Table 5.
Method (fixed or random effects), and data sources | Single zero event trials | Zero total event trials | Continuity correction | Effect estimate (95% CI) | P value | No of trials |
---|---|---|---|---|---|---|
Peto (fixed) | ||||||
IPD only | Included | Excluded | None | OR 1.66 (1.24 to 2.22) | 0.001 | 26 |
IPD+RECORD | Included | Excluded | OR 1.80 (1.46 to 2.22) | 0.000 | 27 | |
Mantel-Haenszel (fixed) | ||||||
IPD only | Included | Excluded | Constant continuity correction of 0.5 | OR 1.60 (1.20 to 2.14) | 0.002 | 26 |
RR 1.57 (1.18 to 2.08) | 0.002 | 27 | ||||
IPD+RECORD | Included | Excluded | OR 1.78 (1.44 to 2.20) | 0.000 | 27 | |
RR 1.74 (1.42 to 2.14) | 0.000 | 27 | ||||
IPD only | Included | Included | OR 1.56 (1.17 to 2.07) | 0.002 | 33 | |
RR 1.53 (1.16 to 2.02) | 0.003 | 33 | ||||
IPD+RECORD | Included | Included | OR 1.75 (1.42 to 2.16) | 0.000 | 34 | |
RR 1.71 (1.40 to 2.10) | 0.000 | 34 | ||||
Dersimonian and Laird (random) | ||||||
IPD only | Included | Excluded | Constant continuity correction of 0.5 | OR 1.54 (1.14 to 2.09) | 0.005 | 26 |
RR 1.52 (1.14 to 2.03) | 0.004 | 26 | ||||
IPD+RECORD | Included | Excluded | OR 1.75 (1.41 to 2.18) | 0.000 | 27 | |
RR 1.72 (1.39 to 2.11) | 0.000 | 27 | ||||
IPD only | Included | Included | OR 1.50 (1.12 to 2.01) | 0.007 | 33 | |
RR 1.48 (1.12 to 1.97) | 0.006 | 33 | ||||
IPD+RECORD | Included | Included | OR 1.72 (1.39 to 2.13) | 0.000 | 34 | |
RR 1.69 (1.37 to 2.08) | 0.000 | 34 | ||||
Mantel-Haenszel (fixed) | ||||||
IPD only | Included | Excluded | Treatment arm correction | OR 1.65 (1.23 to 2.20) | 0.001 | 26 |
RR 1.61 (1.21 to 2.14) | 0.001 | 26 | ||||
IPD+RECORD | Included | Excluded | OR 1.81 (1.46 to 2.24) | 0.000 | 27 | |
RR 1.77 (1.44 to 2.17) | 0.000 | 27 | ||||
IPD only | Included | Included | OR 1.62 (1.21 to 2.15) | 0.001 | 33 | |
RR 1.59 (1.20 to 2.10) | 0.001 | 33 | ||||
IPD+RECORD | Included | Included | OR 1.79 (1.45 to 2.21) | 0.000 | 34 | |
RR 1.75 (1.42 to 2.15) | 0.000 | 34 | ||||
Dersimonian and Laird (random) | ||||||
IPD only | Included | Excluded | Treatment arm correction | OR 1.58 (1.17 to 2.13) | 0.003 | 26 |
RR 1.55 (1.16 to 2.08) | 0.003 | 26 | ||||
IPD+RECORD | Included | Excluded | OR 1.77 (1.42 to 2.20) | 0.000 | 27 | |
RR 1.73 (1.41 to 2.14) | 0.000 | 27 | ||||
IPD only | Included | Included | OR 1.55 (1.15 to 2.09) | 0.004 | 33 | |
RR 1.53 (1.15 to 2.03) | 0.004 | 33 | ||||
IPD+RECORD | Included | Included | OR 1.75 (1.41 to 2.17) | 0.000 | 34 | |
RR 1.72 (1.40 to 2.11) | 0.000 | 34 |
IPD=individual patient level data; OR=odds ratio; RECORD=Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes study; RR=relative risk.
For all analyses, numbers of heart failure events observed among total population of patients was as follows: IPD only, rosiglitazone population: 122 events among 11 837 patients; IPD only, control population: 80 events among 9319 patients; RECORD, rosiglitazone population: 119 events among 2226 patients; RECORD, control population: 61 events among 2232 patients.