ABC | Volume 114, Nº2, February 2020

Original Article Martins et al. Mortality from cardiovascular disease and cancer Arq Bras Cardiol. 2020; 114(2):199-206 over the years. Demographic and epidemiological transitions, differences in access to health care, genetic peculiarities, among other factors, have resulted in the formation of regional population groups with particular characteristics. 14 Analysis of temporal trends in mortality based on population data could further clarify the scenario. This study aimed to compare early and late mortality, by gender, from CAD and CA between 2000 and 2015 in the most populous capital cities of each Brazilian region and in the country as a whole. Methods It was decided to study the mortality rates from DCS and CA in the most populous capital cities, one of each of the five federated states of Brazil. The Federal District was not included. Demographic and mortality data were obtained from the Brazilian Institute of Geography and Statistics (IBGE). The following cities were included: Manaus (Northern region), São Paulo (Southeast), Goiania (Central west), Curitiba (South), and Salvador (Northeast). Early and late mortality was defined using the age ranges of 30-69 years and ≥70 years, according to the definition of early mortality from non-communicable diseases, stratified by gender, adopted by the Brazilian Ministry of Health, and in line with the recommendations published in the United Nations’ World Population Prospects. 14 Data of mortality from 2000 to 2015 were obtained from the Department of Informatics of the Brazilian Unified Health System (DATASUS), and following the International Classification of Diseases (ICD)-10 as follows: mortality from DCS (chapter IX of ICD-10) and specific causes – acute rheumatic fever and chronic rheumatic heart diseases (066), hypertensive diseases (067), ischemic heart diseases (068), acute myocardial infarction (068.1), other forms of heart diseases (069), cerebrovascular diseases (070), atherosclerosis (071), and other and unspecified disorders of the circulatory system (072), and mortality from neoplasms (chapter II of ICD‑10) and specific causes: malignant neoplasms, lip, oral cavity and pharynx (032), malignant neoplasms of esophagus (033), malignant neoplasms of stomach (034), malignant neoplasms of colon, rectum and anus (035), malignant neoplasms of liver and intrahepatic bile ducts (036), malignant neoplasms of pancreas (037), malignant neoplasms of larynx (038), malignant neoplasms of trachea, bronchi and lungs (039), malignant neoplasms, skin (040), malignant neoplasm of breast (041), malignant neoplasm of cervix uteri (042), malignant neoplasm of corpus uteri and uterus, part unspecified (043), malignant neoplasm of ovary (044), malignant neoplasm of prostate (045), malignant neoplasm of bladder (046), malignant neoplasm of meninges, brain and other parts of the central nervous system (047), non-Hodgkin's lymphoma (048), multiple myeloma and malignant plasma cell neoplasms (049), leukemia (050), in situ neoplasms, benign neoplasms and neoplasms of uncertain behavior (051), and other malignant neoplasms (052). The data were obtained from the computerized databases of the death certificate records of the Brazilian Mortality Information System (Vital Statistics System) and “Population estimates: city, gender and age 2000-2015 RIPSA IBGE” (division of Demographic and Socioeconomic sector). All the information was collected from the DATASUS website. 15 Statistical analysis To assess the tendency of the estimated annual percentage change (EAPC) of the mortality rate from DCS and CA during the study period, the Joinpoint regression model (joinpoint software version 4.6.0.0 National Cancer Institute, Bethesda, Maryland, EUA) 16 was used. The Monte Carlo permutation test was used to detect the years when significant changes in the trends occurred. 17,18 Also, the Poisson distribution was used with the JoinPoint regression model. Assuming such distribution, a maximum of two joinpoints were selected. The software calculates the annual percentage change by the parametric method, with a 95% confidence interval for each segment of trend. The program calculated adjusted mortality rates for sex and age using the standard population based on WHO 2000‑2025. Statistical significance was set at 5%. Results Figure 1 shows the curves of mortality from DCS and CA in the five capitals studied. For all groups and age ranges, there was a decrease in DCS, more pronounced in early mortality, especially in women, among whom CA is already the leading cause of mortality. A stability trend or a slight increase in mortality rate was found in the CA curves. In late mortality, there was a striking difference between the rates of mortality from DCS and CA, with higher rates of deaths from DCS compared with deaths from CA. Thus, the intersection point of these curves occurs later as compared with early mortality. Table 1 shows the EAPCs in mortality from DCS and CA in the most populous capital cities of the five regions of Brazil between 2000 and 2015. Between 2000 and 2015, there was a consistent decrease in early and late mortality from DCS, in both sexes, in the most populous capitals, except for late mortality in men in the city of Manaus. The EAPC for mortality from DCS ranged from -6.5% for early mortality in women in São Paulo, to 0.3%, for early mortality among men in Curitiba. This variation, however, was not statistically significant, probably resulted from an inversion of the trend, with increased mortality in the last two years analyzed, i.e., 2014 and 2015. The reductions were comparable between genders and more pronounced in early mortality. Interestingly, the coefficients of late mortality from DCS were at least ten times the coefficients of early mortality. Differently from DCS, the coefficients of mortality from CA showed different behaviors by regions, time periods and genders. In general, there was a decreasing trend in the rates of mortality from CA in São Paulo and Curitiba in all time periods and sexes. In contrast, therewas an increase in the rates of mortality fromCA in all periods inGoiania. In Salvador, there was amixed behavior, characterized by a drop in early mortality in men and women and increment in late mortality in both genders. Manaus was a major exception in terms of mortality rate behavior;mortality fromDCS decreased in all time periods except for the late mortality in men. On the other hand, early mortality from CA exceeded that from DCS in both men and women. 200