Data analysis
Data analysis was performed using a random-effects model within a frequentist method in Stata version 14.0 using the network meta-analysis.12Multiarmtrials were divied into two groups for comparisons and network meta-analyses to combine direct and indirect evidence.13 Because our outcomes are categorical data,so we estimated the pooled odds ratios (ORs) and 95% confidence interval(CI) of clinical efficacy and safety parameters,if the CI for the ORs did not involve one,the comparisions were thought to be with statistical sighificance.We evaluated the heterogeneity of all control groups was assumed to be equal and the correlation of multi-arm studies was considered and use node-splitting method using visual inspection of the forest plots assess local inconsistency .14If the treatments form a closed circle,loop inconsistency should also be tested by using χ2-test and inconsistency factors (IFs).14,15 We calculated the probabilities of ranks and surface under the cumulative ranking (SUCRA) of the treatments. The higher the treatment’s SUCRA value,value ranges from 0 to 1, the higher the likelihood that the treatment was better the curative effect. 16Under the circumstances,hierarchical cluster analyses is the better choice to achieve compare short-term topical teatment efficacy,safety and tolerability of these topical treatments.17Besides,we assessed publication bias using comparsion-adjusted funnel plot.