Table 3.
Consistency of findings when the same trial is examined in more than one clinical population
| Condition and trial | Clinical population | Overall risk of bias | Percentage excluded Median percentage (range for Aaltonen and Vashisht) |
|---|---|---|---|
|
Atrial fibrillation ARISTOTLE (2011) |
Yoon (record review for consenting patients in a single hospital)a | High | 32.3 |
| Lee (primary care electronic medical record population data) | Low | 38.7 | |
| Fanning (record review within multiple hospitals) | High | 39.5 | |
| Desmaele (clinical registry in a single hospital) | High | 54.5 | |
| Hagg (primary care electronic medical record population data) | Low | 71.1 | |
|
Atrial fibrillation ROCKET-AF (2011) |
Yoon (record review for consenting patients in a single hospital)a | High | 34.5 |
| Lee (primary care electronic medical record population data) | Low | 52.5 | |
| Desmaele (clinical registry in a single hospital) | High | 60.7 | |
| Fanning (record review within multiple hospitals) | High | 64.2 | |
|
Atrial fibrillation RE-LY (2009) |
Yoon (record review for consenting patients in a single hospital)a | High | 35.2 |
| Lee (primary care electronic medical record population data) | Low | 36.2 | |
| Fanning (record review within multiple hospitals) | High | 47.4 | |
| Desmaele (clinical registry in a single hospital) | High | 52.4 | |
|
Heart failure MERIT-HF (2000) |
Constantino (record review in a single hospital)a | Unclear | 48.0 |
| Jost (clinical registry in a single hospital) | Low | 58.8 | |
| Masoudi (National Heart Failure Project registry) | High | 82.6 | |
|
Heart failure RALES (1999) |
Masoudi (National Heart Failure Project registry)a | High | 74.7 |
| Costantino (record review in a single hospital) | Unclear | 76.0 | |
|
Acute myocardial infarction GUSTO (1993) |
Krumholz (National Research Registry of Myocardial Infarction)a | High | 84.5 |
| Krumholz (Cooperative Cardiovascular Project registry) | High | 90.6 | |
|
COPD POET-COPD (2011) |
Kruis (seven primary care databases)a | High | 77.0 |
| Halpin (primary care research database) | Low | 88.2 | |
|
COPD UPLIFT (2009) |
Kruis (seven primary care databases)a | High | 58.0 |
| Halpin (primary care research database) | Low | 77.5 | |
|
Rheumatoid Arthritis 9 trialsb |
Aaltonen (National Register for Biologic Treatment)a | High | 80.0 (56.0 to 92.4) |
| Vashisht (RA Investigators’ Network database research registry) | High | 89.6 (74.7 to 91.6) | |
| Vashisht (Veterans’ Affairs Rheumatoid Arthritis research registry) | Unclear | 97.4 (75.6 to 98.4) | |
|
Rheumatoid Arthritis 21 trialsb |
Vashisht (RA Investigators’ Network database research registry)a | Unclear | 89.0 (64.9 to 93.5) |
| Vashisht (Veterans’ Affairs Rheumatoid Arthritis research registry) | High | 97.4 (72.7 to 99.1) |
a Marked trials with most conservative estimate of percentage of patients excluded were analysed
b See supplementary table S14 for individual trial comparisons