World-wide, medical workforces are distinctively homogeneous due to socioeconomic and academic barriers for underrepresented groups, biased recruitment policies, and structural inequalities within medical education and professional bodies (1). This has generated workforces that often do not reflect the diversity of the population they serve, undermining trust with marginalized communities, and reinforcing inequalities within health systems and the broader society (2).

The lack of diversity is particularly evident in the selection of medical specializations, with specialties like surgery seeing an over-representation of male, affluent, and white students (3), and other medical areas such as gynaecology and primary care been disproportionately chosen by female students, often from diverse racial and economic background (4).

The literature suggests that medical students' specialty choices are influenced by personal factors (such as individual interests, values, and demographics), specialty characteristics (lifestyle, work-life balance, income, prestige), and clinical experiences (5). From an economic stand point, known factors include a desire for controllable work hours, geographical location of practice, perceived job opportunities, financial rewards, and guidance from mentors and family (6). All such factors have contributed to scarcity and imbalances in the distribution of specialists, complicating forecasts and training of the medical workforce, ultimately affecting the availability of specialist services in peripheral parts of health systems, particularly for low- and middle-income countries (7).

With a view to increasing access to medical training for individuals from population groups historically underrepresented in the profession, “affirmative action” (AA) policies have been implemented in high- and middle-income countries (LMIC) (8). These policies often involved preferential or compensatory measures (such as admissions quotas, race/ethnicity-based selection, financial subsidies or scholarships), aimed at remedying past inequities, improving diversity in the medical workforce, and enhancing health equity in the society (9) (10). AA measures have often been framed as ‘public policies’ when carried out through public institutions, or private sector ones when implemented through incentives for private schools to recruit and employ from disadvantaged populations (11).

Evidence from the economic literature broadly shows that race- and income-conscious affirmative-action policies increase enrolment and representation of under-represented racial/ethnic groups at selective institutions (12). Studies on law-schools in the United States (US) (13) have reported evidence of a “mismatch” between quota students’ preparation and institutions’ high requirements. Other large empirical studies have found little or mixed evidence that AA causes net harm to beneficiaries (12, 14).

What impact AA policies have in the field of medical education is, however, contested. Studies from the US suggests AA measures help sustain a more racially and socioeconomically diverse pipeline of physicians, which, in observational studies has been associated with greater likelihood of practicing in underserved areas and improved patient-provider (15, 16). On the other hand, evidence that affirmative-action admits perform substantially worse in medical school (a “mismatch” causing harm) is limited and mixed; a study from Brazil comparing social quota students to regular path students found no significant difference in attending medical residency programs, with social quota students often showing strong interest in rural practice and general practice (17).

In Brazil, social and racial quotas gradually expanded in the 2000es with the roll-out of the Student Financing Fund - FIES, 2001- providing low-interest loans for those enrolled in private institutions (18) and fees exemptions by the University for All Program (PROUNI, 2005), offering full and partial scholarships to low-income and non-white students from public high schools (19). In 2012, Federal Law 12.711 mandated that federal universities reserve at least 50% of places for students from public schools, self-declared Black, mixed-race (Pardos), or Indigenous, those with family incomes up to 1.5 minimum wages (20). Admission through quotas requires lower cutoff grades in the National High School Exam (ENEM), which some scholars arguing would affect students’ decisions to use quotas and contribute to stigma (21).

In 2022, 11.5% of the 5.1 million students enrolled participated in quota schemes, while 36.8% of the 3.2 million students in private institutions received some form of student financing. In medicine, however, a traditional course with some of the highest tuition fees in Brazil, these proportions were lower, with only 9% of students entering through quotas and 23.2% receiving financial aid (22).

Medical residency (MR) in Brazil represents the primary pathway for training medical specialists, although it is not mandatory for the practice of medicine as a generalist. In 2024, 47,718 physicians - approximately 8% of all practicing doctors - were enrolled in MR programs (23) (22). These programs are publicly funded through government scholarships, filled via competitive selection processes, and characterized by training lasting two to five years, depending on the specialty.

The rapid expansion of private sector medical schools over the past decade has intensified competition for MR positions, as the number of training slots has not kept pace with the volume of new graduates (24). Within this scenario, data from the 2025 Brazilian Medical Demographics Study indicated that the majority of physicians enrolled in MR in 2024 (62.8%) had graduated from private medical schools (23).

This study examines whether medical students admitted through affirmative action policies in Brazil are less likely to enter medical residency programs than their peers admitted via the regular track, and whether their specialty choices differ, particularly with regards to primary care practice (or Family & Community Medicine, as it is known in Brazil). Because affirmative action measures in Brazil have been implemented through public and private sector policies - quotas for underrepresented groups in public medical schools and financial subsidies for students in private medical schools - we analyse the effects of those policies separately.

By comparing the effect of quota-based and subsidy-based measures on medical students’ likelihood of entering medical residency and their selection of specialties, we aim to provide empirical evidence on the relative effectiveness of affirmative action policies across the public and private medical education sectors in LMICs.

We investigated the association between students’ participation in affirmative action (AA) programs during medical school - specifically admission through quota systems in public schools and receipt of financial aid in private schools - and subsequent entry into medical residency (MR) and choice of primary care specialty. Because socioeconomic profiles differ between students in public and private institutions (25), outcomes were analysed separately across four groups: (1) public school students admitted via quotas vs. regular admissions, and (2) private school students receiving financial aid vs. those without support.

The exposures of interest were:

The outcomes were:

We analysed the first MR admission regardless of subsequent re-enrolments. At the time of data extraction, residents were classified as completed (47.7%), active (43.1%), withdrawn (8.7%), or other status (0.5%). Because only first admissions were included, all programs analysed corresponded to direct-entry specialties.

We adjusted analyses for demographic and institutional factors: sex; self-reported race (IBGE categories: white, black, brown, indigenous, yellow); type of secondary school (public/private); age at medical school entry (≤ 21 or > 21 years); medical school type (public/private); geographic region (North, Northeast, Southeast, South, Central-West); municipal population size (> 300,000 vs. <300,000); years of school operation (≤ 10, 11–20, ≥ 21); and ENADE 2023 score (high: 3–5 vs. low: 1–2). Schools were classified as public if tuition-free; those charging tuition, with or without subsidies, were classified as private.

Our database included 110,911 medical students that graduated between 2018 and 2022. Most students were in medical institutions located in the country’s Southeast and Northeast; the majority of them were self-declared Whites, from a private secondary school background.

Sixty-eight percent of the students in our cohort graduated in private medical schools, and 31.4% from public ones. Among the public medical school students, 28.1% were admitted through racial, disability and income quotas; among the private medical school ones, 44.5% received some form of financial support.

Quota students were considerably more diverse than their peers in public medical schools, with fewer Whites (35.6% Vs 66.8%), more Mixed-race (49.8 Vs 24.6), and Blacks (12.1 Vs 3.9). They also came disproportionately from public secondary schools than their peers (95.3 Vs 20.1).

In private medical schools, although to a less degree, recipients of financial support were also somewhat more diverse than their peers, with fewer Whites (71.2% Vs 78.9%), more Mixed-race (22.9 Vs 16.8), and Blacks (3.1 Vs 1.7). A higher proportion of such scholarship students were from a public secondary school background (22.6 Vs 15.0).

<Insert Table 1 here>

55.7% of all students entered a specialty residency programme – 62.8% of the public medical school ones and 52.5% of the private ones. However, such proportions were lower for the public students admitted through quotas (52.8%), and higher for the private students in receipt of financial aid (55.0%).

Our logistic regression models identified similar factors predicting entrance in a residency programme for public and private medical school students (Table 2). Positive factors were being female, white, and having schooled in a private secondary school. Negative factors for both types of students were being over 21 years at graduation, being black, the medical school being in the Northern region, and in municipalities with fewer than 300,000 inhabitants, age of the medical school (< 21 years of running medical degrees), and lower ENADE scores attributed to the medical school.

Our multivariate models showed that having been admitted in public medical schools through quotas decreased the odds of the student entering a residency programme in comparison to their peers admitted through the regular path (adjusted Odds Ratio [aOR]: 0.73; 95% CI: 0.67; 0.79). On the other hand, being recipient of financial support in private medical schools increased the chance of entering a residency programme for a specialization, in comparison to the other students in private medical schools (aOR:1.23; 95% CI: 1.19; 1.27).

<Insert Table 2 with both multivariate regression models here>

As for the specific specialties chosen by those students that entered a residency programme, the most common ones were similar for public and private medical school students, namely: Internal medicine, General surgery, Paediatrics, Family & Community Medicine, and Obstetrics & Gynaecology (see Fig. 1). Choice of specialty did not differ considerably between quota or financial support students and their peers in public and private medical schools, apart from Family & Community Medicine that was selected by 15.3% of quotas students (Vs 9.7% of their public-school peers) and by 8.2% of financial support students (Vs 6.9% of private school peers).

<Insert Fig. 1 here>

Multivariate logistic regression models for the selection of Family & Community Medicine show that being male is negatively associated with such specialty among public medical school students (aOR: 0.81; 95% CI: 0.73; 0.89) as well as private medical schools ones (aOR: 0.67; 95% CI: 0.61; 0.73). Coming from a private secondary school background is negatively associated with the selection of the primary care specialty among public (aOR: 0.82; 95% CI: 0.71; 0.95), but not for private medical school’s students (aOR: 0.98, 95% CI: 0.88; 1.09).

As for positive associations with entry in Family & Community Medicine, being over 21 years at graduation, and Black or Mixed-race were strong predictive factors for both public and private medical school students. Also having attended a medical school outside Brazil’s Southeast region was positively associated with the selection of Family & Community Medicine for both public and private medical school students.

Being a quota student in public medical schools was strongly associated with the selection of the primary care specialty (aOR: 1.28; 95% CI: 1.09; 1.49). Being recipient of financial subsidies in private universities was also a predictor of primary care choice (aOR: 1.09; 95% CI: 1.00; 1.18), although the magnitude of the effect was smaller.

Our analysis of Brazil’s Higher Education Census and National Medical Residency Commission databases showed that, between 2018–2022, 28.1% of all the students in public medical schools were admitted through AA quotas, and 44.5% of those in private medical schools received financial support from the government. Rate of entry in the Medical Residency training programme for quotas students was lower than for their peers in public medical schools, while for recipients of financial support the rate was higher than for their peers in private medical schools. The rate of selection of the primary care Family & Community Medicine specialty was higher in both quota and financial support students than that in their comparable groups, although it was of a smaller order of magnitude for the latter.

Being female, white, of young age at graduation, and having schooled in private secondary schools, were student characteristics associated with entrance in Medical Residency, as well as higher quality scores attributed to the medical school. Being female, non-white, of older age, and medical school location outside the Southeast region, were the factors associated with choice of the primary care specialty.

The necessity of starting to work before specialization because of financial hardship might in part explain the lower rate of entry in MR for quota students, typically less well-off than their peers in public medical schools (30). By the same token, being in receipt of financial support might have encouraged students in private medical schools to further their education. We acknowledge however that such cohort of students was very diverse, and included individuals from relatively wealthier backgrounds, possibly with access to better support to enter specialisation.

Both quotas and financial support policies appear to have boosted the selection of Family & Community Medicine, although the impact for the former was much larger than for the latter. The different profile of quotas and financial support students might have contributed to such preference, as these tend to come from rural and less wealthy backgrounds, characteristics typically associated with primary care specialties (25).

Students’ academic ability and performance are likely to have been a factor in their decisions to further their medical education and choice of specialty (31). The literature on performance of quota students in medical schools is inconclusive at best (32), with local studies providing evidence that quota students perform as well as their peers (33). As we had only limited information on individual students’ performance, we can only speculate that those in public medical schools must have been of greater ability, as such schools received higher ENADE scores, and it is logical to think the best students would choose and have the ability to enter highly rated and almost fees-free public medical schools (24). This would be consistent with our findings on the association between lower ENADE scores for (both public and private) medical schools and students diminished likelihood to enter medical residency. Contrary to the existing literature on this topic (34, 35), the link between student or school quality and the selection of primary care oriented specialties was not fully supported by our findings, though.

Despite their differences, both types of AA policies in Brazil appear to have borne an impact on students’ likelihood of entering medical residency and the selection of primary care specialties. Governments around the world might therefore be presented with the option of choosing quotas in public universities if they want to increase numbers of primary care specialists and generalists, or financial support to students in private medical schools, if they want to provide students with an incentive to enter medical residency (32). Ultimately, the main difference between the two is that quotas appeared to have increased diversity of medical students to a greater extent than scholarships to private students. And this remains the reason such measures were introduced in Brazil in the first place (36).

The introduction of race and income quotas in private universities as a requirement for the authorisation of new medical courses, or the increase of direct public subsidies to private medical schools, has been criticised as such measures may encourage the expansion of a lower-quality medical education market (37). In turn, our findings appear to suggest that the provision of reserved places for quota students in public undergraduate medical courses would need to be accompanied by additional financial support to ensure both their entry into and retention within Medical Residency programmes.

We acknowledge a few limitations for this study. We included students graduating between 2018 and 2022, with residency data available up to 2024. For students graduating in later years (i.e., 2021–2022), follow-up was limited, so students who delayed their residency to after such dates may not have been captured, potentially underestimating enrolment rates. However, as our previous work (23) has showed that in Brazil just over half of physicians without prior residency enrolment entered a residency program within one year of graduation, and 22.1% began a residency within two years, we believe this risk of underestimation to be limited. Additionally, the early cohorts affected by the 2012 Quota Law were relatively small and institutional adjustments were ongoing, meaning that our logistic regression estimates may reflect early-stage effects and not fully capture the longer-term impact of affirmative action policies. Finally, due to the limited number of individual covariates, hierarchical/contextual models could not be applied, and a black-box adjustment approach was used; as such, residual confounding cannot be ruled out, which may limit causal interpretation.

Affirmative action measures have been used throughout the world to improve access to medical education for underrepresented groups and re-balance the medical workforce. This study provides a comprehensive analysis of the impact of affirmative action policies in Brazil on medical students' entry into residency programs and their specialty choices. By examining the differences between students admitted through quota systems in public medical schools and those receiving financial aid in private institutions, we reveal significant disparities in residency enrolment rates and specialty selections.

Our findings indicate that students from underrepresented backgrounds who are admitted through quotas face lower odds of entering medical residency compared to their peers, while those receiving financial support in private schools exhibit higher chances of residency enrolment. Furthermore, both groups show a notable preference for Family & Community Medicine, highlighting the potential of affirmative action policies to influence the distribution of primary care specialists.

This research underscores the importance of these policies in promoting diversity within the medical workforce and suggests that while quotas enhance student diversity, financial support may serve as an effective incentive for students in private institutions to pursue further medical training. Ultimately, the insights gained from this study may inform policy decisions aimed at addressing the inequities in medical education and workforce distribution in Brazil and similar contexts globally.