Internal migration of physicians who graduated in Brazil between 1980 and 2014

Background The internal migration of physicians from one place to another in the same country can unbalance the supply and distribution of these professionals in national health systems. In addition to economic, social and demographic issues, there are individual and professional factors associated with a physician’s decision to migrate. In Brazil, there is an ongoing debate as to whether opening medicine programmes in the interior of the country can induce physicians to stay in these locations. This article examines the migration of physicians in Brazil based on the location of the medical schools from which they graduated. Methods A cross-sectional design based on secondary data of 275,801 physicians registered in the Regional Councils of Medicine (Conselhos Regionais de Medicina—CRMs) who graduated between 1980 and 2014. The evaluated outcome was migration, which was defined as moving away from the state where they completed the medicine programme to another state where they currently work or live. Results 57.3% of the physicians in the study migrated. The probability of migration ratio was greater in small grouped municipalities and lower in state capitals. 93.4% of the physicians who trained in schools located in cities with less than 100,000 inhabitants migrated. Fewer women (54.2%) migrated than men (60.0%). More than half of the physicians who graduated between 1980 and 2014 are in federative units different from the unit in which they graduated. Individual factors, such as age, gender, time of graduation and specialty, vary between the physicians who did or did not migrate. Conclusions The probability of migration ratio was greater in small municipalities of the Southeast region and strong in the states of Tocantins, Acre and Santa Catarina. New studies are recommended to deepen understanding of the factors related to the internal migration and non-migration of physicians to improve human resource for health policies. Electronic supplementary material The online version of this article (10.1186/s12960-018-0286-8) contains supplementary material, which is available to authorized users.


Background
The internal migration of physicians who leave a city or region of the country to establish themselves in another city or region of the same country can unbalance the distribution of these professionals in the national territories [1]. Relatively few studies on the various forms of internal migration of physicians and other health professionals are available in the international literature.
Studies on the mobility of physicians between countries identified a convergence of individual and professional motivations as well as policy, economic, social and demographic issues [2][3][4]. Similar to the determinants of the mobility of skilled workers in general [5], physicians most commonly migrate from places with worse living and working conditions and with few opportunities for professional development. The attractiveness of specialized education centres, career prospects [6,7], better pay [8,9] and age [10] plays a relevant role in the decision of physicians to migrate. For instance, young physicians are more inclined to change cities or countries.
In Brazil, recent policies have focused on the number, supply, demand, distribution and training of physicians. Since 2013 with the institution of the More Doctors law [11], new training programmes have opened with expanded graduate vacancies in medicine, expanded medical residency programs and the admission of foreign and Brazilian physicians to practice in underserved municipalities and regions.
The "interiorization of medical training" is the decentralization of medical education resources to cities and regions where there are no or few medical schools are considered a possible inducing factor for physicians to stay in these localities [12]. There are no studies in Brazil that support this relationship, although physicians tend to remain in the cities where they conclude their medical residency [13] and these programmes are more concentrated in the South and Southeast regions of Brazil [14].
The objective of the present study is to evaluate the profile and distribution of physicians trained in Brazil between 1980 and 2014. We considered the federative units and the cities where the professionals graduated and where they currently work or reside to contribute to the understanding of the internal migration of physicians in the country.

Methods
This was cross-sectional study using secondary data from the Medical Demographics in Brazil [15]. This survey is the census that includes administrative records from: Regional Medicine Councils (Conselhos Regionais de Medicina-CRMs); National Residency Commission (Comissão Nacional de Residência-CNRM) of the Brazilian Ministry of Education (Ministério da Educação-MEC); and the Brazilian Medical Association (Associação Médica Brasileira-AMB).
The CRMs are responsible for registration and inspection of physicians and their professional activity. All the active physicians in Brazil must be registered in a medical council; there are 27 medical councils, one in each Brazilian Federal Unit, which together integrated the Federal Medicine Council's (Conselho Federal de Medicina-CFM) database.
All physicians registered in the CRMs who completed a degree in medicine between 1980 and 2014 were included in the study. The analysed time interval, which encompassed a significant contingent of graduated and enrolled physicians, included public or private degree programmes in various locations.
The initial database was composed of 388,203 physicians, of which 20% (77,569) did not match the temporal inclusion criteria. Of the 310,634 physicians considered, those with absent or incomplete information concerning current address or graduating school place were excluded, which led to the exclusion of 12.2% of the physicians (34,828). The final database included 275,801 physicians that make up the universe of physicians who graduated in Brazil between 1980 and 2014 ( Table 1).
The physicians' places of graduation (medical school headquarter city), the dates of the first records of the physicians in the CRM (primary enrolment) and the current addresses provided by the physicians to the CRM, which corresponded to a residence and/or work address, were considered.
The evaluated primary outcome of interest was migration. The binary variable (did not migrate/migrated) was defined as a move from one state (state of the school where the physician completed the medical degree) to another state (current living or working location). The qualitative nominal and ordinal variables were evaluated as factors and referred to the characteristics of the physicians (age, gender, date of graduation, and specialty) and the medical programmes (geographical location and public or private institution). The variable "type of municipality of the medical school" was stratified into three levels based on the 2010 census conducted by the Brazilian Institute of Geography and Statistics [15,16] according to the population strata as follows: (1) capitals, (2) large interior municipalities (LM interior) with 100,000 or more inhabitants and (3) small interior municipalities (SM interior) with less than 100,000 inhabitants.
The migration according to the type of municipality of the medical school was analysed using a probability migration ratio (PMR) like indicator obtained from the ratio between the probability of the physician migrating following graduation from a school located in the capitals (reference category) divided by the probability of the physician migrating following graduation from a school located in the interior (LM interior and SM interior). The

Results
Among the 275,801 total physicians considered, 157,998 (57.3%) migrated (i.e., were living and/or working in a different state than the state where the medical school from which they graduated was located at the time of the study) ( Table 1 and Fig. 1).
From the total physicians analysed, 52.9% were male, 22.8% were below 35 years of age, 22.5% had graduated at least 5 years ago and 60% had completed their specialization or medical residence in one of the 53 recognized medical specialties. Most of the physicians Fig. 1 Flowchart of the study population graduated from public schools (57.9%) located in the Southeast region (55.5%) and the state capitals (56.1%).
From 1980 to 2014, more than half of the Brazilian physicians (52.7%) graduated from schools located in the following three states: São Paulo (55,874 physicians or 20.2% of the total), Rio de Janeiro (50,480 or 18.3%) and Minas Gerais (39,065 or 14.1%). The states with the fewest graduated physicians were Acre (241 or 0.09%), Roraima (318 or 0.12%) and Rondônia (908 or 0.33%). The state of Amapá did not have any graduated physicians registered because it did not have a medical programme during the analysed period (Table 1).
A total of 60.7% of the physicians older than 35 to 40 years of age migrated. In the age range below or equal to 30 years, 52.0% migrated ( Table 2). The women migrated less than the men (54.2 and 60.0%, respectively). A greater proportion of physicians who graduated 10 to 15 years before the study migrated (60.1% migrated) compared to the other groups stratified by the time since graduation. Additionally, more physicians without a specialization migrated (59.7%) compared with the specialists (55.7%) and physicians from private medical schools (67.8%) compared with the public medical schools (49.7).
Migration was also analysed according to the location of the graduate schools based on the municipality size and country region. In all regions, PMR was greater in small municipalities and lower in the capitals. Compared to the capitals, PMR of physicians who graduated from schools located in the group of small municipalities was 9.7-fold higher (Table 3).
A total of 93.4% of the physicians who graduated from schools located in municipalities with less than 100,000 inhabitants migrated ( Table 2). The Southeast region had the largest absolute and relative contingent of physicians who graduated from small municipalities, and 93.8% of these physicians migrated (Table 3). Regarding the nature of the schools, physicians who graduated from private schools migrated more often (67.8%) than physicians who graduated from public schools (49.7%), although in the physicians analysed the physicians who graduated from private schools represented 42.1% of the total.
Among the Brazilian states, the largest migration ratios were observed in Tocantins (85.3%), Santa Catarina (70.5%), Acre (60.6%) and Rio Grande do Sul (61.9%), whereas Alagoas (25.3%), Ceará (31.5%), Amazonas (33. 5%) and Piauí (34.8%) had the smallest migration percentages (Fig. 2). Table 4 shows the physician population rate (PPR) and the percent of graduated, migrated and immigrated physicians according Brazilian states and populational stratum. In this table, it is possible to evaluate some phenomena: (I) in almost all situations, small municipalities (SM interior) lost more than 85% of medical graduates; (II) from the 16 municipalities (LG interior) that had the potential to train physicians, only 2 of them lost less than 65% of medical graduates; (III) from the 26 capital cities, only 6 lost more than 65% of medical graduates; and (IV) the capital of São Paulo state receives twice as many physicians as it loses.
There is a high migratory flow of physician from capitals, LG interior and LM interior for the capitals of the Brazilian states, with special focus on the Southeast region or to the capital of the same region of the medical school graduation: 26.1% of the physician from capitals of the North region migrated to capital cities of the Southeast region; 36.1% of the physician from LM interior of the Southeast region migrated to some capitals of the same region. Other similar examples can be seen in Additional file 1: Table S1. In the present study, 176 schools that trained physicians in Brazil between the years considered (1980 and 2014) were headquartered in 103 different municipalities. Of these, 25 were located in the state capitals and the Federal District, 63 were in large interior municipalities and 14 were in small interior municipalities.

Discussion
Notably, among the 14,382 physicians who graduated from medical schools located in small interior municipalities (those with less than 100,000 inhabitants), the large majority (93.4%) did not stay at these locations. Most physicians who graduated from private medical schools also migrated (67.8%) regardless of the size of the municipality where the school was located. Given the trend towards the interiorization and privatization of medical education in Brazil [19], the results of this study allow us to draw preliminary conclusions about the relationship between the location and nature of the school and the non-migration of physicians.
If adopted as an isolated measure, the opening of new, mostly private programmes in municipalities in the interior of the country in regions with a scarcity or a lower density of physicians per inhabitant may not directly affect the permanence of the graduated physicians in these locations. In this sense, the analysis has limitations because the study included a low number of physicians who graduated in small municipalities in the Northeast, North, Central-West and South regions because the internalization of medical schools is a recent phenomenon in Brazil.
The less-intense migration of women compared to men corroborates the literature, which indicates that the female gender is less prone to territorial relocation [20]. Thus, the feminization of medicine in Brazil [21] takes on increased relevance in the different aspects of the observed migration process, with a possible future positive effect on the non-migration of physicians.
The state capitals retained more physicians who graduated there. Although they had a smaller proportion than the capitals, the large cities of the country seemed to be more attractive than small cities to the physicians who graduated in these locations, which experienced a mass migration of physicians who graduated from their schools.
In Brazil, it was observed that the 39 cities with more than 500,000 inhabitants concentrated 30% of the population and 60% of the physicians, which directly affected the low physician/inhabitant ratio in the small municipalities. For example, in locations with 5000 to 10,000 inhabitants, the ratio is 0.28 physicians per 1000 residents. In municipalities with between 10,000 and 20,000 LM interior-interior large municipalities ≥100 000 inhabitants; SM interior-interior small municipalities < 100 000 inhabitants PMR probability migration ratio residents, the ratio is 0.36, whereas in municipalities with between 20,000 and 50,000 residents, the ratio is 0. 64 physicians per 1000 inhabitants [15]. The results presented in Table 4 and in Additional file 1: Table S1 suggest that there is a high migratory flow between the capitals of the Brazilian states, especially with high influence of the state capitals from Southeast region; this region concentrates the largest number of medical residency programs in Brazil [14]. The few municipalities with less than 100,000 inhabitants, with the potential to train physicians, presented little retention capacity, losing their graduates to other regions. This situation has contributed to maintaining the low average per capita medical rate in these localities.
In this study, we also observed that there is a higher migration rate for physicians coming from private schools. This result may raise a warning about the new interiorization policies of medical schools in Brazil, a process that has resulted in an increase in the opening of private schools in the interiors of the states; one of the arguments used by the managers of this policy was that there could be retention of doctors in the host cities of medical schools and their surroundings.
In a systematic review realized by Willis-Shattuck et al. [22] to examine motivating factors that would reduce medical migration both within and across countries, the following factors were listed: financial incentives, career development, continuing education, hospital infrastructure, resource availability, hospital management and personal recognition or appreciation. A more recent systematic review identified similar variables with a high degree of similarity between selected studies; however, workload and autonomy seem to be specific to physicians' choice of workplace between the public and private sectors [23].
A study that analysed geographical mobility of general practitioners in rural Australia found that in small towns of < 5000 residents had the highest risk of leaving rural practice. Mobility rates were significantly higher for GPs who had worked in a location for under 3 years and those working as either contracted or salaried employees, and somewhat higher for international medical graduates [24].
Mobility of US Rural Primary Care Physicians During 2000-2014 was analysed. The researchers noted that county-level physician mobility was higher for counties that lacked a hospital (absolute increase = 5.7%), had a smaller population size, and had lower primary care physician supply, but area level economic and demographic factors had little impact. Their conclusion was that outcomes were notably poorer in the most remote locations and those already having poorer physician supply and professional support [25].
This study may be considered a starting point for the required in-depth analysis of migratory patterns of physicians in Brazil. By using the database of administrative and professional records, it was possible to characterize the variables associated with the medical school and with the physicians who did or did not migrate.
Due to the sole focus on the comparison of the current address and graduation location, the present study did not  analyse the migratory flow of physicians as a whole and did not consider the effect of the birthplace on relocation.
Because this study was a cross-sectional study, it was not possible to capture possible intermediate moves and destinations over time or define whether the current location of the physician was definitive. In a classic study from 1978 on the flows and stocks of nurses and physicians, Mejía [26,27] highlighted the difficulties of defining whether a professional's migration was "permanent" or "temporary" based on secondary data. Another limitation of the analysis is the concept of migration adopted in this study. The movement of a physician In this study, it was not possible to isolate the effect of the location of the medical school in relation to other cultural, employment, and socioeconomic factors that can influence migration.
Future studies on the subject may consider the characteristics of the place of origin (prior to obtaining the medical degree) and the current location of the professionals who migrated. For this purpose, more refined methods will be required, such as methods that use migration networks to establish possible migratory patterns.

Conclusion
More than half of the physicians who graduated between 1980 and 2014 are living and/or working in federative units different from the unit where the medical school from which they graduated was located.
The considerable and dynamic internal migration of physicians in Brazil (defined in this study as the relocation or mobility of professionals from one state to another) seems to contribute to imbalances in the distribution, supply and availability of physicians in the country, whose motives are not yet fully clear.
When compared with physicians who remained in the state in which they graduated, there was a difference in the profiles of the physicians who migrated regarding individual factors, such as age, gender, time since graduation and specialty, which could partially explain the decision to migrate.
The probability of migration ratio was greater in small municipalities of the Southeast region and in the states of Tocantins, Acre and Santa Catarina, which also experienced a more pronounced migration of physicians.
To guide adequate human resource policies in health, new studies are required that consider the socioeconomic level of the location, the establishment of services and employment and continued training opportunities among other possible factors for the non-migration of physicians.
The "push" and "pull" mechanisms that motivate the internal migration of physicians are complex and multifactorial and can have consequences that require more research. For instance, qualitative studies should be performed to examine the migration networks and flows that are able to capture the controlled turnover (i.e., the supply of jobs to physicians from the outside) to meet the needs of health services and to offset retirement and other forms of replacement and exchange of professionals. More studies are also needed on voluntary turnover, which is when the physician leaves a certain place in search of better living conditions, wages or career prospects without a previous offer.
Similarly, the motivations for temporary migration should be explored by human resources policies, which can produce benefits by aggregating value to the professional and the local health system upon the physician's return. This situation is in contrast to permanent migration, which can represent the transfer of human capital from one place to another and thus weaken the capacity of local health systems [2,[27][28][29][30].
Finally, research efforts should focus on monitoring the future movement of physicians in Brazil and other countries to identify factors related to internal migration and non-migration of these professionals and contribute to the conception, planning and assessment of public policies of human resources in health.