Table 1 Results of meta-analyses of trials of nicotine gum to reduce smoking using standard and 'treat-as-one-trial' methods, with risk difference as effect measure
From: Meta-analysis, Simpson's paradox, and the number needed to treat
| Â | Pooled risk difference (95% CI) | P | NNT |
|---|---|---|---|
(a) Two selected trials (see text) | Â | Â | Â |
   Standard meta-analysis (Mantel-Haenszel method) | 0.060 (0.025 to 0.095) | 0.001 | 16.7 |
   Treat-as-one-trial | 0.033 (0.001 to 0.064) | 0.04 | 30.7 |
(b) All 51 trials | Â | Â | Â |
   Standard meta-analysis (Mantel-Haenszel method) | 0.061 (0.050 to 0.071) | <0.001 | 16.4 |
   Treat-as-one-trial | 0.081 (0.070 to 0.092) | <0.001 | 12.3 |