Table 1 Estimates of bias-adjusted pooled effect in five meta-analysis case studies. FE: fixed effects, RE: random effects.
From: Application and investigation of a bound for outcome reporting bias
Example | Cancer | Immunoglobulin | Prostaglandin | Albumin | Asthma |
|---|---|---|---|---|---|
Original pooled logOR: | Â | Â | Â | Â | Â |
FE | -0.360 | -0.204 | -0.071 | 0.366 | -1.258 |
RE | Â | -0.311 | -0.142 | Â | -1.376 |
Tau-squared | 0 | 0.111 | 0.200 | 0 | 0.403 |
N | 5 | 12 | 18 | 18 | 17 |
(Range of number of participants in trials) | (44 to 733) | (66 to 2416) | (20 to 2517) | (10 to 141) | (9 to 50) |
M | 4 | 3 | 1 | 2 | 1 |
(Number of participants per trial) | (67,91,146, 750) | (20,111,235) | (60) | (16,28) | (35) |
Copas-adjusted estimate: | Â | Â | Â | Â | Â |
FE | -0.045 | -0.133 | -0.033 | 0.591 | -1.153 |
RE | Â | -0.135 | -0.050 | Â | -1.241 |
Maximum imputation-adjusted estimate: | Â | Â | Â | Â | Â |
FE (constrained) | 0.229 | -0.111 | -0.027 | 0.589 | -1.034 |
FE (unconstrained) | 0.269 | -0.110 | -0.027 | 0.589 | -1.034 |
RE | Â | -0.154 | -0.049 | Â | -1.160 |