1990 e 2015 Mortality and years of life lost by interpersonal violence and self-harm : in Brazil and Brazilian states : analysis of the estimates of the Global Burden of Disease Study , 1990 and 2015

Rev BRas epidemiol maio 2017; 20 sUppl 1: 142-156 RESUMO: Objetivo: Analisar a mortalidade e os anos de vida perdidos por morte ou incapacidade (DisabilityAdjusted Life Years – DALYs) por violências interpessoais e autoprovocadas, comparando 1990 e 2015, no Brasil e nas Unidades Federadas, utilizando estimativas produzidas pelo estudo Carga Global de Doença 2015 (GBD 2015). Métodos: Análise de dados secundários das estimativas do GBD 2015, com produção de taxas padronizadas de mortes e DALYs. A principal fonte de dados de óbitos foi o Sistema de Informações sobre Mortalidade, submetido à correção do sub-registro de óbitos e redistribuição de códigos garbage. Resultados: De 1990 a 2015, observou-se estabilidade das taxas de mortalidade por homicídios, com variação percentual de -0,9%, passando de 28,3/100 mil habitantes (II 95% 26,9-32,1), em 1990, para 27,8/100 mil (II 95% 24,3-29,8), em 2015. As taxas de homicídio foram mais altas em Alagoas e Pernambuco, e ocorreu redução em São Paulo (-40,9%). As taxas de suicídio variaram em -19%, saindo de 8,1/100 mil (II 95% 7,5-8,6), em 1990, para 6,6/100 mil (II 95% 6,1-7,9), em 2015. Taxas mais elevadas ocorreram no Rio Grande do Sul. No ranking de causas externas por Disability-Adjusted Life Years (DALYs), predominaram as agressões por arma de fogo, seguidas de acidentes de transporte e em sexto lugar lesões autoprovocadas. Conclusões: O estudo aponta a importância das causas externas entre jovens e homens na morte prematura e em incapacidades, constituindo um problema prioritário no país. O estudo Carga Global de Doença poderá apoiar políticas públicas de prevenção de violência.

GBD data indicate reduction in homicide rates worldwide, with regional differences.In Asia and Europe, the rates decreased, but in the Americas, East Africa, and other regions they remained high.Noteworthy is the increase of years of life lost due to death or disability (disability-adjusted life years -DALYs) in Latin America and sub-Saharan Africa, owing to homicides 2 .Suicide is the second leading cause of death from violence globally, the third in Brazil, and a major contributor to DALYs 1, 3 .
In Brazil, external causes correspond to the third leading cause of death.However, in the age groups 1-49 years, they are the main cause and mostly affect young men 4 .Between 2000 and 2010, more than 1.4 million people died from these causes; 38% (545,500 people) from homicides, and 6.5% (92,300 people) from suicide 4 .There are significant regional differences in homicide mortality rates, and the highest rates are observed in the states of the north and northeast regions of the country 4 .
for the physical and emotional effects it causes in people 6 .Homicides and suicides involve young people at full production capacity and result in high individual and collective costs [4][5][6] .
The GBD has innovated by calculating in a comparable way, in all countries of the world, the impact of these events on mortality, in the occurrence of disabilities, in the incidence and prevalence of disease and injury [7][8][9][10][11] .The GBD 2015 has updated estimates and time series from 1990, owing to the inclusion of new data and review of methods.Data were included for Brazil and, for the first time, for the 27 federated units (FUs), enabling comparison between Brazil, FUs, and other countries 12 .
This study aimed at analyzing the mortality and years of life lost due to death or disability (DALYs) from interpersonal violence and self-harm, comparing 1990 and 2015, in Brazil and FUs, using estimates produced by the 2015 GBD study.

MEtHodS
This is a study based on secondary data used in the estimates of 2015 GBD, which applied the methodology proposed by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington, United States of America 13 .
The GBD discloses on its website the data sources used in each country, which, in general, consists of vital registration, verbal autopsy, censuses, surveys, hospital data, police and forensic services records 13 .For violent causes, reports of homicide incidents, road traffic accidents, and suicides recorded by the police are also used.Studies published by national agencies and research institutions, such as the United Nations Surveys on Crime Trends 14 and the Global Status Report on Road Safety 15 were also considered.In countries such as Brazil, which have vital registrations, police data are not used, unless the number of injuries reported in these records exceeds the vital registration.The GBD discloses the sources of information used in each country, state, or other sub-national geographical unit, adopted every year 11,16 .
The main data source on mortality in Brazil was the death registration database of the Mortality Information System (in Portuguese, Sistema de Informações sobre Mortalidade -SIM) of the Ministry of Health, which was subject to adjustments by other national and international sources.
In the calculation of GBD estimates of external causes, a mapping of data sources on disease and injury was carried out first.Second, garbage codes were adjusted, redistributing them to defined causes.Details of the grouping by causes, using the ICD-9 and ICD-10, have been described previously 16 .The GBD 2015 used the International Statistical Classification of Diseases (ICD), ninth and tenth revision: in ICD-9, E000-E999 and codes 800-999; in ICD-10, subgroups V01 to Y98 of Chapter XX and subgroups S00 to T98 of Chapter XIX.Third, statistical models were used and modeling were conducted to estimate data by age, sex, country, year, and cause -the Cause of Death Ensemble Modeling -CODEm.CODEm is an analytical tool that tests the variety of possible statistical models of causes of death and creates a combined set of models that offers the best predictive performance.The DisMod -MR 2.1, a meta-regression calculation tool for simultaneous estimates of incidence, prevalence, remission, disability, and mortality is used 12,17 .
In this study, the following metrics were used: mortality rates and years of life lost due to death or disability, the DALYs.This is a compound indicator that includes the premature death (years of life lost -YLL) and the damage caused by disease, sequelae, or disability, considering different levels of severity of one or more diseases simultaneously years lived with disability -YLD).In this indicator, the weight of lethal and non-lethal diseases is equivalent 18 .The rates were standardized by age, and suicide indicators were calculated for individuals aged over 9 years.
The Global Burden of Disease Study (GBD Brazil 2015) was approved by the Ethics Committee of the Universidade Federal de Minas Gerais (CAAE Project -62803316.7.0000.5149).

rESuLtS
In Brazil, deaths by external causes accounted for 134,931 deaths in 1990 and 168,018 in 2015, and the rates were reduced by 22.8% during the period, from 105.1 (95% UI 105.1-81.2) to 81.2/100,000 (95% UI 77.4-85.4),within the limit of significance.The main groups of causes were homicide, followed by road accidents, other accidental causes, and suicides (Table 1).
Rev BRas epidemiol maio 2017; 20 sUppl 1: 142-156 The DALYs rates from accidents and violence in 1990 and 2015, among men, show that assault by firearms (first position) and pedestrian road injuries (second position) remained with the same classification; road injuries involving motorcyclists moved from ninth positon to third position; and road injuries involving motor vehicle occupants moved from third position to fourth position.Falls remained at the fifth position.Self-harm rose from seventh to sixth position, followed by assault by sharp object (seventh position).Drowning moved from sixth to the eighth position and assaults by other means migrated from fourth to ninth  position.Road injuries involving motorcyclists rose from 17th to 14th position, and unintentional firearms injuries moved from 16th to 18th position (Table 3).Among women, data not shown, falls rose from second to first position of DALYs rates due to external causes, between 1990 and 2015, followed by pedestrians road injuries (second position) and motor vehicle occupants road injuries (third position).Homicides by firearms rose from sixth (1990) to fourth position (2015).Suicide remained at fifth position.
DALYs rates by interpersonal violence and self-harm, in 2015, were higher among men than among women in all FUs, with predominance of assaults by firearms in most states except for Amapá and Roraima, where injuries by sharp objects were predominant.Selfharm was the leading cause of death among women in Roraima, Rio Grande do Sul, Piauí, Mato Grosso do Sul, Santa Catarina, Minas Gerais, São Paulo, and Rio Grande do Norte.The highest rates of DALYs by violence were registered in Alagoas, Pernambuco, and Espírito Santo, and the lowest rates were registered in Santa Catarina, São Paulo, and Piauí.Among women, the highest rates were in Roraima, Espírito Santo, and Pernambuco (Figure 1).
The for age group 0-4 years, in which assaults by other means are prevalent.In other age groups, assaults by the firearm and sharp objects are predominant (Figure 2).

dIScuSSIon
The study shows the magnitude of external causes in the country, which was the third cause of death in 2015.Between 1990 and 2015, these causes were reduced by a quarter.Among the external causes, homicides predominate, followed by road injuries, other accidental causes, and suicides.Homicides remained stable during the period; however, with large variations between states.São Paulo was the only state with a statistically significant reduction during the period.In several other states, although the rates are within the uncertainty interval, they had an increase, such as in Bahia, Rio Grande do Norte, Ceará, and Minas  Gerais.Suicide rates decreased by approximately a quarter, within the uncertainty range.In the ranking of external causes for years of life lost due to death or disability (DALYs), assaults by firearms are predominant, followed by road injuries.Suicides were in the sixth position.The proportional distribution of DALYs by interpersonal violence and self-harm, in 2015, indicate higher frequency among men and in young age groups, especially between 20 and 24 years of age, followed by young people aged 15-19 years.The highest rates of DALYs by violence in men occurred in Alagoas, Pernambuco, and Espírito Santo.
Brazil is simultaneously part of the groups of countries with higher risk of death by homicide and with lower suicide rates, according the World Health Organization reports 19,20 .Despite worldwide phenomenon, homicides have a heterogeneous distribution by geographical areas and social groups.Rates show overall reduction, with large variations depending on the economic status of the countries 20 .In high-income countries, the decrease in homicide rate was 39%; in middle-income countries, the decline was 13%; and in low-income countries, the decrease in the rate was equal to 10%.In a report from the United Nations (UN) in 2012, it was estimated that in Latin America and the Caribbean the majority of deaths from external causes occurred by homicides, and three quarters of them were caused by firearms.In South America, Brazil has the third highest homicide mortality, after Venezuela and Colombia 14 .The risk of homicide is ten times higher than that of high-income countries such as France and the United States of America (rates below 3/100,000), and Latin American countries with low and middle income, such as Argentina and Mexico (3 to 7 per 100,000) 12,21 .
The pattern of occurrence of homicide and suicide shows different aspects in the country, with significant differences by gender, age, and FUs.Violence, in general, is more common among young men, predominantly homicides by firearms in the northeast and north regions.Although within the significance limits, states such as Minas Gerais and Paraná registered rates which differed from that observed in their regions, considering the declining rates in Sao Paulo and stable rates in the other states of the south and southeast regions.A previous study identified increased risk of homicide in cities of Paraná State, which are located close to the border between other countries, possibly owing to greater availability of firearms and illegal drugs 22 .
Homicide victims are usually young adult men (aged 20-34 years) and adolescents (aged 15-19 years) and residing mainly in the northeast and north regions, murdered with firearms.This pattern of mortality are in accordance with other studies on violence, which evidence the main victims as being men, young, black, and poor 21,23,24 .In Brazil, men are 9.2 times more likely to die from homicide than women, which can be partially explained by cultural issues that encourage male violence.The profile of murdered women is also of young victims and, similar to men, death is caused by firearms.In an urban northeast center, the main profile of women murdered is young, black, and single, with low education and income, and they are victims of physical and/or sexual violence 25 .According a UN report, Brazil ranks fifth globally in femicide, or cruel murders resulting from the discriminatory culture of sexual violence and sexism 26 .The violence suffered by adolescents is also associated to school delays and criminal records 27 , trafficking and drug use, alcohol abuse, pregnancy among adolescent girls 28 , and abuses and violence committed by family members 29 .Also among the elderly people, studies reporting abuse perpetrated by their sons, daughters, and other family members are frequent 30 .
Studies on violence, particularly homicides, show that violent behaviors are exacerbated in conditions of structural inequality, in degraded urban areas and outskirts of cities, and are more common among people with low education and income, who use alcohol and drugs, who are involved in trafficking and illegal possession of weapons, and who experience fragile family ties.There is also the victimization by domestic violence, unemployment, precarious education and public safety, as well as police violence.Conflicts in areas of agricultural borders and land disputes are also mentioned 21,[31][32][33][34][35] .
The results presented herein are in accordance with other studies that identified high rates of homicides in the north and northeast regions; and smaller or decreasing rates in southeast region 36 , owing to the decline in rates in Sao Paulo and Rio de Janeiro 37 .Factors that would synergistically influence rates decrease are the following: reduction in the percentage of young people; reduction of unemployment; investments in social policies such as poverty reduction programs; civilian disarmament with an increase of seized weapons; and imprisonment [38][39][40][41][42] .However, security actions become less important as an explanatory factor, after controlling for unemployment and proportion of young people, emphasizing the strength of socioeconomic and demographic factors 43 .It is worth noting that disarmament measures may have positively reflected in reducing homicides 42 ; however, the impact would have been greater among white population than among black population in Brazil 23 .
The presence of armed conflict zones, rivalry, and dispute in trafficking accentuate the risk of deaths from assault, promoting the "ecology of danger" (prevalence of male violence, availability of weapons, coercion, and domination over territories).High homicide rates have been explained by the high concentration of weapons and poor young men 44 .Impunity is also an important factor for the risk of homicide in Brazil 45 and Mexico 46 .
Suicide is among the top ten causes of death worldwide, with more than 800,000 deaths each year.It is estimated 20 attempts for each adult who commits suicide and approximately a third of them have a history of repeated attempts 20,47 .Data from the GBD 2015 indicate the largest global suicide rates in Asia (13.3/100,000) and Europe (17.6/100,000).The Americas have suicide rate of 9.8/100,000, compared with countries such as Russia (36.8/100,000),India (15.8/100,000), and South Africa (14.8/100,000).Brazil is included in the group with lower suicide rates 12 .
Sex, age, culture, and ethnicity have significant implications in the epidemiology of suicide in the world.Consummated suicides among men are more frequent than among women.In this study, the DALYs rates for suicides were higher among men, if compared with women, and among young people aged 15-34 years.The rates of death by suicide are higher among people aged over 70 years 20 , show highest increase among younger people, and indicate more significant risk among indigenous populations 48,49 .Depression is the most relevant factor associated with suicide among elderly people 50 .Survey on violence in Brazil highlights more suicide attempts among women 51 .
Many factors were related to suicide, such as mental disorders (depression, alcoholism, schizophrenia) and sociodemographic characteristics (male, unemployed person, urban residents, retirees, migrants, and singles) 52 .Other conditions are also related to social aspects as follows: personal loss, violence, social isolation, interpersonal conflicts, legal, or labor issues 52,53 , and specifically in childhood and adolescence, physical and sexual abuse, and difficulties with sexual orientation 54,55 .
States of southern Brazil are included in the group with higher incidence of suicide; and Rio Grande do Sul stands out for higher mortality and DALYs rates, which are supported by other studies 56,57 .Deaths are more common in rural areas of the state, and among the predisposing factors are socioeconomic, educational, particularly historical and cultural aspects, as well as those related to working conditions, among others.
Estimates of the GBD 2015 study advanced methodologically in various aspects such as the standardization of data and correction of underreporting and garbage codes, which enabled the comparison between periods, countries, and between the states of Brazil.
There are limitations related to the data sources accessed.Although the SIM has broadened the collection of records and improved their quality in recent years, in the past, and even in some states, there are deaths that were not registered, incomplete records, and high proportion of garbage codes.Studies 58,59 have highlighted the need for SIM correction and recommended using adjustment methodologies, such as the active search for deaths.Other limitation is the difference between SIM and GBD data with respect to the data on legal interventions.GBD did not register deaths from legal intervention in Brazil in 2015, and this needs improvement in future editions.

concLuSIon
The study highlights the differences between the states, the importance of external causes as causes of premature death and disability among young people, and evidences the need to implement policies and measures to reverse this situation.Violence in Brazil is a complex phenomenon that has historical roots and is sustained by the great social inequalities.Any attempt to cope with violence needs to be based on a policy for promotion of life, which requires an interdisciplinary approach with inclusive social policies, implemented in partnership with health area, which may apply its practices of surveillance and care.In Brazil, the actions to manage the violence in recent years have been implemented through coordinated public policies.However, much more must be done to reduce deaths and injuries from violence.Therefore, this needs to be a priority cause of managers, professionals, and society.
The results of the GBD study are a valuable resource for countries seeking to prioritize the management of the main risk factors of deaths and injuries and increase effective actions for the health of the population.

AcKnoWLEdGMEnt
Malta DC thanks the National Council for Scientific and Technological Development (CNPq) for the productivity scholarship.

Table 1 .
Frequency and standardized mortality rate (per 100,000) by accidents and violence with UI (95%), Brazil, in 1990 and 2015.
proportional distribution of DALYs by age, according to interpersonal violence and self-harm, in 2015, indicates higher frequency of cases in the younger age groups, by order of importance, in age groups 20-24 years, 15-19 years, 25-29 years, and 30-34 years.Except Figure 1.Proportional distribution of DALYs by violence, by age group, Brazil, in 2015.