Data Analysis
Stata 15.1 software (network package and the network graphs package) was used to conduct network meta-analysis. 19,20 A network diagram with nodes and lines was constructed to represent different interventions, the size of the nodes represented the number of populations, and the thickness of lines between the nodes represented the number of studies included. For the quit smoking rate (dichotomous variable), the odds ratio (OR) with 95% confidence interval (CI) was used to estimate the effect size. The network meta-analysis results were summarized base on all possible pairwise comparisons, including mixed comparisons (the combined effect of direct and indirect comparisons) and indirect comparisons.
In addition, a traditional meta-analysis was conducted by RevMan software, Similarly, OR with 95% CI was used to calculate the effect size, and the P -value was used to judge whether there was a statistical difference between interventions. Statistical heterogeneity was assessed with the Chi-squared and I -squared (I 2) tests. High heterogeneity in the results was identified if the tests were statistically significant (p< 0.05) and the I 2 > 50%, in which case a random-effects model was selected. Otherwise, a fixed-effects model was used.
The smoking cessation effect of different drugs was estimated based on the surface under the cumulative ranking curve (SUCRA). The value of SUCRA ranged from 0 to 100%, when the SUCRA value was 100%, it indicated that the intervention was the most effective, and the smaller the value, the worse the intervention.