Burden of disease in Brazil and its regions , 2008

The current study estimated DALY (disabilityadjusted life years), an indicator of burden of disease, for Brazil in 2008. The North and Northeast regions showed higher burden of disease. Chronic noncommunicable diseases predominated in all regions of the country, especially cardiovascular diseases, mental disorders (particularly depression), diabetes, and chronic obstructive pulmonary disease. The study also showed a high burden of homicides and traffic accidents. Brazil’s epidemiological profile appears even more complex when one considers the nonnegligible burden of communicable diseases, maternal and perinatal conditions, and nutritional deficiencies. The analyses allowed a more detailed understanding of the Brazilian’s population’s health status, underscoring the need for crosscutting actions beyond specific health sector policies and greater attention to the quality of information on morbidity and mortality. Disability-Adjusted Life Years; DALY; Mortality Resumo No presente estudo, o DALY (anos de vida perdidos ajustados por incapacidade), indicador de estudos de carga de doença, foi estimado para o Brasil em 2008. Entre os principais resultados, observam-se maior carga de doença no Norte e Nordeste e preponderância das doenças crônicas não transmissíveis em todas as regiões do país, em particular as doenças cardiovasculares, os transtornos mentais, com destaque para a depressão, o diabetes e a doença pulmonar obstrutiva crônica. Também chama a atenção a elevada carga dos homicídios e dos acidentes de trânsito. O perfil epidemiológico apresenta-se ainda mais complexo quando se considera a carga não desprezível das doenças transmissíveis, das condições maternas, das condições perinatais e das deficiências nutricionais. As análises empreendidas ao longo do estudo possibilitaram conhecer de forma mais detalhada o status de saúde da população, evidenciando a demanda por ações transversais, que vão além de políticas específicas circunscritas à área de saúde, bem como a necessidade de ampliar o escopo de preocupação com a qualidade das informações sobre morbimortalidade no Brasil. Anos de Vida Perdidos por Incapacidade; DALY; Mortalidade 1551 ARTIGO ARTICLE http://dx.doi.org/10.1590/0102-311X00111614 Leite IC et al. 1552 Cad. Saúde Pública, Rio de Janeiro, 31(7):1551-1564, jul, 2015 Introdução Na segunda metade do século XX, o envelhecimento populacional, antes tido como um fenômeno peculiar dos países desenvolvidos, tornouse aparente em vários países em desenvolvimento. No presente século, a tendência é que se consolide, impulsionado não somente pelo rápido declínio da fecundidade, principal componente desse processo, mas também pelo declínio da mortalidade 1,2. A proporção estimada de idosos − indivíduos com 65 anos e mais – duplicará entre 2010 e 2050, passando de 8% para 16% da população mundial. Esse crescimento será bem mais intenso nos países em desenvolvimento, algo em torno de 250%, do que naqueles desenvolvidos, onde o contingente de idosos deverá se ampliar em cerca de 70% 1,2. De fato, o processo de envelhecimento populacional representa um grande desafio para os países em desenvolvimento no que se refere à elaboração de políticas adequadas para os setores de previdência e saúde. À medida que as pessoas vivem mais, vários benefícios sociais, como aposentadorias e pensões, tendem a se estender por um tempo maior, impondo a necessidade de mudanças no sistema de seguridade social 3. Pressões ainda mais intensas são esperadas sobre o sistema de saúde em decorrência da maior prevalência de doenças crônicas, que requerem longo tempo de acompanhamento e intervenções mais complexas e custosas 4,5,6. Aliado a essa pressão, tem-se ainda o fato de que, nos países em desenvolvimento, persistem desafios importantes a serem vencidos no que diz respeito à obtenção de condições adequadas de saúde da população infanto-juvenil. Assim, para comparar o estado de saúde de diferentes populações e melhor compreender os desafios impostos pelo envelhecimento, faz-se necessário elaborar indicadores sintéticos que permitam avaliar o status de saúde dos segmentos populacionais e suas mudanças ao longo do tempo 7,8. Tradicionalmente, o estado de saúde das populações tem sido mensurado com base em indicadores de mortalidade. Contudo, em um contexto de aumento da expectativa de vida e elevada prevalência de doenças crônicas, é preciso considerar também aspectos relativos à perda de saúde decorrente do tempo vivido com essas doenças 8. Há fortes evidências de que os anos ganhos em virtude do aumento da expectativa de vida não têm se traduzido, necessariamente, em anos vividos com saúde 9,10. Em estudo com base em 187 países, Salomon et al. 11 observaram que aumentos na expectativa de vida são acompanhados de ampliação dos anos vividos com incapacidade. Várias medidas sumárias foram apresentadas com o intuito de mensurar, simultaneamente, o impacto da mortalidade e da morbidade sobre o estado de saúde das populações 7,11,12, sendo a mais popular aquela proposta por Sullivan 7, denominada expectativa de vida saudável ou ativa. Basicamente, trata-se de um método que informa quantos anos um indivíduo esperaria viver em bom estado de saúde, sendo frequentemente utilizado para monitoramento do estado de saúde de uma população. Todavia, o método é pouco informativo no que se refere ao provimento de insumos para elaboração de políticas, visto que não incorpora dados sobre as etiologias associadas à perda de saúde. No final do século passado, Murray & Lopez 13 propuseram uma nova medida sintética denominada DALY (disability-adjusted life years – anos de vida perdidos ajustados por incapacidade), indicador do estudo de carga de doença que combina informações de mortalidade e morbidade, sendo originalmente calculado para uma lista de mais de 100 doenças e agravos. O DALY permite mensurar o impacto de cada doença ou agravo sobre o estado de saúde da população, constituindo-se em ferramenta fundamental para elaboração de políticas voltadas para a redução da carga de doença 14. A importância do DALY como indicador sintético de saúde vai além da identificação dos principais agravos e doenças que afetam a saúde das populações, incluindo aplicações em estudos de custo-efetividade, nos quais são comparados os benefícios obtidos por meio da utilização de diferentes intervenções 8. Essa medida tem sido ainda amplamente utilizada na identificação de fatores socioambientais que aumentam o risco de ocorrência de doenças e agravos 15. A Organização Mundial da Saúde (OMS) adotou o DALY como um dos elementos mais importantes no processo de avaliação do estado de saúde das populações 16. Inicialmente utilizado para estimar a carga de doença de grandes regiões do mundo 13, o indicador tem sido empregado na estimativa da carga de doença de países 14,17,18, regiões de países 19, cidades 20 e para avaliar a carga de doenças específicas 21. Além disso, sendo uma medida bastante sensível às iniquidades em saúde, o DALY também pode ser usado para identificar grupos mais vulneráveis da população 22. A estimativa feita pelo DALY e sua utilização para cotejar diferentes aspectos relacionados à saúde da população revestem-se de importância e complexidade maiores para países em desenvolvimento, como o Brasil, onde o processo de envelhecimento ocorre em ritmo acelerado. Nos últimos 50 anos, a expectativa de vida no Brasil aumentou 25 anos, atingindo 73,4 anos em CARGA DE DOENÇA NO BRASIL 1553 Cad. Saúde Pública, Rio de Janeiro, 31(7):1551-1564, jul, 2015 2010 23. Segundo projeções das Nações Unidas, a proporção de idosos no país passará de 7% para 14%, entre 2010 e 2030, mudança que levou mais de um século para ocorrer nos países desenvolvidos, como a França 24. Logo, o Brasil precisa de políticas públicas que se adaptem rapidamente a essa nova realidade, sendo fundamental o conhecimento das demandas do setor de saúde. O objetivo do presente trabalho é identificar as principais doenças e agravos que acometem a população brasileira, segundo as macrorregiões do país, com base na metodologia empregada nos estudos de carga de doença, tendo como referência o ano de 2008. Espera-se que o perfil epidemiológico aqui retratado constitua informação relevante para as novas políticas a serem desenvolvidas para a área de saúde no Brasil. Metodologia da carga global de doença O DALY mede, simultaneamente, os impactos da mortalidade e da morbidade que afetam o status de saúde de determinada população, utilizando como métrica comum o tempo 13. Um DALY representa um ano de vida saudável perdido, sendo calculado como a soma de dois componentes: o de mortalidade, representado pelos anos de vida perdidos em decorrência de morte prematura (years of life lost – YLL), e o de morbidade, caracterizado pelos anos de vida saudável perdidos em virtude de incapacidade (years lost due to disability – YLD). DALY = YLL + YLD Em sua forma mais simples, o YLL para determinada causa de morte é calculado como o produto entre o número de óbitos relacionado a essa causa e a expectativa de vida estimada para a idade na qual o óbito ocorreu. Levando-se em conta os diferenciais de mortalidade por sexo, o cálculo do YLL para uma causa c, idade i e sexo s pode ser expresso da seguinte forma: YLL(c,i,s) = N(c,i,s) x E(i,s) Em que: N(c,i,s) é o número de óbitos devido à causa c para a idade i e sexo s; enquanto E(i,s) compreende a expectativa de vida à idade i e sexo s. O YLL é calculado utilizando-se como padrão expectativas de vida referentes aos níveis 25 e 26 das tábuas-modelo de mortalidade desenvolvidas por Coale & Guo 25, que correspondem, respectivamente, a 80 anos para homens e 82,5 anos para as mulheres. Esse padrão foi idealizado como a maior expectativa de vida alcançável quando o DALY foi proposto, e seu uso permite a comparabilidade entre os resultados. Tendo em vista que a mortalidade é um evento incidente, o cálculo do YLD é feito com base na incidência, ou seja, casos novos de determinada doença/a

nated in all regions of the country, especially cardiovascular diseases, mental disorders (particularly depression), diabetes, and chronic obstructive pulmonary disease.The study also showed a high burden of homicides and traffic accidents.Brazil's epidemiological profile appears even more complex when one considers the nonnegligible burden of communicable diseases, maternal and perinatal conditions, and nutritional deficiencies.The analyses allowed a more detailed understanding of the Brazilian's population's health status, underscoring the need for crosscutting actions beyond specific health sector policies and greater attention to the quality of information on morbidity and mortality.

Introduction
In the second half of the 20 th century, population aging, previously considered a phenomenon of developed countries, became apparent in various developing countries.The trend is for population aging to consolidate in the 21 st century, fueled by both the rapid decline in fertility (the principal component in this process) and the decline in mortality 1,2 .The estimated proportion of elderly − individuals 65 years or older -will double between 2010 and 2050, from 8% to 16% of the world population.This growth will be far more intense in the developing countries, on the order of 250%, when compared to the developed ones, where the elderly contingent is expected to grow by 70% 1,2 .
Population aging poses a major challenge for developing countries to elaborate adequate social security and health policies.As people live longer, various social benefits like retirement and pensions tend to last longer, thus requiring changes in the social security system 3 .Even more intense pressures are expected on the health system due to the greater prevalence of chronic diseases, which require long follow-up and more complex and costly interventions 4,5,6 .This pressure is aggravated by the fact that developing countries face important and persistent challenges for achieving adequate health conditions in the pediatric and adolescent population.Thus, in order to compare the health status of different populations and better understand the challenges raised by aging, summary indicators are needed evaluating the health status of different population groups and their changes over time 7,8 .
The health status of populations has traditionally been measured by mortality indicators.However, in a context of increasing life expectancy and high prevalence of chronic diseases, it is also necessary to investigate the loss of health due to the time lived with these diseases 8 .Strong evidence indicates that the years gained from increasing life expectancy have not necessarily been translated as years lived with health 9,10 .In a study of 187 countries, Salomon et al. 11 observed that increasing life expectancy is accompanied by more years lived with disability.
Various summary indicators have been proposed to simultaneously measure the impact of mortality and morbidity on the health status of populations 7,11,12 , the most popular of which proposed by Sullivan 7 , called healthy or active life expectancy.The method basically predicts how many years an individual would expect to live in good health, and is frequently used to monitor a population's health status.Still, the method provides little policy input, since it fails to incorporate data on the etiologies associated with loss of health.
In the late 20 th century, Murray & Lopez 13 proposed a new summary measurement called DALY (disability-adjusted life years), an indicator for studying burden of disease that combines information on both mortality and morbidity, originally calculated for a list of more than 100 diseases and health conditions.DALY allows measuring the impact of each disease or health condition on the population's health status, serving as a fundamental tool for policies to reduce burden of disease 14 .The importance of DALY as a summary health indicator extends beyond the identification of the principal conditions and diseases that affect the populations' health, including applications to cost-effectiveness studies that compare the benefits obtained from different interventions 8 .DALY has also been widely used to identify socio-environmental factors that increase the risk of diseases and health conditions 15 .
The World Health Organization (WHO) adopted DALY as one of the most important elements in the evaluation of populations' health status 16 .Initially used to estimate burden of disease in major world regions 13 , DALY has been employed to estimate burden of disease of countries 14,17,18 , regions of countries 19 , and cities 20 , and even to evaluate the burden of specific diseases 21 .In addition, since DALY is quite sensitive for measuring health inequalities, it can also be used to identify the population's most vulnerable groups 22 .
The estimates made by DALY and its use to compare a population's different health-related aspects are even more important and complex for developing countries like Brazil, where population aging is occurring at a rapid pace.In the last 50 years, life expectancy in Brazil increased by 25 years, reaching 73.4 in 2010 23 .According to United Nations forecasts, the proportion of elderly in the country will increase from 7% to 14% between 2010 and 2030, a change that took more than a century in developed countries like France 24 .Brazil thus needs public policies that adapt rapidly to this new reality, an essential component for interpreting the demands on the health sector.
The current study aims to identify the main diseases and health conditions affecting the Brazilian population, according to the country's major geographic regions, based on the methodology used in burden of disease studies, with 2008 as the reference year.The epidemiological profile depicted here is expected to provide relevant input for new health policies in Brazil.
Cad. Saúde Pública, Rio de Janeiro, 31(7):1-13, jul, 2015 Global burden of disease methodology DALY simultaneously measures the impacts of mortality and morbidity on a given population's health status, using time as the common metric 13 .One DALY represents a year of healthy life lost, and is calculated as the sum of two components: mortality, represented by the years of life lost due to premature death (YLL), and morbidity, the years lost due to disability (YLD).DALY = YLL + YLD In its simplest form, YLL from a given cause of death is calculated as the product of the number of deaths related to this cause and the estimated life expectancy for the age at which the death occurred.Considering the differences in mortality by gender, the calculation of YLL for cause c, age a, and sex s can be expressed as follows: YLL(c,a,s) = N(c,a,s) x E(a,s) Where: N(c,a,s) is the number of deaths due to cause c for age a and sex s; while E(a,s) is life expectancy for age a and sex s.
YLL is calculated based on the life expectancies for levels 25 and 26 of the model life tables developed by Coale & Guo 25 , corresponding to 80 years for men and 82.5 years for women.This standard was conceived as the highest attainable life expectancy when DALY was proposed, and its use allows comparing the results.
Since mortality is an incident event, calculation of YLD is based on incidence, that is, new cases of a given disease/health condition in a specific year 13,26 .In order to add the two components of DALY, Murray & Lopez 13 estimated a set of weights to quantify the loss of health during the time lived with the disease/health condition, varying from 0, considered full health status, to 1, defined as the worst degree, equivalent to death.Calculation of YLD for a cause c, age a, and sex s can be expressed simply as follows: YLD(c,a,s) = I(c,a,s) x D(c,a,s) x W(c,a,s) Where: I(c,a,s) is the number of incident cases due to cause c, for age a and sex s; D(c,a,s) refers to the mean duration of disability from for cause c, for age a and sex s; and W(c,a,s) expresses the weight of disability from for cause c, for age a and sex s.
The current study basically used the weights proposed by Murray & Lopez 13 , but for some diseases or health conditions we used weights elaborated for the burden of disease study in Australia 27 .Calculation of DALY in our study also incorporated a 3% discount rate, such that the first year of life is lost entirely, while the others correspond to 97% of the previous one.Thus, in ten years, one year of healthy life gained due to health interventions in the current year is 24% less than the gain in the first year, thus decreasing the differences between years of life lost by young people and elderly 26,28 .The age-weight function, sometimes used in burden of disease studies, was not employed here, because it introduces greater complexity into the method when compared to its impact on the estimation of DALY 29 .

Classification of diseases or health conditions
The causes of diseases or health conditions were classified in three main groups: group Iinfectious and parasitic diseases, maternal and perinatal conditions, and nutritional deficiencies; group II -noncommunicable diseases; and group III -external causes 13 .
The main groups were subdivided into 21 subgroups of diseases and health conditions (Table 1), which in turn were subdivided into 107 specific causes 13 .The current study considered 22 subgroups by separating neurological and psychiatric diseases.The list of causes was adjusted to Brazil's epidemiological profile, with inclusions and exclusions that resulted in 100 specific diseases and health conditions.

• Data source
The data source for estimating YLL was the Mortality Information System (SIM), available on the website of the Brazilian Health Informatics Department (DATASUS; http://www.datasus.gov.br).Since the reference year for burden of disease in Brazil was 2008, the study used the mean number of deaths in 2007-2009, disaggregated by State, sex, age group, and cause of death coded according to the 10 th revision of the International Classification of Diseases (ICD-10).

• Underreporting of deaths
The degree of coverage of death reporting was corrected for Brazil as a whole, regions and States, sex, and age group (< 1 year, ≥ 1 year).The degree of coverage of deaths from external causes was assumed to be close to 100%, so only the so-called natural deaths were corrected.This process was conducted such that the sum of the number of deaths estimated by State was equal to the estimated number of deaths in each region, which when summed would reproduce the total deaths in Brazil.To ensure more reliable estimates, the correction factors for both the infant age group (< l year of age) and individuals 1 year Cad.Saúde Pública, Rio de Janeiro, 31( 7):1-13, jul, 2015 or older were estimated independently by two researchers.In case of disagreement, the estimates were evaluated by a third researcher.
For infants (< l year), the factors were estimated with a variant of the Brass child-surviving method proposed by Trussell, with the corresponding temporal allocation 30 .The coverage rate for deaths in individuals one year of age or older was estimated using a variety of demographic methods: Brass growth balance equation 31 ; Bennett & Horiuchi 32 ; Hill general growth balance equation 33 ; and adjusted synthetic extinct generation 34 (Table 2).

• Ill-defined causes and garbage codes
Ill-defined causes, referring to ICD-10 Chapter XVIII (symptoms, signs, and abnormal findings of clinical and laboratory findings, not elsewhere classified), and septicemias were distributed proportionally in all causes of death, except for external causes, according to State, sex, age group, and cause of death.
According to the Global Burden of Disease methodology, some ICD-10 codes, called "garbage codes", should also be distributed, corre-sponding to ICD-10 categories that do not furnish sufficient information on the underlying disease or cause of the lesion, for example "malignant neoplasm without specification of site", or "heart failure".A study by Mathers et al. 35 to evaluate the quality of mortality databases highlighted six categories of garbage codes: cardiovascular diseases, cancers, and external causes.In this study, another three categories of garbage codes were identified and distributed proportionally in a set of specific causes of deaths: diseases of the digestive system (K920, K921, K922); diseases of the respiratory system (J961, J98); leishmaniasis (B559); hepatitis (B559, B189, B19); meningitis (G038, G039); malignant neoplasm of the uterus, unspecified (C55).

• Data source
Calculation of YLD depends on information on incidence and duration of the diseases, which is usually not available, so that it is necessary to estimate them based on other clinical and epidemiological parameters.This estimation uses Table 1 Major groups and subgroups of diseases evaluated in the global burden of diseases study.

Major groups Subgroups
Cad. Saúde Pública, Rio de Janeiro, 31( 7):1-13, jul, 2015 the Dismod II software package, based on three clinical epidemiological parameters to estimate incidence and duration of diseases, evaluating their internal consistency 36 .Thus, through analyses of national databases and a literature review, for each disease or health condition considered in the study, the available data were collected on incidence, prevalence, mortality, case-fatality, relative risk of death, remission, and duration.In order to produce estimates that reflected regional characteristics, the preferred information sources on morbidity were national databases and surveys, with data disaggregated by region.When these sources were not available, information was obtained from surveys conducted in one or more cities of Brazil.Data from the international literature were employed when there were no Brazilian national data on the given parameters.Expert consensus meetings were convened for some diseases with scarce information.
For the literature review, we elaborated quality criteria for article selection, prioritizing population-based studies that used methods to guarantee the estimates' internal validity and minimize the risk of bias.

• Estimation of residual categories of YLD
In the estimation of burden of disease, YLL -the mortality component -can be calculated for all causes of death.However, given the complexity of the morbidity estimation process, YLD is only measured for the set of diseases considered most important for the burden of morbidity and mortality.The other diseases or health conditions are classified, according to the groups to which they belong, in so-called residual categories.
Based on the above, the residual morbidity categories were estimated in three stages.In the first, the residual burden of morbidity for each major group was obtained by applying to the residual YLL the YLD/YLL ratio for the set of diseases from the same group whose YLD was estimated 13,26 .Next, we applied this same ratio to the groups with relevant burden of mortality.In the third stage, the difference between these two estimates was distributed among the diseases with low mortality, proportionally to the burden of YLL from each.

• Interpretation of the results
The results were interpreted as proportions, rates, and adjusted rates, the latter obtained by standardization using the Brazilian population for both sexes in 2008 as the standard.This adjustment aimed to eliminate the age structure effect.

Results
In 2008, the estimate for Brazil was 36,957,662 DALY, of which 18,260,990 YLL and 18,696,672 YLD, resulting in 195 DALY, 96 YLL, and 99 YLD per thousand inhabitants (Table 3).Higher adjusted rates were observed in the Northeast and North regions, with 218 and 206 DALY per thousand inhabitants, respectively.
In Brazil, the DALY rates were higher in men (208 DALY/1,000) than in women (183 DA-LY/1,000).Men showed higher YLL rates, while women recorded higher YLD, indicating higher mortality in the male population and longer time lived with disability among women.Excess male mortality, present in all regions of Brazil, was higher in the Northeast, where mortality accounted for 61.5% of total DALY.
Table 4 shows the proportional distribution and crude and adjusted YLL, YLD, and DALY according to regions of the country and major groups of causes.Of the total DALY for Brazil, 13.2% were due to infectious and parasitic diseases, maternal and perinatal conditions, and nutritional deficiencies (group I); 77.2% to noncommunicable health conditions (group II); and 9.5% to external causes (group III).
The rates for Brazil were 25.8, 150.6, and 18.6 DALY per 1,000 inhabitants for groups I, II, and III, respectively.The North and Northeast showed the highest adjusted rates in group I, with 31.4 and 31.8DALY per 1,000 inhabitants, 22% and 23% higher than Brazil, respectively.These same two regions also showed higher rates in group II.The highest adjusted rates in group III (external causes) were in the South and Central, approximately 4% and 15% above the national mean, respectively (Table 4).
Adjusted YLL rates showed wider regional variation than YLD.The widest variation in YLL was in group I, and the adjusted rate in the North exceeded that of the South by 85%.The highest adjusted YLL rate in group II was in the Northeast, 35% higher than in the South (Table 4).
Table 5 lists the 15 leading specific causes of DALY in Brazil, accounting for more than 50% of the global burden of disease in men (52.8%) and women (51.9%).# The national rates were used as the reference (100) for the adjusted rates ratio.
Ischemic heart disease was the leading cause in men, with an adjusted rate of 15.4 DALY per thousand, and the second leading cause in women, with 11.3 DALY per thousand.For men, the second leading cause was homicide/violence, with 13.4 DALY per thousand, but this cause did not even appear among the 15 leading causes in women.Alcohol abuse and dependence ranked third in men (10.1 DALY) and 13 th in women (2.1 DALY).Stroke ranked third in both men and women.Fifth place was occupied by traffic accidents in men and chronic obstructive pulmonary disease in women.These two causes presented higher adjusted rates in the male population, and this difference was more striking for traffic accidents, whose adjusted rate was approximately four times higher than in women.
Cad. Saúde Pública, Rio de Janeiro, 31( 7):1-13, jul, 2015 Importantly, depression was the leading cause of burden of disease in women, with 25.2 DALY per 1,000, and the seventh leading cause in men, with a much lower rate (7.1 DALY).Diabetes ranked third in the female population and sixth in the male population.Despite the difference in order, the DALY rate for diabetes was similar in men and women, with 9.0 and 9.4 DALY per 1,000, respectively.As for regional variations in the five leading specific causes, the pattern was quite similar to that observed in Brazil as a whole.In the North, Northeast, and Central, for men there was a change in order between the two leading causes, with homicide and violence in first place and ischemic heart disease in second.The Northeast showed the highest adjusted rate from homicide and violence, with 16.4 DALY per thousand in- habitants, exceeding the estimated rates for the Southeast and South by 40%.Diabetes ranked from fourth to sixth among the different regions.
Interregional variations in the ranking of specific causes were wider in women.The principal difference compared to the national pattern for the five leading causes in women was the pres-ence of bipolar affective disorder in fifth place in the North and Northeast, where it outstripped chronic obstructive pulmonary disease and dementias, although with DALY rates similar to other regions of the country.

Discussion
The current study used DALY, a summary indicator for burden of disease studies, to estimate the magnitude of the Brazilian population's leading health problems.
Studies have criticized the lack of standardized application of the methodology in burden of disease studies and the absence of information on the correction of underreporting on mortality 37 .To minimize these problems, despite some adaptations of the analysis to Brazil's reality, the current study aimed to follow precisely the methodology described by Murray & Lopez 13 .In addition, underreporting of mortality was corrected and the garbage codes were distributed proportionally in a set of specific causes of mortality.
The main results feature the preponderance of chronic noncommunicable diseases in all regions of the country, particularly cardiovascular diseases, mental disorders, diabetes, and chronic obstructive pulmonary disease.The high burden of external causes also called attention, especially homicides and traffic accidents, with men as the main victims, accounting for a considerable share of the excess male mortality.The epidemiological profile is even more complex when one considers the non-negligible burden of communicable diseases, maternal and perinatal conditions, and nutritional deficiencies, especially in the North and Northeast.Corroborating other research 8,26,27 , the current study showed higher burden of mortality in men and higher morbidity in women.In this sense, there was a particularly high burden of psychiatric disorders in Brazilian women.
The comparison between regions indicates a higher global burden in the North and Northeast, reflecting earlier deaths and higher burden of disability due to health problems.The highest burden is due not only to the health conditions in group I, but also to chronic diseases.The high burden from this group of diseases may reflect worse living conditions and worse access to health services to control risk factors (such as hypertension) and treat the diseases that lead to higher incidence of disability and earlier deaths.The higher homicide rates in the North and Northeast also called attention, reflecting the increase in violence in these two regions in the last decade 38 .
Chronic noncommunicable diseases are now the leading cause of death in the world, with increasing the Brazilian population prevalence, especially in low and middle-income countries, due to aging and changes in behavioral, occupational, and environmental risk factors 1,39 .The importance of increasing access to cost-effective interventions to reduce morbidity and mortality from these diseases was documented in a recent proposal by the WHO, emphasizing a health system based on primary care to guarantee equity in the continual and sustainable supply of care 40 .In Brazil, upgrading primary care and organizing a services network that guarantees the continuity of care at levels of higher complexity remain as challenges for administration of the Brazilian Unified National Health System (SUS) 41 .
The high burden of psychiatric disorders, especially depression and alcohol dependence, poses a major challenge for health services, since only a small proportion of individuals with these disorders seek care, and these diseases are frequently diagnosed when a patient seeks treatment for another comorbidity 42,43 .Primary health care services could thus play an important role in the identification and treatment of these disorders 44 .However, such a strategy requires the development of treatment protocols as well as training and follow-up of the services provided by general practitioners, since evidence shows that such professionals lack the necessary knowledge to determine the most adequate treatment for these patients 43,45 .
The presence of homicides/violence as the second leading cause of death in men calls attention to this serious public health problem, the pattern of which differs from what is observed in the low and middle-income country group, where it is not among the ten leading causes of DALY 26 .To overcome this problem requires long-term health, public security, educational, and economic policies, including those focused on increasing income levels and reducing inequalities 46,47 .
In relation to the Brazilian national databases that served as the principal source of information in this study, it is important to increase coverage, improve standardization of information, with the use of the same categories in the different systems, increase data completeness, and facilitate access to the databases and to the results of nationwide health surveys, all essential measures to allow better estimates of the Brazilian population's health conditions, with greater efficiency and at lower costs.Important measures in this area feature investments to improve the coverage of the Information System for Notifiable Diseases (SINAN), access to data from the Brazilian National Agency for Supplementary Health or ANS (for private health plans, including services outsourced by the SUS), and ICD-10 classification of the information collected on injuries and their causes in the survey entitled Surveillance System for Violence and Accidents (Viva) 48 .Given the scarcity of epidemiological data on psychi-Cad.Saúde Pública, Rio de Janeiro, 31(7):1-13, jul, 2015 atric disorders, which represent a high burden of disease, nationwide studies to investigate its prevalence by geographical areas of the country should be conducted.
The estimation of diseases is a highly dynamic process, and thus the strategies to evaluate diseases and health conditions are constantly undergoing improvement, so that evaluations of burden of disease over time should be conducted with caution.In a recent study on global burden of disease by Murray et al. 8 with the year 2010 as the reference, various methodological changes were proposed in comparison to that performed in 1990 13 : the number of target diseases and health conditions that were evaluated, and the respective disabilities, more than doubled; a new reference life table was used; there was a change in the weights of diseases and disabilities; and YLD was expressed as prevalence rather than incidence.
Chan 49 calls attention to the need for cooperation between different research groups on burden of disease in order to establish standards for documenting estimation strategies, datasharing, and greater transparency of methods.Such information-sharing is highly important for assessing the proposed methodological changes in the estimation of DALY and for the development of new burden of disease studies in Brazil.
The analyses conducted over the course of this study allow at least two relevant considerations, given the study's primary objective of identifying the Brazilian population's main diseases and health conditions.The first is the clear contribution by this type of study to the knowledge on the population's health status, revealing the demand for crosscutting measures that go beyond specific health sector policies.The second, no less important and complementary to the first, is the clear evidence of the need to expand the scope of concern with the quality of morbidity and mortality data in Brazil.Finally, the regional differences identified here emphasize the need for policies adapted to each region's realities, with the aim of minimizing the inequalities that remain as a striking trait of Brazilian society.

Años de Vida Perdidos por Incapacidad; DALY; Mortalidad
ContributorsI. C. Leite, J. G. Valente, J. M. A. Schramm and R. P. Daumas participated in all stages of the article.R. N. Rodrigues were responsible for estimation of mortality underreporting, and participated in the elaboration of the article.M. F. Santos, A. F. Oliveira, R. S. Silva, M. R. Campos and J. C. Mota were responsible for the estimation of a set of disease and reviewed the article.

Table 2
Correction factors for underreporting of deaths in infants (< 1 year) and individuals 1 year and older.Regions ofBrazil, 2008.

Table 3
Absolute numbers and crude and adjusted YLL, YLD, and DALY in men and women.Regions ofBrazil, 2008.
DALY: disability-adjusted life years; YLD: years lost due to disability; YLL: years of life lost.Source: Research Center for Applied Methods in Global Burden of Disease studies, National School of Public Health, Oswaldo Cruz Foundation.* Per 1,000; ** Adjusted rates, with the Brazilian population as the standard, in both sexes, in 2008.

Table 4 :
Distribution and crude and adjusted rates, YLL, YLD, and DALY, according to major groups of causes.Regions ofBrazil, 2008.
DALY: disability-adjusted life years; YLD: years lost due to disability; YLL: years of life lost.Source: Research Center for Applied Methods in Global Burden of Disease studies, National School of Public Health, Oswaldo Cruz Foundation.* Major groups: I − infectious and parasitic diseases, maternal and perinatal conditions, and nutritional deficiencies; II -noncommunicable diseases; III -external causes.** Per 1,000; *** Adjusted rates, with the Brazilian population as the standard, in both sexes, in 2008.

Table 5
Distribution and crude and adjusted DALY rates by sex and the 15 leading specific causes.Regions of Brazil, 2008.

Table 5 (
continued) : Research Center for Applied Methods in Global Burden of Disease studies, National School of Public Health, Oswaldo Cruz Foundation.* Adjusted rates, with the Brazilian population as the standard, in both sexes, in 2008.