Placebo-controlled trials: there were 90 trials with aspirin (6,567 participants) versus placebo (6497 participants), 4 trials with aspirin plus codeine (148 participants) versus placebo (145 participants), and 3 trials with aspirin and caffeine (215 participants) versus placebo (222 participants).
Head-to-head comparisons: there were 4 trials of aspirin plus codeine (148 participants) versus aspirin (140 participants), and 2 trials of aspirin plus caffeine (156 participants) versus aspirin (163 participants).
The pooled estimate for the effect (TOTPAR) of aspirin at a single dose of 650 mg was 14.05% (95% CI: 12.27, 15.83); the NNT was 3.57 (95% CI: 3.08, 4.23). A visual inspection showed no obvious dose-response relationship for aspirin, whereas the weighted regression analysis using the inverse of the response variance as the weight showed a significant effect (P<0.00001).
The corresponding estimate for the aspirin (650 mg)-codeine (60 mg) combination was 27.25% (95% CI: 19.77, 34.74); the NNT was 2.67 (95% CI: 1.67, 6.55). A further study showed another aspirin (650 mg)-codeine (30 mg) combination to have a positive effect relative to placebo (TOTPAR 20.03%, 95% CI: 7.18, 32.90).
The positive effects of TOTPAR (%) were confirmed by the pooled estimates obtained when using SPID (%) and the proportion of patients responding with at least moderate pain relief as the outcome measure. All three formulations (aspirin alone, aspirin plus codeine, and aspirin plus caffeine) were more effective than placebo.
The combination aspirin (650 mg)-codeine (60 mg) was more effective than aspirin (650 mg) in the indirect comparison using TOTPAR as an efficacy end point: TOTPAR was 27.25% (95% CI: 19.77, 34.74) for the combination compared with 14.05% (95% CI: 12.27, 15.83) for aspirin alone. The combination was not more effective when SPID (%) or the proportion of patients responding with moderate to excellent pain relief were used. Aspirin (650 mg)-caffeine (65 mg) was no more effective than aspirin (650 mg) on any of the efficacy end points studied; the NNT was 2.55 (95% CI: 1.89, 3.93).
In head-to-head comparisons, neither the aspirin-codeine nor the aspirin-caffeine combinations were more effective than aspirin alone for any of the efficacy variables. None of the three formulations showed any more side-effects than placebo, when expressed as a rate ratio with a 95% CI.