Public Medical Schools in Brazil Lead the Pack in New National Proficiency Exam

Brazil has one of the fastest-growing medical education systems in the world, and in 2025 the country introduced a new national examination designed to measure whether that growth is producing competent new physicians. The National Medical Education Assessment Examination, known as ENAMED, evaluates graduating medical students and, crucially, allows regulators to judge each medical course not by its average score but by the proportion of its graduates who reach a predefined proficiency threshold. A new nationwide analysis published in BMC Medical Education has now used the first year of ENAMED results to ask a pointed question: which kinds of institutions are actually getting the majority of their graduates over the proficiency bar?

The study, led by Bruno B. Andrade of the Oswaldo Cruz Foundation and Clariens Educação together with co-first authors Klauss Villalva-Serra and Rodrigo C. Menezes, took a retrospective observational approach covering medical programs across the entire country. The team linked ENAMED-2025 results to institutional and regulatory characteristics of each medical program, including its administrative category, the year it was established, and the amount of enrollment it is authorized to offer. The outcome they focused on was deliberately strict: a course met the benchmark only if at least 60 percent of its graduating examinees were classified as proficient.

What makes the analysis statistically distinctive is its use of hierarchical Bayesian modeling. Rather than treating every school’s result as an isolated number, the models estimated each program’s posterior probability of meeting the benchmark while accounting for course size and for geographic clustering at both the state and municipality levels. This matters because medical schools within the same state, or within the same municipality, are exposed to shared regional conditions, from the local labor market for faculty to the capacity of teaching hospitals. The researchers used Integrated Nested Laplace Approximation, an efficient computational method for fitting such spatially structured models, and reported credible intervals that express the uncertainty around each estimate.

The headline finding is a clear stratification by institutional type. Federal and state public medical schools showed the highest probabilities of meeting the proficiency benchmark, followed by community-based institutions. Private programs fared worse overall, and for-profit schools had the lowest probabilities of all. In a country where the private sector has absorbed much of the recent expansion in medical training, this gradient is likely to intensify debates about whether market-driven growth can be reconciled with uniform professional standards.

Timing also emerged as a significant factor. Programs established during Brazil’s recent expansion period, when federal policies pushed to increase physician supply and reduce regional disparities, were less likely to meet the benchmark than older, more established schools. This pattern suggests that new programs may need time to mature, to recruit experienced faculty, and to build the clinical training infrastructure that proficiency examinations reward. It also raises questions about the pace at which new medical courses were accredited during the expansion years.

Perhaps the most technically interesting result concerns scale. Greater authorized enrollment was associated with a lower probability of reaching the 60 percent proficiency threshold, and the strongest inverse association appeared among for-profit institutions. In other words, the bigger the intake a school is authorized to train, the harder it appears to bring the majority of its graduates to proficiency, with the effect most pronounced where commercial incentives are strongest. The finding aligns with a long-standing concern in medical education: that rapid scaling of class sizes can outstrip the availability of clinical placements, supervision, and formative assessment.

The authors are careful about what these results do and do not show. Because the analysis is cross-sectional, the associations between institutional characteristics and benchmark attainment cannot be interpreted as causal. A for-profit school with large enrollment and low proficiency rates may differ from a small public school in many unmeasured ways, including student selection, faculty qualifications, and local health system integration. The Bayesian framework quantifies uncertainty honestly, but it cannot manufacture causal inference from a single year of observational data.

Even so, the study arrives at a consequential moment for Brazilian health policy. The country’s Unified Health System, the SUS, depends on a steady supply of physicians distributed across a continent-sized territory, and expansion policies were designed partly to correct the concentration of doctors in wealthy urban centers. ENAMED was introduced to provide a proficiency-based measure for institutional evaluation, complementing older instruments such as the Preliminary Course Concept, an aggregate performance indicator calculated by the National Institute for Educational Studies and Research Anísio Teixeira on behalf of the Ministry of Education. By shifting attention from averages to the share of graduates who are genuinely proficient, ENAMED makes it harder for a strong minority of high scorers to mask a weak majority.

The methodological choices also carry lessons for other countries wrestling with medical workforce expansion. Hierarchical models that respect geographic clustering and course size can produce more stable institutional rankings than raw pass rates, which are notoriously noisy for small cohorts. Reporting posterior probabilities and credible intervals, rather than binary pass-fail labels, gives regulators a graded view of risk: a school with a 55 percent posterior probability of meeting the benchmark next year warrants different attention than one sitting at 5 percent. The same architecture could be applied to licensure and accreditation data in other large, regionally diverse education systems.

For now, the study’s practical message is that administrative category, the timing of a program’s establishment, and authorized enrollment are all associated with substantial differences in the probability of meeting Brazil’s new proficiency benchmark. As ENAMED accumulates years of data, repeated analyses of this kind will show whether newly established programs converge toward the standards of older schools, whether large-enrollment institutions adjust their capacity, and whether the public-private gap narrows or widens. The first year of the examination has already provided something Brazil previously lacked: a nationwide, course-level, proficiency-based map of where its medical education system is succeeding and where it is under strain.

Subject of Research: Institutional predictors of medical graduates meeting the proficiency benchmark on Brazil's national medical education assessment

Article Title: Institutional characteristics associated with meeting the proficiency benchmark in Brazil’s National Medical Education Assessment (ENAMED): a nationwide Bayesian hierarchical analysis

Article References: Andrade, B. B., Villalva-Serra, K., Menezes, R. C., Brito, Q. H., Quintanilha, L. F., & Avena, K. M. (2026). Institutional characteristics associated with meeting the proficiency benchmark in Brazil’s National Medical Education Assessment (ENAMED): a nationwide Bayesian hierarchical analysis. BMC Medical Education. https://doi.org/10.1186/s12909-026-10604-0

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10604-0

Keywords: ENAMED, medical education, Brazil, proficiency benchmark, Bayesian hierarchical model, for-profit medical schools, physician workforce, higher education assessment, authorized enrollment, public universities, health policy, accreditation