Medical competence, diagnostic accuracy and treatment accuracy among health workers in Africa: A systematic review protocol

The shortage, both quantitative and qualitative, of the health workforce in Africa has been widely reported. Recent regional reports (Organization, 2021; Asamani et al., 2024) have shown an increasing number of the stock of health workers in Africa as a result of increasing production capacity. Paradoxically, the fiscal space constraints of the respective member countries do not allow for the recruitment of all the qualified health professionals, creating high-health worker unemployment in Africa. As such, there are critical gaps in the health workforce needed for universal health coverage. Additionally, recent discourse has reported inconsistencies in the quality of the recruited health workforce across member countries, health professions, diseases, and assessment methods. Thus, this review sought to synthesize the competencies, diagnostic accuracy and treatment accuracy of the health workforce in the African region. The primary synthesis will be structured around competence construct, clinical domain, geography, cadre and measurement method. Studies will first be grouped by outcome type: clinical knowledge or vignette competence; diagnostic accuracy; treatment accuracy or correct management; guideline adherence and prescribing; observed practice or standardised-patient performance; and intervention-related change. Within each group, results will be summarised by clinical domain, country, cadre and study design. Large representative studies will be distinguished from small facility-level or programme studies because their inferential value differs. Meta-analysis will be considered only when at least three studies report sufficiently comparable outcome definitions, denominators and populations. For binary outcomes, proportions or risk ratios may be pooled using random-effects models after variance stabilisation where appropriate. For continuous competence scores, standardised mean differences may be considered if instruments are conceptually comparable. Statistical heterogeneity will be assessed using I2, prediction intervals where useful and visual inspection of forest plots. If pooling would obscure more than clarify, results will be reported narratively with structured tables and descriptive visualisations. Where meta-analysis is not appropriate, SWiM-informed synthesis will describe the grouping logic, report study-level results transparently, avoid vote counting by statistical significance alone and explain how study size, representativeness, risk of bias, directness and consistency influence interpretation (Campbell et al., 2020). Quantitative findings will be reported with denominators, percentages, confidence intervals and effect estimates as available. The final review will avoid producing a single continental competence estimate unless the underlying evidence supports that inference. Planned Subgroup and Sensitivity Analyses • Cadre: physicians, nurses, midwives, clinical officers, non-physician clinicians, community health workers and mixed cadres. • Clinical domain: maternal and child health, malaria, HIV and infectious diseases, cervical cancer screening, emergency care, general primary care and other domains. • Measurement method: vignette, written test, standardised patient, direct observation, record review, surveillance data and diagnostic reference standard. • Setting: country, region, facility level, facility ownership, urban or rural location and public or private sector where reported. • Risk of bias: primary interpretation including all eligible studies, with sensitivity analyses excluding high-risk studies and separating large, representative studies from smaller, selected samples.