In Reply We appreciate the responses by Chen and Li and Chen and Liu to our meta-analysis.1 Chen and Li suggest that the databases we searched may have resulted in selection bias and possibly false conclusions. We carefully chose PubMed, Web of Science, and PsycINFO for our search. These databases cover biomedical, social, behavioral, and mental health research. This is important because our research question is not solely medical but rather has a multidisciplinary character. Despite considerable overlap, we agree EMBASE is a valuable addition to PubMed for biomedical research. However, because of the multidisciplinary research question, we deemed a search across multiple disciplines more important than searching multiple database in the same field. An earlier meta-analysis searched both EMBASE and PubMed but identified no additional unique cohorts compared with ours.2 A 2018 meta-analysis3 covering the same research question included 44 studies compared with 71 studies in our meta-analysis while searching the same databases. Altogether, we believe we did as much as possible to minimize risk of selection bias and we feel confident our study covered the most relevant studies available. Moreover, our results are in line with previous meta-analyses.2,3 Nevertheless, we acknowledge that selection bias is an important concern in meta-analysis that received our full attention.