2.4. Study selection and data extraction

We will import the literature extracted from the database into EndnoteX9 software, and delete duplicates through the preliminary screening of Endnote software and literature. By reading the titles and abstracts, we will eliminate the literatures that is repeatedly published and does not meet the inclusion criteria. Then, we will consult the full text, cross- checked, exclude the literature such as nonrandomized controlled trials and conference summaries, and make a record of the reasons for the deletion of the literature. Next, according to the pre-designed information extraction table, 2 reviewers will independently extract the data and cross-check it. In the event of a disagreement, the final decision will be made by a third reviewer. We will extract the information included basic information of the study, research methods, subjects, details of the intervention and control groups, outcome indicators and measurement methods, as well as the number of events and total number of dichotomous variables, and the total sample size, mean and standard deviation of the continuous variables. If information affecting NMA results is lacking in the literature, we will attempt to contact the original researchers for relevant information. The PRISMA flow chart of the selection process is shown in Figure ​Figure11.

PRISMA flow chart.