IJCS | Volume 31, Nº3, May/ June 2018

251 Santos et al. Mortality due to hypertensive diseases in Brazil International Journal of Cardiovascular Sciences. 2018;31(3)250-257 Original Article contributes direct or indirectly to 50% of deaths due to cardiovascular diseases (CVD). 4 In low- and medium- income countries, where treatment and control are lower than in developed countries, it is estimated that the prevalence of HBP-related diseases is of about 80%. 8 The impact of this illness reflects in high costs for the Brazilian Unified National Health System (SUS), with an annual treatment cost of approximately US$ 398.9 million, that is, about 1.43% of overall healthcare expenditures. 9 The objective of this research is to estimate the impact of arterial hypertension in the Brazilian territory in a period of five years. For such purpose, we analyzed trends in mortality rates associated with hypertensive diseases in Brazil, between 2010 and 2014, stratified according to skin color and age group, both for states and regions. Methods An epidemiological study was performed from aggregate data obtained in population strata and combined with a spatial analysis. Data regarding the organization of the Brazilian territory, including the coordinates and the estimated population for the years studied, were taken from the Brazilian Institute of Geography and Statistics (IBGE) website in shapefile format. 10,11 Epidemiological information regarding mortality due to arterial hypertension was obtained from the Ministry of Health database, DATASUS. 12 These data refer to notifications from the Mortality Data System (SIM). 13 Deaths were filtered by category I.10 of the International Classification of Diseases (ICD-10). Then, the aggregate data were obtained per year, state, sex, age and skin color. The selection of the analysis period, between 2010 and 2014, occurred due to the following reasons. First, because we consider that more recent analysis provides greater reliability in data collection, due to progressive improvements in the process of computerization with technological advances. Second, because it potentially portrays the transition scenario resulting from the introduction of losartan, an effective antihypertensive drug, which has been distributed free of charge by the “Popular Pharmacy Program” since 2010, and became in 2014 the most demanded medication in units of the Unified Health System (SUS), including in the countryside of Brazil. 14,15 However, one must consider that the program’s effect may occur unevenly across the regions, which would potentially influence the analysis. Finally, the end of the periodunder consideration, the year of 2014, arose fromthe fact that it is themost recent date available in DATASUS to obtain vital statistics all over the Brazilian territory. The selected variables were year, sex, age, skin color, state, region and number of deaths. Since it is a chronic illness, the selected age groups used to calculate the rate of mortality associated with arterial hypertension were as follows: 50-59 years; 60-69 years; 70-79 years; 80 or more years. This calculation was computed for the twenty-six Brazilian states and the Federal District. The variable “skin color” basically portrays the skin color and ethnic traits, based on the death certificate data, and may be classified as “white”, “yellow”, “brown”, “black”, “indigenous” or “ignored”. Statistical analysis The categorical variables were presented as absolute number and percentage. The numerical variables were presentedas average andstandarderror. Regressionmodels (Poisson and negative binomial) for analyzing countable data have been used in longitudinal studies to estimate future mortality rates. Due to overdispersion, negative binomial regression was preferred. The estimate of the “effect size” was adjusted to sex, age groups, skin color, country region and year, and presented in the form of incidence rate ratio (IRR) and confidence intervals at 95%. In order tominimize distortions resulting from spatial and temporal differences between the populations, random-effect models included annual population estimates for each state as an “exposition” factor, that is, the coefficient was restricted, producing an IRR equal to 1, with standard error (virtually) zero, adjusting the calculation for the other coefficients. In order to select the model that provides the best predictive adequacy, the Akaike information criterion (AIC) was used. For the spatial analysis, the “spmap” command was used to draw choropleth maps containing the Brazilian states and the Federal District, and representing the distribution of mortality rates associated with arterial hypertension in quintiles. Statistical significance was considered as a two-tailed p value < 0.05. The statistical calculations and the spatial analysis were conducted in Stata, version 14.2 (College Station, Texas, USA). Ethnical aspects Since these were public data, and there were no elements of identification of the individuals studied, there was no need to use an informed consent term.