Study Identification

A search strategy of potentially eligible clinical trials was prespecified for MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials (CENTRAL) databases from inception to June 30, 2019. The terms and rules used for the search strategy are summarized in Table 1.

Summary of search strategy

The exact combinations and criteria that were specified in each of the three databases are listed in Tables S1–S3. Studies were filtered using all three layers of conditions in Table 1. The types of included literature were research articles, short communication, and abstracts having the corresponding trial registrations affiliated. The search process was repeated using sentence-wise and uppercase spellings.

After dropping duplicate studies, two reviewers independently screened the title and abstracts of the search results for initial inclusion. A third reviewer would be involved had there been a discrepancy between the first two reviewers. The rest of the texts were examined for further inclusion in parallel by two groups, each of which consisted of two reviewers. Within-group discrepancies were resolved by a cross-group reviewer, and cross-group discrepancies were discussed by all reviewers to reach a unanimous decision.

The studies were considered eligible for inclusion if they (1) were randomized clinical trials; (2) were conducted on individuals with type 2 diabetes mellitus; (3) compared GLP-1RAs of interest with each other or with a control group (restricted to placebo or no treatment) with or without the same add-on therapy in all arms; (4) had a follow-up between 24 and 30 weeks; (5) contained results of at least one of the prespecified primary and secondary endpoints; and (6) were published in English language. The final list of included studies was manually supplemented with two studies on loxenatide (Table S4) [14–16].