ABSTRACT

OBJECTIVE

To analyze the association between municipal rates of ambulatory care sensitive conditions (ACSC) hospitalization and the quality of primary health care (PHC), socioeconomic, and demographic variables and those related to local characteristics of the health system from 2010 to 2019.

METHOD

Ecological time series study in Brazilian municipalities analyzing the correlation of ACSC hospitalization rates with PHC quality measured by the three cycles of the Primary Care Access and Program for improving primary care access and quality (PMAQ-AB). The study included municipalities whose teams participated in 80% or more of at least two PMAQ-AB cycles. The correlation between standardized ACSC hospitalization rates and PHC quality and other variables was analyzed. Spearman’s test was used between the response variable and numerical explanatory variables. Generalized equations estimation was used as a multivariate model associating ACSC hospitalization rates with the other variables over the years.

RESULTS

A total of 3,500 municipalities were included in the models. The quality of PHC (PMAQ-AB score) showed an inverse association with the variation in ACSC hospitalization rates. Hospitalization rates fell by -2% per year every ten-point increase in the PMAQ-AB score, adjusted by the remaining variables. A one-unit increase in the beds per 1,000 inhabitants variable had an impact of approximately +6.4% on ACSC hospitalization rates. Regarding population size, larger municipalities had lower ACSC hospitalization rates. Increased PHC coverage and lower socioeconomic inequality were also associated with the reduction in hospitalizations.

CONCLUSIONS

The reduction in ACSC hospitalization rates over time was associated with an increase in the quality of PHC. It was also associated with a reduction in the number of hospital beds and municipalities with better socioeconomic indicators.

Primary Health Care; Quality of Health Care; Family Health

INTRODUCTION

Since 1994, when the Family Health Program as created, primary health care (PHC) in Brazil has expanded greatly throughout the country ¹1. Malta DC, Santos MAS, Stopa SR, Vieira JEB, Melo EA, Reis AAC. A Cobertura da Estratégia de Saúde da Família (ESF) no Brasil, segundo a Pesquisa Nacional de Saúde, 2013. Cien Saude Colet. 2016;21(2):327-38. https://doi.org/10.1590/1413-81232015212.23602015
https://doi.org/10.1590/1413-81232015212... . Since the creation of the Unified Health System (SUS), various government policies have sought to make PHC the system’s main gateway, as there is solid evidence that strengthening PHC lowers health costs, impacts on various indicators, and contributes to reducing health inequalities ²2. Starfield B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milkbank Q. 2005;83(3):457-502. https://doi.org/10.1111/j.1468-0009.2005.00409
https://doi.org/10.1111/j.1468-0009.2005... .

Evaluating the expansion of PHC and improving its quality is essential for health planning in the country. One of the pillars of health quality assessment is the evaluation of results ³3. Donabedian A. Evaluación de la calidad de la atención médica. Rev Calidad Asistencial. 2001;16:S11-27. . The indicator of hospitalizations for primary care-sensitive conditions is used in many countries as an instrument to evaluate their results ⁴4. Billings J, Zeitel L, Lukomnik J, Carey TS, Blank AE, Newman L. Impact of socioeconomic status on hospital use in New York City. Health Aff (Millwood). 1993;12(1):162-73. https://doi.org/10.1377/hlthaff.12.1.162
https://doi.org/10.1377/hlthaff.12.1.162...

5. Alfradique ME, Bonolo PF, Dourado I, Lima-Costa MF, Macinko J, Mendonça CS, et al. Internações por condições sensíveis à atenção primária: a construção da lista brasileira como ferramenta para medir o desempenho do sistema de saúde (Projeto HPCSC-Brasil). Cad Saude Publica. 2009;25(6):1337-49. https://doi.org/10.1590/s0102-311x2009000600016
https://doi.org/10.1590/s0102-311x200900... - ⁶6. Rosano A, Loha CA, Falvo R, van der Zee J, Ricciardi W, Guasticchi G, et al. The relationship between avoidable hospitalization and accessibility to primary care: a systematic review. Eur J Public Health. 2013;23(3):356-60. https://doi.org/10.1093/eurpub/cks053
https://doi.org/10.1093/eurpub/cks053... . Since the creation of the national list of hospitalizations for ambulatory care sensitive conditions (ACSC) in 2008, there has been an increase in studies on this subject in Brazil, seeking to understand its behavior and applicability ⁷7. Ministério da Saúde (BR). Portaria nº 221, de 17 de abril de 2008 [cited 2023 Aug 18]. Publica em forma do anexo a Lista Brasileira de Internações por Condições Sensíveis à Atenção primária. Available from: https://bvsms.saude.gov.br/bvs/saudelegis/sas/2008/prt0221_17_04_2008.html
https://bvsms.saude.gov.br/bvs/saudelegi... , ⁸8. Medina MG, Aquino R, Vilasbôas ALQ, Nunes CA. A pesquisa em Atenção Primária à Saúde no Brasil. In: Mendonça MHM, Matta GC, Gondim R, Giovanella L. Atenção primária à saúde no Brasil: conceitos, práticas e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. Chapt. 16. .

Time series of ACSC hospitalization showed a drop between 1999 and 2007 in all Brazilian regions and in the country as a whole. In general, an association was also found between this drop in hospitalizations and the expansion of PHC coverage in the country ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... . However, this association has not always proved to be true, especially in studies that analyze intra-urban or municipal data in isolation ¹⁰10. Malvezzi E. Internações por condições sensíveis à atenção primária: revisão qualitativa da literatura científica brasileira. Saúde Redes. 2018;4(4):119-34.11. , ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. . Other factors that also showed an association with ACSC hospitalization rates were socioeconomic conditions ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ¹³13. Mendonça CS, Leotti VB, Dias-da-Costa JS, Harzheim E. Hospitalizations for primary care sensitive conditions: association with socioeconomic status and quality of family health teams in Belo Horizonte, Brazil. Health Policy Plan. 2017;32(10):1368-74. https://doi.org/10.1093/heapol/czx103
https://doi.org/10.1093/heapol/czx103... , hospital beds available in the municipality ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. , ¹⁴14. Busby J, Purdy S, Hollingworth W. A systematic review of the magnitude and cause of geographic variation in unplanned hospital admission rates and length of stay for ambulatory care sensitive conditions. BMC Health Serv Res. 2015;15:324. https://doi.org/10.1186/s12913-015-0964-3
https://doi.org/10.1186/s12913-015-0964-... and health insurance coverage ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. . Some studies have tried to measure the correlation between PHC quality and ACSC hospitalization rates using the Primary Care Assessment Tool (PCATool) ¹³13. Mendonça CS, Leotti VB, Dias-da-Costa JS, Harzheim E. Hospitalizations for primary care sensitive conditions: association with socioeconomic status and quality of family health teams in Belo Horizonte, Brazil. Health Policy Plan. 2017;32(10):1368-74. https://doi.org/10.1093/heapol/czx103
https://doi.org/10.1093/heapol/czx103... , ¹⁵15. Gonçalves MR, Hauser L, Prestes IV, Schmidt MI, Duncan BB, Harzheim E. Primary health care quality and hospitalizations for ambulatory care sensitive conditions in the public health system in Porto Alegre, Brazil. Fam Pract. 2016;33(3):238-42. https://doi.org/10.1093/fampra/cmv051
https://doi.org/10.1093/fampra/cmv051... or the Program for Improving Access and Quality in Primary Care (PMAQ-AB) ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ¹⁶16. Araújo WRM, Queiroz RCS, Rocha TAH, Silva NC, Thumé E, Tomasi E, et al. Estrutura e processo de trabalho na atenção primária e internações por condições sensíveis. Rev Saude Publica. 2017;51(75). https://doi.org/10.11606/S1518-8787.2017051007033
https://doi.org/10.11606/S1518-8787.2017... .

The PMAQ-AB is a nationwide program that began in 2011 and has been carried out over three cycles, the last one ending in 2019. Its main objective was to induce the expansion of access and improve the quality of PHC in the country, but it also fulfilled the role of evaluating family health teams through quality standards. PMAQ-AB stands out for its scope and capillarity, reaching around 39,000 teams and 96% of Brazilian municipalities in its last cycle ¹⁷17. Giovanella L, Mendonça MHM, Medina MG, Lima JG, Fausto MCR, Seidl HM, et al. Contribuições dos estudos PMAQ-AB para a avaliação da APS no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (Organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. Chapt. 19. , ¹⁸18. Facchini LA, Tomasi E, Thumé E. Acesso e qualidade na atenção básica brasileira: análise comparativa dos três ciclos da avaliação externa do PMAQ-AB, 2012-2018. São Leopoldo: Oikos; 2021. . It also stands out for maintaining aspects linked to: processes in the area of care, results in indicators and structures in basic units as the basis of its evaluation process, considering the context of Donabedian’s classic triad ³3. Donabedian A. Evaluación de la calidad de la atención médica. Rev Calidad Asistencial. 2001;16:S11-27. .

However, there is still a lack of national studies evaluating the association between variation in ACSC hospitalization rates and PHC quality by municipality and its evolution over the years, as most studies only evaluate one municipality ¹³13. Mendonça CS, Leotti VB, Dias-da-Costa JS, Harzheim E. Hospitalizations for primary care sensitive conditions: association with socioeconomic status and quality of family health teams in Belo Horizonte, Brazil. Health Policy Plan. 2017;32(10):1368-74. https://doi.org/10.1093/heapol/czx103
https://doi.org/10.1093/heapol/czx103... , ¹⁵15. Gonçalves MR, Hauser L, Prestes IV, Schmidt MI, Duncan BB, Harzheim E. Primary health care quality and hospitalizations for ambulatory care sensitive conditions in the public health system in Porto Alegre, Brazil. Fam Pract. 2016;33(3):238-42. https://doi.org/10.1093/fampra/cmv051
https://doi.org/10.1093/fampra/cmv051... or a single year ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ¹⁶16. Araújo WRM, Queiroz RCS, Rocha TAH, Silva NC, Thumé E, Tomasi E, et al. Estrutura e processo de trabalho na atenção primária e internações por condições sensíveis. Rev Saude Publica. 2017;51(75). https://doi.org/10.11606/S1518-8787.2017051007033
https://doi.org/10.11606/S1518-8787.2017... . Thus, the aim of this study was to analyze the association between municipal rates of ACSC hospitalization and quality of PHC, socioeconomic and demographic variables and those related to local characteristics of the health system between 2010 and 2019.

METHODS

This work is characterized as an ecological time series study, in which correlations were analyzed between standardized ACSC hospitalization rates in Brazilian municipalities from 2010 to 2019 and the quality of care in PHC, characteristics of municipal healthcare, socioeconomic and demographic conditions.

Standardized ACSC hospitalization rates were calculated for each municipality using data from the Sistema de Internação Hospitalar (SIH – Hospital Admission System) from 2010 to 2019. Such rates were standardized by sex and age groups (0–4, 5–19, 20–59, 60–79 years), using as a reference the Brazilian population estimated by the Brazilian Institute of Geography and Statistics ( Instituto Brasileiro de Geografia e Estatística – IBGE) in 2014, as it was the midpoint of the period. The year of hospital discharge was considered to be the year of the patient’s hospitalization; and the municipality of residence defined the place where the hospitalization was counted ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... . Hospitalizations for diseases derived from the 10 ^th revision of the International Classification of Diseases (ICD-10) and validated as primary care-sensitive conditions in Brazil were considered, consisting of 20 diagnostic groups divided into 120 ICD-10 subgroups, excluding groups related to childbirth ⁷7. Ministério da Saúde (BR). Portaria nº 221, de 17 de abril de 2008 [cited 2023 Aug 18]. Publica em forma do anexo a Lista Brasileira de Internações por Condições Sensíveis à Atenção primária. Available from: https://bvsms.saude.gov.br/bvs/saudelegis/sas/2008/prt0221_17_04_2008.html
https://bvsms.saude.gov.br/bvs/saudelegi... . As recommended by the Projeto ICSAP-Brasil ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... , people aged 80 or over were also excluded from the study.

The study variables are presented in the Chart and include the quality of PHC; characteristics of local health systems such as: number of beds per 1,000 inhabitants, percentage of health insurance coverage and coverage of family health teams; socio-economic characteristics such as the Brazilian Deprivation Index ( Índice Brasileiro de Privações – IBP) and the Municipal Human Development Index ( Índice de Desenvolvimento Humano Municipal – MHDI), region and demographic characteristics such as sex and population size (small size I – up to 20,000 inhabitants; small size II – from 20,001 to 50,000 inhabitants; medium size – from 50,001 to 100,000 inhabitants; large size – from 100,001 to 900,000 inhabitants; metropolis – more than 900,000 inhabitants). The municipal size strata were defined according to a study by Castro et al. ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981...

The main explanatory variable, PHC quality, was calculated using scores from the evaluation of the PMAQ-AB PHC teams, taken from the Program’s national database (Chart). The simple mean of the scores of family health teams from the same municipality received in each of the three PMAQ-AB cycles was calculated, which was different for each cycle. In the first and second cycles, the score prior to the certification process and the weighting by extracts were used to allow for better comparisons. In the third cycle, the score was multiplied by ten, since it was the only cycle in which the score varied from 0.0 to 10.0 and not from 0.0 to 100.0, thus allowing comparisons with the other cycles. In order to analyze the data, scores from the first PMAQ-AB cycle were repeated in 2010, 2011, and 2012 and correlated with the other variables or the same years; and so were those from the second cycle in 2013, 2014, and 2015 and those from the third cycle in 2016, 2017, 2018, and 2019 (Chart).

As an inclusion criterion for participation in the study, only municipalities that had at least two cycles with more than 80% of family health and primary care teams participating in the PMAQ-AB study were selected. A total of 3,800 municipalities were initially included in the study. The distribution pattern of ACSC hospitalization rates in these municipalities was checked and those which, in any of the years, had standardized rates lower than 15.0 or higher than 600.0 were excluded as they were discrepant values and could be related to errors in the information and surveillance system. As a result, another 280 municipalities were excluded from the analysis, leaving approximately 3,500 municipalities, with some variations from year to year, due to the creation of new municipalities in the period and missing data on hospitalization in a given year.

Considering that each municipality was in the ten years of analysis (2010 to 2019), and the stratification of rates by sex, there were around 70,000 units of analysis (3,500 municipalities* ten years * 2).

The distribution pattern of the variables was checked and, given that a non-normal pattern was observed, the relationship between the response variable and the numerical explanatory variables was verified using Spearman’s correlation ¹⁹19. Hollander M, Wolfe DA, Chicken E. Nonparametric statistical methods. 3rd ed. Chichester: John Wiley & Sons; 1999. . IBP quintiles is an ordinal categorical variable, but for the purposes of analysis it was considered numerical.

In order to assess the influence of the variables on ACSC hospitalization rates over time, the generalized equations estimating (GEE) model was chosen ²⁰20. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):13-22. https://doi.org/10.1093/biomet/73.1.13.
https://doi.org/10.1093/biomet/73.1.13... . Since the variables had a non-normal distribution, with continuous and positive values, the gamma pattern was assumed. Given that the correlations among the MHDI, IBP quintiles and percentage of health insurance coverage variables were strong, it was decided to run three different GEE models to avoid problems of multicollinearity.

To build the model, i.e., select the variables, the Stepwise method was used, which consists of adding (forward) and removing (backward) variables to define which ones will compose the final analysis ²¹21. Efroymson MA. Multiple regression analysis. In: Ralston A, Wilf HS (editors). Mathematical methods for digital computers. New York: John Wiley; 1960. p. 58-76. . In the forward step, the independent variables were univariately analyzed and those with a p-value < 0.20 were selected for the multivariate model. In the backward step, the variable with the highest p-value was removed from the analysis at a time, and the procedure was repeated until only significant variables remained in the final model. A 5% significance level was adopted for the multivariate model. All analyses were carried out in the R software, version 3.6.0, with the help of the geepack package.

The study was carried out in accordance with Conselho Nacional de Saúde (CNS – National Health Council) Resolution 466 of December 12, 2012. The SIH information is available on the Datasus website, without identifying patients.

RESULTS

Table 1 shows the distribution pattern of the ACSC hospitalization rates and the other numerical variables over the ten years. It can be seen that the PHC quality variable showed the greatest variation in the units analyzed (5,670 to 7,004) due to the fact that some municipalities participated in only two PMAQ-AB cycles.

Mean hospitalization rates declined over the period, from 157.72 hospitalizations per 10,000 inhabitants in 2010 to 107.69/10,000 inhabitants in 2019. It is important to note that this is not the rate for Brazil, but a simple mean among the municipalities in the study. The mean score for the quality of PHC built fell from the first (58.95) to the second cycle (53.80) and the highest average was in the third cycle (59.55). The mean number of beds per 1,000 inhabitants also fell, from 1.58 in 2010 to 1.31 in 2019. The mean percentage of health insurance coverage remained between 7.67% and 9.40% and the mean percentage of FHT coverage remained between 0.85 and 0.91 (85% to 91%). It should also be noted that the data related to vulnerability indicators (MHDI, IBP quintiles) did not vary over the years as they are related to the 2010 Census ( Table 1 ).

When Spearman’s correlations between the ACSC hospitalization rates and the numerical variables year by year were made, a negative correlation was observed with the PHC quality indicator scores, with statistical significance in the respective municipalities, both in 2010 and in the years of the third cycle. In 2011 and 2012 there was no correlation and in the second cycle there was a positive correlation ( Table 2 ).

The correlation of rates with socio-economic indicators (MHDI and IBP) year after year on a univariate basis showed that the highest hospitalization rates were in municipalities with best indices ( Table 2 ). Regarding the variables of municipal characteristics, the strongest positive association of ACSC hospitalization rates was observed with the variable beds per 1,000 inhabitants (0.31 to 0.40) ( Table 2 ).

The results of the final model, which correlated over the years using the GEE, are shown in Tables 3 and 4. In the univariate analysis ( Table 3 ), all the variables were significant (p-values < 0.05); it should be noted that the numerical variables that showed negative changes with the variation in hospitalization rates over the years, when increased by one unit were: quality of PHC – score, percentage of health insurance coverage and percentage of FHT coverage. The variables MHDI and IBP quintiles and beds per 1,000 inhabitants had positive changes.

In the multivariate analysis, in the first model, using the variables in Table 1 (except for the IBP quintiles and the percentage of health insurance coverage), it was observed that municipal hospitalization rates for ACSC hospitalization declined as the quality of PHC improved over the years, and it was found that for every 10-point increase in the quality score, there was a -2.10% drop in the annual hospitalization rate; the 1% increase in FHT coverage reduced hospitalization by -11.81% each year. Increasing the MHDI by 0.10 resulted in a reduction in hospitalization rates of -4.90%. On the other hand, increasing the beds per 1,000 inhabitants variable resulted in an increase in hospitalization rates for primary care-sensitive conditions ( Table 4 ).

In the second model, including the IBP quintiles variable and excluding the MHDI and percentage of health insurance coverage, the IBP showed a positive change. In municipalities with higher vulnerability quintiles, there was a positive change in hospitalization rates, while values of the other variables remained the same ( Table 4 ).

In the third multivariate model, excluding MHDI and also IBP quintiles, and using the percentage of health insurance coverage, the association found with health insurance coverage was -11.06% in rates for each 1% increase in coverage, adjusted for the variables in the model ( Table 4 ). The other numerical variables in models 2 and 3 kept their associations with hospitalization rates similar to model 1.

Among the categorical variables, negative associations were found with medium, large, and metropolitan areas, when compared to small I municipalities, while the other variables remained fixed in all models ( Table 4 ).

In terms of region, the number of hospitalization rates in the Midwest had a negative variation compared to the North in all three models of around -17%. In the South, hospitalization rates for ACSC hospitalization were higher than in the North, ranging from 11.91% to 27.51% in the study models. In the Southeast and Northeast there was a positive association only in model 3. Regarding sex, male hospitalization rates were lower than female hospitalization rates by -11.05% in all models ( Table 4 ).

DISCUSSION

The study showed a downward trend in the mean of standardized ACSC hospitalization rates over the ten-year period, corroborating the findings of time series studies in previous periods ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... , ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. . Three models were used in the multivariate analysis and factors that showed an association with the reduction in hospitalizations were: improvement in the quality of PHC (as measured by PMAQ scores), increased coverage of the family health teams (FHT), reduction in the number of hospital beds linked to the SUS and increased coverage of health plans. In addition, municipalities with better socioeconomic indicators (measured by the IBP and MHDI), medium-sized, large, and metropolitan municipalities, living in the Midwest region and being male were associated with lower rates, while an increase in hospitalization rates was observed as the number of beds per inhabitant got bigger.

The study found an association between lower ACSC hospitalization rates in municipalities with better performance in the PMAQ-AB. These results confirm previous studies which showed a reduction in ACSC hospitalization rates when evaluating the first two cycles of PMAQ-AB ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ¹⁶16. Araújo WRM, Queiroz RCS, Rocha TAH, Silva NC, Thumé E, Tomasi E, et al. Estrutura e processo de trabalho na atenção primária e internações por condições sensíveis. Rev Saude Publica. 2017;51(75). https://doi.org/10.11606/S1518-8787.2017051007033
https://doi.org/10.11606/S1518-8787.2017... , the increase in the number of teams joining over the three cycles ²²22. Russo LX, Powell-Jackson T, Barreto JOM, Borghi J, Kovacs R, Gurgel Junior GD, et al. Pay for performance in primary care: the contribution of the Program for Improving Primary Care Access and Quality of Primary Care (PMAQ) on avoidable hospitalizations in Brazil, 2009–2018. BMJ Glob Health. 2021;6:e005429. http://dx.doi.org/10.1136/bmjgh-2021-005429
http://dx.doi.org/10.1136/bmjgh-2021-005... , or which compared municipalities that joined PMAQ-AB with those that did not ²³23. Soares C, Ramos M. Uma avaliação dos efeitos do PMAQ-AB nas internações por condições sensíveis à Atenção Básica. Saude Debate. 2020;44(126):708-24. https://doi.org/10.1590/0103-1104202012609
https://doi.org/10.1590/0103-11042020126... . This study advances by indicating that, over a ten-year period, even among municipalities that participated with at least 80% of their teams, a ten-point increase in scores over the course of the PMAQ-AB cycles is associated with a reduction in rates of -2% per year, adjusted for other variables.

However, the association with PHC quality was not found by Gonçalves et al. ¹⁵15. Gonçalves MR, Hauser L, Prestes IV, Schmidt MI, Duncan BB, Harzheim E. Primary health care quality and hospitalizations for ambulatory care sensitive conditions in the public health system in Porto Alegre, Brazil. Fam Pract. 2016;33(3):238-42. https://doi.org/10.1093/fampra/cmv051
https://doi.org/10.1093/fampra/cmv051... in a cohort study in Porto Alegre, nor by Mendonça et al. ¹³13. Mendonça CS, Leotti VB, Dias-da-Costa JS, Harzheim E. Hospitalizations for primary care sensitive conditions: association with socioeconomic status and quality of family health teams in Belo Horizonte, Brazil. Health Policy Plan. 2017;32(10):1368-74. https://doi.org/10.1093/heapol/czx103
https://doi.org/10.1093/heapol/czx103... in a study in Belo Horizonte. Both studies used the PCATool as a quality indicator and evaluated time series in a single municipality.

When relating the fall in ACSC hospitalization rates to FHT coverage, municipalities that had a 1% increase in the indices showed reductions in the rates for the period by -11.5 to -12.5%. This correlation was also observed in previous studies, which considered that increased FHT coverage would be an explanatory factor for this improvement ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... , ²⁴24. Macinko J, Dourado I, Aquino R, Bonolo PF, Lima-Costa MF, Medina MG, et al. Major expansion of primary care in Brazil linked to decline in unnecessary hospitalization. Health Aff. 2010;29(12):2149-60. https://doi.org/10.1377 / hlthaff.2010.0251.
https://doi.org/10.1377 / hlthaff.2010.0... .

The univariate analysis over a one-year period showed that higher FHT coverage rates were not associated with lower hospitalization rates. This differs from the increase in coverage over the years, as this movement may be related to a real increase in investment, strengthening and expansion of PHC ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. , ²⁵25. Magalhães ALA, Morais Neto OL. Desigualdades intraurbanas de taxas de internações por condições sensíveis à atenção primária na região central do Brasil. Cien Saude Colet. 2017; 22(6):2049-62. https://doi.org/10.1590/1413-81232017226.16632016
https://doi.org/10.1590/1413-81232017226... .

Regarding socio-economic variables, it was found that an increase in the MHDI, associated with better living conditions, resulted in a drop in hospitalization rates. The IBP also identified a correlation in the quintiles of greater vulnerability, greater poverty, and increased hospitalization. These findings are in line with other studies that correlated vulnerability and hospitalization and that places with greater deprivation have worse health indicators ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ¹³13. Mendonça CS, Leotti VB, Dias-da-Costa JS, Harzheim E. Hospitalizations for primary care sensitive conditions: association with socioeconomic status and quality of family health teams in Belo Horizonte, Brazil. Health Policy Plan. 2017;32(10):1368-74. https://doi.org/10.1093/heapol/czx103
https://doi.org/10.1093/heapol/czx103... , ²⁶26. Allik M, Leyland A, Ichihara MYT, Dundas R. Creating small-area deprivation indices: a guide a guide for stages and options. J Epidemiol Community Health. 2020;74(1):20–5. doi:10.1136/jech-2019-213255
https://doi.org/10.1136/jech-2019-213255... .

The variable beds per 1,000 inhabitants showed the highest correlation in all years in the univariate analysis and maintained a positive association in the three multivariate models investigated. The increase in the beds per 1,000 inhabitants’ indicator affected the growth in ACSC hospitalization rates, adjusted for the other variables. This association has already been found in other studies and may be related to the greater demand for hospitalization in places with more beds ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ²³23. Soares C, Ramos M. Uma avaliação dos efeitos do PMAQ-AB nas internações por condições sensíveis à Atenção Básica. Saude Debate. 2020;44(126):708-24. https://doi.org/10.1590/0103-1104202012609
https://doi.org/10.1590/0103-11042020126... , ²⁷27. Castro MSM, Travassos C, Carvalho MS. Efeito da oferta de serviços de saúde no uso de internações hospitalares no Brasil. Rev Saude Publica. 2005;39(2):277-84. https://doi.org/10.1590/S0034-89102005000200020
https://doi.org/10.1590/S0034-8910200500... , ²⁸28. Carmo, M. Hospitais de pequeno porte e rede de atenção à Saúde: um estudo de sua inserção e avaliação em Minas Gerais no período de 2004 a 2014 (Doctoral thesis). Belo Horizonte: Universidade Federal de Minas Gerais; 2016. .

The study found that the increase in health insurance coverage resulted in a drop in ACSC hospitalization. The findings of a reduction in ACSC hospitalizations related to an increase in private health insurance coverage have also been described ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... . It is possible that this indicator reflects better socio-economic conditions in these municipalities. It may also be related to the absence of hospitalization data in the non-SUS sector when calculating ACSC hospitalization rates.

Regarding demographic characteristics, in all three models there was a decrease in rates for medium, large-sized municipalities and metropolitan areas compared to small-sized municipalities. This finding is consistent with results of the study by Carmo ²⁸28. Carmo, M. Hospitais de pequeno porte e rede de atenção à Saúde: um estudo de sua inserção e avaliação em Minas Gerais no período de 2004 a 2014 (Doctoral thesis). Belo Horizonte: Universidade Federal de Minas Gerais; 2016. (2016), which pointed out that hospitals with up to 49 beds, considered small, accounted for 35% of the country’s ACSCs hospitalizations in 2015. The author stated that this profile results from the low complexity of hospitals, little diversity in the specialties offered, and that, in the state of Minas Gerais, almost half of them were located in cities of up to 10,000 inhabitants and such hospitals present greater demand and pressure for hospitalization ³3. Donabedian A. Evaluación de la calidad de la atención médica. Rev Calidad Asistencial. 2001;16:S11-27. , ²⁸28. Carmo, M. Hospitais de pequeno porte e rede de atenção à Saúde: um estudo de sua inserção e avaliação em Minas Gerais no período de 2004 a 2014 (Doctoral thesis). Belo Horizonte: Universidade Federal de Minas Gerais; 2016. .

Men have lower hospitalization rates. Studies show that women generally have a better perception of the signs and symptoms of illnesses and seek health services more often ²⁹29. Malta DC, Gomes CS, Prates EJ, Santos AFP, Almeira WS, Stopa SR, et al. Análise da demanda e acesso aos serviços nas duas semanas anteriores à Pesquisa Nacional de Saúde 2013 e 2019. Rev Bras Epidemiol. 2021;24(supl. 2):E210002.SUPL.2. https://doi.org/10.1590/1980-549720210002.supl.2
https://doi.org/10.1590/1980-54972021000... , which can result in greater opportunities for treatment and hospitalization.

These findings are important to reinforce the need to invest in PHC quality and evaluation programs, given the discontinuity of the PMAQ-AB. It is suggested that new follow-up studies be carried out in places that have invested in PHC quality, as well as studying the impact of variation over time in socio-economic conditions. Studies should also investigate the correlation between smaller municipalities and greater demand for hospitalization in these locations.

Limitations

This is an ecological study and may contain association biases due to the concurrence of many interrelated explanatory factors. The use of ACSC hospitalization rates requires careful interpretation, as they may not only reflect PHC performance. This is an indicator that relates to the organization of health systems and is influenced by various other factors ¹¹11. Mendonça CS, Nedel FB, Batista SR, Medina MG. A utilização do indicador internações por condições sensíveis à atenção primária no Brasil. In: Mendonça MHM, Matta GC, Gondin R, Giovanella L (organizers). Atenção primária à saúde no Brasil: conceitos, prática e pesquisa. Rio de Janeiro: Editora Fiocruz; 2018. p. 527-68. . There may also be problems related to the hospitalization database because it does not include data from hospitals not affiliated with the SUS, which would interfere with the analysis and interpretation, especially in cities presenting greater health insurance coverage ¹²12. Castro DM, Oliveira VB, Andrade ACS, Cherchiglia ML, Santos AF. Impacto da qualidade da atenção primária à saúde na redução das internações por condições sensíveis. Cad Saúde Pública. 2020:36(11): e00209819. https://doi.org/10.1590/0102-311X00209819
https://doi.org/10.1590/0102-311X0020981... , ²³23. Soares C, Ramos M. Uma avaliação dos efeitos do PMAQ-AB nas internações por condições sensíveis à Atenção Básica. Saude Debate. 2020;44(126):708-24. https://doi.org/10.1590/0103-1104202012609
https://doi.org/10.1590/0103-11042020126... . Even so, the SIH is a significant, reliable source of information, enabling relevant epidemiological analyses of hospital morbidity ⁹9. Universidade Federal de Minas Gerais. Faculdade de Medicina. Núcleo de Educação em Saúde Coletiva. Projeto HPCSC-BRASIL: avaliação do impacto das ações do programa de saúde da família na redução das internações hospitalares por condições sensíveis à atenção básica em adultos e idosos: Relatório final de pesquisa. Belo Horizonte: UFMG; 2012 [cited 2022 Apr 7]. Available from: https://www.nescon.medicina.ufmg.br/biblioteca/imagem/3261.pdf
https://www.nescon.medicina.ufmg.br/bibl... . The IBP and MHDI variables are based on the 2010 Census and cannot measure the socioeconomic variation of municipalities over the period, but they remained in the model as they measure social inequalities. It should also be noted that adherence to the PMAQ-AB by municipalities and teams to be evaluated was voluntary, so there may be a bias in the selection of teams, especially in the first cycle.

CONCLUSIONS

The study showed that better results in the quality of PHC in Brazil resulted in a reduction in ACSC hospitalization rates, reaffirming the need for progress in the country beyond the expansion of primary care coverage. It was also observed that places with greater socioeconomic deprivation had higher hospitalization rates, confirming that social inequality has an important relationship with health outcomes. The availability of beds per 1,000 inhabitants, especially in municipalities with fewer than 20,000 inhabitants, seems to be related to more hospitalizations.

The study also revealed the need to invest in evaluative programs considering the Donabedian triad on a national scale that can play a role in inducing improvements in PHC and indicating which directions to take towards a more effective and resolutive health system.

REFERENCES

Publication Dates

History