Skip to main content

Table 1 Conditional power

From: Conditional power of antidepressant network meta-analysis

     NCP=20% NCP=80%
     Sample size Sample size
  Trial design Effect size Event probability median [range] median [range]
Efficacy
Main analysis r = 1/0 fxyN pcN = 0.17 250 [49 - 485] 1586 [894 - 4190]
Sensitivity analysis r = 1/1 fxyN pcN = 0.17 126 [10 - 384] 1198 [720 - 3464]
  r = 1/2 fxyN pcN = 0.17 117 [10 - 222] 994 [624 - 2586]
  r = 1/0 fxy d = 0.01 pcN = 0.17 250 [49 - 483] 1576 [916 - 4176]
  r = 1/0 fxy d = 0.1 pcN = 0.17 245 [49 - 509] 1612 [866 - 4371]
  r = 1/0 fxy d = 0.2 pcN = 0.17 242 [49 - 538] 1644 [817 - 4682]
  r = 1/0 fxy d = 0.3 pcN = 0.17 238 [49 - 566] 1628 [773 - 9660]
  r = 1/0 fxy d = 0.5 pcN = 0.17 236 [47 - 622] 1668 [695 - 9772]
  r = 1/0 fxy d = 0.8 pcN = 0.17 228 [44 - 699] 1754 [595 - 9880]
  r = 1/0 fxyN pc = 0.01 706 [51 - 1407] 3678 [2471 - 10000]
  r = 1/0 fxyN pc = 0.1 373 [50 - 741] 2384 [1370 - 9927]
  r = 1/0 fxyN pc = 0.2 212 [48 - 416] 1367 [781 - 3793]
  r = 1/0 fxyN pc = 0.3 136 [43 - 318] 1044 [599 - 3217]
  r = 1/0 fxyN pc = 0.5 96 [36 - 271] 874 [503 - 2877]
Tolerability
Main analysis r = 1/0 fxyN pcN = 0.57 198 [40 - 320] 791 [521 - 1246]
Sensitivity analysis r = 1/1 fxyN pcN = 0.57 104 [64 - 278] 738 [414 - 1258]
  r = 1/2 fxyN pcN = 0.57 102 [81 - 168] 675 [480 - 1161]
  r = 1/0 fxy d = 0.01 pcN = 0.57 207 [41 - 318] 766 [619 - 1202]
  r = 1/0 fxy d = 0.1 pcN = 0.57 192 [40 - 321] 782 [569 - 1283]
  r = 1/0 fxy d = 0.2 pcN = 0.57 181 [40 - 343] 791 [513 - 1373]
  r = 1/0 fxy d = 0.3 pcN = 0.57 170 [39 - 366] 817 [471 - 1462]
  r = 1/0 fxy d = 0.5 pcN = 0.57 146 [39 - 420] 908 [409 - 1782]
  r = 1/0 fxy d = 0.8 pcN = 0.57 121 [40 - 615] 1027 [320 - 2497]
  r = 1/0 fxyN pc = 0.01 319 [41 - 523] 1258 [824 - 1991]
  r = 1/0 fxyN pc = 0.1 153 [41 - 277] 662 [442 - 1046]
  r = 1/0 fxyN pc = 0.2 55 [37 - 117] 371 [257 - 582]
  r = 1/0 fxyN pc = 0.3 40 [32 - 72] 277 [207 - 435]
  r = 1/0 fxyN pc = 0.5 34 [31 - 55] 227 [192 - 365]
  1. Listed are median [range] of sample sizes at 20% and 80% conditional power (NCP=20%, NCP=80%) with respect to the three parameters considered in the sensitivity analysis, i.e., trial design (r), effect size (fxy), and event probability. The first row for each outcome lists results obtained in the main analysis considering a trial design with a ratio of direct/indirect evidence of r = 1/0, with anticipated effect sizes set equal to the network estimates (fxyN), and anticipated event probabilities set equal to the average network event probabilities (pcN). The remaining rows list results obtained in the sensitivity analyses considering trial designs with ratios of direct/indirect evidence of r = 1/1 and r = 1/2, varying anticipated effect sizes (fxy in terms of Cohen’s d), and varying anticipated event probabilities (pc). The supplementary appendix provides details on sensitivity analyses (Supplement 1, Fig. S3, Tab. S3)