This study found the PMM was not associated with a reduction in IMR or NMR on average, however there were small associated reductions in IMR in municipalities with high IMR prior the programme. There appeared to be little effect of the PMM on other outcomes or across socio-economic strata of municipalities, including programme prioritisation criteria.
The general absence of an impact of the PMM on aggregate corroborates previous studies on the PMM [17, 18], yet is in contrast to evidence from the ESF in Brazil [10] that shows primary care doctor density is inversely associated with infant mortality. There are multiple reasons that could explain the lack of association. Firstly, the health impacts of the PMM may take longer to be realized. The PMM was implemented in 2013, and only five years of post-implementation data are used in this analysis. Furthermore, the additional number of doctors provided may have been too small to substantially impact infant health outcomes. Secondly, there is evidence that the PMM doctors were allocated to non-priority municipalities likely limiting their effectiveness and potential to reduce IMR [14, 15]. Many PMM doctors were allocated to areas with already low levels of IMR or where there were higher levels of non-PMM primary care doctors working. Diminishing returns from increasing doctor density may have also reduced the effectiveness of PMM doctors in decreasing IMR [10]. Thirdly, PMM doctors may have been limited in their ability to improve infant health given the large roles other health professionals (such as nurses and community health workers) play in the Brazilian health system and also the reliance on hospitals for birth.
This study found a lack of association between the PMM and infant health outcomes across strata when stratifying by municipal socio-economic characteristics including poverty and urbanisation. This suggests despite poor targeting of PMM doctors to priority municipalities [14, 15], the relationship between the PMM and IMR was not substantially affected by municipal socio-economic factors other than infant mortality. This is generally supported by other studies which report no association between PMM and infant mortality in municipality subgroups of mother’s education, mother’s age and marital status [18]. However, it does appear that baseline IMR is a determinant of the relationship between the PMM and IMR as there were IMR reductions associated with the PMM in municipalities with the highest IMR at baseline. This finding corroborates the results of research on the ESF [9] which found IMR reductions from ESF expansion were greatest in municipalities with the highest IMR. This finding suggests that PMM doctors may deliver health benefits where underlying health outcomes are poor and also programmes such as the PMM can contribute to improvements in health inequalities when targeting is based on health indicators rather than socio-economic factors.
The PMM was not associated with changes in prenatal care outcomes in most models. The PMM was associated with an increase in the proportion of expectant mothers receiving over seven prenatal care visits but only in municipalities with the lowest IMR at baseline and the highest density of CHWs and non-PMM doctors in 2012. This finding indicates wealthier municipalities with a higher density of HRH prior to the programme may have more effectively integrated PMM doctors to local health services, allowing nurses and CHWs to focus on prenatal care provision. Therefore, the PMM can introduce minor improvements to services and processes, but in areas that are considered least in need of new doctors.
This study has several limitations. First, the data collated from the Brazilian Ministry of Health and publicly available sources may have administrative errors—including possible underreporting of infant deaths in some municipalities. However, the data was collated from established sources which have been used for several evaluations of the PMM programme [13, 17, 18]. Additionally, statistical approaches employing time and municipal fixed effects would have likely accounted for some sources of bias. Second, the absence of adequate reporting systems for maternal and health outcomes in some municipalities prior to PMM may have skewed findings. Evidence demonstrates the expansion of health services in Brazil has reduced under-reporting and therefore expansion of health professionals could be associated with increases in mortality rates [22]. This may have masked some of the associations between the PMM and reductions in IMR. Third, although the data includes the entire five-year period of PMM implementation (2013–2018), the programme may have long-term impacts on infant health beyond 2018 which cannot be measured with this data. Fourth, the ecological design of the study restricts causal inference and prevents the exploration of heterogeneity within municipalities. Fifthly, there may have been systematic differences between PMM- recipient and non-recipient municipalities that could have biased the findings, although the use of IPTW-RA aimed to minimize these biases and represent the most robust method for observational studies.
The impact of the PMM in municipalities with high IMR at baseline indicates that HRH programmes can deliver improvements where the health needs are the greatest. There are wider policy implications from this work. More comprehensive targeting arrangements are necessary to maximize the health gains of scarce resources—both in Brazil and in other settings. This includes up-to-date data on health outcomes and socio-economic characteristics of local populations. Evidence indicates the effectiveness of the PMM was diluted due to inappropriate targeting of PMM doctors to the most needed areas [14]. The results from this study also suggest that the targeting criteria that were in place were of limited benefit as no effect of the PMM was found when stratifying by criteria. Policymakers may need to balance different and conflicting health gains from different sub-populations (e.g. infant health gains versus morbidity improvements in older populations) when distributing resources. Additionally, PMM doctors were often allocated to the most deprived regions within municipalities [14], implying that municipal-level health metrics and impact studies (including this one) are of only limited value for effective resource targeting. More detailed analyses on the sub-populations within municipalities most benefitting from the PMM is necessary to understand the complexity of the impact of PMM on a localised level.
In August 2019, President Jair Bolsonaro announced the creation of the Médicos pelo Brasil (Doctors for Brazil) Program, to replace the PMM. However as of March 2021, the new program has not yet been implemented and longer-term strategies to address the lack of primary care doctors in underserved areas are absent. Furthermore, the increased pressure on Brazil’s healthcare system due to COVID-19 led the federal government to publish new calls for PMM doctors in 2020, including in non-priority municipalities. Under the original PMM, there was a planned expansion of 15,000 extra medical schools places to train primary care doctors, who, in 2021 would be filling unserved positions in primary care. However, these extra places were not fully expanded and coupled with the expansion of PMM positions into non-priority municipalities, the challenges in the provision and distribution of primary care doctors in Brazil remain. Tackling the ongoing lack of doctors and their inappropriate distribution is key for further strengthening the health system, providing access to high-quality care, and tackling the large social inequalities that exist in the country.