Additional analyses

We considered three sources of heterogeneity: methodological, clinical, and other sources of heterogeneity [61, 73]. These sources were selected a priori based on information from previous systematic reviews on this research topic and through discussions between the reviewers [45–48, 74]. Our “a priori” defined potential sources of heterogeneity were listed in our published protocol [14]. The type of stakeholders, i.e., patients, clinicians, and office staff, was excluded as a source of diversity because outcomes were analyzed separately for each type of stakeholder. We reported when “post hoc” defined sources of heterogeneity were investigated.

The presence of statistical heterogeneity was investigated by calculating Cochran’s Q, the degrees of freedom based on the number of eligible studies, and the pertinent p value [64, 75–77]. We also calculated the following statistics: Kendall’s τ ², τ, and I ² [64, 75, 77–80]. These calculations, their use, and strategies for dealing with heterogeneity were explained in our published protocol [14, 61].

Our protocol described our planned methods for conducting subgroup analyses, meta-regressions, and sensitivity analyses [14]. Criteria, rationales, and caveats for undertaking such research procedures were also outlined in this protocol [61, 81, 82].