JMIR Publications

ABSTRACT

Background:

Objective measurement of physician workload is essential for understanding how clinical and administrative activities are distributed in primary care and evaluating changes associated with digital technologies. However, measurement approaches differ in the activities they capture and often require substantial manual observation and coding.

Objective:

This scoping review aimed to map the methods and metrics used to objectively measure physician workload in primary care using primary data and identify components of workload measurement potentially suitable for automation.

Methods:

Following Joanna Briggs Institute guidance and the PRISMA extension for Scoping Reviews, we searched MEDLINE via PubMed, Scopus, Web of Science, and CINAHL for English-language, peer-reviewed studies published between January 1, 2016, and June 30, 2026. Reference lists and forward citations were also searched. Eligible studies reported objective workload measures derived from direct observation, recordings, study-specific digital activity capture, or prospectively timestamped activity records. Studies based exclusively on subjective measures or secondary analyses of routinely collected electronic health record (EHR) audit logs were excluded. Reviewers independently screened studies, and data extracted by one reviewer were verified by two others. Findings were synthesized descriptively without formal risk-of-bias appraisal.

Results:

Thirteen studies across six countries were included. Eight used direct observation, and 12 reported at least one time-based measure. Measures captured consultation and direct patient care, EHR and documentation activities, task volume and workflow, and total working time, including work outside scheduled hours. Task counts, sequences, transitions, and operational failures provided information beyond activity duration. Two studies examined AI-assisted documentation. Measurement remained heavily dependent on manual activity classification and temporal coding. Differences in task definitions, observation boundaries, and outcome measures limited comparability across studies. Potential automation targets included activity classification, identification of task boundaries and workflow transitions, and calculation of workload metrics from observational data. These applications represented proposed methodological extensions rather than validated approaches.

Conclusions:

Objective workload measurement in primary care remains dominated by time-based indicators that requires substantial human effort. Standardized task definitions and reporting are needed to improve comparability. Future research should evaluate whether automated coding can reduce measurement effort while preserving temporal accuracy and clinical context, using independently human coded observations as reference standards and retaining human oversight for ambiguous activities.