Saúde Income and vocational integration of Brazilian physicians since the establishment of the National Health System

RESULTADOS: Quanto ao perfi l demográfi co observou-se uma tendência de ampliação da presença de mulheres e de profi ssionais com mais de 55 anos, além da preservação da alta proporção de brancos. Com relação à ocupação e à renda, observou-se um aumento do empresariamento médico e a manutenção de elevados rendimentos, em termos relativos, especialmente para aqueles que mesclavam ocupações de empregado e empregador.


INTRODUCTION
Among the technological, managerial, institutional and policy changes that have occurred within the Brazilian healthcare system since the implementation of the Sistema Único de Saúde (SUS -National Health System), the employment market for healthcare professionals -or, in broader terms, the healthcare workforce -has been analyzed from various databases and methodological perspectives.This topic has been prominent within the agenda for healthcare systems in Latin America since the 1950s.At that time, among other important sectoral transformations, the creation of the Chilean National Health Service in 1952, inspired by the British universal model that was instituted in 1948, imposed theoretical, conceptual and institutional challenges on the training of human resources, especially in relation to training for physicians that would be considered appropriate for the expansion of the public provision of healthcare services. 6 Brazil, according to Nunes, 12 the pioneering work of Donnangelo on the relationships between the pattern of state intervention and the employment market for physicians, based on a survey among physicians in the city of São Paulo in 1971, defi ned three different types of participation in the employment market, either separately or as combinations of these types: self-employed, salaried workers and owners of private companies.The predominance of combined forms (56.8% of the sample) and the observations of an association between higher income levels and control over the means of production signaled changes in the stratifi cation patterns among physicians, in terms of income, prestige and professional qualifi cation.The concept of social organization of medical practice formulated by Donnangelo 4 was also the inspiration for studies looking back on the historical nature of changes in medicine and medical employment.Nogueira, 10,11 Schraiber, 15 Campos 1 and Paim, 13 among others, detailed the singularities of the process of technical division of medical employment and its social, economic, legal, political and ideological repercussions, thereby systematizing the process of linking employment with the organization and institutionalization of medical practice.
Beginning in the mid-1980s, information on the medical employment market started to be gathered from official sources, coming from systematic surveys and registers with national coverage, such as the Pesquisa Assistência Médico-Sanitária (AMS -Medical-Health Care Survey) of the Instituto Brasileiro de Geografi a e Estatística (IBGE -Brazilian Institute for Geography and Statistics), the Relação Anual de Informações Sociais (RAIS -Annual Report on Social Information) of the Ministry of Labor and Employment and the Cadastro Nacional de Estabelecimentos de Saúde (CNES -National Register of Health Establishments).The observation units for these sources -health establishments or establishments in general -made it possible to quantify and categorize physicians' employment positions.Specifi c surveys, also of national scope, have been conducted by teaching and research institutions and by professional bodies, such as those by the Fundação Oswaldo Cruz (Oswaldo Cruz Foundation) and the Conselho Federal de Medicina, (CFM -Federal Medical Council) in 1995 a and in 2002-2003. b There have also been localized surveys, such as the one by that have brought out individualized information on the integration of healthcare professionals into the employment market.
The recent reappearance of the problem of human resources on the agenda of sectoral priorities within the context of implementation of SUS and reorganization of private healthcare services has focused on transformations in the ways in which healthcare professionals are hired.Furthermore, the conjugated effects from adopting the uniform principles of the unifi ed legal framework for hiring public employees, which has been implemented within a scenario of budget restrictions, fi scal adjustments and increased fl exibility of the rules for bringing workers into private healthcare establishments, has resulted in growth of outsourcing and of various forms of unregulated occupation within the sector. 7us, accompanying the changes in the Brazilian healthcare system, the focus of studies has settled on the magnitude and detailing of the medical employment market; changes in hiring methods; increased precariousness; and conflicts generated through tensions between spheres of power and professional bodies.More recently, the debate among government institutions, professional representative bodies and private companies regarding the forecasts for the need for specialists and the polemic regarding the quantity and quality of institutions for training physicians in Brazil has had input from the deepening of the analyses on the provision, demand for and ways of integrating physicians in the employment market.
The rich and diversifi ed national production of knowledge on human resources and healthcare, the creation of the Department of Employment and Education Management within the scope of the Ministry of Health in 2003 and the establishment of a network for monitoring human resources have reaffi rmed the importance of this topic and stimulated the emergence of local research groups.In recognition of the importance and specifi c features of investigations on employment and healthcare professionals, the databases on medical employment have almost always been analyzed by researchers and technical experts who are qualifi ed to make inferences regarding the complex social processes underlying the statistics, starting from partial empirical evidence.
The two sources of information most consulted regarding the employment market for physicians are the  1995 and 2002-2003 and the registration data held by the medical councils, along with information coming from isolated investigations, complement the analyses on the predominant traits of the structure and dynamics of physicians' employment.These sources, especially the offi cial ones, record distinct dimensions of the employment market.Whereas AMS counts the employment positions that exist, RAIS gives the numbers of workers (physicians) with formal links.
Therefore, given the differences in the nature and coverage of the records on the numbers of physicians, the frequencies recorded by these three sources (IBGE, RAIS and CFM) are not automatically comparable.The number of employment positions recorded by the AMS/2005 (476,604), b which excludes employment in private offi ces, was signifi cantly greater than the approximately 360,000 physicians recorded by CFM c and than those 297,724 formally employed recorded by RAIS/2005.d These discrepancies coming from the different types of observation of the three sources make it recommendable to acquire prior recognition of their advantages and disadvantages regarding coverage and adequacy for analyses on the medical employment market.
The National Household Sampling Surveys (Pesquisas Nacionais por Amostra de Domicílios, PNADs), which also include information on occupation, types of linkage to the employment market, income and other demographic and social data, broken down at the individual level, are less often accessed for analyses on the healthcare workforce than the AMS and RAIS data are.Researchers probably make little use of PNAD data on the integration of healthcare professionals, for methodological reasons.The sampling design of the PNAD is not aimed at obtaining representativeness for specifi c professional categories.Secondarily, there are operational, instrumental and theoretical problems relating to the processing and analysis of large and complex databases.From the PNADs of 1988, 1993, 1998 and 2003, it can be estimated that there were, respectively, 172,412, 178,755, 228,882 and 263,604 physicians in Brazil.These numbers indicate lower frequencies of physicians than shown by the databases cited above.
The aim of the present study was to examine employment trends and the income makeup of Brazilian physicians based on the PNADs, and to determine the magnitude of the impact of decentralization of the public healthcare system on the links between physicians and the public sector.Thus, considering the continuous nature and reorganization of private healthcare entrepreneurialism following the implementation of SUS, the focus of the analysis is turned towards examining the integration and remuneration of Brazilian physicians, as a proxy for the limits and perspectives for putting the proclaimed universal right to healthcare into effect.The initial choice of years to investigate was mainly due to the existence of special information on healthcare in the PNADs of 1998 and 2003 (health supplement), thus determining a time interval of fi ve years.It was deemed methodologically convenient to keep to this time spacing, and thus the PNADs of 1993 and 1988 were also included.The 1988 survey coincided with the implementation of SUS and therefore represents a "t0" that was adequate for the purposes of the study.

Microdata from the
The database comprised records of individuals whose principal occupation (when there was more than one) or sole occupation was "physician".Through selection of only the code for "physician" as the primary occupation (code "2231" for the PNAD of 2003 and "151" for the others), the four surveys generated a sample consisting of 1,886 physicians: 382 in 1988, 400 in 1993, 531 in 1998 and 573 in 2003.Another option used for selecting physicians was through interviewees' declarations that their second occupation was "physician", in the PNADs of 1993, 1998 and 2003 only.The 1988 survey did not detail any second occupation.In this case, by selecting the records with the code of "physician" for the fi rst or second occupation, a database of 1,586 physicians was obtained: 426 in 1993, 559 in 1998 and 601 in 2003.
Thus, comparing this with the option of selecting only the fi rst occupation, 26 physicians were incorporated for 1993, 28 for 1998 and 28 for 2003.In accordance with the investigations of interest, both sample sets were analyzed: one with the 1988 PNAD (1,886 physicians) and the other without it (1,586 physicians).
The following social, demographic and occupational variables were selected: sex; color or race, dichotomized as white and non-white (mixed, black, Asian and indigenous); location among the fi ve Brazilian regions; whether more than one occupation was held; the position held as the fi rst or second occupation, using the categories of "employee", "employer" and "self-employed" as the categories for medical work and, residually, "other non-remunerated worker"; remuneration from the fi rst or sole occupation; and total income, including all the interviewee's sources.The interviewees' ages were categorized as up to 29 years, 30 to 54 years and 55 years and over.For physicians in the "employee" category, the employment sector was analyzed (whether private, federal, state or municipal).Finally, the physicians were classifi ed in terms of the origin of their overall income, as "exclusively public", "exclusively private" and "mixed" (both public and private).The chi-square test was used to analyze the categorical variables of interest.The analyses were carried out considering the "relative weight" of the physicians in the samples from each year.After correction to the year 2003, the interviewees' remuneration from their fi rst or sole occupation and their income from all sources were basically explored by calculating mean and median values for the categories of interest.

RESULTS
Table 1 presents a trend of expanding presence of women among the physicians and a trend of increasing numbers of professionals aged 55 years and over.On the other hand, it indicates preservation of high proportions of whites and, in terms of distribution between the regions, continuation of strong concentration in the southeastern region.The participation of women among the physicians grew signifi cantly: 30.9% in 1988, 35.8% in 1993, 46.7% in 1998 and 47.6% in 2003.The presence of physicians of white skin color remained stable over the four survey years studied (86.4%, 84.8%, 88.5% and 88.0%) and much higher than the general Brazilian population of white color/ race (around 53% according to the 2003 PNAD).The proportion of physicians with more than one occupation decreased during the 1990s and did not change at the beginning of the 2000s (61.0% in 1988, 52.0% in 1993, 47.8% in 1999 and 46.3% in 2003).The proportion of physicians who were employers also increased signifi cantly, from 6.3% in 1988 to 15.4% in 2003.
Regarding the classifi cation of the physicians according to the sources of their income, there was no statistically signifi cant variation in its distribution.Around 30% of the physicians declared that their income came exclusively from public sources.The proportion whose income was exclusively from private sources was between 40% and 47%.Considering only the physicians in the "employee" category, most of them remained linked to the public sector (more than 65% in the three years evaluated).While the private sphere remained stable, there was a movement within the public sector from federal and state links to municipal links.The municipal sphere employed 16.2% of the physicians in 1993 and this proportion rose to 34.5% in 2003 (Table 2).
Considering the number of inhabitants per physician, a slight overall improvement for Brazil was observed.The data for the northern, southern and central-western regions indicated a declining trend, especially from 1993 onwards, while the northeastern region showed stability.However, it cannot be said that there was convergence of the indicators of number of inhabitants per physician between the Brazilian regions.The northern and northeastern regions always had much higher indicators than those recorded for the southeastern region (around three times higher, or more) (Figure 1).The data on the physicians' median income categorized according to different types of integration in the employment market suggest that the professionals with more than one occupation, and especially those who mixed the conditions of employee and employer or self-employed, had much higher incomes (more than R$ 4,600.00)than did those who declared that they had a single occupation (maximum of R$ 2,700.00).For the different combinations of integration, no trend of declining or growing income over time was observed.Except for the category "employed (one occupation)", median income considering all income was shown to be considerably higher than the income relating to the fi rst occupation (around twice as high), although less markedly so in the category of "self-employed or employer" (Figure 2).
Comparing the median income of Brazilian physicians with that of two segments of the population -the occupied Brazilian population with 12 or more years of schooling (i.e. with university-level education) and the entire economically active Brazilian population with an income -it was observed that even if only the main or sole occupation was considered, the physicians had much higher income levels, especially those with more than one employment linkage.Between 1988 and 1993, the income of physicians and the population with at least 12 years of schooling declined.However, from 1993 onwards, while the median income of the population with greater schooling continued to decline, physicians' incomes did not (Figure 3).
In Table 3, separation of the income sources according to the physicians' employment links (public, private or mixed) shows that the income of those with exclusively private links was considerably higher than the income of those with exclusively public links in 1993 and 1998, but the values became close in 2003.It can also be seen that the private portion of the mixed links was higher than the public portion.The sum of income from public sources (public portion of the mixed group and the exclusively public group) was lower than the sum of income from private sources (private portion of the mixed group and the exclusively private group).This pattern was observed independent of the summary measurement used (mean or median).Mean values systematically higher than median values indicated asymmetrical distribution to the right for the physicians' income.

DISCUSSION
The analysis on the extracted PNAD data regarding physicians' occupations and income raises questions about the process of juxtaposing changes or accentuating historical trends in this employment market.Some of these questions about certain dimensions of the employment market in the healthcare sector have been answered based on the processing of the PNAD information.In the study by Dedecca et al, 3 the PNAD data from 1992 and 1997 provided support for measuring employment through stratifi cation of the healthcare sector based on establishments.The basis was the size and degree of closeness to fi nal healthcare activities.
Póvoa & Viegas 14 used a location-based decision model that was fed by data from the 1997 and 2001 PNADs, to detect variables associated with Brazilian physicians' geographical distribution.Neither of these studies went deeply into examining the relationships between physicians' occupations and income.
Thus, the scarcity of studies based on PNAD with the aim of obtaining support for inferences regarding the medical employment market, and the limitations inherent to the design of such investigations, means that no defi nitive conclusions or structured hypotheses can be put forward.Nonetheless, they provide a basis for making conjectures.The need for caution is reinforced by the limitation to only the PNADs of 1988, 1993, 1998 and 2003.
Thus, examination of the PNAD data suggests that there has been a set of strategies for protecting physicians' a Physicians who declared these sources for their fi rst or second occupation, excluding those without income information b Refers only to physicians in the "employee" category in the fi rst occupation incomes.Such strategies seem to have been based on bringing together the preservation of structural determinants, such as the immunity of the access barriers to medical schools faced by the majority of young Brazilians, as seen through the low access and continuing exclusion of the black population, with traits relating to the recent dynamics of the integration of physicians in the employment market.The maintenance of the primacy of white physicians and the exacerbation of the income differentials between physicians and other segments of the population, and the signifi cant proportion of physicians with several linkages, seem to confi rm the high symbolic or material distinction of the value of physicians' employment.Brazilian physicians' mean income was 8.7 times greater than the per capita gross domestic product (GDP) for the same year.
Around the world, there is no rule regarding income disparities between physicians and other segments of the population.The various studies on physicians' integration in the employment market have not focused on the relationships between multiple occupations and income.The PNAD data suggest that, for Brazilian physicians to assure their relatively high income levels, within the context of underfunding of SUS, they have emphasized their entrepreneurial leanings.It is plausible to suppose that physicians' entrepreneurialism results from the need to regularize their links with private establishments and with health plan and insurance companies.a In turn, this expedient took away pension rights and required physicians to remain active in the employment market for longer times.
Even though the PNAD data present limitations, their individualized characteristics of occupation and income, and especially the detailing of the relationships between income and physicians' multiple links that are made available, contribute towards deepening the understanding of Brazilian physicians' complex public-private integration.The estimates of the numbers of physicians and other characteristics such as the trends towards feminization of the profession and multiple employment links are compatible with those coming from other data sources, especially the CFM's records and surveys.
The movement of physicians' employment within the public sector from the federal to subnational sphere, because of the decentralization of SUS, was also detected in the IBGE's AMS.The dimensions of other attributes such as physicians' color and prolonged service in the employment market, extracted from the PNADs, seem to be in line with reality, although not included as variables of interest in other information sources.
Although none of the PNADs was methodologically designed for the purposes of the present study, it can be believed that the individuals selected constituted a representative sample of Brazilian physicians, in each of the years.However, for this, the sample weights (expansion factors) available in the database, resulting from the sampling design implemented by the IBGE, were considered.In the four PNADs, the sample weights ranged from 127 to 975, thus indicating that each physician sampled may be representing between 127 and 975 other Brazilian physicians.On the other hand, the intention of the design was concentrated more on statistical comparisons between the years than on specifi c expansion of the results to the whole population of physicians.In this respect, instead of sample weights, the main analyses were developed considering the relative sample weights for each year, by dividing the sample weights for each analysis unit (physicians) by the mean sample weight for that year.This strategy ensured representativeness for the sample size, 8 thus making it possible to calculate the statistical signifi cances of the comparisons and analyses of interest.Hence, this procedure corrected potential bias of selection in the separate estimates and allowed the use of inferential statistical procedures, since it maintained the original sample sizes.However, the inferences may still have been biased because of the sampling complexity of the PNADs, given that complex samples tend to bias the estimated standard errors, 16,17 although there is no evidence that this might have been the case in the present study.
The data analyzed indicate the effects of decentralization on the physicians' employment market, with the trend towards decreased remuneration from salaried work and involvement in private entrepreneurialism.
a SN, Carvalho CL, Girardi JB, Araújo JF.Confi guração do mercado de trabalho dos assalariados em saúde no Brasil.In: Barros AFR, Santana JP, Santos Neto M, organizadores.Observatório de recursos humanos em saúde no Brasil: estudos e análise.Brasília: Ministério da Saúde, 2004.v.2. a The public sources refer to income from public employment (federal, state or municipal); the private sources refer to income from the occupations of private employee, employer and self-employed.
The physicians' remuneration was corrected monetarily, and this also took into account that in 1988 and 1993, there were different currencies in circulation in Brazil (cruzado and cruzeiro real, respectively).Thus, the remuneration values in the PNADs of 1988, 1993 and 1998 were converted to values for the year 2003.This conversion could be done by multiplying the interviewees' remuneration in the years cited, by a correction factor derived from infl ation/defl ation and from currency conversion.The IBGE's Índice de Preço ao Consumidor Ampliado (IPCA -Expanded Consumer Price Index) was used to simultaneously perform the calculations of infl ation and currency conversion.The physicians' income in 1988, 1993 and 1998 was corrected to 2003 values by means of the following multiplications, taking September as the reference month: a 1988 2003: income in 1988 x 0.013561424; 1993 2003: income in 1993 x 0.027014545; 1998 2003: income in 1998 x 1.5116135.
These frequencies refer only to physicians who declared these positions as their sole or main occupation PNAD: Pesquisa Nacional por Amostra de Domicílios (National Household Sampling Survey) a Mixed, black, Asian or indigenous b Excludes 15 physicians with an occupation described as "other non-remunerated worker" (1988 = 1; 1993 = 3; 1998 = 2; 2003 = 9)

Table 2 .
Distribution of Brazilian physicians' links regarding income sources and employment sector.Brazil, 1988-2003