Reliability across content areas in progress tests assessing medical knowledge: a Brazilian cross-sectional study with implications for medical education assessments.

Abstract

Background: Brazilian medical schools equitably divide their medical education assessments into five content areas: internal medicine, surgery, pediatrics, obstetrics and gynecology, and public health. However, this division does not follow international patterns and may threaten the examinations' reliability and validity.

Objective: To assess the reliability indices of the content areas of serial, cross-institutional progress test examinations.

Design and settings: This was an analytical, observational, and cross-sectional study conducted at nine public medical schools (mainly from the state of São Paulo) with progress test examinations conducted between 2017 and 2023.

Methods: The examinations covered the areas of basic sciences, internal medicine, surgery, pediatrics, obstetrics and gynecology, and public health. We calculated reliability indices using Cronbach's α, which indicates the internal consistency of a test. We used simple linear regressions to analyze temporal trends.

Results: The results showed that the Cronbach's α for basic sciences and internal medicine presented lower values, whereas gynecology, obstetrics, and public health presented higher values. After changes in the number of items and the exclusion of basic sciences as a separate content area, internal medicine ranked highest in 2023. Individually, all content areas except pediatrics remained stable over time.

Conclusions: Maintaining an equitable division in assessment content may lead to suboptimal results in terms of assessment reliability, especially for internal medicine. Therefore, content sampling of medical knowledge for general assessments should be reappraised.