SLIDE 1
Title Page Brazils Family Health Strategy: factors associated with - - PDF document
Title Page Brazils Family Health Strategy: factors associated with - - PDF document
Title Page Brazils Family Health Strategy: factors associated with program uptake and coverage expansion over 15 years (1998-2012) Monica Viegas Andrade, PhD 1,3 ; Augusto Quaresma Coelho, MD candidate 2 ; Mauro Xavier Neto, MD candidate 2 ;
SLIDE 2
SLIDE 3
3 RESEARCH IN CONTEXT Evidence before this study We searched on PubMed for relevant articles using the “Family Health Strategy” as search term present in Title or Abstract up to May, 2016. We found 429 articles analyzing Family Health Strategy experience in Brazil; only nine were related to FHS coverage implementation and expansion. Eight of those nine studies focused on specific urban municipalities, and the remaining study was a national comparison of different data sources for FHS coverage. Combined, these studies indicated that potential barriers to the uptake and coverage expansion of the Family Health Strategy (FHS), launched in Brazil in 1994, included shortage of healthcare professionals, municipal budget constraints, lack of proper infrastructure, and availability of other healthcare providers from the private sector. No study conducted a comprehensive analysis of potential factors contributing to the uptake and expansion of the FHS over time and across all Brazilian municipalities, including an assessment of if/how these factors may change over time. Added value of this study From 1998 to 2012 population coverage of FHS increased from 4·4% to 54%, but the expansion was highly heterogeneous, suggesting two groups of municipalities (i) early adopters, mostly smaller municipalities in less developed areas that faster attained universal coverage; and (ii) laggards, mostly larger municipalities that presented a lower expansion of FHS coverage. The uptake and expansion of the FHS from years 1998 to 2012 were positively associated with small population size of a municipality, low population density, low coverage of private health insurance, low level of economic development, alignment of the political party of the Mayor and the state Governor, and the availability of healthcare staff. Regionally, the Northeast presented the faster coverage expansion independently of population size. Implications for all the available evidence Establishing a primary health care program with high coverage is the first step towards achieving universal health coverage. The Brazilian experience shows that scaling up the FHS program is feasible in a context of large socioeconomic heterogeneity, but achieving universal coverage requires designing policies according to population size and economic development. For small and poor municipalities funding mechanisms are likely to guarantee UHC. For larger and richer ones, competing sources of health care are often an obstacle, and thus policies should include mechanisms that encompass both public and private sectors.
SLIDE 4
4 INTRODUCTION Universal Health Coverage (UHC) is one of the Sustainable Development Goals (SDG).1 Achieving UHC requires strong health systems that promote and deliver equitable and integrated services thorough primary health care (PHC).2 Expansion of PHC is a critical first step towards UHC,1 and it has been associated with better population health outcomes (lower infant and maternal mortality, reduction of mortality from heart and cerebrovascular diseases, reduced hospitalizations, and lower premature deaths from asthma, heart and cerebrovascular diseases, and pneumonia).3-5 Countries that have successfully expanded PHC have achieved UHC with improved health system
- utcomes.6,7
The Brazilian National Unified Health System (Sistema Unificado de Saúde – SUS), launched in 1988, was designed as a public policy to overcome health inequities.8 PHC is delivered by SUS through the Family Health Strategy (FHS), created in 1994 (described in the panel). The FHS has been associated with declines in infant mortality;9 declines in avoidable hospitalizations;10,11 better health care access and utilization;12 and reductions in social inequalities in healthcare access.13,14 Despite these results, two decades after the FHS inception almost 50% of the Brazilian population was not
- covered. Proposed reasons for this gap include shortage of professionals, municipal
budget constraints, lack of proper infrastructure, and availability of private insurance.15-
18 However, no comprehensive municipal-level analysis of factors associated with the
uptake and the expansion of coverage of the FHS over time has been done. To addresses this issue we assembled a 15-year time-series of municipal data from varied sources, and used a multilevel model for change to identify factors associated with the FHS uptake and expansion across 5,419 Brazilian municipalities from 1998 to
- 2012. We considered eight domains that could potentially affect implementation and
coverage expansion: economic development, health care supply, health care needs/access, availability of private insurance, political context, geographical isolation, regional characteristics, and population size. METHODS Data We used several sources to create a longitudinal dataset by municipality covering the years 1998 to 2012. In 1998, Brazil had 5,507 municipalities, and 58 new ones were created until 2012. We used the 1998 political division as reference, and aggregated data for the new municipalities back into the administrative unit they originated from. We excluded 88 municipalities with inconsistent mortality records. Thus, final analysis considered 5,419 municipalities (99·8% of total population). Data on the proportion of the population covered by the FHS were obtained from the Brazilian Ministry of Health. As data are available monthly, we used July (mid-year period) as the reference. Although the FHS was launched in 1994, data are available
- nly from 1998 onwards. This is not a major limitation, as municipalities needed time to
hire health professionals. By 1998, the FHS covered 4·4% of the population.
SLIDE 5
5 As for factors that could be associated with the uptake and expansion of the FHS we considered eight domains: economic development, health care supply, health care needs/access, availability of private insurance, political context, geographical isolation, regional characteristics, and population size. Table 1 summarizes the variables included in each domain, their source, and expected relationship with the FHS uptake and expansion. A map of municipalities was obtained from IBGE, and the SIRGAS 2000-Mercator geographical projection was used to facilitate calculation of distances without
- distortions. All mapping was done in ArcGIS 10.2 (ESRI; Redlands, CA).
Statistical Analysis Coverage density curves were represented by Epanechnikov kernel density estimators utilizing the function kdensity in STATA v.12 (Stata Corp., College Station, TX, USA). To investigate factors associated with the FHS implementation and expansion, we used a longitudinal multilevel model for change19 considering the municipality as the unit of
- analysis. The first level addressed within-municipality changes in coverage, and the
second level appraised between-municipality differences in coverage change. The composite model is: 𝑧𝑗𝑢 = [𝛿00 + 𝛿10𝑢𝑗𝑛𝑓𝑗𝑢 + 𝜷𝒍𝒀𝑗𝑢 + 𝜸𝒍(𝒂𝑗𝑢𝑢𝑗𝑛𝑓𝑗𝑢)] + [𝜂0𝑗 + 𝜂1𝑗𝑢𝑗𝑛𝑓𝑗𝑢 + 𝜁𝑗𝑢] where: 𝑧𝑗𝑢 is the proportion of population covered by the FHS in the municipality i at time t; 𝛿00 is the average initial coverage status; 𝛿10 is the conditional rate of change of coverage; time represents the years of analysis, and is centered on 1998 so that model estimates represent initial status; 𝜷𝒍 is a vector of coefficients for each of the k covariates affecting the initial status; Xit is the vector of k covariates affecting the initial status; 𝜸𝒍 is a vector of coefficients for each of the k covariates affecting the rate of change; Zit is the vector of k covariates affecting the rate of change; 𝜁𝑗𝑢 represents the residuals of the first level (fraction of coverage in municipality i that is unpredicted on
- ccasion t); lastly, 𝜂0𝑗 and 𝜂1𝑗 are residuals of the second level (portion of initial status
and rate of change of coverage, respectively, not explained by the model). Three different sets of models were run: (i) data from years 1998 to 2012 including variables from all eight domains, except for the proportion of people covered by BF, and the proportion of people with private health insurance, since these variables do not have data prior to 2004; (ii) data from years 2004 to 2012, to assess the possible association of poverty and private health insurance with coverage; and (iii) models stratified by population size. Goodness of fit was analyzed using pseudo-R2 and Deviance statistics. All calculations were performed in STATA v.12. This study was approved by the institutional review board of the Harvard T.H. Chan School of Public Health, Protocol # IRB16-0157. Role of funding source No funding.
SLIDE 6
6 RESULTS Table 2 presents descriptive statistics for municipalities for 1998, 2004 and 2012. FHS coverage increased significantly over time; while in 1998 50% of the municipalities had not implemented the program, in 2012, this figure was 5%. On average, FHS municipal coverage was 0·06 in 1998 and 0·81 in 2012 (Figure 1 shows the FHS geographic expansion). GDP per capita also grew during the period revealing higher level of
- development. The proportion of deaths diagnosed as non-defined declined, suggesting
improvements in the local healthcare system. The coverage of BF expanded from around 22% in 2004 to 36% in 2012. Regarding population size, there is considerable heterogeneity across municipalities. The supply of hospital beds per 1,000 inhabitants decreased from 2·32 in 1998 to 1·82 in 2012, reflecting a reorganization in the health system that started in the end of the 1990s.20-22 In contrast, the number of doctors per 1,000 inhabitants increased from 1·15 in 1998 to 2·11 in 2012.8 Figure 2 shows density curves of the FHS coverage across municipalities, stratified by region, in 1998, 2002, 2008 and 2012. In 1998, most municipalities had not launched the FHS (distribution highly concentrated near zero). Four years later the distribution became bimodal (except for the Center-West), indicating two distinct groups of municipalities: (i) ‘early adopters’, which had near full coverage, and (ii) ‘laggards’, in which the FHS was either not implemented or had coverage lower than 20%. Coverage expansion continued, and by 2012 the distribution was skewed towards high levels, albeit with a long left tail indicating remaining coverage gaps. The Northeast region had the fastest FHS uptake, reaching near universal coverage in 2012. In contrast, the North, South and Southeast had major coverage gaps. The FHS coverage expansion also showed distinct patterns considering the size of the
- municipality. Figure 3 shows coverage expansion according to population size and
stratified by regions. Smaller municipalities tended to reach > 80% coverage by 2012 regardless of region, whereas medium and larger municipalities had a slower pace in coverage expansion. The Northeast was the only region to present a different pattern for large municipalities, as it reached around 70% coverage in 2012. The period of fastest expansion occurred from 1998 to 2002, regardless of municipal size. On average, FHS coverage increased by 590·7% between 1998 and 2002, and by 77·8% between 2002 and 2012. Coverage density curves by population size (Figure 4) suggested that the largest coverage gap remains among municipalities with population size 50,000 inhabitants, reflecting both a slow pace in the initial uptake of the program, and a slow expansion of
- coverage. In 2012, 49·4% and 90·1% of the population in larger and smaller
municipalities, respectively, was covered by the FHS. The bimodal pattern revealed in the regional analysis was also observed for municipalities with < 20,000 people. Table 3 shows the results for the first set of models (Table S1 in supplementary appendix has the taxonomy of models). Model A indicated that 68% of the variation in coverage was attributable to differences within municipalities. Model B estimated that 55% of the within-municipality variation in coverage was associated with changes over
- time. Population size was inversely associated with both the initial uptake and the
expansion of the program – larger municipalities tended to start with a lower coverage
SLIDE 7
7 level and to progress with a slower rate of change. Regarding the magnitude of the coefficients, population size presented a strong and monotonic effect for the program
- uptake. Small municipalities started the program with a coverage level 25% higher than
a municipality with 50,000 people. These findings corroborate the pattern shown in Figures 3 and 4. Mayor-Governor political alignment favored both the adoption and the expansion of the program, whereas Mayor-President alignment was only important for the rate of change. Municipalities with higher gaps in health care access showed a negative association with coverage. Also, dummy variables were positive and significant for 13 states, eight
- f them in the Northeast, which experienced faster implementation, and attained almost
universal coverage (Figure 2). The second set of models is shown in Table 4 (Table S2 in supplementary appendix has the taxonomy of models). The proportion of the population receiving BF was positively associated with FHS coverage levels in 2004, but inversely associated with the rate of
- change. Since high BF coverage is expected in poor municipalities, FHS coverage was
already high in 2004. With regards to private health insurance, a negative rate of change coefficient suggests that this factor is a disincentive for expansion. The stratified models (Table S3 in supplementary appendix) did not reveal any new findings, except that the political alignment between Mayors and Governors was only significant for municipalities with < 10,000 inhabitants. Regional characteristics highlighted in Figure 3 show that the Northeast undertook a major effort to implement the program regardless of municipality size. The pseudo-R2 showed that each set of models explained a significant portion of the variability in coverage, and the deviance indicated that final models (Tables 2 and 3) were the best fit among the taxonomy of models tested (Tables S1 and S2). DISCUSSION From 1998 to 2012, the uptake and coverage expansion of the FHS in Brazilian municipalities was not homogeneous, with coverage density curves indicating two distinct groups: (i) early adopters, mostly smaller municipalities in less developed areas; and (ii) laggards, mostly larger municipalities. By 2012, 54% of the population was covered by the FHS, and 58% of the municipalities had coverage 95%. Longitudinal models showed that factors associated with the uptake and expansion of the program included population size, regional characteristics, availability of private insurance, Mayor-Governor political alignment, and health care needs/poverty. An important aspect for FHS uptake is financing.23,24 Financial mechanisms that allowed municipalities to implement the FHS were introduced in 1998. Indeed, while in 1998 the national FHS coverage was < 5%, it increased by 60% from 1998 to 1999, and by 128% from 1998 to 2000. However, while the financing mechanism may have acted as an incentive to facilitate uptake, it raises concerns of sustainability, as municipalities
SLIDE 8
8 are responsible for an important part of PHC expenditure, which may cause instabilities and discontinuities in these policies.19 The size of the municipal population had a large effect on the initial coverage and expansion over time. Larger municipalities tended to start FHS with a lower coverage, and to expand it slowly. Since each FHS team is expected to cover up to 4,000 people, in order to expand, and eventually achieve universal coverage, more populous municipalities have to contract a larger number of professionals to form family health teams, and must manage the family health units where those teams are based. This suggests the presence of diseconomies of scale in program management. Also, larger municipalities often show heterogeneity in the supply of primary care,15,18 and are likely to have high covered by private health insurance. To address these difficulties, in December 2003 the government sought financing for the Family Health Extension Program (PROESF – Programa de Expansão e Consolidação do Saúde da Família),25 to expand coverage in municipalities with > 100,000 inhabitants. By 2007, FHS coverage in 184 municipalities included in PROESF increased to 34·4% from 25·7% in 2003.25 However, analysis is needed to assess if the coverage expansion in these municipalities was significantly different from that observed in municipalities not included in PROESF. High access to private health insurance is a disincentive for Mayors to implement and expand the FHS, as middle- and high-income classes26 favor the use of the private
- sector. We argue that the dual health system in Brazil is an important obstacle for public
primary care and expansion of the FHS as Mayors, mainly in larger cities, choose not to implement FHS due to low demand. Municipalities with high BF coverage may not only demand more social programs, but may be more likely to implement them. With FHS, the marginal benefits are higher, as poor municipalities only have public primary care. The supply of primary care will not
- nly improve population welfare, but also improve a Mayor’s political support. The
negative coefficient of the rate of change is most likely a consequence of the high level
- f coverage already observed for these municipalities in 2004; the higher the level of
coverage, the harder it is to further increase it. In Brazil, although municipalities are responsible for the provision of primary care, policies and regulation mechanisms are defined and managed by Federal Units.27 Hence, political alignment between the Mayor and the Governor can make municipal management of primary care easier. Larger municipalities, which are more able to develop their policies, are more policy-independent from Federal Units. The FHS expansion presented a different pattern in the Northeast region, particularly for bigger municipalities, indicating that the FHS is a policy priority in this region probably related to the likelihood of achieving higher marginal benefits, compared to other regions that have better socio-economic indicators.28 The supply of family doctors is undoubtedly one of the most important challenges for achieving higher levels of FHS coverage. In 2013, the Brazilian Government launched the More Doctors Program (Programa Mais Médicos) aimed at increasing the number
SLIDE 9
9
- f family doctors in underserved regions of the country.29 The effect of this program on
the FHS coverage is yet to be evaluated. To the best of our knowledge, this is the first examination of the mechanisms associated with the FHS implementation and expansion across municipalities. The study has many
- strengths. First, it used the longest time series and the best scale for which FHS data are
publicly available. Second, it gathered data from varied sources to capture a multitude
- f domains that could affect FHS uptake and expansion. Third, the unit of analysis,
municipality, is the decision-unit for the FHS, since the choice to adopt/expand the program rests on the Mayor. Thus, study results do offer evidence to inform recommendations aimed at increasing program coverage. Limitations were related to data availability. The 15-year time series was the longest period of analysis we could undertake, since no data on the FHS coverage are available prior to 1998, and since the majority of the covariates included in the longitudinal model were not available post 2012. No annual time series by municipality was available for infant mortality and access to sanitation. However, since these two variables can be considered as proxy for poverty, their potential effects were partially assessed by BF coverage. The uptake and expansion of the FHS increased substantially in Brazil from 1998 to
- 2012. Yet, gaps remain. Small municipalities need financial support to uptake FHS, and
are likely to expand coverage faster. They could be helped by federal fund transfers to develop infrastructure and contract personnel. In larger municipalities, which often have high access to private healthcare, FHS expansion is stymied and policies are needed to achieve an optimal public-private mix.
SLIDE 10
10 Panel: Characteristics of the Family Health Strategy (FHS) in Brazil The FHS uses a community-based approach, with services are provided by a team that comprises one physician, one nurse, one nurse assistant, and up to six community health
- agents. Each team is responsible for providing care for a maximum of 4,000 people
living in a determined geographic catchment area. Usually teams are based in health units, with infrastructure to provide ambulatory care and to serve as the primary care reference center for the population living in the catchment area. In higher density areas more than one team can be located in the same
- unit. Community health agents play a crucial role, acting as a bridge between health
units and the population. Each household should be visited at least once per month by community health agents, who are responsible for household enrollment and data collection, identification of potential risk factors, monitoring the uptake of prescriptions, and scheduling visits to the family health units. The FHS follows a decentralized health care model, in which the responsibility for management and provision of health rests with the local levels of government - municipalities are responsible for the overall management of primary care, including contracting and paying health care providers, and managing and supplying adequate
- infrastructure. Ultimately, the implementation and progressive coverage expansion of
the FHS rests with the Mayor. The financing scheme of the FHS, created in 1998, uses a framework of incentives with two components.30 First, each municipality receives a fixed amount from the federal government, based on the number of inhabitants, to finance primary care expenses. Second, municipalities receive a variable amount conditioned on performance indicators, and on the development of some primary health care programs, such as the
- FHS. All transfers from the federal government to municipalities are conditioned on the
performance in the management of primary health care, monitored through information systems and regulation mechanisms.
SLIDE 11
11 CONTRIBUTORS MVA and MCC conceptualized the study, proposed and carried out data analysis and interpretation, and prepared the first draft of the manuscript. LC contributed in data collection, and preparation of figures and tables. AQC and MXN contributed in data collection and cleaning. RA contributed in data interpretation. All authors reviewed and contributed to subsequent drafts and approved the final version for publication. ACKNOWLEDGMENTS We thank the programming support provided by Simo Goshev, from the Institute for Quantitative Social Science (IQSS), Harvard University. MVA thanks the CAPES Foundation for a scholarship, and acknowledges the Takemi Program, Harvard T.H. Chan School of Public Health. AQC and MXN were supported by a fellowship from the Science Without Borders Program, CAPES, Brazil. LC thanks the Foundation for Supporting Research in the State of Minas Gerais (FAPEMIG) for a scholarship. MCC and RA thank the support from the Department of Global Health and Population, Harvard T.H. Chan School of Public Health. DECLARATION OF INTERESTS We declare no competing interests.
SLIDE 12
12 REFERENCES 1.
- WHO. Health in 2015: from MDGs, Millennium Development Goals to SDGs,
Sustainable Development Goals. Geneva: World Health Organization, 2015. 2. Shi L. The impact of primary care: a focused review. Scientifica 2012; 2012: 1- 22. 3. Starfield B. Primary care: an increasingly important contributor to effectiveness, equity, and efficiency of health services. SESPAS report 2012. Gac Sanit 2012; 26 Suppl 1: 20-6. 4. Macinko J, Guanais FC. Population experiences of primary care in 11 Organization for Economic Cooperation and Development countries. Int J Qual Health Care 2015; 27(6): 442-9. 5. Macinko J, Starfield B, Shi L. The contribution of primary care systems to health outcomes within Organization for Economic Cooperation and Development (OECD) countries, 1970-1998. Health Serv Res 2003; 38(3): 831-65. 6. Atun R, Aydın S, Chakraborty S, et al. Universal health coverage in Turkey: enhancement of equity. The Lancet 2013; 382(9886): 65-99. 7. Atun R, de Andrade LOM, Almeida G, et al. Health-system reform and universal health coverage in Latin America. The Lancet 2015; 385(9974): 1230-47. 8. Grignolati M, Lindelow M, Coutolenc B. Twenty Years of Health System Reform in Brazil. In: Bank W, editor. An Assessment of the Sistema Único de Saúde. Washington, DC: World Bank; 2013. p. 1-112. 9. Rocha R, Soares RR. Evaluating the impact of community-based health interventions: evidence from Brazil's Family Health Program. Health Econ 2010; 19 Suppl: 126-58. 10. Ceccon RF, Meneghel SN, Viecili PR. Hospitalization due to conditions sensitive to primary care and expansion of the Family Health Program in Brazil: an ecological study. Rev Bras Epidemiol 2014; 17(4): 968-77. 11. Dourado I, Oliveira VB, Aquino R, et al. Trends in primary health care-sensitive conditions in Brazil: the role of the Family Health Program (Project ICSAP-Brazil). Med Care 2011; 49(6): 577-84. 12. Lima-Costa MF, Turci MA, Macinko J. A comparison of the Family Health Strategy to other sources of healthcare: utilization and quality of health services in Belo Horizonte, Minas Gerais State, Brazil. Cad Saúde Pública 2013; 29(7): 1370-80. 13. Andrade MV, Noronha K, Barbosa AC, et al. Equity in coverage by the Family Health Strategy in Minas Gerais State, Brazil. Cad Saúde Pública 2015; 31(6): 1175-87. 14. de Santiago AX, Barreto IC, Sucupira AC, Lima JW, de Andrade LO. Equitable access to health services for children aged 5 to 9 in a medium city of northeasth of Brazil: a result of Family Health Strategy. Rev Bras Epidemiol 2014; 17 Suppl 2: 39- 52. 15. d'Avila Viana AL, Rocha JS, Elias PE, Ibanez N, Bousquat A. Primary health care and urban dynamics in large cities in Sao Paulo State, Brazil. Cad Saúde Pública 2008; 24 Suppl 1: S79-90. 16. de Mendonça MH, Martins MI, Giovanella L, Escorel S. Challenges for human resources management from successful experiences of Family Health Strategy
- expansion. Cien Saúde Colet 2010; 15(5): 2355-65.
17. Giovanella L, de Mendonca MH, de Almeida PF, et al. Family health: limits and possibilities for an integral primary care approach to health care in Brazil. Cien Saúde Colet 2009; 14(3): 783-94.
SLIDE 13
13 18. Machado CV, Lima LD, Silva Viana L. Organization of traditional Primary Health Care and the Family Health Program in large cities in Rio de Janeiro State,
- Brazil. Cad Saúde Pública 2008; 24 Suppl 1: S42-57.
19. Mendes Á, Marques RM. O financiamento da Atenção Básica e da Estratégia Saúde da Familia no Sistema Único de Saúde. Saúde Debate 2014; 38(103): 900-16. 20. Boing AF, Vicenzi RB, Magajewski F, et al. Reduction of ambulatory care sensitive conditions in Brazil between 1998 and 2009. Revista de saude publica 2012; 46(2): 359-66. 21. Mendes Ada C, Sa DA, Miranda GM, Lyra TM, Tavares RA. The public healthcare system in the context of Brazil's demographic transition: current and future
- demands. Cad Saúde Pública 2012; 28(5): 955-64.
22. Duarte SL, Garcia MLT. Psychiatric Reform: the path of psychiatric beds reduction in Brazil. Emancipação 2013; 13(1): 39-54. 23. de Sousa MF, Hamann EM. Family Health Program in Brazil: an incomplete agenda? Cien Saúde Colet 2009; 14 Suppl 1: 1325-35. 24. Mendonça CS. Family Health, more than never! Cien Saúde Colet 2009; 14 Suppl 1: 1493-7. 25. World Bank. Implementation completion and Results Report on a Loan in the amount of US$68 million to the Federative Republic of Brazil in support of the first phase of the Family Health Extension Adaptable Lending Program (IBRD -71050 JPN- 55399). Washington, DC: World Bank, 2007. 26. Macinko J, Harris MJ. Brazil's family health strategy - delivering community- based primary care in a universal health system. N Engl J Med 2015; 372(23): 2177-81. 27. Ministério da Saúde. Norma Operacional da Assistência a Saúde/SUS – NOAS. Brasília: Ministério da Saúde, Gabinete do Ministro, 2002. 28. Rocha S. Desigualdade Regional e Pobreza no Brasil: a evolucao 1981/95. Rio de Janeiro: Instituto de Pesquisa Econômica Aplicada, 1998. 29. Ministerio da Saúde. Programa Mais Médicos - Dois anos: mais saúde para os
- Brasileiros. Brasília, DF: Ministério da Saúde - Secretaria de Gestão do Trabalho e da
Educação na Saúde, 2015. 30.
- Brasil. Ministério da Saúde. Secretaria Executiva. Programa de Atenção Básica
(PAB): parte física. 2001; (1): 32.
SLIDE 14
14 FIGURE LEGENDS Figure 1: Proportion of the population covered by the Family Health Strategy in each Brazilian municipality - 1998, 2002, 2008 and 2012. Maps indicate the boundaries of states (federal units) and regions in Brazil.
SLIDE 15
15 Figure 2: Density curves of the Family Health Strategy coverage across Brazilian municipalities stratified by regions - 1998, 2002, 2008 and 2012. Density curves are represented by Epanechnikov kernel density estimators. The bandwidth estimation is shown below each curve.
SLIDE 16
16 Figure 3: Annual distribution of the proportion of the population covered by the Family Health Strategy, 1998 to 2012. (A) National distribution by population size of the municipality in 1998. Regional distribution for municipalities (B) with less than 5,000 people; (C) with 5,000 to 9,999 people; (D) with 10,000 to 19,999 people; (E) with 20,000 to 49,999; and (F) with 50,000 people or more.
SLIDE 17
17
SLIDE 18
18 Figure 4: Density curves of the Family Health Strategy coverage across Brazilian municipalities stratified by population size - 1998, 2002, 2008 and 2012. Density curves are represented by Epanechnikov kernel density estimators. The bandwidth estimation is shown below each curve.
SLIDE 19
19 Table 1: Data sources and variables by domain (all data accessed on November 2015)
DOMAIN 1: ECONOMIC DEVELOPMENT - Variable 1.1. Municipal gross domestic product (GDP) Brazilian Institute of Geography and Statistics (IBGE) http://www.sidra.ibge.gov.br/bda/pesquisas/PIBMun/default.asp Nominal GDP (1999-2012), deflated with implicit price deflator. 1998 GDP estimated with 1999-2000 growth rate. Hypothesis: richer municipalities have better financial resources and supply of health care providers, facilitating implementation/expansion; poorer municipalities implement social policies to improve local welfare. DOMAIN 1: ECONOMIC DEVELOPMENT - Variable 1.2. Proportion of the population covered by Bolsa Família (BF - conditional cash transfer launched in 2003 Brazilian Social Development Ministry http://mds.gov.br/assuntos/bolsa-familia/dados and IBGE: http://www.ibge.gov.br/home/estatistica/populacao/censo2010/, http://www.ibge.gov.br/home/estatistica/populacao/censo2000/, http://www.ibge.gov.br/home/estatistica/populacao/contagem2007 /default.shtm. (number of families covered by BF * average household size)/estimated population. Since average household size is available only for 2000, 2007, and 2010, the household size in year 2000 was used to calculate the variable for years 2004 to 2006, 2007 data used for years 2007 to 2009, and 2010 data used for years 2010 to 2012. Hypothesis: BF expansion could have a learning effect for Mayors, since it is a municipal decision to implement a national registry that makes it possible for individuals to apply for the cash transfer benefit. DOMAIN 2: HEALTH CARE SUPPLY - Variable 2.1. Number of doctors per 1,000 inhabitants DOMAIN 2: HEALTH CARE SUPPLY - Variable 2.2: Number of hospital beds per 1,000 inhabitants (excluding psychiatric beds) IBGE http://www.ibge.gov.br/home/estatistica/populacao/condicaodevid a/ams/2009/microdados.shtm, and National Registry of Health Establishments (CNES) http://cnes.datasus.gov.br/ No dramatic changes were observed in these variables over time, thus 1999 survey data was applied to years 1998 and 1999, 2002 survey data applied to years 2000 to 2002, and 2005 survey data applied to years 2003 to 2005. Hypothesis: Shortage of healthcare professionals and infrastructure are negatively associated with the FHS uptake DOMAIN 3:HEALTH CARE NEEDS/ACCESS - Variable 3.1: proportion of deaths with non-defined cause Ministry of Health http://www2.datasus.gov.br/DATASUS/index.php?area=0205&id =6937&VObj=http://tabnet.datasus.gov.br/cgi/deftohtm.exe?sim/c nv/obt1 number of deaths with non-defined causes / total number of deaths Hypothesis: Communities that have a high proportion of deaths with non-defined cause are expected to have lower access to healthcare services and consequently lower health status. DOMAIN 4: AVAILABILITY OF PRIVATE INSURANCE - Variable 4.1 Proportion of population covered by private health insurance Brazilian Regulatory Agency http://www.ans.gov.br/perfil-do- setor/dados-e-indicadores-do-setor No reliable data were available prior to 2004. Hypothesis: Competing sources of health care may impose barriers to the expansion of the FHS. DOMAIN 5: POLITICAL CONTEXT - Variables 5.1 and 5.2: Party affiliations of Mayors, state Governors and the President Electoral Tribune http://www.tse.jus.br/eleicoes/eleicoes- anteriores/eleicoes-anteriores, and IBGE http://www.ibge.gov.br/home/estatistica/economia/perfilmunic/def ault.shtm Two dummy variables: the Mayor’s party is the same as the state Governor’s, the Mayor’s party is the same as the President’s. Hypothesis: The Mayor is responsible for the FHS implementation, thus political alignment may act as an incentive to expand the FHS. DOMAIN 6: GEOGRAPHICAL ISOLATION - Variable 6.1: Distance to the closest municipality with a hospital having > 100 beds; Variable 6.2: Population density Data on total population and hospitals with more than 100 beds were mapped, and population densities (per km2) and distances (in meters) were calculated using ArcGIS 10.2. DOMAIN 7: REGIONAL CHARACTERISTICS 27 dummy variables, one for each of the federal units (or states) in the country DOMAIN 8: POPULATION SIZE Five dummy variables that categorized the size of each municipality in 1998: < 5,000, 5,000-9,999, 10,000-19,999, 20,000-49,999, and 50,000 inhabitants.
SLIDE 20
20 Table 2: Selected characteristics of Brazilian municipalities (N=5,419) for years 1998, 2004, and 2012
Variables 1998 2004 2012 Mean (Standard Deviation) Median (IQR) Mean (Standard Deviation) Median (IQR) Mean (Standard Deviation) Median (IQR) Proportion of population covered by the FHS 0·06 (0·19) 0·0 (0·0) 0·64 (0·38) 0·76 (0·71) 0·81 (0·29) 1·00 (0·31) Population size 29,392 (177,686) 10,316 (15,591) 32,565 (193,873) 10,787 (17144) 35,422 (207,281) 11,382 (18683) Hospital beds per 1,000 inhabitants 2·32 (2·74) 1·87 (3·55) 1·96 (2·77) 1·57 (2·95) 1·82 (2·12) 1·45 (2·69) Population density (inhabitants per km2) 91·60 (498·88) 23·47 (36·23) 102·31 (547·2) 24·18 (38·44) 110·88 (581·53) 24·76 (41·62) Distance to closest municipality with a hospital with 100 beds or more (meters) 40·20 (43·59) 29·88 (31·44) 44·9 (43·98) 34·80 (34·89) 43·1 (43·55) 32·50 (33·91) Municipal GDP per capita (R$) 25,372 (27,183) 19,020 (20,859) 32,743 (39,076) 24,540 (27,784) 37,944 (42,716) 29,358 (29,731) Proportion of deaths with non-defined cause 0·28 (0·24) 0·20 (0·345) 0·20 (0·19) 0·14 (0·24) 0·09 (0·09) 0·06 (0·1) Doctors per 1,000 inhabitants 1·15 (1·51) 0·95 (0·93) 1·85 (1·55) 1·52 (1·76) 2·11 (1·92) 1·47 (1·71) Proportion of population covered by Bolsa Familia
- 0·22
(0·15) 0·18 (0·24) 0·36 (0·22) 0·31 (0·40) Proportion of population with private health insurance coverage
- 0·05
(0·09) 0·02 (0·045) 0·08 (0·12) 0·04 (0·10) Note: Statistical indicators presented in the table were calculated using values of the variables for each municipality. Thus, the mean corresponds to the average of the variable among all municipalities in each year (e.g., in 1998 the average municipal coverage of the FHS was 6%, while the national coverage was 4·4%).
SLIDE 21
21 Table 3: Multilevel longitudinal models of change considering data from years 1998 to 2012
Variables Model A Model B Full Model (*) coefficient p-value coefficient p-value coefficient p-value Fixed effects Intercept 0·5802 < 0·001 0·2147 < 0·001
- 0·2374
0·21 Municipal GPD per capita (R$) 0·0000016 < 0·001 Number of doctors per 1000 inhabitants 0·024795 < 0·001 Distance to closest munic. with 100 beds 0·0005 < 0·001 Population density
- 0·000014
0·046 Political Alignment Mayor-Governor 0·0090 0·05 Mayor-President
- 0·0084
0·13
- Prop. of deaths with non-defined cause
- 0·5463
< 0·001 Population size Less than 5,000 0·2547 < 0·001 5,000-9,999 0·1665 < 0·001 10,000-19,999 0·0986 < 0·001 20,000-49,999 0·0497 < 0·001 Rate of change Intercept 0·0522 < 0·001 0·0355 < 0·001 Municipal GPD per capita (R$)
- 0·0000001
< 0·001 Number of doctors per 1000 inhabitants
- 0·0025
< 0·001 Distance to closest munic. with 100 beds
- 0·00005
< 0·001 Population density
- 0·0000015
0·10 Political Alignment Mayor-Governor 0·0011 0·07 Mayor-President 0·0027 < 0·001
- Prop. of deaths with non-defined cause
0·0859 < 0·001 Population size Less than 5,000 0·0127 < 0·001 5,000-9,999 0·0146 < 0·001 10,000-19,999 0·0133 < 0·001 20,000-49,999 0·0070 < 0·001 Variance components Level 1 within σ 0·1176 0·0527 0·0489 Level 2 rate of change 0·0005 0·0005 initial status 0·0552 0·0635 0·0488 covariance
- 0·0021
- 0·0029
Goodness of fit Pseudo-R2 0·4668 Deviance 67970·1 13951·7 5847·4 Test deviance 54018·4 114·7 Observations 5419 5419 5419 (*) Includes 27 dummy variables, one for each federal unity (results not shown).
SLIDE 22
22 Table 4: Multilevel longitudinal models of change considering data for from years 2004 to 2012
Variables Model A Model B Full Model coefficient p-value coefficient p-value coefficient p-value Fixed effects Intercept 0·7488 < 0·001 0·5363 < 0·001
- 0·0965
0.63 Municipal GPD per capita (R$) 0·00000078 < 0·001 Number of doctors per 1000 inhabitants
- 0·0157
< 0·001 Distance to the closest munic. with 100 beds
- 0·00039
< 0·001 Population density
- 0·000126
0·32 Political Alignment Mayor-Governor
- 0·0186
0·02 Mayor-President
- 0·0229
0·14 Proportion of deaths with non-defined cause
- 0·4658
< 0·001 Population size Less than 5,000 0·4572 < 0·001 5,000-9,999 0·3098 < 0·001 10,000-19,999 0·1398 < 0·001 20,000-49,999 0·0351 0·21
- Prop. population covered by Bolsa Família
0·4082 < 0·001
- Prop. population with private health insurance
0·1153 0·13 Rate of change Intercept 0·0213 < 0·001 0·0177 < 0·001 Municipal GPD per capita (R$)
- ·00000005
< 0·001 Number of doctors per 1000 inhabitants 0·0017 < 0·001 Distance to the closest munic. with 100 beds 0·000025 0·03 Population density
- 0·0000004
0·60 Political Alignment Mayor-Governor 0·0017 0·04 Mayor-President 0·0025 0·09 Proportion of deaths with non-defined cause 0·0424 < 0·001 Population size Less than 5,000
- 0·0083
< 0·001 5,000-9,999
- 0·0002
0·93 10,000-19,999 0·0089 < 0·001 20,000-49,999 0·0088 < 0·001
- Prop. population covered by Bolsa Família
- 0·0242
< 0·001
- Prop. population with private health insurance
- 0·0279
< 0·001 Variance componentes Level 1 within σ 0·0317 0·0192 0·0188 Level 2 rate of change 0·0012 0·0011 initial status 0·0792 0·2529 0·1931 covariance
- 0·0147
- 0·0130
Goodness of fit Pseudo-R2 0·3585 Deviance
- 12776·9
- 26314·0
- 30668·1