Direct and indirect approaches to addressing bias from missing data in primary care electronic health records: a rapid scoping review

Biases in primary care electronic health records (EHRs), often introduced by missing data for underserved groups, may affect the validity of research findings and contribute to health inequalities. This rapid scoping review identified and synthesised evidence on methods used to identify, report and mitigate bias arising from missing data in primary care EHRs. Seventy-six studies were included. Most used machine learning techniques, medical statistics approaches, or additional data sources to address missingness, while relatively few explicitly reported patterns of missing data. The review highlights current approaches and identifies opportunities to improve reporting and mitigation of bias in routine primary care data