Estimates of hypertension and diabetes mellitus prevalence according to Health Vulnerability Index in Belo Horizonte, MG, Brazil

Crizian Saar Gomes Regina Tomie Ivata Bernal Alexandra Dias Moreira Renato Azeredo Teixeira Laís Santos de Magalhães Cardoso Antonio Luiz Pinho Ribeiro Deborah Carvalho Malta About the authors

ABSTRACT:

Objective:

To estimate the prevalences of hypertension and diabetes for small areas in Belo Horizonte, according to the Health Vulnerability Index (HVI).

Methods:

Ecological study with data from the Surveillance of Risk and Protection Factors for Chronic Diseases by Telephone Survey (Vigitel) conducted in Belo Horizonte, from 2006 to 2013. The self-reported diagnosis of diabetes and hypertension were evaluated. The estimates of prevalence and the 95% confidence interval (95%CI) were calculated using the direct and indirect method by HVI grouped into four categories: low, medium, high and very high health risk.

Results:

During the period evaluated, 26% (95%CI 25.2 – 26.8) and 6.1% (95%CI 6.7 – 6.5) of the adult population from Belo Horizonte reported being hypertensive and diabetic, respectively. According to the indirect method to obtain estimates of hypertension and diabetes prevalences per HVI, it was found that areas of very high risk had a higher prevalence of adults with hypertension (38.6%; 95%CI 34.8 – 42.4) and diabetes (16.2%; 95%CI 13.1 – 19.3) when compared to the low risk (28.2%; 95%CI 27.0 – 29.4 and 6%; 95%CI 5.4 – 6.7, respectively).

Conclusion:

The adult population living in areas at high risk for health had a higher prevalence of hypertension and diabetes compared to those with a lower risk.

Keywords:
Diabetes mellitus; Hypertension; Small-area analysis; Logistic models; Health surveys

INTRODUCTION

Noncommunicable diseases (NCD) are the main causes of morbidity and mortality worldwide, in addition to causing premature deaths, disabilities and high costs and financial burdens for individuals, societies, and health systems11. World Health Organization. Noncommunicable diseases [Internet]. Genebra: World Health Organization; 2018 [accessed on Oct. 15, 2020]. Available at: https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases
https://www.who.int/news-room/fact-sheet...
. In Brazil, NCD accounted for 75% of the causes of death in 2015, with four major groups of diseases standing out: cardiovascular, cancer, diabetes, and chronic respiratory diseases22. Malta DC, França E, Abreu DMX, Perillo RD, Salmen MC, Teixeira RA, et al. Mortality due to noncommunicable diseases in Brazil, 1990 to 2015, according to estimates from the Global Burden of Disease study. Sao Paulo Med J 2017; 135(3): 213-21. https://doi.org/10.1590/1516-3180.2016.0330050117
https://doi.org/10.1590/1516-3180.2016.0...
.

Studies point out that these diseases are multifactorial and emphasize the importance of social determinants in their occurrence, with worse indicators in the most vulnerable and socially marginalized population33. Barreto ML. Desigualdades em Saúde: uma perspectiva global [Internet]. Ciênc Saúde Coletiva 2017 [accessed on Oct. 15, 2020]; 22(7): 2097-108. https://doi.org/10.1590/1413-81232017227.02742017
https://doi.org/10.1590/1413-81232017227...
,44. Pearce N, Ebrahim S, McKee M, Lamptey P, Barreto ML, Matheson D, et al. Global prevention and control of NCDs: Limitations of the standard approach [Internet]. J Public Health Policy 2015 [accessed on Oct. 15, 2020]; 36: 408-25. https://doi.org/10.1057/jphp.2015.29
https://doi.org/10.1057/jphp.2015.29...
. An example is the fact that NCD more frequently affect low-income populations, as they are more vulnerable, exposed to risks, and have less access to health services and health promotion and disease prevention practices55. Abegunde DO, Mathers CD, Adam T, Ortegon M, Strong K. The burden and costs of chronic diseases in low-income and middle-income countries [Internet]. Lancet 2007 [accessed on Oct. 15, 2020]; 370(9603): 1929-38. https://doi.org/10.1016/S0140-6736(07)61696-1
https://doi.org/10.1016/S0140-6736(07)61...
.

NCD and their risk factors have been monitored in the country by large national surveys, which are essential to aggregate evidence on individual and collective health66. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Departamento de Análise em Saúde e Vigilância de Doenças Não Transmissíveis. Vigitel Brasil 2019: vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico: estimativas sobre frequência e distribuição sociodemográfica de fatores de risco e proteção para doenças crônicas nas capitais dos 26 estados brasileiros e no Distrito Federal em 2019. Brasília: Ministério da Saúde; 2020.,77. Instituto Brasileiro de Geografia e Estatística. Coordenação de Trabalho e Rendimento. Pesquisa Nacional de Saúde (PNS) 2013: acesso e utilização dos serviços de saúde, acidentes e violências: Brasil, grandes regiões e unidades da federação. Rio de Janeiro: IBGE; 2015.; however, national surveys do not allow estimates by municipalities or for smaller areas within the sampled cities. Thus, it is necessary to advance the monitoring of these diseases in small geographic areas, in order to identify the existing inequalities, which are often masked by national or state average statistics88. Bernal RTI, de Carvalho QH, Pell JP, Leyland AH, Dundas R, Barreto ML, et al. A methodology for small area prevalence estimation based on survey data [Internet]. Int J Equity Health 2020 [accessed on Oct. 15, 2020]; 19: 124. https://doi.org/10.1186/s12939-020-01220-5
https://doi.org/10.1186/s12939-020-01220...
,99. Allik M, Brown D, Dundas R, Leyland AH. Developing a new small-area measure of deprivation using 2001 and 2011 census data from Scotland. Health Place 2016; 39: 122-30. https://doi.org/10.1016/j.healthplace.2016.03.006
https://doi.org/10.1016/j.healthplace.20...
.

Estimates for small areas can help managers in public health in planning, defining intervention priorities and allocating resources, in addition to identifying possible decentralized solutions to the diagnosed problems, which may include social, cultural, legal, political or health-related approaches in small areas88. Bernal RTI, de Carvalho QH, Pell JP, Leyland AH, Dundas R, Barreto ML, et al. A methodology for small area prevalence estimation based on survey data [Internet]. Int J Equity Health 2020 [accessed on Oct. 15, 2020]; 19: 124. https://doi.org/10.1186/s12939-020-01220-5
https://doi.org/10.1186/s12939-020-01220...
,99. Allik M, Brown D, Dundas R, Leyland AH. Developing a new small-area measure of deprivation using 2001 and 2011 census data from Scotland. Health Place 2016; 39: 122-30. https://doi.org/10.1016/j.healthplace.2016.03.006
https://doi.org/10.1016/j.healthplace.20...
. The use of composite indicators in the health field, such as the Health Vulnerability Index (HVI), makes it possible to identify health inequalities and helps in the redesign of health care network and promotion of population development in various geographic scales of action, contributing to guide public policies in the definition of priorities for the allocation of resources99. Allik M, Brown D, Dundas R, Leyland AH. Developing a new small-area measure of deprivation using 2001 and 2011 census data from Scotland. Health Place 2016; 39: 122-30. https://doi.org/10.1016/j.healthplace.2016.03.006
https://doi.org/10.1016/j.healthplace.20...
,1010. Belo Horizonte. Secretaria Municipal de Saúde de Belo Horizonte. Gerência de Epidemiologia e Informação. Índice de Vulnerabilidade da Saúde 2012 [Internet]. Belo Horizonte: Secretaria Municipal de Saúde; 2013 [accessed on Oct. 15, 2020]. Available at: https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
https://prefeitura.pbh.gov.br/sites/defa...
.

The present study investigated arterial hypertension (AH), as it is considered the most prevalent risk factor for cardiovascular diseases, being responsible, in Brazil, for 2,283.48 (95%CI 2,050.77 – 2,496.19) disability-adjusted life years (DALYs)/100 thousand, in 20191111. Institute for Health Metrics and Evaluation. GBD Compare | IHME Viz Hub 2019. Institute for Health Metrics and Evaluation; 2019 [accessed on Nov. 20, 2020]. Available at: https://vizhub.healthdata.org/gbd-compare/
https://vizhub.healthdata.org/gbd-compar...
; and diabetes mellitus (DM), for being responsible for 1,076.51 (95%CI 915.36 – 1,256.42) DALYs/100 thousand, in 20191111. Institute for Health Metrics and Evaluation. GBD Compare | IHME Viz Hub 2019. Institute for Health Metrics and Evaluation; 2019 [accessed on Nov. 20, 2020]. Available at: https://vizhub.healthdata.org/gbd-compare/
https://vizhub.healthdata.org/gbd-compar...
, in addition to both presenting high prevalence in the Brazilian population66. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Departamento de Análise em Saúde e Vigilância de Doenças Não Transmissíveis. Vigitel Brasil 2019: vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico: estimativas sobre frequência e distribuição sociodemográfica de fatores de risco e proteção para doenças crônicas nas capitais dos 26 estados brasileiros e no Distrito Federal em 2019. Brasília: Ministério da Saúde; 2020.,77. Instituto Brasileiro de Geografia e Estatística. Coordenação de Trabalho e Rendimento. Pesquisa Nacional de Saúde (PNS) 2013: acesso e utilização dos serviços de saúde, acidentes e violências: Brasil, grandes regiões e unidades da federação. Rio de Janeiro: IBGE; 2015.. Thus, the study aimed to estimate the prevalences of AH and DM in small areas of Belo Horizonte, according to the HVI.

METHODS

This is an ecological study that used data from the Surveillance of Risk and Protection Factors for Chronic Diseases by Telephone Survey (Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico – Vigitel) system carried out in Belo Horizonte, from 2006 to 2013. More detailed information on data collection and Vigitel samples are available in a previous publication1212. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Secretaria de Gestão Estratégica e Participativa. Vigitel Brasil 2010: Vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico. Brasília: Ministério da Saúde; 2011..

The present study evaluated the self-reported medical diagnosis of DM and AH through positive replies to the questions:

  • Has any doctor ever told you that you have diabetes?;

  • Has any doctor ever told you that you have hypertension?

The HVI was considered in this study as the “small areas”. It was developed in 1998 and updated in 2012 with data from the 2010 census, by the Municipal Health Secretariat of Belo Horizonte to point out priority areas for intervention and allocation of resources, in addition to allowing the analysis of characteristics of population groups residing in census tracts1010. Belo Horizonte. Secretaria Municipal de Saúde de Belo Horizonte. Gerência de Epidemiologia e Informação. Índice de Vulnerabilidade da Saúde 2012 [Internet]. Belo Horizonte: Secretaria Municipal de Saúde; 2013 [accessed on Oct. 15, 2020]. Available at: https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
https://prefeitura.pbh.gov.br/sites/defa...
. It is a composite indicator that combines socioeconomic (residents per household, percentage of illiterate people, percentage of private households with a per capita income of up to half the minimum wage, average nominal income of the head of the household, percentage of mixed and black people, and indigenous) and environmental variables (sewage, water supply, and solid waste destination).

HVI ranged from 0 to 1 — values close to 1 indicate high social vulnerability and values close to 0 indicate low or nonexistent social vulnerability — and is categorized according to the following cut-off points:

  • low: HVI ≤ 0.1957;

  • medium: 0.1957 < HVI ≤ 0.2865;

  • high: 0.2865 < HVI ≤ 0.3782;

  • very high: HVI > 0.3782.

In 2012, the 3,830 sectors of Belo Horizonte were grouped into the following categories:

  • low risk: 1,330 – 34.7%;

  • médium risk: 1,460 – 38.1%;

  • high risk: 737 – 19.2%;

  • very high risk: 303 – 7.9%1010. Belo Horizonte. Secretaria Municipal de Saúde de Belo Horizonte. Gerência de Epidemiologia e Informação. Índice de Vulnerabilidade da Saúde 2012 [Internet]. Belo Horizonte: Secretaria Municipal de Saúde; 2013 [accessed on Oct. 15, 2020]. Available at: https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
    https://prefeitura.pbh.gov.br/sites/defa...
    .

In order to include the census tracts in the Vigitel databases, a linkage was made with the National Register of Addresses for Statistical Purposes of the 2010 census by Postal Codes1313. Instituto Brasileiro de Estatísticas. CNEFE: Cadastro Nacional de Endereços para Fins Estatísticos. Censo 2010 [Internet]. Instituto Brasileiro de Estatísticas [accessed on Aug. 10, 2020]. Available at: https://www.ibge.gov.br/estatisticas/downloads-estatisticas.html
https://www.ibge.gov.br/estatisticas/dow...
. Then, the HVI information by census tract was inserted. This procedure was performed in a data center with a high level of security — physical and virtual.

The prevalence of DM and AH and the respective 95% confidence intervals (95%CI) were estimated, according to the HVI, using the direct and indirect estimation methods for small areas1414. Rao JNK, Molina I. Small Area Estimation. 2ᵃ ed. Hoboken: John Wiley & Sons; 2015..

The direct estimation method consists of using the sample design variables to obtain the estimates in a smaller area. For joint analysis of Vigitel data, it was necessary to calculate post-stratification weights, using the rake method, to adjust the distribution of the Vigitel sample to the 2010 census population by HVI1515. Cervantes IF, Brick JM, Jones ME. Weighting for nontelephone household in the 2001 California Health Interview Survey. Joint Statistical Meetings – Section on Survey Research Methods; 2002.. The weights were calculated in the STATA program using the SURVWGT1616. Nick W. SURVWGT: Stata module to create and manipulate survey weights. Statistical Software Components. Boston: Boston College Department of Economics; 2018. package, requiring the sample weight information to execute the package:

(weight=number of adults in the householdnumber of telephones in the household)

The indirect estimation method consists of using statistical models to input the variable dichotomous response (Y), in the set of census tracts without any Vigitel interview. Of the total of 3,830 census tracts in Belo Horizonte, 513 (13.3%) had no Vigitel interview. Therefore, imputations of the outcome variables in these sectors were necessary.

For the construction of the logistic regression model, the census tracts with a single interview were selected (535 sectors). This criterion is due to the similarity in the distribution of sectors according to HVI in the group with one and no Vigitel interview. The sample of 535 sectors was divided into two subsamples of equal sizes, the first (sample 1) being used for the development of the logistic regression model and the second (sample 2) for its validation, in order to ensure that the model obtained in the first sample was robust.

The following covariables from the census tract, extracted from the 2010 census, were used to build the models: percentage of households without water supply, percentage of households without a bathroom, percentage of households with literate people, percentage of households with females, percentage of households with residents over the age of 55, percentage of households with non-white residents, percentage of households with income above one minimum wage, percentage of households with six or seven residents, percentage of households with eight or more residents.

For the imputation of the outcome variable in the set of sectors without any Vigitel interview, a cutoff point of 0.260 was considered for AH and 0.062 for DM. The adult resident in the sector with a predicted probability greater than or equal to the cutoff point was classified as 1 or 0, otherwise. This cut-off point was obtained in the group of census tracts with a single Vigitel interview.

To evaluate the model performance, the 2 × 2 classification matrix was used, with the proportion of correct classification given as: the answer is AH diagnosis and the model classified the subject as hypertensive, so it was classified as true positive (TP); the subject is not hypertensive and the model classified them as non-hypertensive, so they were classified as true negative (TN). On the other hand, the proportion of incorrect classification was given as: the response category is hypertensive and the model classified the individual as non-hypertensive, that is, as a false negative (FN); and the response category is non-hypertensive and the model classified it as hypertensive, called false positive (FP). The sensitivity of the model was defined by TPTP+FN, specificity by TNTN+FP, and accuracy by TP+TNTP+FN+TP+FN.

After imputing data on outcomes in the census tracts without interviews, the prevalence of outcomes by HVI was estimated. Post-stratification weights were calculated to adjust the sample distribution for the 2010 census population by HVI, using the rake method1515. Cervantes IF, Brick JM, Jones ME. Weighting for nontelephone household in the 2001 California Health Interview Survey. Joint Statistical Meetings – Section on Survey Research Methods; 2002.. These weights were calculated in the STATA version 14.0 program using the SURVWGT1616. Nick W. SURVWGT: Stata module to create and manipulate survey weights. Statistical Software Components. Boston: Boston College Department of Economics; 2018. package, requiring the sample weight information to execute the package. In this study, data from the N1, N2, and N populations extracted from the 2010 Belo Horizonte census were considered for calculating the weight of the group of sectors with and without Vigitel interviews — (weight=NN1) and (weight=NN2), respectively —, with N being the total adult population aged 18 years old and older, N1 the total of adults in the sectors with Vigitel interviews, and N2 the total of adults in the sectors without Vigitel interviews.

More detailed information on the direct and indirect methods of estimation for small areas employed can be obtained in another publication88. Bernal RTI, de Carvalho QH, Pell JP, Leyland AH, Dundas R, Barreto ML, et al. A methodology for small area prevalence estimation based on survey data [Internet]. Int J Equity Health 2020 [accessed on Oct. 15, 2020]; 19: 124. https://doi.org/10.1186/s12939-020-01220-5
https://doi.org/10.1186/s12939-020-01220...
.

To assess the differences between the prevalence of AH or DM by IVS, the Student's t test was used, with a significance level of 5%.

The present study was developed according to Resolution No. 466/2012, of the National Health Council, and is integrated with the research project entitled “Inequalities in small geographical areas of the indicators of noncommunicable diseases, violence, and its risk factors (Desigualdades em pequenas áreas geográficas dos indicadores de doenças crônicas não transmissíveis, violências e seus fatores de risco)”, approved by the Research Ethics Committee of Universidade Federal de Minas Gerais.

RESULTS

Between 2006 and 2013, Vigitel interviewed 15,833 adults living in the city of Belo Horizonte, with 14,174 (90%) geocoded interviews.

During the study period, 26% (95%CI 25.2 – 26.8) and 6.1% (95%CI 6.7 – 6.5) of the adult population in Belo Horizonte reported being hypertensive and diabetic, respectively. The prevalences over the years are shown in Table 1.

Table 1
Estimated prevalence of arterial hypertension and diabetes mellitus in adults aged 18 years old and older, according to year, Belo Horizonte, MG, Brazil. Vigitel, 2006 to 2013.

According to the adjusted logistic regression model, the chance of an adult being classified as hypertensive is 0.006 (exponential −5.103). Three out of a total of nine variables showed associations with AH. The percentage of households with eight or more residents and the percentage of households with females were associated with an increased likelihood that an adult in the census tract would be classified as hypertensive. In contrast, the variable percentage of households with residents over the age of 55 decreased the likelihood of an adult in the census tract being classified as hypertensive (Table 2).

Table 2
Analysis of logistic regression for hypertension and diabetes mellitus, Belo Horizonte, MG, Brazil. Vigitel, 2006 to 2013.

The chance of an adult being classified as diabetic is 0.008 (exponential of −4.745). Three variables were associated with DM — two (percentage of households with eight or more residents and percentage of households with non-white residents) increased and one (percentage of households with six or seven residents) decreased the likelihood of an adult from the census tract to be classified as diabetic (Table 2).

In assessing the adjusted model for AH, the cut-off point of 26% was used to classify the census tracts without an interview as hypertensive or non-hypertensive. In sample 1, sensitivity was 64%, specificity 53%, and accuracy 56%. In sample 2, the sensitivity, specificity, and accuracy values were 51, 55, and 54%, respectively. In assessing the accuracy of the adjusted model, it was noted that the results of the two samples are very close. This indicates that the model is consistent.

In sample 1, the model performed better in the hypertensive than in the non-hypertensive category, while in sample 2 the performance in the non-hypertensive category was greater when compared to the result of sample 1. On the other hand, in sample 2, it was observed that the performance in the hypertensive category was lower than that found in sample 1. This variation was expected due to the use of one of the samples to build the model. In general, the performance of the model's correctness in the validation sample is lower when compared to the result of the training sample. These results suggest that the adjusted logistic regression model is consistent and reasonably accurate, 56 and 54%, for samples 1 and 2, respectively (Table 3).

Table 3
Accuracy of the logistic regression model for hypertension and diabetes mellitus, Belo Horizonte, MG, Brazil. Vigitel, 2006 to 2013.

For DM, the cut-off point of 6.2% was used to classify census tracts without interview as diabetic or non-diabetic. In sample 1, the sensitivity was equal to 62%, with specificity of 65% and accuracy of 64.8%. In sample 2, sensitivity, specificity, and accuracy were 53, 67, and 66.4%, respectively. In assessing the accuracy of the adjusted model, it was noted that the results of the two samples are very close. This indicates that the model is consistent.

In sample 1, the model performed well in the non-diabetic and diabetic classification, while in sample 2 the performance in the non-diabetic category was higher when compared to the result of sample 1. On the other hand, it was observed that the performance in the diabetic category reduced in sample 2. This variation was expected by the use of one of the samples to build the model. In general, the performance of the model's correctness in the validation sample is lower when compared to the result of the training sample. These results indicate that the adjusted logistic regression model is consistent and has good accuracy, 64.8 and 66.4% for samples 1 and 2, respectively (Table 3).

Tables 4 and 5 compare the prevalence estimates for AH and DM according to the HVI, using the direct and indirect methods of estimation in small areas. According to the indirect method, in the 2006 to 2013 period, the areas of very high risk had a higher prevalence of hypertensive adults (38.6%; 95%CI 34.8 – 42.4) when compared to the low (28.2%; 95%CI 27.0 – 29.4) and medium risk ones (33.4%; 95%CI 32.1 – 34.6). In turn, the areas of very high risk had a higher prevalence (16.2%; 95%CI 13.1 – 19.3) of DM when compared to the medium (9.7%; 95%CI 8.9 – 10.5) and low risk areas (6%; 95%CI 5.4 – 6.7). In the direct method the prevalences did not differ statistically among the HVI strata, and it was observed that the prevalence of AH ranged from 26% (95%CI 24 – 27) for low risk to 30% (95%CI 28 – 32) for high risk, and DM ranged from 5.9% (95%CI 5.2 – 6.6) to 7.7% (95%CI 6.5 – 9.0).

Table 4
p-value of the differences between the prevalence of hypertension and diabetes mellitus between the categories of the health vulnerability index, Belo Horizonte, MG, Minas Gerais. Vigitel, 2006 to 2013. Estimated prevalence of hypertension and diabetes mellitus according to the health vulnerability index using the direct and indirect estimation methods in small areas, Belo Horizonte, MG, Brazil. Vigitel, 2006 to 2013.
Table 5
p-value of the differences between the prevalence of hypertension and diabetes mellitus between the categories of the health vulnerability index, Belo Horizonte, MG, Minas Gerais. Vigitel, 2006 to 2013.

DISCUSSION

The present study estimated the prevalence of AH and DM, according to the HVI, in Belo Horizonte. The direct and indirect methods of estimation in small areas were compared. It was found that areas with high risk had a higher prevalence of AH and DM compared to those with low risk, according to the indirect method of estimation in small areas. The direct method showed no prevalence differences among the areas.

Previous studies have shown that less economically privileged areas have worse health conditions when compared to more privileged areas both between countries55. Abegunde DO, Mathers CD, Adam T, Ortegon M, Strong K. The burden and costs of chronic diseases in low-income and middle-income countries [Internet]. Lancet 2007 [accessed on Oct. 15, 2020]; 370(9603): 1929-38. https://doi.org/10.1016/S0140-6736(07)61696-1
https://doi.org/10.1016/S0140-6736(07)61...
and within the same country, revealing the unequal geographical distribution of diseases1717. Matozinhos FP, Velásquez-Meléndez G, Pessoa MC, Mendes LL, Gomes CS, Costa MA. Distribuição espacial da obesidade em área urbana no Brasil [Internet]. Ciênc Saúde Coletiva 2015 [accessed on Oct. 15, 2020]; 20(9): 2779-86. https://doi.org/10.1590/1413-81232015209.00442015
https://doi.org/10.1590/1413-81232015209...
2020. Lai S, Shen C, Yang X, Zhang X, Xu Y, Li Q, et al. Socioeconomic inequalities in the prevalence of chronic diseases and preventive care among adults aged 45 and older in Shaanxi Province, China. BMC Public Health [Internet] 2019 [accessed on Oct. 15, 2020]; 19: 1460. https://doi.org/10.1186/s12889-019-7835-5
https://doi.org/10.1186/s12889-019-7835-...
.

Inequalities in the health status between groups with different socioeconomic status are configured as a public health problem in several locations around the world2020. Lai S, Shen C, Yang X, Zhang X, Xu Y, Li Q, et al. Socioeconomic inequalities in the prevalence of chronic diseases and preventive care among adults aged 45 and older in Shaanxi Province, China. BMC Public Health [Internet] 2019 [accessed on Oct. 15, 2020]; 19: 1460. https://doi.org/10.1186/s12889-019-7835-5
https://doi.org/10.1186/s12889-019-7835-...
. It is known that worse socioeconomic conditions are associated with unhealthy behaviors, lack of knowledge about self-care2020. Lai S, Shen C, Yang X, Zhang X, Xu Y, Li Q, et al. Socioeconomic inequalities in the prevalence of chronic diseases and preventive care among adults aged 45 and older in Shaanxi Province, China. BMC Public Health [Internet] 2019 [accessed on Oct. 15, 2020]; 19: 1460. https://doi.org/10.1186/s12889-019-7835-5
https://doi.org/10.1186/s12889-019-7835-...
2222. Grintsova O, Maier W, Mielck A. Inequalities in health care among patients with type 2 diabetes by individual socio-economic status (SES) and regional deprivation: a systematic literature review. Int J Equity Health [Internet] 2014 [accessed on Oct. 15, 2020]; 13: 43. https://doi.org/10.1186/1475-9276-13-43
https://doi.org/10.1186/1475-9276-13-43...
, barriers to healthy eating and unsafe spaces that perpetuate structural inequalities2323. Durfey S, Kind A, Buckingham WR, DuGoff EH, Trivedi AN. Neighborhood disadvantage and chronic disease management. Health Serv Res 2019; 54(Supl. 1): 206-16. https://doi.org/10.1111/1475-6773.13092
https://doi.org/10.1111/1475-6773.13092...
, which could explain the higher occurrence of NCD in these areas. Previous studies also point out that the unfavorable socioeconomic situation and the regional deprivation of individuals with DM and AH are related to the poorer quality of care, which directly interferes with the results of the control of these diseases, leading to a higher incidence of complications2222. Grintsova O, Maier W, Mielck A. Inequalities in health care among patients with type 2 diabetes by individual socio-economic status (SES) and regional deprivation: a systematic literature review. Int J Equity Health [Internet] 2014 [accessed on Oct. 15, 2020]; 13: 43. https://doi.org/10.1186/1475-9276-13-43
https://doi.org/10.1186/1475-9276-13-43...
2424. Rebolledo ES, Chiaravalloti Neto F, Meza MTE, Giatti LL. Variación espacial de la cobertura del tratamiento de hipertensión arterial, en el municipio de Santiago de Chile. Salud Colect [Internet] 2018 [accessed on Oct. 15, 2020]; 14(1): 5-17. https://doi.org/10.18294/sc.2018.1181
https://doi.org/10.18294/sc.2018.1181...
.

Research that identifies regional health disparities is essential for promoting social programs and policies aimed at socially vulnerable groups. In addition to individual factors, environmental factors are directly associated with lifestyle and the occurrence of NCD2323. Durfey S, Kind A, Buckingham WR, DuGoff EH, Trivedi AN. Neighborhood disadvantage and chronic disease management. Health Serv Res 2019; 54(Supl. 1): 206-16. https://doi.org/10.1111/1475-6773.13092
https://doi.org/10.1111/1475-6773.13092...
. Thus, analyses that consider the context in which the individual is inserted are fundamental for the recognition of social contexts in the determination and control of diseases, contributing to the planning of prevention and treatment actions.

The indirect method of estimation for small areas used in this study is highlighted, which used the imputation of missing data in areas that do not have Vigitel telephone interviews, avoiding the under or overestimation of the prevalences of chronic diseases by HVI obtained by the direct method, which considers only the sectors with Vigitel interviews. An example is the presence of AH and DM, which is strongly influenced by age, and in the areas of greater health vulnerability there is a smaller number of aged people compared to the most economically privileged locations2525. Moreira APL, Malta DC, Vianna RPT, Moreira PVL, Carvalho AT. Risk and protection factors for self-reported hypertension and diabetes in João Pessoa, Brazil. The VIGITEL survey, 2014. A cross-sectional study. São Paulo Med J [Internet] 2017 [accessed on Oct. 15, 2020]; 135(5): 450-61. https://doi.org/10.1590/1516-3180.2017.0044250517
https://doi.org/10.1590/1516-3180.2017.0...
2727. Malta DC, Bernal RTI, Andrade SSCA, Silva MMA, Velasquez-Melendez G. Prevalence of and factors associated with self-reported high blood pressure in Brazilian adults. Rev Saúde Pública [Internet] 2017 [accessed on Oct. 15, 2020]; 51(Supl. 1): 11s. https://doi.org/10.1590/S1518-8787.2017051000006
https://doi.org/10.1590/S1518-8787.20170...
.

The HVI for estimating disease prevalence in small areas also deserves to be highlighted. This index takes into account the census tracts, which are the smallest territorial division adopted by the Brazilian Institute of Geography and Statistics. As it is composed of socioeconomic and environmental variables, the HVI allows the identification of inequities in census tracts and the prioritization of areas in the planning of health actions by managers1010. Belo Horizonte. Secretaria Municipal de Saúde de Belo Horizonte. Gerência de Epidemiologia e Informação. Índice de Vulnerabilidade da Saúde 2012 [Internet]. Belo Horizonte: Secretaria Municipal de Saúde; 2013 [accessed on Oct. 15, 2020]. Available at: https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
https://prefeitura.pbh.gov.br/sites/defa...
.

In addition to this, primary health care (PHC) emerges as a facilitator in monitoring individuals and reducing regional inequalities, enabling the performance of health professionals in the territories in different contexts. The territories are subdivided into coverage areas composed of contiguous census tracts, allowing Family Health Strategy teams to hold information on the population and propose local interventions.

This study has some limitations. Firstly, the exclusion of 10.5% of Vigitel's interviews due to the non-identification of census tracts by linkage, which could lead to a selection bias, however the use of post-stratification weights, according to the 2010 census population by HVI, minimizes potential bias. Secondly, the accuracy of the adjusted model for imputing the variable hypertensive or non-hypertensive adults in the set of sectors without any Vigitel interview was between 50 and 60%, which is considered acceptable, but subject to under or overestimation of the outcome. Thirdly, the information on DM and AH is self-reported by the participants, which can overestimate or underestimate the prevalence of the outcomes, but direct diagnostic measures, obtained by laboratory and clinical tests, are difficult to be performed in large populations, in addition to being costly, and studies of validation comparing self-reported and clinical measures indicate good reliability results2828. Fontanelli MDM, Teixeira JA, Sales CH, Castro MAD, Cesar CLG, Alves MCGP, et al. Validation of self-reported diabetes in a representative sample of São Paulo city. Rev Saúde Pública 2017; 51: 20. https://doi.org/10.1590/S1518-8787.2017051006378
https://doi.org/10.1590/S1518-8787.20170...
. Fourthly, to estimate the prevalence of the indicators, data from 2006 to 2013 were aggregated, which reflects the mean in the period and, consequently, the loss of the trend over the years.

Finally, the estimates obtained by the indirect method require external validation of the results found in this article. It is emphasized that Vigitel's external validation study, compared to the household survey, has already been carried out for the outcome of tobacco use by HVI and has identified acceptable values88. Bernal RTI, de Carvalho QH, Pell JP, Leyland AH, Dundas R, Barreto ML, et al. A methodology for small area prevalence estimation based on survey data [Internet]. Int J Equity Health 2020 [accessed on Oct. 15, 2020]; 19: 124. https://doi.org/10.1186/s12939-020-01220-5
https://doi.org/10.1186/s12939-020-01220...
. Thus, it is also recommended to proceed with an external validation study to estimate the prevalence of chronic diseases for small areas using Vigitel data.

It was concluded that high health risk areas have higher AH and DM prevalences than low risk areas. The results demonstrate that investigations carried out for all municipalities, without taking into account regional differences, can bring biased estimates of the health situation in vulnerable areas and may not identify inequalities in the population. Producing reliable estimates for small regions allows to know and locate differences in the distribution of health events in the territory. Thereby, it contributes to the understanding of local realities, being an important strategy for the identification of areas of greater vulnerability and for the redesign of a health care network and social promotion, at different geographical scales, considering the territory, a principle that governs PHC, in addition to assisting managers in planning public policies according to the community needs.

  • Financial support: Secretariat of Health Surveillance, Ministry of Health, TED 148-2018, project “Desigualdades em pequenas áreas geográficas dos indicadores de doenças crônicas não transmissíveis, violências e seus fatores de risco”; Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (310679/2016-8 and 465518/2014-1); Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) (PPM-00428-17 and RED-00081-16); Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).

REFERENCES

  • 1
    World Health Organization. Noncommunicable diseases [Internet]. Genebra: World Health Organization; 2018 [accessed on Oct. 15, 2020]. Available at: https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases
    » https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases
  • 2
    Malta DC, França E, Abreu DMX, Perillo RD, Salmen MC, Teixeira RA, et al. Mortality due to noncommunicable diseases in Brazil, 1990 to 2015, according to estimates from the Global Burden of Disease study. Sao Paulo Med J 2017; 135(3): 213-21. https://doi.org/10.1590/1516-3180.2016.0330050117
    » https://doi.org/10.1590/1516-3180.2016.0330050117
  • 3
    Barreto ML. Desigualdades em Saúde: uma perspectiva global [Internet]. Ciênc Saúde Coletiva 2017 [accessed on Oct. 15, 2020]; 22(7): 2097-108. https://doi.org/10.1590/1413-81232017227.02742017
    » https://doi.org/10.1590/1413-81232017227.02742017
  • 4
    Pearce N, Ebrahim S, McKee M, Lamptey P, Barreto ML, Matheson D, et al. Global prevention and control of NCDs: Limitations of the standard approach [Internet]. J Public Health Policy 2015 [accessed on Oct. 15, 2020]; 36: 408-25. https://doi.org/10.1057/jphp.2015.29
    » https://doi.org/10.1057/jphp.2015.29
  • 5
    Abegunde DO, Mathers CD, Adam T, Ortegon M, Strong K. The burden and costs of chronic diseases in low-income and middle-income countries [Internet]. Lancet 2007 [accessed on Oct. 15, 2020]; 370(9603): 1929-38. https://doi.org/10.1016/S0140-6736(07)61696-1
    » https://doi.org/10.1016/S0140-6736(07)61696-1
  • 6
    Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Departamento de Análise em Saúde e Vigilância de Doenças Não Transmissíveis. Vigitel Brasil 2019: vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico: estimativas sobre frequência e distribuição sociodemográfica de fatores de risco e proteção para doenças crônicas nas capitais dos 26 estados brasileiros e no Distrito Federal em 2019. Brasília: Ministério da Saúde; 2020.
  • 7
    Instituto Brasileiro de Geografia e Estatística. Coordenação de Trabalho e Rendimento. Pesquisa Nacional de Saúde (PNS) 2013: acesso e utilização dos serviços de saúde, acidentes e violências: Brasil, grandes regiões e unidades da federação. Rio de Janeiro: IBGE; 2015.
  • 8
    Bernal RTI, de Carvalho QH, Pell JP, Leyland AH, Dundas R, Barreto ML, et al. A methodology for small area prevalence estimation based on survey data [Internet]. Int J Equity Health 2020 [accessed on Oct. 15, 2020]; 19: 124. https://doi.org/10.1186/s12939-020-01220-5
    » https://doi.org/10.1186/s12939-020-01220-5
  • 9
    Allik M, Brown D, Dundas R, Leyland AH. Developing a new small-area measure of deprivation using 2001 and 2011 census data from Scotland. Health Place 2016; 39: 122-30. https://doi.org/10.1016/j.healthplace.2016.03.006
    » https://doi.org/10.1016/j.healthplace.2016.03.006
  • 10
    Belo Horizonte. Secretaria Municipal de Saúde de Belo Horizonte. Gerência de Epidemiologia e Informação. Índice de Vulnerabilidade da Saúde 2012 [Internet]. Belo Horizonte: Secretaria Municipal de Saúde; 2013 [accessed on Oct. 15, 2020]. Available at: https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
    » https://prefeitura.pbh.gov.br/sites/default/files/estrutura-de-governo/saude/2018/publicacaoes-da-vigilancia-em-saude/indice_vulnerabilidade2012.pdf
  • 11
    Institute for Health Metrics and Evaluation. GBD Compare | IHME Viz Hub 2019. Institute for Health Metrics and Evaluation; 2019 [accessed on Nov. 20, 2020]. Available at: https://vizhub.healthdata.org/gbd-compare/
    » https://vizhub.healthdata.org/gbd-compare/
  • 12
    Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Secretaria de Gestão Estratégica e Participativa. Vigitel Brasil 2010: Vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico. Brasília: Ministério da Saúde; 2011.
  • 13
    Instituto Brasileiro de Estatísticas. CNEFE: Cadastro Nacional de Endereços para Fins Estatísticos. Censo 2010 [Internet]. Instituto Brasileiro de Estatísticas [accessed on Aug. 10, 2020]. Available at: https://www.ibge.gov.br/estatisticas/downloads-estatisticas.html
    » https://www.ibge.gov.br/estatisticas/downloads-estatisticas.html
  • 14
    Rao JNK, Molina I. Small Area Estimation. 2ᵃ ed. Hoboken: John Wiley & Sons; 2015.
  • 15
    Cervantes IF, Brick JM, Jones ME. Weighting for nontelephone household in the 2001 California Health Interview Survey. Joint Statistical Meetings – Section on Survey Research Methods; 2002.
  • 16
    Nick W. SURVWGT: Stata module to create and manipulate survey weights. Statistical Software Components. Boston: Boston College Department of Economics; 2018.
  • 17
    Matozinhos FP, Velásquez-Meléndez G, Pessoa MC, Mendes LL, Gomes CS, Costa MA. Distribuição espacial da obesidade em área urbana no Brasil [Internet]. Ciênc Saúde Coletiva 2015 [accessed on Oct. 15, 2020]; 20(9): 2779-86. https://doi.org/10.1590/1413-81232015209.00442015
    » https://doi.org/10.1590/1413-81232015209.00442015
  • 18
    Höfelmann DA, Gonzalez-Chica DA, Peres KG, Boing AF, Peres MA. Chronic diseases and socioeconomic inequalities in quality of life among Brazilian adults: findings from a population-based study in Southern Brazil [Internet]. Eur J Public Health 2018 [accessed on Oct. 15, 2020]; 28(4): 603-10. https://doi.org/10.1093/eurpub/ckx224
    » https://doi.org/10.1093/eurpub/ckx224
  • 19
    Kauhl B, Maier W, Schweikart J, Keste A, Moskwyn M. Exploring the small-scale spatial distribution of hypertension and its association to area deprivation based on health insurance claims in Northeastern Germany. BMC Public Health [Internet] 2018 [accessed on Oct. 15, 2020]; 18: 121. https://doi.org/10.1186/s12889-017-5017-x
    » https://doi.org/10.1186/s12889-017-5017-x
  • 20
    Lai S, Shen C, Yang X, Zhang X, Xu Y, Li Q, et al. Socioeconomic inequalities in the prevalence of chronic diseases and preventive care among adults aged 45 and older in Shaanxi Province, China. BMC Public Health [Internet] 2019 [accessed on Oct. 15, 2020]; 19: 1460. https://doi.org/10.1186/s12889-019-7835-5
    » https://doi.org/10.1186/s12889-019-7835-5
  • 21
    Hunter DJ, Reddy KS. Noncommunicable diseases. N Engl J Med [Internet] 2013 [accessed on Oct. 15, 2020]; 369(14): 1336-43. https://doi.org/10.1056/NEJMra1109345
    » https://doi.org/10.1056/NEJMra1109345
  • 22
    Grintsova O, Maier W, Mielck A. Inequalities in health care among patients with type 2 diabetes by individual socio-economic status (SES) and regional deprivation: a systematic literature review. Int J Equity Health [Internet] 2014 [accessed on Oct. 15, 2020]; 13: 43. https://doi.org/10.1186/1475-9276-13-43
    » https://doi.org/10.1186/1475-9276-13-43
  • 23
    Durfey S, Kind A, Buckingham WR, DuGoff EH, Trivedi AN. Neighborhood disadvantage and chronic disease management. Health Serv Res 2019; 54(Supl. 1): 206-16. https://doi.org/10.1111/1475-6773.13092
    » https://doi.org/10.1111/1475-6773.13092
  • 24
    Rebolledo ES, Chiaravalloti Neto F, Meza MTE, Giatti LL. Variación espacial de la cobertura del tratamiento de hipertensión arterial, en el municipio de Santiago de Chile. Salud Colect [Internet] 2018 [accessed on Oct. 15, 2020]; 14(1): 5-17. https://doi.org/10.18294/sc.2018.1181
    » https://doi.org/10.18294/sc.2018.1181
  • 25
    Moreira APL, Malta DC, Vianna RPT, Moreira PVL, Carvalho AT. Risk and protection factors for self-reported hypertension and diabetes in João Pessoa, Brazil. The VIGITEL survey, 2014. A cross-sectional study. São Paulo Med J [Internet] 2017 [accessed on Oct. 15, 2020]; 135(5): 450-61. https://doi.org/10.1590/1516-3180.2017.0044250517
    » https://doi.org/10.1590/1516-3180.2017.0044250517
  • 26
    Santos CES, Rech CR, Antes DL, Schneider IJC, d’Orsi E, Benedetti TRB. Incidence and prevalence of diabetes self-reported on elderly in south of Brazil: results of EpiFloripa Ageing Study. Ciên Saúde Coletiva [Internet] 2019 [accessed on Oct. 15, 2020]; 24(11): 4191-200. https://doi.org/10.1590/1413-812320182411.31092017
    » https://doi.org/10.1590/1413-812320182411.31092017
  • 27
    Malta DC, Bernal RTI, Andrade SSCA, Silva MMA, Velasquez-Melendez G. Prevalence of and factors associated with self-reported high blood pressure in Brazilian adults. Rev Saúde Pública [Internet] 2017 [accessed on Oct. 15, 2020]; 51(Supl. 1): 11s. https://doi.org/10.1590/S1518-8787.2017051000006
    » https://doi.org/10.1590/S1518-8787.2017051000006
  • 28
    Fontanelli MDM, Teixeira JA, Sales CH, Castro MAD, Cesar CLG, Alves MCGP, et al. Validation of self-reported diabetes in a representative sample of São Paulo city. Rev Saúde Pública 2017; 51: 20. https://doi.org/10.1590/S1518-8787.2017051006378
    » https://doi.org/10.1590/S1518-8787.2017051006378

Publication Dates

  • Publication in this collection
    16 Apr 2021
  • Date of issue
    2021

History

  • Received
    20 Oct 2020
  • Reviewed
    02 Dec 2020
  • Accepted
    12 Dec 2020
  • Preprint
    15 Dec 2020
Associação Brasileira de Pós -Graduação em Saúde Coletiva São Paulo - SP - Brazil
E-mail: revbrepi@usp.br