0271/2024 - Persistência temporal e espacial de casos de tuberculose nos municípios brasileiros entre 2001 e 2022
Temporal and spatial persistence of tuberculosis cases in Brazilian municipalities between 2001 and 2022
Autor:
• Naanda Kaanna Matos de Souza - Souza, N. K. M. - <naanda.kaanna@gmail.com>ORCID: https://orcid.org/0000-0001-5392-175X
Coautor(es):
• Lucas Dias Soares Machado - Machado, L. D. S. - <lucasdsmachado@hotmail.com>ORCID: https://orcid.org/0000-0003-4450-3796
• Denis Fernandes Alves - Alves, D. F. - <denis_fernandes@outlook.com>
• Luis Abel Silva Filho - Silva Filho, L. A. - <abeleconomia@hotmail.com>
• Viviane Martins Silva - Silva, V. M. - <viviane.silva@ufc.br>
ORCID: https://orcid.org/0000-0002-8033-8831
Resumo:
A tuberculose (TB) é um problema de saúde pública por sua distribuição persistente a nível mundial, estando fortemente influenciada pelos determinantes sociais da saúde. Objetivou-se analisar a associação entre variáveis socioeconômicas e a persistência temporal e espacial das taxas de TB notificadas nos municípios brasileiros entre os anos 2001 e 2022. Estudo ecológico de séries temporais e análise espacial realizado a partir de dados secundários de fontes públicas. Buscou-se a correlação espacial entre as taxas de TB e variáveis socioeconômicas nos municípios brasileiros e os seus efeitos diretos e indiretos sobre essas taxas. Recorreu-se aos Índices de Moran Global e Local univariado e bivariado, bem como ao Modelo de Erro Espacial de Durbin. Os resultados mostram que um aumento no PIB per capita e nos gastos com saúde impactam na redução da taxa de casos de TB. Outrossim, há uma correlação direta entre as taxas de TB e as variáveis densidade demográfica e taxa de profissionais de saúde. Concluiu-se que, apesar de sua multicausalidade, a pobreza é um forte determinante no aumento das taxas de TB. A redução dos casos da doença nos municípios brasileiros depende de políticas públicas em saúde e ações intersetoriais que sejam direcionadas a cada espaço geográfico com base em suas particularidades.Palavras-chave:
Saúde Pública. Doenças Negligenciadas. Tuberculose. Estudos Ecológicos.Abstract:
Tuberculosis (TB) is a public health problem due to its persistent distribution worldwide, being strongly influenced by social determinants of health. The objective was to analyze the association between socioeconomic variables and the temporal and spatial persistence of TB rates reported in Brazilian municipalities between the years 2001 and 2022. An ecological study of time series and spatial analysis was conducted using secondary datapublic sources. Spatial correlation was sought between TB rates and socioeconomic variables in Brazilian municipalities and their direct and indirect effects on these rates. Global and Local Moran\'s Indices, both univariate and bivariate, as well as the Durbin Spatial Error Model, were used. The results show that an increase in per capita GDP and health expenditure impacts the reduction of TB cases. Additionally, there is a direct correlation between TB rates and demographic density and the rate of healthcare professionals. It was concluded that, despite its multifactorial nature, poverty is a strong determinant in the increase of TB rates. The reduction of TB cases in Brazilian municipalities depends on public health policies and intersectoral actions that are targeted to each geographical area based on its specific characteristics.Keywords:
Public Health. Neglected Diseases. Tuberculosis. Ecological StudiesConteúdo:
Acessar Revista no ScieloOutros idiomas:
Temporal and spatial persistence of tuberculosis cases in Brazilian municipalities between 2001 and 2022
Resumo (abstract):
Tuberculosis (TB) is a public health problem due to its persistent distribution worldwide, being strongly influenced by social determinants of health. The objective was to analyze the association between socioeconomic variables and the temporal and spatial persistence of TB rates reported in Brazilian municipalities between the years 2001 and 2022. An ecological study of time series and spatial analysis was conducted using secondary datapublic sources. Spatial correlation was sought between TB rates and socioeconomic variables in Brazilian municipalities and their direct and indirect effects on these rates. Global and Local Moran\'s Indices, both univariate and bivariate, as well as the Durbin Spatial Error Model, were used. The results show that an increase in per capita GDP and health expenditure impacts the reduction of TB cases. Additionally, there is a direct correlation between TB rates and demographic density and the rate of healthcare professionals. It was concluded that, despite its multifactorial nature, poverty is a strong determinant in the increase of TB rates. The reduction of TB cases in Brazilian municipalities depends on public health policies and intersectoral actions that are targeted to each geographical area based on its specific characteristics.Palavras-chave (keywords):
Public Health. Neglected Diseases. Tuberculosis. Ecological StudiesLer versão inglês (english version)
Conteúdo (article):
Persistência temporal e espacial de casos de tuberculose nos municípios brasileiros entre 2001 e 2022Temporal and spatial persistence of tuberculosis cases in Brazilian municipalities between 2001 and 2022
AUTHORS:
Naanda Kaanna Matos de Souza – Universidade Federal do Ceará (UFC). Email:
Lucas Dias Soares Machado – Instituto Federal de Educação, Ciência e Tecnologia da Paraíba (IFPB). Email:
ORCID:
Denis Fernandes Alves – Universidade Federal do Pernambuco (UFPE). Email:
Luis Abel da Silva Filho – Universidade Regional do Cariri (URCA). Email:
Viviane Martins da Silva – Universidade Federal do Ceará (UFC). Email:
RESUMO
A tuberculose (TB) é um problema de saúde pública por sua distribuição persistente a nível mundial, estando fortemente influenciada pelos determinantes sociais da saúde. Objetivou-se analisar a associação entre variáveis socioeconômicas e a persistência temporal e espacial das taxas de TB notificadas nos municípios brasileiros entre os anos 2001 e 2022. Estudo ecológico de séries temporais e análise espacial realizado a partir de dados secundários de fontes públicas. Buscou-se a correlação espacial entre as taxas de TB e variáveis socioeconômicas nos municípios brasileiros e os seus efeitos diretos e indiretos sobre essas taxas. Recorreu-se aos Índices de Moran Global e Local univariado e bivariado, bem como ao Modelo de Erro Espacial de Durbin. Os resultados mostram que um aumento no PIB per capita e nos gastos com saúde impactam na redução da taxa de casos de TB. Outrossim, há uma correlação direta entre as taxas de TB e as variáveis densidade demográfica e taxa de profissionais de saúde. Concluiu-se que, apesar de sua multicausalidade, a pobreza é um forte determinante no aumento das taxas de TB. A redução dos casos da doença nos municípios brasileiros depende de políticas públicas em saúde e ações intersetoriais que sejam direcionadas a cada espaço geográfico com base em suas particularidades.
Palavras-chave: Saúde Pública. Doenças Negligenciadas. Tuberculose. Estudos Ecológicos.
ABSTRACT
Tuberculosis (TB) is a public health problem due to its persistent distribution worldwide, mainly in developing countries, as it is strongly influenced by social determinants of health. The objective was to analyse the association between socio-economic variables and the temporal and spatial persistence of TB rates reported in Brazilian municipalities between 2001 and 2022. An ecological study of time series and spatial analysis was conducted using secondary data from public sources. Spatial correlation was sought between TB rates and socio-economic variables in Brazilian municipalities, and their direct and indirect effects on these rates. Global and Local Moran Indexes, both univariate and bivariate, as well as the Durbin Spatial Error Model, were used. The results show that an increase in GDP per capita and health expenditure impacts the reduction in TB cases. Additionally, there is a direct correlation between TB rates and population density and the number of healthcare professionals. It was concluded that, despite its multifactorial nature, poverty is a strong determinant of increased TB rates. Reduction of TB cases in Brazilian municipalities depends on public health policies and intersectoral action targeting each geographical area based on its specific characteristics.
Keywords: Public Health. Neglected Diseases. Tuberculosis. Ecological Studies
INTRODUCTION
Tuberculosis (TB) is an infectious disease, considered a public health problem due to its persistent distribution in the world, especially in developing countries, as it is strongly influenced by social determinants of health1.
About 7.5 million were diagnosed with TB in 2022, representing the highest record of diagnoses for the infection during the 27 years of World Health Organisation (WHO) monitoring. Regarding mortality in that year, approximately 1.3 million deaths by TB were recorded. It was estimated that at least 40% of those deaths stemmed from interaction with COVID-19 between 2020 and 20222.3.
The 2023 Global Tuberculosis Report 4 also pointed out a case reduction of 900,000 by 2022, as compared to 2021 and 2020. In the panorama covered by that report, Brazil appeared on the list of 30 countries with a high TB incidence, and this would serve as a reference for the WHO in the 2021-2025 period. In this country, the proportion of new TB cases in 2022 was 32 for every 100,000 inhabitants, thus exceeding the goal of less than 10 for this same population reference, as determined by the national plan for the eradication of TB as a public health problem 4.
Despite experiencing a transitional epidemiological process, with reduction of infectious and parasitic diseases in parallel to increased non-transmissible chronic diseases, TB is still a matter of concern in Brazilian public health policies. This is due to the characteristics of heterogeneity in ways of life and quality of healthcare provided across the vast expanse of the country, demarcated by iniquities such as high poverty indexes in certain localities and social exclusion5.
Moreover, the high TB morbidity and mortality rates in Brazil are associated with social and economic issues, such as poverty, malnutrition, bad sanitary conditions, low educational level, unhealthy homes, high population density and an ageing population5.6.
Appropriating the factors that constitute risk of persistence of TB in countries such as Brazil, it is essential to have epidemiological confrontation of this disease, effectively directing the decision-making. In this sense, the space-time approach aids understanding of the dynamics of the presence of TB, allowing analysis of its singularities and variations over time, and in different geographical localities. In addition, this approach, combined with statistical and economic strategies, expands the scope of analysis by adding hypothetically-related variables to the phenomenon, such as economic and demographic conditions5.
From this perspective, the aim was to analyse the association between socio-economic variables and the temporal and spatial persistence of TB rates, notified in Brazilian municipalities between 2001 and 2022, emphasising an explanatory approach to the phenomenon.
METHOD
Ecological study was conducted of temporal series and spatial analysis based on secondary data from public sources, considering information from Brazilian municipalities from 2001 to 2022. Due to being secondary data available in public access information systems, it was supported by Resolution No. 510/2016, and there was no need for approval by the Research Ethics Committee. To understand the persistence of TB in Brazilian municipalities, variables related to health, demography and economics were extracted from the data sources, as illustrated in Chart 1.
The values of Gross Domestic Product (GDP) and the numbers of professionals engaged in healthcare for the years 2021 and 2022 were constructed via a moving mean, since this information had not yet been made available by the Brazilian Institute of Geography and Statistics (IBGE) and the Annual Social Information Report (RAIS), respectively.
As shown in Chart 1, for the construction of the variables, proportions were calculated from the gross data extracted from the sources. Initially, the transformation into proportions did not follow a Gaussian distribution. Thus, following what Clayton and Schifflers (1987)7, Hu et al. (2015)8, Hradsky and Komarek (2021)9 had done, the variables in proportions in this study were logarithmatised, thus allowing approximation to a normal distribution.
The conversion of the variables into logarithms based on 10 had the effect of reducing bias (specifically, the outliers) from extremely high values (long tails), decreasing the effects on measurements, such as correlation or regression of this bias10.11, as well as allowing analysis in terms of proportions, making the latter more understandable12.
Initially, for data analysis, the Exploratory Analysis of Spatial Data (AEDE) was used. This method aims to describe the distribution and spatial association of a particular variable among the units evaluated, as well as identify patterns, forms of spatial instability, and possible outliers. This analysis presupposes adoption of a spatial weighting matrix (W), whose elements represent the degree of spatial connection, following proximity criteria such as contiguity and/or geographical distance13.
Contiguity matrixes have binary spatial weights that can be constructed according to the notion of neighbourhood based on contiguity, where two municipalities are considered adjoined if they share a physical boundary. In this study, a contiguity matrix of the queen type was used. The definition of this matrix was based on the characteristics of the disease and its forms of transmission13.
The analysis incorporated the statistics of the univariate and bivariate Moran Index (I), which allows observation of the existence of global spatial regimes in the data, classifying the data autocorrelation as positive or negative. If positive, the observation of high values in a variable of interest tends to be surrounded by high values of this same variable in neighbouring regions (represented by Wy), indicating spatial similarity. This relationship also applies to low values14.
The Global Moran I statistic was applied to the bivariate context for the purpose of investigating the association between a variable observed in a certain region with another variable in adjacent regions. Positive spatial autocorrelation is evidence of a link between the values of the variables under analysis and their geographical positions14,15.
The adoption of these techniques aimed to provide an evaluation of the degree of linear association (positive or negative) between the values of a variable in a specific municipality (i) and the average of another variable in neighbouring municipalities. Through this, it is feasible to statistically map significant probability values. Thus, the Moran Index dispersion diagram, the map of the significance of Local Indicators of Spatial Autocorrelation (LISA) and the map of LISA clusters can be generated.
In the econometric analysis process, spatial dependence on estimations involving variables in space should be considered as a basic hypothesis. In this sense, the estimations of the models presented in this study began with the basic Ordinary Least Squares (OLS) model. However, in the knowledge that the spatial data generating process does not bear the simplification of the hypothesis of non-spatial heterogeneity, there was, therefore, a need for additional tests such as Moran Index statistics of Regression Residues, as well as the tests, Lagrange Multiplier ( LM) and robust LM for spatial-error and lag-spatial processes, their choice having met all the criteria suggested by the specialised literature16,17. Thus, to use the general structure of a space model as presented in Elhorst (2014)14: DY_t=ρWY_t+Xβ +WXδ+ε, ε=λWε+u where DY_t is the dependent variable, WY_t denotes the effects of endogenous interaction between the dependent variable, WX the effects of exogenous interaction between independent variables and Wε the interaction effects between the degree of disturbance of the different units. The others ρ, β, Δ and λ are parameters.
In the general model of spatial regression equation, W represents a matrix of spatial weights, normalised by the line, with zeros in the main diagonal (the queen matrix was used). By this mechanism, bias in the estimators is corrected, given by omission of spatially correlated variables, as well as spatial heterogeneity, as suggested by Lesage and Pace (2009)18. Thus, when δ = λ = 0 , the Spatial Autocorrelation Model (SAR) is indicated; if δ = ρ = 0 the Spatial Error Model (SEM) is indicated; and, if δ = 0, the Spatial Autoregressive Model with Autocorrelation Error (SARAR) is indicated. In this study, the cases in which only λ = 0 were also considered, in which case Durbin’s Spatial Model (SDM) was indicated; and, the case where ρ = 0, Durbin\'s Spatial Error Model (SDEM)19,20. The SDEM model was indicated by the tests performed and the estimations were performed by it.
In order to test the presence of spatial autocorrelation in the residues, Moran Index tests and LM tests were applied. Definition of the confidence interval of the Moran Index tests was performed through Monte Carlo simulations. The LM tests followed the chi-square distribution with degrees of freedom equal to the number of restricted spatial parameters19,21.
The choice of best specification came from tests on the residues of an OLS model. The models were estimated using the statistical package of the Software R (R Core Team, 2021) spatialreg20. The spatial equation in a cross-sectional temporal cut for 2001-2005, 2006-2010, 2011-2016, 2017-2022 (level regression) was estimated. The choice of temporal cuts instead of binary variables was due to the aim to follow the coefficients of the explanatory variables on the variable explained at distinct points in time.
It is noteworthy that the coding used for data processing purposes and presentation of the results consists of abbreviation of the variable followed directly by the years of the period to which they refer, for example, pib_0610 corresponding to information on the GDP in the period 2006 to 2010. It is also noteworthy that, like all the variables of this study, they were worked on in base 10 logarithm analysis, when referring to any variable, such as the “tuberculosis case rate”, it is understood as its logarithm, that is, the “logarithm of the tuberculosis case rate”.
RESULTS
The spatial analysis adopted to visualise the global spatial autocorrelation is based on the Moran dispersion diagram. Given that the variable of interest is the logarithmic rate of TB cases, it was observed that the angular coefficient was positive, considering the characteristics of this disease’s transmission. It is noteworthy that the Global Moran Index had a high value, exceeding 0.44 in the first temporal cut and 0.419 in the last, which indicates a substantially high level of spatial autocorrelation, forming the basis of spatial analysis and its geographical implications for the economy, demographic dynamics, and, above all, the health sector of the municipalities.
Figure 1 presents the dispersion diagram in the four temporal intervals investigated. The High-High (HH) quadrant denotes that the regions exhibit high TB rates per thousand inhabitants, as represented in the first quadrant of the diagram. The second quadrant, Low-High (LH) refers to a group in which any region with a low TB rate is surrounded by regions with high rates. In turn, the third quadrant, Low-Low (LL) describes a spatial grouping where the regions present values below the mean, surrounded by regions that also have low rates. Finally, the fourth quadrant, High-Low (HL) addresses the scenario in which a region with high rates is adjacent to regions with low ones.
The global I. de Moran bivariate was constructed, and all the spatial autocorrelations presented significance from the statistical point of view, showing a negative relationship with the number of health professionals per thousand inhabitants and with the expenditure on health per capita. Therefore, the municipalities that have a high rate of notified TB cases tend to be surrounded by municipalities with a low number of health professionals and low per capita expenditure on health. The reverse is also valid, that is, municipalities that have a low rate of notified TB cases tend to be surrounded by municipalities with a high number of health professionals and high expenditure on health, that is, higher investments in this sector provide reduction and control of the disease in municipalities and their neighbours.
In this same context, in the case of GDP per capita, the results point out that a municipality with a low case rate is surrounded by municipalities with a high GDP per capita. The reverse also applies, showing that poorer municipalities tend to suffer a higher incidence and transmission of the disease. Regarding population density, a positive relationship is evident, that is, when there is a high rate of notified TB cases in the municipality, it is surrounded by municipalities with high population density, which confirms expectations, given the transmission characteristics.
High spatial autocorrelation rates reveal the need to examine the LISA maps and their significance in the bivariate context. Initially, the relationship between TB case rates, health professional figures and health expenditure per capita (Figure 2), followed by case rates with population density and GDP per capita (Figure 3).
Figure 2 illustrates the map of LISA bivariate clusters for the variables, notified TB case rate and number of health professionals per thousand inhabitants. It is noteworthy that the AB HL type of cluster in the states of Acre, Amazonas, Pará and Maranhão, indicate an association between a high case rate and a low number of health professionals in these locations, which underlines the need for a greater presence of the latter in these states. In addition, the HH type of cluster is more prominent in states such as São Paulo, Rio de Janeiro and Rio Grande do Sul, indicating a complex relationship between high TB rates and a large number of health professionals in these areas. Furthermore, the BA LH relationship evidenced by clusters in these regions, i.e. low case rate and high health professional figures in the neighbouring municipality.
Finally, it is possible to identify a large amount of LL type clusters in municipalities located in the MATOPIBA region (Maranhão, Tocantins, Piauí and Bahia), a region enjoying an expanding agribusiness frontier in recent times. In addition, LH-type clusters, i.e. municipalities with a low case rate near municipalities that have a great number of health professionals , are also located in the immediate vicinity of the Centre-South states.
As for health expenditure per capita, Figure 2 displays the local Moran bivariate for the occurrence of TB in the years selected. It is important to highlight the HL bivariate cluster, predominant in the North and Northeast region, which indicates that municipalities with a high rate of TB cases have reduced health expenditure per capita . In the space-temporal analysis, such a cluster demonstrates a growing trend and stabilisation in the last period. In the first temporal cut (2001-2005), 557 clusters were identified, rising to 604 in the second (2006-2010), 690 in the third (2011-2016), and, finally, remaining at 690 in the recent years of investigation (2017-2022).
Meanwhile, HH clusters predominate in the Midwest region, showing that the increase in disease rates is correlated with health expenditure, this space-temporal relationship varying between reductions and increases, suggesting the possibility of greater expenditure as a way of improving health in these states. In turn, LL-type clusters reveal a more dispersed distribution throughout the country. Finally, LH-type clusters are also important in the Centre-West Midwest and the western part of São Paulo state, which indicates that there is a low case rate associated with high expenditure on per capita health in certain municipalities.
Figure 3 presents the local Moran bivariate for the occurrence of TB and the population density of Brazilian municipalities in the years selected. The data show that, in the first cut, the TB case rate is related to the high density populations of their neighbours. However, the fact that the HL cluster concentrates 554 municipalities, especially in the North, Northeast and Midwest, revealing that population density can be a risk factor for transmission, but the endemic areas in the country are those considered less densely populated, in terms of inhabitants per Km². Thus, the municipalities with a high case rate are close to municipalities with low population density.
In the cuts of 2011-2016 and 2017-2022, the composition of clusters in this dimension shows patterns similar to those in other cuts. A pattern in time and space between the TB case rate relationships in a municipality and the population density of neighbouring municipalities is recognised.
In the TB case rate and GDP per capita dimension of the municipalities, Figure 3 shows that there is an inverse relationship between these two variables. Municipalities with high TB case rates are associated with neighbouring municipalities with low GDP per capita, and this pattern predominated among those with statistical significance, according to the respective maps. In cuts that covered 2011-2016 and 2017-2022, the time-space pattern of previous cuts is also maintained. From this, it can be inferred that there is a strong space correlation between the high rates of TB cases in a municipality and the low GDP per capita of its neighbours.
The results presented in Table 1 were estimated via the SDEM model. These results show that, in the first two cuts, the variable, number of health professionals per thousand inhabitants had negative impacts, but without statistical significance for the diagnostic rate, also in terms of numbers per thousand inhabitants. In the last two cuts the variable showed a negative sign, showing that the increase in the number of health professionals by 1% in the municipality contributed to the reduction of the rate of notified TB cases by 1.5% and 2.1% in the penultimate and last cuts, respectively. Moreover, the variable in lag shows the effect that the variable has on neighbouring municipalities. In this sense, it is possible to state that in all cuts presented in the table, the increase in the number of health professionals in a municipality had an impact on the reduction in case rate in the neighbouring municipality, the indirect effects being statistically significant and greater than the direct effects, as can be viewed in the Table.
The variable, population density in all cuts, when it comes to direct effects, has a positive sign, pointing out that the increase in the number of inhabitants per Km² in the municipality impacts the increase in the confirmed TB case rate. In lag (indirect effects - effects on the neighbouring municipality), the first two cuts show that it has a negative impact on neighbouring municipalities, and, in the last two, the variable did not have statistical significance.
The variable, expenditure on health and per capita sanitation in all temporal cuts had a positive sign, showing that there is a relationship between increased health and sanitation expenditure and increased TB cases in the same municipality. The lag of this variable also presented a positive, significant sign for all the cuts analysed.
In contrast, the GDP per capita variable showed a negative sign in all cuts. The 1% increase in a municipality\'s GDP per capita had an impact on reducing the TB rate by 5.6% and 6.4% in the first two cuts, and 9% and 10.8% in the last two analysed, respectively. Moreover, the 1% increase in the GDP per capita of a municipality impacted the reduction by 7.1%, 7.1%, 7.9% and 6.6% in the TB rate in the neighbouring municipality, in the temporal cuts presented, respectively.
DISCUSSION
During the years studied, it was observed that municipalities with high TB case rates were surrounded by other municipalities with the same number of cases, and this pattern was identified mainly in the North region of the country, especially in Pará State ; in the Northeast region, especially Maranhão, Ceará and northern Bahia, the coastal region of Paraíba, extending to Paraná and Mato Grosso do Sul, maintaining the spatial pattern throughout all the years studied. The number of municipalities with high case rates has greatly decreased over the years, but still represent high rates, far from reaching the goals of decreased rates or eradication established by the WHO and the Ministry of Health (MS)22,23.
Studies show that the spatial distribution of TB indicates that areas with similar incidence rates tend to be close to each other, i.e. grouped24. This reinforces the relationship between spatial agglomerates of the TB endemic , noting that these places are important in the transmission chain of the disease25.
A recent study, with the objective of identifying the determinants of the TB variables and their trend during 2005 and 2016 in the various regions of Brazil, identified that the North of Brazil is the region with the highest mean annual temperature, lower coverage of Primary Health Care (PHC), lower concentrations of doctors and nurses and a low level human development index (HDI - second lowest of Brazil)26, and may be important indicators for the highest concentration of cases in northern Brazil.
There are several factors that can contribute to the spread of TB, as it is a disease with biological, clinical and socio-economic determinants such as poverty, agglomeration, HIV co-infection, malnutrition and insufficient access to health care27. Thus, it is important to highlight that a single isolated variable does not have the ability to determine the increase or reduction of TB cases, but understanding the behaviour of these variables in time and space can aid explanation of the set of factors that determine the population\'s infection by the disease.
The strong relationship of socio-economic factors with TB assists understanding its persistence in developing countries such as Brazil, considering their population heterogeneity and territorial expansion26,28.
In the case of the variables analysed in this study, the association between TB cases and poverty and problems triggered by it, such as hunger and malnutrition, deficits in hygienic conditions, difficulties in access to basic sanitation, bad housing conditions, among others. In the case of Brazil, these situations are observed mainly in remote, more internalised places. In addition, the availability of occupied health professionals can contribute to access to public health policies and health care opportunities, such as health surveillance and early diagnosis29.
Complex results were found regarding bivariate analysis between TB case rates and the variables, “rate of occupied health professionals per thousand inhabitants” and “expenditure on health and sanitation per capita”. Municipalities with high rates of TB cases are surrounded by municipalities with a high proportion of occupied health professionals per1,000 inhabitants, a pattern evident in large urban centres in São Paulo and Rio de Janeiro States. On the other hand, in some regions of Amazonas, Acre and MATOPIBA, municipalities with high case rates of the disease are surrounded by those with low numbers of occupied health professionals. It is suggested that the phenomenon may be linked to other factors, such as population density, urban mobility and other specific characteristics of towns and cities.
Timely diagnosis and complete treatment of individuals with TB involves, in addition to the availability of occupied health professionals, access to health services in which these professionals work, especially PHC30.
In the case of per capita health and sanitation expenditure, it is observed that municipalities with a high rate of TB cases are surrounded by municipalities with low health and sanitation expenditure, a pattern evidenced in the North and Northeast regions. This scenario illustrates public policy that does not provide solutions, and this has an impact on the health and goals of control and elimination of TB, which, in turn, contributes to inclusion of indicators related to sustainable development objectives (SDOs) that aim to stimulate universal access to health and intersectoral actions that focus on social determinants to eradicate this disease by 203027.31.
Econometric analysis showed that, in the first two periods studied, when referring to the municipality itself, the increase of one percent in the mean of the number of health professionals impacts the mean number of notified TB cases in the municipality itself. However, when referring to the impact on neighbouring municipalities, it is observed, in all periods studied, that a 1% increase in the health professionals rate in a city leads to a reduction of 9.1% (2017-2022) to 18.9% (2001-2005) in TB case rates, which becomes increasingly lower over the years.
It is undeniable that diagnosing and properly treating TB cases are fundamental measures for their control25. Thus, although a reduction of cases is necessary and expected, this phenomenon does not always represent the real number of cases present in the population, as there may be underreporting. Thus, a larger supply of health services, a greater number of trained professionals and better diagnostic and treatment conditions for TB cases may reflect on the increase in notification records and better recognition of cases24.
PHC is configured as a gateway to health services, constituting a timely locus for the investigation and diagnosis of TB. A study on PHC accessibility in Brazil noted that there is difficulty in the population’s access to health professionals (doctors and nurses), more markedly in the North and Northeast regions31.
Early diagnosis is one of the most important factors for disease control and may be related to the type of health service sought by the user as a gateway. Although the active search for respiratory symptoms is a multiprofessional activity with the purpose of diagnosing TB cases early and interrupting its transmission chain, the Community Health Agent (CHA) stands out as the main professional in identifying these cases in a community33,34.
Thus, it is evident there is a need for preparation and training of these professionals and recognition of the importance of involvement of the entire health team, especially the nurses and doctors in the team\'s permanent, continuous education. In addition, access should be easy, paying proper, resolute attention and ensuring the continuity of care in specialised services, when necessary33.
The Northeast and the North are marked by a high turnover of health professionals in PHC services, which reinforces the precariousness and absence of bonds with patients and community. This fact interferes directly in the process of training and continuing education of professionals, as the actions are not completed33.
Econometric analysis showed association between population density and presence of TB, while increasing one percent in the average population density increases by up to 6.1% the average of cases recorded per thousand inhabitants in the municipality itself, and the same increased population density reduces the case rate by up to 5.7% in neighbouring municipalities. This can be explained by the migration process, where there is a movement of people from one place to another, whether leaving or coming. At the time, it is suggested that the movement of individuals from smaller towns surrounding them to those with greater opportunities for employment, study, health, among other conditions, can cause this phenomenon of increased population density in the municipality of arrival and consequent reduction of cases in neighbouring municipalities where the emigration occurred.
GDP per capita stood out throughout the study as an important determining variable for the incidence and persistence of TB all over the country, because municipalities located in areas of low GDP per capita, even when presenting low population density, still showed a high number of TB cases, either in the municipality itself or in neighbouring municipalities, such as in the North and Northeast regions. This phenomenon highlights poverty as perhaps one of the most important factors determining the presence of the disease.
A study that evaluated the effects of poverty on TB transmission dynamics identified that disease transmission is more common in communities affected by poverty than in rich ones. Besides this, the results suggested that even when all other determining factors are equal, the rate of poor contacts will have more impact than the rate of rich contacts, supporting the argument that TB is a disease of the poor35.
It is estimated that ending extreme poverty will result in a reduction in the worldwide TB incidence of 33.4% (95% CI) by 2035, and the expansion of social protection coverage will result in a 76.1% reduction in the incidence; both paths together will result in a reduction in incidence of 84.3%36.
The dependent relationship between TB and poverty occurs due to the cyclical effect resulting from the association between poverty and the precariousness of health conditions, reducing opportunities for jobs and subsistence. It is evident that the association between poverty and urban agglomerates (shantytowns), lack of basic health services, inadequate food, low education, alcohol abuse, tobacco and other drugs. All of these factors are directly linked to the presence of the disease, which further exacerbates the relationship between poverty and TB37.
This study stands out for presenting robust analyses of a long temporal series over the last 20 years, addressing the pre- and post-pandemic years of Covid-19, with a large sample throughout the country. However, as it is a study with secondary data from Brazilian databases, it can be pointed out as a limitation on the quality of the data collected through epidemiological surveillance in the entire country, and may be permeated by form-filling bias and undernotification. However, it is recognised as a safe data source that has been adopted to aid public health policies in Brazil and enable comparisons with other locations.
Another limitation is that it does not perform moderation and confusion analyses, given the content of the data available, which allows development of hypotheses, but there are no causal delimitations. Furthermore, there is also as a potential source of confusion, the impossibility of considering the different age structures of the data due to the absence of these records in the sources in most Brazilian municipalities, which can be pointed out as an aspect to be perfected in public information systems , so that new analyses can be conducted. Finally, it is noteworthy that it is impossible to quantify the difference and dependence over the period analysed, given the stratified analyses.
Faced with this challenge, it is recognised that the study has the potential to contribute to public health by associating it with econometric analysis in its analytical methods, ensuring recognition of the economy as a social determinant of health.
CONCLUSION
TB’s temporal and spatial persistence in Brazil is complex and associated with different social determinants of health. The Brazilian regions that most stood out for the persistence in notified TB rates per thousand inhabitants were the North, Northeast and Midwest. It was also identified that these rates are directly related to the number of health occupied professionals per thousand inhabitants and the population density. Whereas, it is inversely related to GDP per capita and expenditure on health per capita. Among the variables analysed, we highlight the GDP per capita in the determination of notified TB case rates, since regions with low GDP, such as the North, even with low population density, showed high persistence of TB in the years studied. These data emphasise poverty as a strong influence on the presence of the disease.
Reflections refer to the need for public health policies and intersectoral actions that are directed to each geographical space based on its particularities and the main determinants and social conditions present in the localities.
Given the multicausality of TB and its relationship with geographical space, studies are suggested that explore the behaviour of this disease and its determinants in specific Brazilian areas, and address the 2020s. In addition, the investigations should deal with the importance of each Brazilian region’s singularities and how these impact the existence of TB.
REFERENCES
Taylan, M; Dogru, S; Sezgi, C; Yılmaz, S1. Epidemiological Trends and Seasonal Dynamics of Tuberculosis in Southeastern Turkey. Nigerian Journal of Clinical Practice 2023; 26:928-33. https://doi.org/10.4103/njcp.njcp_629_22.
Sveinbjornsdottir GM, Kamowa D, Katundu PN, Gizurarson S, Compliance and illiteracy when treating tuberculosis. International Health 2023; ihad077:1-3. https://doi.org/10.1093/inthealth/ihad077.
World Health Organization. Global tuberculosis report 2023. Geneva: World Health Organization; 2023. Licence: CC BY-NC-SA 3.0 IGO. https://iris.who.int/bitstream/handle/10665/373828/9789240083851-eng.pdf?sequence=1
Pinto PFPS, Teixeira CSS, Ichihara MY, Rasella D, Nery JS, Sena SOL, Brickley EB, Barreto ML, Sanchez MN, Pescarini JM. Incidence and risk factors of tuberculosis among 420 854 household contacts of patients with tuberculosis in the 100 Million Brazilian Cohort (2004–18): a cohort study. Lancet Infect Dis 2023. https://doi.org/10.1016/S1473-3099(23)00371-7.
Lima SVMA, dos Santos AD, Duque AM, Goes MAO, Peixoto MVS, Araújo DC, Ribeiro CJN, Santos MB, Araújo KCGM, Nunes MAP. Spatial and temporal analysis of tuberculosis in an area of social inequality in Northeast Brazil. BMC Public Health 2019; 873. https://doi.org/10.1186/s12889-019-7224-0.
Pereira AGL, Escosteguy CC, Valencia LIO, Magalhães MAFM, Medronho RA. Análise espacial de casos de tuberculose e associação com fatores socioeconômicos: uma experiência no município do Rio de Janeiro. Cad Saúde Colet 2018; 26 (2): 203-210. https://doi.org/10.1590/1414-462X201800020013.
Clayton D, Schifflers E. Models for temporal variation in cancer rates. I: Age-period and age-cohort models. Stat Med. 1987; 6(4), 449-67. https://doi.org/10.1002/sim.4780060405
Hu Y, van Lenthe FJ, Mackenbach JP. Income inequality, life expectancy and cause-specific mortality in 43 European countries, 1987-2008: a fixed effects study. Eur J Epidemiol. 2015; 30(8), 615-25. https://doi.org/10.1007/s10654-015-0066-x
Hradsky O, Komarek A. Demographic and public health characteristics explain large part of variability in COVID-19 mortality across countries. Eur J Public Health. 2021; 31 (1), 12-16. https://doi.org/10.1093/eurpub/ckaa226
Mucelin CA. A teoria dos logaritmos: modelos e aplicações. Rev Tecnologia Humanismo 2001; 15 (21), 34-45.
Chen Y. On the four types of weight functions for spatial contiguity matrix. Letters in Spatial and Resource Sciences 2012; 5 (2), 65-72. https://doi.org/10.1007/s12076-011-0076-6
Gomes HMG, Pavanelo E. Uma história a respeito dos Logaritmos. Revista Brasileira de História da Matemática 2022; 22 (43), 102-123. https://doi.org/10.47976/RBHM2022v22n43102-123
Getis A, Aldstadt J. Constructing the spatial weights matrix using a local statistic. Geographical analysis 2004; 36 (2), 90-104. https://doi.org/10.1111/j.1538-4632.2004.tb01127.x
Elhorst JP, organizer. Spatial econometrics: from cross-sectional data to spatial panels. Heidelberg: Springer, 2014.
Anselin, L. Lagrange multiplier test diagnostics for spatial dependence and spatial heterogeneity. Geographical Analysis 1988; 20: 1-17. https://doi.org/10.1111/j.1538-4632.1988.tb00159.x
Le Gallo, J. Cross-Section Spatial Regression Models. M. M. Fischer, P. Nijkamp (Eds.). Handbook of regional science, p. 1511, 2014.
Lesage J.; Pace; K.P. Introduction to Spatial Econometrics. Boca Raton: CRC Press; 2009.
Anselin L, Bera AK. Spatial Dependence in Linear Regression Models with an Introduction to Spatial Econometrics. In: Lesage J.; Pace; K.P. Introduction to spatial econometrics. Handbook of applied economic statistics, Boca Raton: CRC Press; 1998.
Bivand R, Millo G, Piras G. A review of software for spatial econometrics in R. Mathematics 2021; 9:1276. https://doi.org/10.3390/math9111276
Burridge P. On the Cliff‐Ord test for spatial correlation. Journal of the Royal Statistical Society: Series B (Methodology) 2018; 42: 107-108. https://doi.org/10.1111/j.2517-6161.1980.tb01108.x
World Health Organization. The End TB strategy. Geneva: World Health Organization; 2015. https://www.who.int/teams/global-tuberculosis-programme/the-end-tb-strategy
Ministério da Saúde. Brasil Livre da Tuberculose: Plano Nacional pelo Fim da Tuberculose como Problema de Saúde Pública: estratégias para 2021-2025. Brasília: Ministério da Saúde, 2021.
Lima SS, Vallinoto ACR, Machado LFA, Ishak MOG, Ishak R. Análise espacial da tuberculose em Belém, estado do Pará, Brasil. Rev Pan-Amaz Saude 2017; 8:55-63. http://dx.doi.org/10.5123/s2176-62232017000200007.
Maciel ELN, Sales CMM. A vigilância epidemiológica da tuberculose no Brasil: como é possível avançar mais?. Epidemiol Serv Saúde 2016; 25:175-178. http://dx.doi.org/10.5123/S1679-49742016000100018.
Cortez AO, Melo AC, Neves LO, Resende KA, Camargos P. Tuberculosis in Brazil: one country, multiple realities. J Bras Pneumol 2021; 47: e20200119. http:/doi.org/10.36416/1806-3756/e20200119
Trajman A, Saraceni V, Durovni B. Os Objetivos do Desenvolvimento Sustentável e a tuberculose no Brasil: desafios e potencialidades. Cad Saúde Pública 2018; 34: e00030318. https://doi.org/10.1590/0102-311X00030318
Lima SVMA, dos Santos AD, Duque AM, Goes MAO, Peixoto MVS, Araújo DC, Ribeiro CJN, Santos MB, Araújo KCGM, Nunes MAP. Spatial and temporal analysis of tuberculosis in an area of social inequality in Northeast Brazil. BMC Public Health 2019; 873. https://doi.org/10.1186/s12889-019-7224-0
Macedo Junior AM, Silva CDD, Araújo EM, Silva JD, Gomes JT, Granjeiro JSC, Rocha MS. Perfil epidemiológico e fatores determinantes na saúde ambiental da tuberculose no Brasil. Revista Ibero Americana de Ciências Ambientais 2020; 11: 243-52. http://doi.org/10.6008/CBPC2179-6858.2020.007.0022.
Paiva JPS, Brito AB, Bezerra-Santos M, Carmo RF, Souza CDF. Temporal trend of Tuberculosis incidence in northeastern Brazilian municipalities according to Social Vulnerability Index parameters: An ecological study. J Bras Pneumol 2023;49:e20220353. http://doi.org/10.36416/1806-3756/e20220353
Wei X, Fu T, Chen D, Gong W, Zhang S, Long Y, Wu X, Shao Z, Liu K. Spatial-temporal patterns and influencing factors for pulmonary tuberculosis transmission in China: an analysis based on 15 years of surveillance data. Environ Sci Pollut 2023; 30: 96647–59 (2023). https://doi.org/10.1007/s11356-023-29248-4
Figueiredo DCMM, Shimizu HE, Ramalho WM. A Acessibilidade da Atenção Básica no Brasil na avaliação dos usuários. Cad Saúde Colet 2020; 28:288-301. https://doi.org/10.1590/1414-462X202000020288
Cecilio HPM, Teston EF, Marcon SS. Acesso ao diagnóstico de tuberculose sob a ótica dos profissionais de saúde. Texto Contexto Enferm 2017; 26:e0230014. https://doi.org/10.1590/0104-07072017000230014
Matos LP, Oliveira LAT, Ribeiro AF. Educação permanente como ferramenta estratégica para melhoria dos indicadores de busca ativa dos sintomáticos respiratórios de tuberculose em uma unidade de saúde da família no município de Guarulhos – SP. Rev APS 2020; 23: 775-90. https://doi.org/10.34019/1809-8363.2020.v23.32464
Bhunu CP, Mushayabasa S, Smith RJ. Assessing the effects of poverty in tuberculosis transmission dynamics. Applied Mathematical Modelling 2012; 36: 4173-85. https://doi.org/10.1016/j.apm.2011.11.046
Carter DJ, Glaziou P, Lönnroth K, Siroka A, Floyd K, Weil D, Raviglione M, Houben RMGJ, Boccia D. The impact of social protection and poverty elimination on global tuberculosis incidence: a statistical modelling analysis of Sustainable Development Goal 1. Lancet Global Gealth 2018; 6: e514-22. https://doi.org/10.1016/S2214-109X(18)30195-5
Moreira ASR, Kritski AL, Carvalho ACC. Determinantes sociais da saúde e custos catastróficos associados ao diagnóstico e tratamento da tuberculose. J Bras Pneumol 2020; 46 :e20200015. https://doi.org/10.36416/1806-3756/e20200015
