Recruitment and participation of general practices in research networks: quantitative evaluation of the first joint study in six German practice-based research networks

Background Recruiting general practices for research is challenging. Practice-based research networks (PBRNs) may improve participation, but little is known about how newly established PBRNs function operationally in multicentre studies. Aim To evaluate recruitment effort, participation, data completeness, and acceptability in the first joint study of six German PBRNs. Design and setting Multicentre cross-sectional feasibility study across 18 university sites in six German PBRNs. Method GPs completed an online REDCap survey covering practice characteristics, aggregated patient numbers manually extracted from practice management systems (PMS), and attitudes towards post-COVID/post-vaccination-related symptoms. Sites documented contacts, reminders, and time investment. A two-step cluster analysis characterised site-level recruitment strategies, and a generalised linear mixed model examined predictors of data completeness. Results Of 906 contacted GPs, 156 participated (17.2%); 69 (7.6%) provided near-complete datasets (≥90% item response). Recruitment required 134 staff-hours overall, equivalent to about 2 hours and 13 contacted GPs per near-complete dataset, and varied substantially between sites. Two recruitment clusters emerged: an email-oriented, low-effort strategy that recruited more practices overall, and a phone-intensive, high-effort strategy that achieved higher response rates but lower acceptability. Data completeness was lowest for PMS-derived patient data. Completeness varied by site, but not by PBRN or recruitment cluster, and was higher among younger and female GPs, in multi-doctor practices, and among those with prior research experience. Conclusion In this first joint German PBRN project, participation and data completeness were limited despite network affiliation. Site-level capability, practice research experience, and easier extraction of routine data appear crucial for future PBRN studies.