Spatial pattern and temporal trend of mortality due to tuberculosis 10

ABSTRACT Objectives: To describe the epidemiological profile of mortality due to tuberculosis (TB), to analyze the spatial pattern of these deaths and to investigate the temporal trend in mortality due to tuberculosis in Northeast Brazil. Methods: An ecological study based on secondary mortality data. Deaths due to TB were included in the study. Descriptive statistics were calculated and gross mortality rates were estimated and smoothed by the Local Empirical Bayesian Method. Prais-Winsten’s regression was used to analyze the temporal trend in the TB mortality coefficients. The Kernel density technique was used to analyze the spatial distribution of TB mortality. Results: Tuberculosis was implicated in 236 deaths. The burden of tuberculosis deaths was higher amongst males, single people and people of mixed ethnicity, and the mean age at death was 51 years. TB deaths were clustered in the East, West and North health districts, and the tuberculosis mortality coefficient remained stable throughout the study period. Conclusions: Analyses of the spatial pattern and temporal trend in mortality revealed that certain areas have higher TB mortality rates, and should therefore be prioritized in public health interventions targeting the disease.


Introduction
Tuberculosis (TB) is a global public health problem and the main cause of death due to infectious disease (1) .
In Brazil, the incidence of TB was around 41.0 cases per 100,000 inhabitants in 2015, and the mortality rate was 2.4 deaths per 100,000 inhabitants.
In the last two decades there has been a decline in the incidence of TB and the TB mortality rate, although the disease still influences the economy and the health systems. In 2014, in accordance with the Sustainable Development Goals, the World Health Organization (WHO) established the 'End TB strategy', whose target is to reduce TB mortality by 95.0% and TB incidence by 90.0% relative to the 2015 figures by 2035 (1)(2) .
There are barriers to achieving these targets; and its association with hepatitis. Gender is also factor, as mortality is higher in men (3) . In addition, cultural factors come into play, for example the social stigma attached to the disease, which means that individuals avoid attending clinics for fear of receiving a TB diagnosis, and then search for alternative therapies (4) .
The efficiency and quality of healthcare delivery also influence TB mortality (5)(6) .
Due to the available diagnostic and treatment technologies, TB deaths can be considered an unfair event, which highlights the need to reduce its occurrence.
Thus, in order to reduce the mortality rate due to TB in Brazil, social as well as political, human and economic changes are needed.
Spatial and temporal patterns in TB mortality have been studied (7)(8)

Methods
This was an ecological study (9) . The study was carried out in Natal, the capital of the state of Rio Grande do Norte, in Northeast Brazil. The city is divided into 36 neighborhoods and one area of environmental reserve and healthcare is administered through five health districts: North I, North II, South, East and West (10) ( Figure 1). The city has a Human Development Index (HDI) of 0.7 and a Social Exclusion Index (SEI) of about 0.6, a Poverty Index of 40.86% and Gini Index of 0.6 (11) .
The justification for choosing Natal as the study context is that the Ministry of Health has made it a priority city in relation to TB control. In 2015, the incidence of TB in Natal was 37.1 cases per 100,000 inhabitants and the mortality rate was 2.6 deaths per 100,000 inhabitants according to local data from Municipal Health Secretary of the city.
The study population consisted of all cases of death due to TB for which an underlying and associated cause were registered in the Mortality Information System    First we carried out a point density analysis using the Kernel estimation method, an exploratory interpolation method based on defining circular areas of influence around the points where a phenomenon occurs, and then used these to produce a density map that identifies vulnerable areas (12)(13) . The density map provides an overview of the disease distribution and can be used to guide exploration of the point pattern of the health data. Therefore, a radius of 1,000 meters was considered. The density distribution maps of TB deaths were produced using ArcGIS 10.2 software.
Next, the mortality rates due to TB were standardized in each neighborhood per gender and age range using the direct method (9) and considering Natal's population as the standard. The age ranges chosen were based on the disease distribution in the study population: zero to 15 years, 16 to 59 years and 60 years or older.
Annual TB mortality rates were smoothed using a local empirical Bayes model, with a view to minimize the instability caused by oscillations in small population and underreporting of TB deaths. As a result of applying this method, a weighted average was obtained between the gross rate of the neighborhoods and taking the regional rate of the closest neighbors for reference. This rate considered the population density and the local mean rate, starting from a spatial proximity matrix (14) .
Terraview version 4.2.2 was used to calculate the smoothed rates. Next, ArcGIS version 10.2 was used to produce distribution maps of the local empirical Bayes rates, grouped by quintile.
In addition, coefficients for mortality due to TB were expressed as logarithms in order to classify the temporal trend in the disease between 2008 and 2014 as downward, stationary or upward. The time series statistics in StataSE 13 were used for calculating this, applying the Prais-Winsten generalized linear regression method. This procedure corrects the first-order temporal autocorrelation in analyses of organized time series. The annual variance in the measure and its 95% confidence intervals (95%CI) were also calculated (15) .

Discussion
The study aimed to describe the epidemiological profile of TB mortality, to analyze the spatial pattern of these deaths and to investigate the temporal trend in TB mortality in Northeast Brazil.
The study showed that in most of the cases where TB was listed as the cause of death, the deceased was male, between 15 and 59 years of age, single and of mixed ethnicity. In spatial terms, deaths from TB were concentrated in the West, North and East health districts.
With regard to the epidemiological profile, the results do not differ from those of other studies of TB mortality in Brazil and around the world (6) . The gender distribution of mortality (a higher proportion of deaths in males) is similar to the incidence pattern of the disease, with higher morbidity among men (6) and people of mixed ethnicity (3) . This may be due to the relationship between biological and social factors, or to gender differences in exposure factors and prevalence of infection with evolution to disease, amongst other issues related to access to health services.
The study showed that most of the cases had a low level of education. Other authors have reported that low education, unemployment and income are individual-level factors associated with increased incidence of TB and with lower treatment compliance, and therefore they may also be related to access to health services and the quality of diagnosis. People with less education and lower incomes are less likely to perceive that they are at risk and to comply with the treatment, because they present individual and unequal access to information, to benefits deriving from knowledge, to consumer goods and mainly to health services (16).
Moreover, regarding the age of TB mortality as a basic and associated cause, the disease affected patients who were in the economically active age range (15 to 59 years), a finding which is in line with another Brazilian study (17) . This is an important issue, since TB mortality affects economic and social development at a regional level, and is both a cause and consequence of poverty (18) .
Regarding an associate cause of TB mortality, AIDS cases in Brazil are higher in individuals between 25 and 39 years of age for men and women (16) . Furthermore, an earlier study showed a larger number of cases of TB/    (6) .
Another aspect verified in the study was the mortality due to the association between TB and HIV.
TB is the most frequent opportunistic disease in HIV patients, and several studies have demonstrated that it is also one of the main causes associated with death in that population. A study carried out in Africa (19) reported that 47.8% of all deaths investigated were related to

TB/HIV co-infection, which is consistent with global
statistics showing that TB is the cause of death in one out of three Acquired Immune Deficiency Syndrome (AIDS) patients.
It has been shown that characteristics related to treatment history such as treatment abandonment, multiple drug resistance and TB/HIV co-infection are associated with the death of TB cases ( 20) .
In contrast, death from TB may be considered an unfair and avoidable event, as the UHS has all the resources required to diagnose and treat patients, and treatment is freely and universally available (21) .
An important issue is whether all Brazilian populations affected by TB have access to care; some groups, mostly vulnerable groups (homeless people, prisoners, drugs users and the unemployed, amongst others) still face many barriers to care? (22) .
Considering the Universal health system has been adopted in Brazil under a social right perspective and equality, it would be mandatory to provide actions or TB care according to the needs of the population. Health equality is an important index to verify when each individual has a fair opportunity to achieve his or her full health potential (23) . When differential mortality can be linked to differences in social conditions, it is clear that health equality has not been achieved (21) . TB mortality in Brazil is more seriously affected by social inequality than by the availability of medical technology for diagnosis and treatment (1) .
The spatial distribution of TB is affected by socioeconomic inequalities across the study area. In this study, it could be proven that the space was relevant in investigating and understanding the occurrence and distribution of mortality in the city, as it is the environment where the infectious agent circulates, and which, under specific conditions, provokes the disease and even death as a result (24) .
The spatial distribution of cases indicated that TB North I, and North II has 63.00% of the population covered by the FHS (10) .
The West health district is classified as the poorest based on family income data, and it is also the second most populous and has the highest density of TB cases (25) . In addition, it has the highest number of people per household, concentrating the largest number of subnormal clusters, the second highest percentage of slums (25) and 69.00% of the population have FHS coverage (10) . of high social vulnerability (25) . About 37.00% of the population has FHS coverage in the East health district, which may represent weakness in terms of TB care either for diagnosis or treatment (10) .  (20) .
There is a global downward trend in TB mortality.
The number of cases in Brazil has dropped by 2% per year on average over the last ten years (24) , which suggests that mortality rates in Brazil follow the WHO proposal concerning the priorities in terms of early case detection, patient treatment and its conclusion with cure as the outcome (2) .

According to the WHO's Global Tuberculosis
Report (1) , Brazil reached all the TB-related millennium development goals (stop and reverse the upward trend However, in order to achieve the WHO goal of reducing TB deaths by 75% by 2025 and 95% by 2035 through the 'End TB strategy', the incidence rate would have to drop by 10% per year for the next 20 years (24) .

Various strategies have been proposed to improve
Brazilian TB patients' access to healthcare, and it has been suggested that healthcare decentralization is the best way of achieving health equality and equal access to care for all social strata. Thus, the literature has shown FHS advances in relation to qualification of the system, the regional specificity of FHS policies and programs that represent a way of tackling the fragmented nature of the healthcare system, but as yet it has not had an impact on disease rates (27) .