Eighteen relevant RCTs were identified (n=8,014 participants, range 62 to 1,686). Only RCTs with high-level scientific reliability, using placebo patches in control groups, were included.
The pooled odds ratio showed a significant benefit in the use of nicotine patches (OR 1.753, 95% CI 1.520 to 2.021). There was no evidence for heterogeneity or for publication bias. The combined risk ratio also showed a significant benefit for stopping smoking (RR 1.594, 95% CI 1.406 to 1.806), and there was no evidence for heterogeneity or for publication bias. However, the funnel plots for both odds ratios and risk ratios were not symmetrical indicating possible publication bias.
The combined weighted risk difference was also calculated giving a significant positive benefit for nicotine patches of 5.33% (95% CI 3.99 to 6.66), but in this case there was evidence for heterogeneity (p=0.04), and evidence for possible publication bias (Kendall correlation coefficient=0.2810; p=0.103). Repeat of the analysis of combined weighted risk difference using a random-effects model also gave a significant benefit for nicotine patches of 6.17% (95% CI 4.12 to 8.22%).
The pooled analysis for number of people needing treatment gave 18.8 (95% CI 15.0 to 25.1).
The visual checking of funnel plots implied that some studies which did not report a benefit for nicotine patches may not have been published.