Measures of treatment effect
The RevMan V.5.3 will be used for the statistical analysis and the synthesis of the efficacy data. For dichotomous data, a risk ratio with 95% CIs will be used for analysis. For continuous data, a mean difference or a standard mean difference (SMD) with 95% CIs will be used for analysis. SMD will be used if different assessment tools were used.
Unit of analysis issues
Due to the lack of individual patient data, an analysis will be based on aggregated outcome data.
Assessment of heterogeneity
χ2 test with a significance level of p<0.1will be used in the forest plot for investigating statistical heterogeneity. Moreover, an I² test will be calculated for quantifying inconsistency. If the heterogeneity tests show little or no statistical heterogeneity in these trials(I²<50%), the fixed effects model will be used for the pooled data. The random-effects model will be used with heterogeneous data (50% <I2<75%). If there is considerable heterogeneity, meta-analysis will not be performed.
Assessment of reporting biases
A funnel plot will be generated to observe the reporting biases when more than ten studies are included.25
Data synthesis
The RevMan software (V.5.3) will be used for data synthesis. Depending on the level of statistical heterogeneity, data will be synthesised and analysed. The fixed-effects model will be used for the pooled data if the statistical heterogeneity is detected as little or low. Otherwise, the random-effects model will be used. If necessary, subgroup analysis will be performed, or the potential reasons will be analysed. The meta-analysis will not be performed on the condition of considerable heterogeneity in included trials.
Subgroup analysis and investigation of heterogeneity
.Subgroup analysis will be conducted if data is available. Variations will be considered in the characteristics of the treatments for acupuncture, participants, control types. And there will be subgroups to interpret the heterogeneity.
Sensitivity analysis
Sensitivity analysis will be used to monitor the robustness of the primary decision taken during the review process. Several decision nodes will be considered during the systematic review process to carry out a sensitivity review, such as small studies, methodological weaknesses and missing data. As suggested by the Cochrane Handbook, the sensitivity analysis will involve two steps: first, including all the studies as the primary meta-analysis does, and second, including those definitely known to be eligible. The sensitivity analysis results will be provided in summary tables. The risk of bias will be discussed in the review process, as indicated by the results of the sensitivity analysis.
Summary of evidence
The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system approach will be used to measure the quality of main outcomes by two reviewers. And the results will be present in “Summary of findings” tables. The assessments of evidence quality will be rated “high”, “moderate”, “low” or “very low”. The evidence of a particular study’s evidence will be assessed based on the risk of bias, inaccuracy, inconsistency, indirectness, publication bias, effect size or dose-response relation.