Table 3. Variation in terms of statistical conclusion for Peto fixed-effect meta-analyses.
| RevMan method | |||||
Fixed-effect ( ) |
Fixed-effect ( ) |
||||
| Counts (cell percentages) | Counts (cell percentages) | ||||
| Stata method | NS | Sig | NS | Sig | |
| FE (Peto) | NS | 1367(62.0%) | 0(0.0%) | 1184(55.5%) | 0(0.0%) |
| Sig | 4(0.2%) | 835(37.9%) | 0(0.0%) | 950(44.5%) | |
| DL | NS | 1367(62.0%) | 0(0.0%) | 1178(55.2%) | 318(14.9%) |
| Sig | 4(0.2%) | 835(37.9%) | 6(0.3%) | 632(29.6%) | |
| DLb | NS | 1367(62.0%) | 36(1.6%) | 1177(55.2%) | 359(16.8%) |
| Sig | 4(0.2%) | 799(36.2%) | 7(0.3%) | 591(27.7%) | |
| VC | NS | 1367(62.0%) | 17(0.8%) | 1181(55.3%) | 343(16.1%) |
| Sig | 4(0.2%) | 818(37.1%) | 3(0.1%) | 607(28.4%) | |
| MVa | NS | 1366(61.9%) | 117(5.3%) | 1178(55.2%) | 448(21.0%) |
| Sig | 4(0.2%) | 718(32.5%) | 6(0.3%) | 502(23.5%) | |
| No comp* | 1(0.0%) | 0 (0.0%) | 0(0.0%) | 0 (0.0%) | |
| MVb | NS | 1367(62.0%) | 39(1.8%) | 1180(55.3%) | 344(16.1%) |
| Sig | 4(0.2%) | 796(36.1%) | 4(0.2%) | 606(28.4%) | |
| BP | NS | 1369(62.1%) | 179(8.1%) | 1178(55.2%) | 462(21.6%) |
| Sig | 2(0.1%) | 656(29.7%) | 6(0.3%) | 488(22.9%) | |
| B0 | NS | 1367(62.0%) | 13(0.6%) | 1182(55.4%) | 266(12.5%) |
| Sig | 4(0.2%) | 822(37.3%) | 2(0.1%) | 684(32.1%) | |
| VC2 | NS | 1367(62.0%) | 4(0.2%) | 1177(55.2%) | 362(17.0%) |
| Sig | 4(0.2%) | 831(37.7%) | 7(0.3%) | 588(27.6%) | |
| DL2 | NS | 1367(62.0%) | 0(0.0%) | 1180(55.3%) | 333(15.6%) |
| Sig | 4(0.2%) | 835(37.9%) | 4(0.2%) | 617(28.9%) | |
| ML | NS | 1362(61.7%) | 1(0.0%) | 1174(55.0%) | 156(7.3%) |
| Sig | 4(0.2%) | 826(37.4%) | 2(0.1%) | 790(37.0%) | |
| No conv† | 5(0.2%) | 8(0.4%) | 8(0.4%) | 4(0.2%) | |
| REML | NS | 1343(60.9%) | 3(0.1%) | 1177(55.2%) | 293(13.7%) |
| Sig | 4(0.2%) | 819(37.1%) | 6(0.3%) | 653(30.6%) | |
| No conv† | 24(1.1%) | 13(0.6%) | 1(0.0%) | 4(0.2%) | |
| PL | NS | 1364(61.8%) | 135(6.1%) | 1174(55.0%) | 382(17.9%) |
| Sig | 2(0.1%) | 690(31.3%) | 1(0.0%) | 564(26.4%) | |
| No conv† | 5(0.2%) | 10(0.5%) | 9(0.4%) | 4(0.2%) | |
*
MVa fails to produce an estimate when all effects are equal, a scenario not unheard of when meta-analysing dichotomous data.
†
Iterative method failed to convergence to a solution.