlled trial / PSI / 16 Hospitals, France.
2002-2003 / 925 / Mean 66 / 28 day mortality / 10.6% / Randomised controlled trial using PSI to determine site of care
Renaud, B[51]
Pneumocom 2 / Prospective cohort study / PSI / 14 hospitals, Catalonia, Spain,
2003 / 853 / Mean 65 / 28 day mortality / 6.3% / Validation of PSI in European population
Querol-Ribelles JM[52] / Prospective cohort study / PSI / Valencia, Spain, 2000 / 243 / 63 +/- 19 / 30-day mortality / 6.2% / Validation of PSI
Restrepo, M[53] / Retrospective chart review / PSI / 2 hospitals, San Antonio, Texas, USA
1999-2002 / 730 / 59.2 +/- 16.2 / 30-day mortality / 8.1% / Compare patients admitted to the ward and intensive care units with CAP
Reyes Calzada S[54] / Prospective observational study / PSI / Multicentre, Valencia, Spain Dates not stated / 425 / 69 +/- 16 / 30-day mortality / 8.2% / Evaluate adherence to community-acquired pneumonia guidelines
Roson, B[55] / Prospective cohort study / PSI / Barcelona, Spain
1995-1997 / 533 / Mean 64 / 30-day mortality / 6.6% / Use of PSI to determine site of care
Schuetz , P[33] / Randomised controlled trial / PSI, CURB65, CRB65 / Basel, Switzerland, 2002-2005 / 373 / 73 (59-82) / 30-day mortality / 11% / Validation and recalibration of severity scores
Shindo, Y[34] / Retrospective cohort study / CURB65 / Handa City, Japan 2005-2007 / 329 / 75 +/- 15.7 / 30-day in hospital mortality / 9.4% / Compare CURB65 with ADROP (Alternative scoring system)
Tejera, A[35] / Prospective cohort study / PSI, CURB65 / Tenerife, Spain
Dates not available / 226 / 74 (61-82) / In hospital mortality / 12.4% / Assessment of TREM-1 as a prognostic marker in CAP
Van der Eerden[56] / Prospective cohort study / PSI / Alkmaar, Netherlands, 1998-2000 / 260 / Mean 64 / 30-day mortality / 10% / Validation of PSI
Zuberi, FF[36] / Prospective chart review / CURB65, CRB65 / Karachi, Pakistan. 2006-2007 / 137 / 60.4 +/- 18.5 / 30-day mortality / 13.1% / Validation of CURB65/CRB65 in a developing country
TABLE E2- Characteristics of studies included in the meta-analysis (note- references refer to those in the main document).
The following studies contained data on PSI, CURB65 or CRB65 but were excluded from the main analysis due to duplicate publication of data, or failure to meet the inclusion criteria of non-selected CAP populations. The reasons for exclusion are listed.
First author name / year / score(s) assessed / Reason for exclusionAngus DC[5] / 2002 / PSI / Data contained in (1)
Arnold FW[6] / 2003 / PSI / Low risk patients only
Bont J[7] / 2008 / CRB65 / Outpatient only
Bruns, AH[8] / 2008 / PSI / Limited to severe CAP (PSI IV and V)
Buising KL[9] / 2007 / CURB65, PSI / Data contained in (10)
Cabre M[11] / 2004 / PSI / Not consecutive, unselected CAP patients. Data not presented.
Campbell, SG[12] / 2006 / PSI / Patients discharged from emergency department only.
Chalmers JD[13] / 2008 / PSI, CURB65, CRB65 / Data contained in (14)
Chalmers, JD[15] / 2008 / PSI, CURB65, CRB65 / Data contained in (14)
Cham, G[16] / 2009 / PSI / No mortality data included.
Christ-Crain, M[17] / 2006 / PSI / Data reported in (18)
Curran, A[19] / 2008 / PSI / HIV positive patients only
Dremsizov T[20] / 2009 / PSI / Data reported in (1)
Escobar GJ[21] / 2008 / PSI / Modified version of the PSI excluding some parameters
Espana PP[22] / 2006 / PSI, CURB65 / Data contained in (23)
Ewig S[24] / 1999 / PSI / Limited to elderly patients (aged >65 years only)
Gotoh, S[25] / 2008 / PSI / Data not reported
Haeuptle J[26] / 2009 / PSI / Limited to Legionella pneumonia.
Hohenthal, U[27] / 2009 / PSI / No mortality data presented.
Huang, DT[28] / 2009 / PSI, CURB65 / Data contained in (29)
Ioachimescu OC[30] / 2004 / PSI / Limited to Streptococcus pneumoniae pneumonia patients only.
Kollef KE[31] / 2008 / CURB65 / Single organism only (MRSA)
Kruger, S[32] / 2008 / CRB65 / Data contained in (33)
Lin CC[34] / 2005 / PSI / Data reported in (35)
Masia, M[36] / 2005 / PSI / Data not reported.
Muller, B[37] / 2007 / PSI / Not limited to CAP. Data reported in (17)
Myint, PK[38] / 2006 / CURB65 / Limited to elderly patients only
Naito T[39] / 2006 / PSI / Elderly patients aged >80 years only.
Pauls S[40] / 2008 / CRB65 / Did not report mortality data.
Pilotto, A[41] / 2009 / PSI / 1 year mortality as end-point. Age limited to >65 years.
Prat, C [42] / 2006 / PSI / Not exclusively CAP (including tuberculosis and PCP).
Salluh JI[43] / 2008 / CURB65 / Limited to severe CAP in ICU
Sanders KM[44] / 2006 / PSI / Immunocompromised patients only
Sanz, F [45] / 2009 / PSI / Low risk (PSI I-III) only.
Schaaf, B[46] / 2007 / CRB65 / Limited to a single organism (Streptococcus Pneumoniae)
Spindler, C[47] / 2006 / CURB65 / Limited to a single organism (Streptococcus Pneumoniae)
Teramoto, S [48] / 2008 / CURB65 / Data not reported
Valencia, M[49] / 2007 / PSI, CURB65 / Pneumonia severity index class V patients only.
Vecchiarino, P [50] / 2004 / PSI / Data not presented.
Wilson PA[51] / 2005 / PSI / ICU admitted patients only.
Yealy DM[52] / 2005 / PSI / Data contained in (53)
Table E3- Studies of severity scores excluded from the meta-analysis.