Table 1 Characteristics of Included Studies
Author & Year | Study Design | Country | Number of subunit compared | Total study population | Quality score | POC, Outcome or Both | Variation in Patient population between units | Variation in hospital type / subunit | Variation in disease factors |
|---|---|---|---|---|---|---|---|---|---|
Subunit of variation: Geographical region or country | |||||||||
Arnold et al. [12] 2013 | Retrospective Cohort | International – 16 countries across USA, Canada, Europe and Latin America | 70 hospitals across 3 geographical regions (USA/Canada, Europe, Latin America) | 6371 | 9.5 | Both | Significant differences in baseline populations. Latin America lowest prevalence of every co-morbidity. | Variation between hospitals grouped by continents. International variation in healthcare practice and resources. | Europe - fewest low severity scoring patients, greatest number of high severity scoring patients. |
Blasi et al. [13] 2013 | Retrospective Cohort | International - Europe | 10 countries (128 sites) | 2039 | 6.5 | Outcome | Not reported | Not reported | Included HCAP in addition to CAP |
Lave et al. [23] 1996 | Retrospective Analysis of Administrative data | USA | 7 geographical regions | 36,222 | 7 | Both | Not reported | All hospitals part of a larger non-profit organisation. Bed size varies 80–500 beds. Teaching and non-teaching facilities. | Not reported |
Remond et al. [27] 2010 | Mixed Prospective / Retrospective Cohort | Australia | 2 regions (7 hospitals) | 293 | 6.5 | Both | Different ethnicity between cohorts | Six small regional hospitals in the Kimberley, one tertiary hospital in Central Australia | Regional differences in isolated causative organisms. |
Subunit of variation: Hospital | |||||||||
Aelvoet et al. [11] 2016 | Retrospective Analysis of Administrative data | Belgium | 111 hospitals | 108,213 | 7 | Outcome | Not reported | All hospitals in Belgium | Not reported |
Cabre et al. [14] 2004 | Retrospective Cohort | Spain | 27 hospitals | 1920 | 6.5 | Both | The number of comorbidities varied among hospitals. | All community hospitals - urban and rural | Proportion of patients belonging to each risk class (by PSI) varied widely among hospitals |
Capelastegui et al. [15] 2005 | Retrospective Cohort | Spain | 5 hospitals | 1498 | 6 | Both | Statistically significant differences in patient demographic factors between hospitals. | All teaching general hospitals with similar resources | Statistically significant differences in PSI score classification between hospitals |
Dedier et al. [16] 2001 | Retrospective Cohort | USA | 38 hospitals | 1062 | 5 | Both | Not reported | All academic hospitals | Not reported |
Feagan et al. [17] 2000 | Retrospective Cohort | Canada | 20 hospitals | 858 | 6.5 | Both | Only comparison reported between teaching and general hospital populations | 11 teaching hospitals, 9 community hospitals | Not reported |
Fine et al. [10] 1993 | Prospective Cohort | USA | 4 hospitals | 552 | 9.5 | Both | Mean number of comorbid conditions per patient varied significantly among hospitals. | 2 university hospitals, one veterans hospitals, one community hospital | Disease severity and aetiology similar across hospitals |
Garau et al. [18] 2008 | Retrospective Cohort | Spain | 10 hospitals | 3233 | 8 | Outcome | Not reported | All tertiary hospitals | Proportion of patients belonging to each PSI class varied widely across hospitals, as did the proportion with an aetiological diagnosis. |
Gilbert et al. [19] 1998 | Prospective Cohort | USA/Canada | 4 hospitals | 1328 | 9.5 | Both | Significant differences in mean age, gender, racial distribution and comorbidities among the 4 sites. | Three university teaching hospitals, one community teaching | Statistically significant differences in causative organisms identified and severity of illness. |
Hedlund et al. [20] 2002 | Retrospective Cohort | Sweden | 17 hospitals | 982 | 5 | Outcome | Seven university hospitals, 10 county hospitals. | The mean PSI varied between 0.9 and 1.9 at different sites | |
Iroezindu et al. [21] 2016 | Prospective Case control | Nigeria | 4 hospitals | 400 | 6 | Outcome | Not reported | All tertiary hospitals | Not reported |
Klausen et al. [31] 2012 | Retrospective Analysis of Administrative data | Denmark | 22 hospitals | 11,322 | 8.5 | Outcome | Not reported | All Danish public health hospitals | Not reported |
Laing et al. [22] 2004 | Prospective Cohort | New Zealand | 2 hospitals | 474 | 7 | Both | Similar demographics between the two populations except significant differences in ethnicity and rates of COPD. | “Similar institutions” | No significant differences in disease severity by PSI. |
Malone et al. [24] 2001 | Retrospective Cohort | USA | 5 hospitals | 330 (52 severe) | 5.5 | POC | Not reported | All acute care facilities (Centura) | Not reported |
McCormick et al. [25] 1999 | Prospective Cohort | USA/Canada | 4 hospitals | 1188 | 9 | Both | A younger more mixed-race population identified at one site. The proportion admitted from a nursing home varied from 9 to 16%. | Three university teaching hospitals, one community teaching | Severity of illness and symptom profiles were similar across hospitals. One hospital had fewer “high risk” aetiology. |
Menendez et al. [26] 2003 | Prospective Cohort | Spain | 4 hospitals | 425 | 7 | NA | Not reported | Not reported | Not reported |
Reyes Calzada et al. [28] 2007 | Prospective Cohort | Spain | 4 hospitals | 425 | 6 | Both | No significant differences in co-morbidity, age and sex. Smoking significantly more frequent in two hospitals. | One tertiary and 3 district general hospitals | Not reported |
Schouten et al. [29] 2005 | Analysis of baseline population from RCT | Netherlands | 8 hospitals | 436 | 6.5 | POC | Not reported | Eight medium sized hospitals in the south-east of the Netherlands | Not reported |
Sow et al. [30] 1996 | Prospective Cohort | France and New Guinea | 2 hospitals | 333 | 5 | Outcome | Mean age and pre-existing illness rate was significantly lower in Guinea than France. | One hospital in the Republic of Guinea compared to one in France | Similar severity between cohorts (clinical definition not validated severity score) |