Abstracts
PURPOSE:
To carry out a study of association between socioeconomic and demographic factors and homicides in general population, in the state of Bahia, in 2009.
METHODS:
This is an ecological study. The data were collected from the database of the Information System about Mortality of the Ministry of Health, from the Brazilian Institute of Geography and Statistics and the Institute of Applied Economic Research. The Global Moran index was calculated for the detection of spatial autocorrelation, and the Local Moran index was calculated for the detection of spatial Clusters. The transformation in the variable answer (homicides rates) was performed and it was shaped using the Conditional Autoregressive Model.
RESULTS:
The data showed spatial autocorrelation. Two clusters of municipalities with high rates of homicides were identified, one located predominantly in the Greater Metropolitan Region of Salvador and the other in the South Region of Bahia, especially Eunápolis and Lauro de Freitas, which had the highest rates. The Average Residents Variables, local GDP and the Percentage of Illiteracy presented an inverse association with homicide rates, and the variables Firjan's municipal development index of work and income. Enrolment in high school and the Average of Bolsa Família were directly associated.
CONCLUSIONS:
The urbanization process, in most cases, not controlled by the State, in most cases, made the cities bigger and with better socioeconomic conditions, attraction centers for people with different socioeconomic levels, increasing the social inequality among the residents of these regions, with parallel increase in homicide rates.
Violence; Epidemiologic factors; Residence characteristics; Statistical analysis; Spatial analysis; Socioeconomic factors
OBJETIVO:
Realizar estudo de associação entre fatores socioeconômicos e demográficos e homicídios na população geral, no estado da Bahia, em 2009.
MÉTODOS :
Trata-se de um estudo ecológico. Os dados foram coletados da base de dados do Sistema de Informações sobre Mortalidade do Ministério da Saúde, do Instituto Brasileiro de Geografia e Estatística, e do Instituto de Pesquisas Econômicas Aplicadas. Foi calculado o índice de Moran Global e foi calculado o índice de Moran Local para a detecção de clusters espaciais. Realizou-se a transformação na variável resposta taxa de homicídios, e a mesma foi modelada utilizando o modelo autorregressivo condicional.
RESULTADOS:
Os dados apresentaram autocorrelação espacial. Foram identificados dois conglomerados de municípios com altas taxas de homicídio, um localizado predominantemente na Região Metropolitana de Salvador e o outro na região do Sul Baiano, com destaque para Eunápolis e Lauro de Freitas, que apresentaram as maiores taxas. As variáveis Média de Moradores, Produto Interno Bruto municipal e Percentual de Analfabetismo apresentaram associação inversa às taxas de homicídio, e as variáveis Índice Firjan de Desenvolvimento Municipal - trabalho e renda, Matrícula no Ensino Médio e Média do Bolsa Família apresentaram associação direta.
CONCLUSÕES:
O processo de urbanização, na maior parte das vezes, não controlado pelo poder público, tornou as cidades maiores e com melhores condições socioeconômicas, pólos de atração de pessoas de diferentes níveis socioeconômicos, aumentando a desigualdade social entre os habitantes dessas regiões, havendo paralelamente, aumento das taxas de homicídio.
Violência; Fatores epidemiológicos; Distribuição espacial da população; Análise estatística; Análise especial; Fatores socioeconômicos
INTRODUCTION
External causes ranked third in overall mortality in Brazil in 2007, and homicide
accounted for 36.4% of these deaths11. Reichenheim ME, Souza ER, Moraes CL, Mello J MHP, Silva CMFP, Minayo
MCS. Violência e lesões no Brasil: efeitos, avanços alcançados e desafios futuros.
The Lancet, 2011; 377(9781): 1962-75.. Moreover, this group of causes is also significant in hospitalizations,
representing 8.6% of hospital admissions for all causes in the country in 2011, and with
about 5.0% higher cost than other causes22. DATASUS. Banco de dados do Sistema Único de Saúde. Disponível em
www.datasus.gov.br (Acessado em 15 de maio de 2011).
Disponível em
www.d...
.
Compared to other World Health Organization (WHO) member countries, Brazil was ranked sixth in the ranking of homicide mortality rates in 2005, with 25.8 deaths per 100,000 inhabitants. Compared to other Latin American countries, only El Salvador and Colombia had higher rates, with 50.1 and 45.4 deaths per 100,000 inhabitants, respectively. In general, European countries like France, Germany and Spain presented much smaller rates, bellow 1.0 deaths per 100,000 inhabitants, which also occurs in countries like Japan and Australia33. Waiselfisz JJ. Mapa da violência 2010: Anatomia dos homicídios no Brasil. São Paulo: Instituto Sangari; 2010..
By analyzing the deaths by homicide in Brazil, one can notice an increase in rates until
2003, and a subsequent fall, which was not consistent22. DATASUS. Banco de dados do Sistema Único de Saúde. Disponível em
www.datasus.gov.br (Acessado em 15 de maio de 2011).
Disponível em
www.d...
. In the following years, the observed decline has been due to the decrease in
homicides, especially in the state of São Paulo, between 1999 and 2007, which represents
a decrease of 65.0% in this rate44. Peres MFT, Vicentin D, Nery MB, de Lima RS, de Souza ER, Cerda M,
Cardia N, Adorno S. Queda dos homicídios em São Paulo, Brasil: uma análise
descritiva. Rev Panam Salud Publica 2011; 29(1): 17-26.. In contrast, the rates in the North and Northeast regions showed an increase in
this period. The latter region had greater proportional growth (10.5%) compared to the
whole country, from 8,739 deaths by homicide (20.0% of the total deaths) in 1997 to
15,432 deaths (31% of total homicides in Brazil) in 200722. DATASUS. Banco de dados do Sistema Único de Saúde. Disponível em
www.datasus.gov.br (Acessado em 15 de maio de 2011).
Disponível em
www.d...
.
The State of Bahia exemplifies this trend by presenting a sharp rise in homicide rates, which increased from 15.6 per 100,000 inhabitants in 1997 to 25.8 per 100,000 inhabitants in 2007, showing the state's importance in the context of social violence in the region and, consequently, in Brazil. During this period, the percentage of deaths by homicide in the state of Bahia, in relation to all deaths by homicide in the Northeast region, increased from 22.64% in 1997 to 23.51% in 2007, and the state's percentage of deaths by homicide in relation to Brazil increased from 0.47 to 1.09% during the same period.
Given the importance of this phenomenon in the country and in the Northeast region, this study aimed to examine possible associations between socioeconomic and demographic variables and homicide rates in the state of Bahia, in 2009. The study is relevant, as the literature indicates discrepancies between the findings of studies that seek to associate variables or socioeconomic and demographic factors to deaths by homicide55. de Lima MLC, Ximenes RAA, de Souza ER, Luna CF, de Albuquerque MFPM. Análise espacial dos determinantes socioeconômicos dos homicídios no estado de Pernambuco. Rev Saúde Pública 2005; 39(2): 176-82. , 66. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Impacto da violência na saúde dos brasileiros. Brasília: Ministério da Saúde; 2005.. This not only indicates the existence of gaps in knowledge regarding the importance of these variables in determining homicides, but especially in the way they act and interact77. Land KC, McCall PL, Conhen LE. Structural covariates of homicide rates: are there any invariances across time and social space? Am Soc Rev 1990; 95: 922-63..
Another issue that arises is the unit of analysis in these studies, for most of them88. Macedo AC, Paim JS, Silva LM, Costa Md Mda C. Violência e desigualdade social: mortalidade por homicídios e condiçöes de vida em Salvador, Brasil. Rev Saúde Pública 2001; 35(6): 515-22. , 99. Barata RB, Ribeiro MC, Guedes MB, de Moraes JC. Intra-urban differentials in death rates from homicide in the city of São Paulo, Brazil, 1988-1994. Soc Sci Med 1998; 47(1): 19-23. refers to municipalities, with few focusing on the federative unit and comparing similarities and differences between their municipalities. In addition, only a few of them perform spatial data modeling, being restricted most of the times to describing and spatially identifying regions with high and low risk of death by homicide.
METHODS
Study Design and Variables
This is an ecological study, which aimed to evaluate the associations between socioeconomic and demographic variables and homicide rates in 2009 in the State of Bahia. The outcome variable is the rate of deaths by homicide per 100,000 inhabitants for each municipality, which was calculated by dividing the total number of homicides of Bahia residents, recorded in each municipality in the state, for the total estimated for each of the 417 municipalities.
The number of homicides and the population per municipality were extracted from the Brazilian Ministry of Health's Mortality Information System (SIM/DATASUS). Homicides, according to Chapter XX of the International Statistical Classification of Diseases and Related Health Problems, 10th Revision of November, 2006 (ICD-10), classified as "External Causes of Morbidity and Mortality", grouping X85-Y09, are given the generic title of Aggressions, and the population used was the one estimated for the year of 2009.
In the state, with variations between municipalities, it was observed that, in recent years, after the growth of deaths by homicide, the growth of the death classification group categorized as "external causes of undetermined intent" also occurred. It is noteworthy that the increase in this group, whose intentionality is unknown, affects the strength of the associations. In the present study, it is considered that its growth may have reduced the magnitude of the observed associations, given that several studies66. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Impacto da violência na saúde dos brasileiros. Brasília: Ministério da Saúde; 2005. , 1010. Mello Jorge MHP, Gotlieb SLD, Laurenti R. O sistema de informações sobre mortalidade: problemas e propostas para o seu enfrentamento II - Mortes por causas externas. Rev bras epidemiol 2002; 5(2): 212-23. have shown that a portion of the "undetermined intent" death classification group is composed of homicides.
Exposure variables, according to the theoretical model proposed in the
literature55. de Lima MLC, Ximenes RAA, de Souza ER, Luna CF, de Albuquerque MFPM.
Análise espacial dos determinantes socioeconômicos dos homicídios no estado de
Pernambuco. Rev Saúde Pública 2005; 39(2): 176-82.
,
1111. Kleinschmitt SC, Wadi YM, Stadutto JA. Análise espacial dos
homicídios: um estudo sobre o Estado do Paraná. In: Encontro Nacional da Associação
Brasileira de Estudos Regionais e Urbanos(VII ENABER); 2009, São Paulo (Br). São
Paulo: FAE-USP; 2009. p. 1-20.
,
1212. Cruz OG. Homicídios no Estado do Rio de Janeiro: Análise da
distribuição espacial e sua evolução [dissertação de mestrado]. São Paulo: Faculdade
de Saúde Pública da USP; 1996., are of the socioeconomic and demographic type. Data were collected from the
Brazilian Institute of Geography and Statistics (IBGE)1313. IBGE - Instituto Brasileiro de Geografia e Estatística. Censo
demográfico 2010, Disponível em
http://www.sidra.ibge.gov.br/cd/cd2010universo.asp?o=5&i=P (Acessado em 30 de
outubro de 2011).
Disponível em
http:...
and the Institute of Applied Economic Research (IPEA)1414. IPEA - Instituto de Pesquisas Econômicas Aplicadas. Disponível em
http://www.ipeadata.gov.br/ (Acessado em 30 de outubro de 2011).
Disponível em
http:...
. The variables collected by IBGE are reported below:
-
Demographic Density - The ratio between the total resident population and area of the municipality, in 2010 (inhabitants/km²);
-
Percentage of the population aged 15 to 24 years - People between 15 and 24 years compared to the total population, in 2010;
-
Average of residents per household - Sum of the number of residents per private, permanent households, divided by the number of private, permanent households;
-
Proportion of poor - is the ratio between the number of individuals with per capita family income below 50% of the minimum wage relative to the total sum of all individuals with per capita family income in all income brackets, in 2010;
-
Percentage of illiteracy - Percentage of people aged 15 years and older, who cannot read and write a simple note, in the language they know, in the total resident population of the same age group in a given geographical area, in 2010.
Variables collected from IPEA and their definitions are as follows:
-
Average receipt of Bolsa Família - Sum of the total value of the benefit received, divided by the number of benefits received in 2010;
-
Municipal Gross Domestic Product (Municipal GDP) - Sum of all goods and services produced by the municipality in 2008;
-
Number of students enrolled in high school - Total enrollment in high school in the municipality in 2010;
-
FIRJAN Municipal Development Index (IFDM) on education - Varies from 0 to 1. The closer to 1, the greater is the development. It considers the following information: tuition in early childhood education, dropout rate, age-grade distortion rate, percentage of teachers with higher education, average of daily hours in class and the result of IDEB (Basic Education Development Index). Its source is the Brazilian Ministry of Education, and refers to each municipality in 2007;
-
IFDM on employment and income - Varies from 0 to 1 The closer to 1, the greater is developing. It considers the following information: formal employment generation, formal employment stock and average wages in formal employment. Its source is the Brazilian Ministry of Labor and Employment, and data were collected for each municipality in 2007.
Data Analysis
Because the data are spatially localized, i.e., from the municipalities of the state studied, the possible importance of their spatial arrangement in the analysis was considered1515. Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex: Longman Scientific; 1995..
Exploratory and descriptive data analyses, as well as exploratory spatial data
analyses, were performed in order to observe the behavior of homicide rates in the
space. To this end, shaded maps were used to detect possible municipalities with an
excess of the issue under study. Digital meshes containing scanned maps of the
municipalities of Bahia were downloaded from the IBGE site, and the software used to
analyze the damage was R1616. R Development Core Team (2008). R: A language and environment for
statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN
3-900051-07-0. Disponível em http://www.R-project.org.
Disponível em http://www.R-project.org...
, version 2.14.0, as well as Spdep, xlsx, maptools, spatstat, foreign and
RcolorBrewer libraries.
Another way to view the spatially located data is through smoothed maps that re-estimate the rates of the issue according to the behavior of their neighbors. This study used the Global Empirical Bayes estimator, which re-estimates homicide rates in each municipality by making use of a global average1515. Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex: Longman Scientific; 1995..
This re-estimation process considers the population size of each municipality, making the largest corrections in municipalities with smaller populations. The criteria used to establish the neighborhood between the municipalities was the connectivity between the borders, i.e., given two towns, I and J, they are neighbors if they share at least one side in common.
To test the hypothesis that there is a spatial correlation between the data, we used
the Moran Index1717. Câmara G, Druck S, Monteiro AM, Carvalho MS. Análise Espacial de
Dados Geográficos. Brasília: EMBRAPA, 2004. Disponível em
http://www.dpi.inpe.br/gilberto/livro/analise/cap5-areas.pdf (Acessado em 31 de
janeiro de 2011).
Disponível em
http:...
, which checks for spatial independence. The Local Moran Index1717. Câmara G, Druck S, Monteiro AM, Carvalho MS. Análise Espacial de
Dados Geográficos. Brasília: EMBRAPA, 2004. Disponível em
http://www.dpi.inpe.br/gilberto/livro/analise/cap5-areas.pdf (Acessado em 31 de
janeiro de 2011).
Disponível em
http:...
was used to detect spatial patterns of the issue, seeking, locally, the
existence of municipalities clusters with similar behavior regarding homicide
rates.
For the data modeling, the Conditional Autoregressive Model (CAR) was used for the transformed outcome variable. The Global Empirical Bayes estimator was applied to the rates and, sequentially, the log was applied to the re-estimated rates, in order to obtain data normality1515. Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex: Longman Scientific; 1995..
The CAR model is expressed by Z = Xβ + ε, with ε = λ Wε + ξ, where Wε is the error component with spatial effects, λ s the autoregressive coefficient and ξ is the error component with constant and uncorrelated variance1515. Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex: Longman Scientific; 1995..
The variables were tested individually, in order to assess which of them would be initially incorporated into the model, and they were later tested jointly, to assess the effects when they were in the presence of each other, as a way of control. Thus, they were removed from the model one by one, first the least significant (at a significance level of 10%) and so on, until only the significant variables were left in the model. The use of a significance level of 10% in the initial stage of the study sought not to delete variables that could prove significant in the joint analysis, aiming to expand the analysis possibilities.
To assess the quality of the adjusted model, the graphical analysis of the residuals
was made, searching for signs of violation of the independence assumptions. For a
quantitative test, the Moran's I index was used on the residuals1818. Sá Carvalho M, Bivand R. Exploring and analysing rates of Hansen's
Disease in Olinda, Brazil. 2003. Disponível em:
http://leg.ufpr.br/lib/exe/fetch.php/pessoais:marilia:bivandmarilia.pdf (Acessado em
19 de novembro de 2011).
Disponível em:
http...
.
RESULTS
In 2009, of the 5,210 homicides recorded in the state of Bahia, in 4,876 of them, the victim was male, representing 93.6% of deaths; and 334 cases involved females (6.4%). The most affected age group was 15 - 24 years old, which totaled 2,374 deaths (45.5%), followed by the 25 - 34 years old group, with 1,592 cases (30.5%). These data show a profile known from other studies1919. Barata RB, Ribeiro MCSA, de Sordi M. Desigualdades sociais e homicídios na cidade de São Paulo, 1998. Rev Bras Epidemiol 2008; 11(1): 3-13. , 2020. Gawryszewski VP, Costa LS. Homicídios e desigualdades sociais no Município de São Paulo. Rev Saúde Pública 2005; 39(2): 191-7., in which young males are the main victims of homicides.
It is noteworthy that of the 417 municipalities in the study, 112 (26.8%) presented number, and consequently a homicide rate, equal to zero.
Figure 1 shows the distribution of homicide rates in the space, in which the colored bar represents the value of homicide rates. Therefore, the municipalities represented by the lighter color present lower rates, while those with a darker color are the ones where rates are the highest.
Map of homicide rates (per 100,000 inhabitants) in Population (A) and smoothed homicide rates according to behavior of neighbors (B) in the municipalities of the state of Bahia in 2009.
You can clearly see the existence of spatial concentration of homicide rates: coastal municipalities have the highest values, while inland municipalities have the lowest values. The smoothing of the rates further highlights this phenomenon, since municipalities with initially high rates showed lower rates after smoothing, such as some municipalities in the Northern Bahia.
The Moran Index, which assesses the existence of a spatial structure, obtained the value 0,49 (p < 5%), indicating a direct correlation between the observations and corroborating the hypothesis of spatial correlation, observed in the graphical analysis of homicide rates.
Figure 2 shows the Local Indicator of Spatial Association (LISA), which highlighted the existence of spatial clusters, i.e., in the metropolitan region of Salvador, as well as in the southern region of the state, around Itabuna and Porto Seguro, homicide rates follow a pattern of interpersonal violence that spread among neighboring municipalities.
Local Indicator of Spatial Association to detect spatial clusters of homicide rates (per 100,000 inhabitants) in municipalities in the state of Bahia, in 2009.
Table 1 shows twenty municipalities belonging to these two clusters, where homicide rates in the general population are higher.
Homicide rates (per 100.000) of the twenty municipalities with higher rates, which make up the two clusters identified in the study, in the municipalities of Bahia in 2009.
The twenty municipalities presented are responsible for 3,021 deaths by homicide in the state of Bahia, which represents 87% of all deaths by homicide in the 417 municipalities of the state, especially the city of Salvador, with 1,837 deaths by homicide, which represents 53% of cases among the total number of deaths due to homicide.
On the other hand, of the first five municipalities with the highest homicide rates, three of them are located in the middle region of the Southern Bahia, which certainly presents itself as a potentially increased risk of death by homicide when compared to other regions of the state, mainly inland regions. The municipalities located outside the clusters have an average rate of 13.32 deaths per 100,000 inhabitants, well below the average of 54.55 deaths per 100,000 inhabitants of the municipalities that make up the cluster.
The individual data modeling, in which the transformed variable homicide rate was used (normality obtained and verified at 3%) resulted to be significant for all exposure variables. In the multivariate modeling stage, the following variables were not significant: Demographic Density, Percentage of Young Inhabitants, Percentage of Poor Inhabitants and the IFDM on education.
Table 2 shows the results of the modeling with regard to the other variables, jointly adjusted in the model. The variables are the following: Average of Residents per Household, Municipal GDP and Illiteracy Rate, which showed an inverse association with mortality by homicide. On the other hand, the variables IFDM on Work and Income, Average Receipt of Bolsa Família and Enrollment in High School were positively associated with the rate of mortality by homicide.
Estimate, Standard Deviation, Statistics z and p-value of the variables used to predict transformed homicide rates.
The diagnosis of the adjustment of the model indicated that the presence of a spatial autocorrelation in homicide rates was contemplated and handled properly by the methodology, for the Moran Index applied on the residuals was approximately zero, with p = 0.741.
DISCUSSION
The study enabled the discovery of clusters of homicide rates in the vicinity of the city of Salvador and in municipalities of the southern region of the state of Bahia. These areas are places with higher economic dynamic, and are characterized as tourist areas, which attract a working population, resulting in high population density. However, this variable was not significant in the final model.
The associations were: positive between homicides and IFDM on work and income, and a negative between homicides and municipal GDP. This finding seems to indicate that income or wealth alone is not enough variables to explain the occurrence of homicides. Several authors2121. Briceño-León R. Urban violence and public health in Latin America: a sociological explanatory framework. Cad Saúde Pública 2005; 21(6): 1629-64. , 2222. Santana FS. Homicídios em Salvador, 1998-2001: Efeitos das desigualdades socioeconômicas na mortalidade por violência [dissertação de mestrado]. Rio de Janeiro: Escola Nacional de Saúde Pública FIOCRUZ; 2004., emphasize that such violence would be more related to inequality than poverty. In Brazil, the richest 10% hold 45% of the country's wealth.
Latin American countries are the ones that experience the greatest inequalities in the world, and that may be the reason why they present high homicide rates. In contrast, in countries in Africa, with high percentages of poor people, and in Europe, where inequality is lower, the picture is reversed, and they present low rates of these events.
Increased and unplanned urbanization has also been implicated as a determinant factor for homicides. In this study, an inverse association was observed between Illiteracy Rates and Average of Residents per Household and homicide rates. Unlike the population density, which evaluates how much the municipality is urbanized, the average number of residents per household addresses the family composition. In this sense, family composition seems to act as a protective factor for homicide in inland cities, which has the highest average number of residents per household.
A recent survey, published by IBGE, has shown that fertility rates in inland cities in the states are higher than those in larger municipalities.
Inverse relationship between the rate of illiteracy and homicides was also found. One hypothesis that could be thought of in relation to this finding is: municipalities with the highest percentage of illiteracy are the poorest, which present a lower GDP compared to other municipalities in the state, and in turn, they do not constitute as a place of interest for criminal actions.
There was a direct association between homicides and enrollment in high school. Previous studies1111. Kleinschmitt SC, Wadi YM, Stadutto JA. Análise espacial dos homicídios: um estudo sobre o Estado do Paraná. In: Encontro Nacional da Associação Brasileira de Estudos Regionais e Urbanos(VII ENABER); 2009, São Paulo (Br). São Paulo: FAE-USP; 2009. p. 1-20. indicate that enrollment in high school would be a protective factor for deaths by homicide. However, this association was not assessed jointly in the presence of the other study variables. It is also important to note that this variable does not measure school attendance, nor is compared with the total of young people in each municipality, i.e., it is not possible to observe the percentage of school-age children who are regularly enrolled, and school attendance of these children.
The direct relationship between the Average Value Received by the Bolsa Família Program and homicides suggests that the areas that receive the highest average values are those that have more social protection resources, i.e., greater coverage of social protection bodies, as well as those which are socially vulnerable.
Epidemiological studies diverge when it comes to associating socioeconomic variables with homicide rates. Some studies55. de Lima MLC, Ximenes RAA, de Souza ER, Luna CF, de Albuquerque MFPM. Análise espacial dos determinantes socioeconômicos dos homicídios no estado de Pernambuco. Rev Saúde Pública 2005; 39(2): 176-82. , 1212. Cruz OG. Homicídios no Estado do Rio de Janeiro: Análise da distribuição espacial e sua evolução [dissertação de mestrado]. São Paulo: Faculdade de Saúde Pública da USP; 1996. found a positive association between homicides and the variables that measure improved living conditions, as in the present analysis. One factor that has also been mentioned1212. Cruz OG. Homicídios no Estado do Rio de Janeiro: Análise da distribuição espacial e sua evolução [dissertação de mestrado]. São Paulo: Faculdade de Saúde Pública da USP; 1996. is unplanned urbanization, leading to high rates of homicide.
It should be underscored that one limitation to this study is the limited availability of access to free data that enable greater understanding of the violent processes. To this end, it would be important to have Public Safety information on firearms and drugs, as well as the monitoring and dissemination of social actions related to the promotion of peace and prevention of violence that could be provided by the municipalities.
Thus, the issue in question has generated a wide range of hypotheses to be verified. In other studies44. Peres MFT, Vicentin D, Nery MB, de Lima RS, de Souza ER, Cerda M, Cardia N, Adorno S. Queda dos homicídios em São Paulo, Brasil: uma análise descritiva. Rev Panam Salud Publica 2011; 29(1): 17-26. , 2323. Nadanovsky P, Celeste RK, Wilson M, Daly M. Homicide and impunity: an ecological analysis at state level in Brazil. Rev Saúde Pública 2009; 43(5): 733-42., it was found that investments in the areas of Public Security and Education are explanatory factors for the decrease in homicide rates in São Paulo. Therefore, one should evaluate the investment currently applied in these areas as a possible beginning of a process to reverse the increase in homicides in the state of Bahia.
Also shown as good alternatives to the problem are successful examples, such as the cities of New York44. Peres MFT, Vicentin D, Nery MB, de Lima RS, de Souza ER, Cerda M, Cardia N, Adorno S. Queda dos homicídios em São Paulo, Brasil: uma análise descritiva. Rev Panam Salud Publica 2011; 29(1): 17-26., United States, and Cali2424. World report on violence and health: a summary [Relatório mundial sobre violência e saúde: um resumo]. Genebra, Organização Mundial da Saúde, 2002., Colombia, where the reduction in crime rates was initiated with a comprehensive plan for the prevention and repression of even the simpler criminal attitudes, as well as actions to guide communities on the importance of its relationship with the physical space, preserving plazas, monuments, keeping the streets clean and lessons on civism in schools.
CONCLUSION
By using the spatial analysis model, it was possible to assess the existence of associations between variables studied and homicides, given the assumption of controlling the effect of spatial autocorrelation in these associations. But the findings also indicate that the explanations for this phenomenon are complex and need to be better understood.
This study corroborates the findings obtained in several national55. de Lima MLC, Ximenes RAA, de Souza ER, Luna CF, de Albuquerque MFPM. Análise espacial dos determinantes socioeconômicos dos homicídios no estado de Pernambuco. Rev Saúde Pública 2005; 39(2): 176-82. , 1212. Cruz OG. Homicídios no Estado do Rio de Janeiro: Análise da distribuição espacial e sua evolução [dissertação de mestrado]. São Paulo: Faculdade de Saúde Pública da USP; 1996. e and international2525. Messner SF, Anselin L, Baller RD, Hawkins DF, Deane G, Tolnay SE. The spatial patterning of county homicide rates: an application of exploratory spatial data analysis. Journal of Quantitative Criminology 1999; 15(4): 423-50. , 2626. Breetzke GD. A socio-structural analysis of crime in the city of Tshwane, South Africa. S Afr J Sci 2012; 106(11/12): 4-7. studies, which point to the need for systematically reducing social inequality in cities with high homicide rates.
It is important to highlight some limitations of this study, such as the lack of more detailed information about the municipalities that enable the understanding of violent processes. Another limitation relates to the fact that the information is aggregated in the municipality's unit of analysis, making it impossible to capture existing differences within municipalities.
Moreover, this is an ecological study, and its results cannot be explained at an individual level2727. Rothman KJ, Greenland S, & Lash TL. Modern Epidemiology. Philadelphia: Lippincott, Williams & Wilkins; 2008.. It is also necessary to point out that the quality of information available in secondary databases is characterized as another limitation of the studies.
It is worth remembering that violence is a social phenomenon that is part of human history, and it assumes specific features according to time and place2828. Minayo MCS. A violência social sob a perspectiva da saúde pública. Cad Saúde Pública 1994; 10 (Suppl 1): 7-18.. Thus, we are far from any definitive solutions to the problem, with any measure adopted being seen as a challenge to the government and civil society, in order to jointly seek auspicious results.
REFERENCES
-
1Reichenheim ME, Souza ER, Moraes CL, Mello J MHP, Silva CMFP, Minayo MCS. Violência e lesões no Brasil: efeitos, avanços alcançados e desafios futuros. The Lancet, 2011; 377(9781): 1962-75.
-
2DATASUS. Banco de dados do Sistema Único de Saúde. Disponível em www.datasus.gov.br (Acessado em 15 de maio de 2011).
» Disponível em www.datasus.gov.br -
3Waiselfisz JJ. Mapa da violência 2010: Anatomia dos homicídios no Brasil. São Paulo: Instituto Sangari; 2010.
-
4Peres MFT, Vicentin D, Nery MB, de Lima RS, de Souza ER, Cerda M, Cardia N, Adorno S. Queda dos homicídios em São Paulo, Brasil: uma análise descritiva. Rev Panam Salud Publica 2011; 29(1): 17-26.
-
5de Lima MLC, Ximenes RAA, de Souza ER, Luna CF, de Albuquerque MFPM. Análise espacial dos determinantes socioeconômicos dos homicídios no estado de Pernambuco. Rev Saúde Pública 2005; 39(2): 176-82.
-
6Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Impacto da violência na saúde dos brasileiros. Brasília: Ministério da Saúde; 2005.
-
7Land KC, McCall PL, Conhen LE. Structural covariates of homicide rates: are there any invariances across time and social space? Am Soc Rev 1990; 95: 922-63.
-
8Macedo AC, Paim JS, Silva LM, Costa Md Mda C. Violência e desigualdade social: mortalidade por homicídios e condiçöes de vida em Salvador, Brasil. Rev Saúde Pública 2001; 35(6): 515-22.
-
9Barata RB, Ribeiro MC, Guedes MB, de Moraes JC. Intra-urban differentials in death rates from homicide in the city of São Paulo, Brazil, 1988-1994. Soc Sci Med 1998; 47(1): 19-23.
-
10Mello Jorge MHP, Gotlieb SLD, Laurenti R. O sistema de informações sobre mortalidade: problemas e propostas para o seu enfrentamento II - Mortes por causas externas. Rev bras epidemiol 2002; 5(2): 212-23.
-
11Kleinschmitt SC, Wadi YM, Stadutto JA. Análise espacial dos homicídios: um estudo sobre o Estado do Paraná. In: Encontro Nacional da Associação Brasileira de Estudos Regionais e Urbanos(VII ENABER); 2009, São Paulo (Br). São Paulo: FAE-USP; 2009. p. 1-20.
-
12Cruz OG. Homicídios no Estado do Rio de Janeiro: Análise da distribuição espacial e sua evolução [dissertação de mestrado]. São Paulo: Faculdade de Saúde Pública da USP; 1996.
-
13IBGE - Instituto Brasileiro de Geografia e Estatística. Censo demográfico 2010, Disponível em http://www.sidra.ibge.gov.br/cd/cd2010universo.asp?o=5&i=P (Acessado em 30 de outubro de 2011).
» Disponível em http://www.sidra.ibge.gov.br/cd/cd2010universo.asp?o=5&i=P -
14IPEA - Instituto de Pesquisas Econômicas Aplicadas. Disponível em http://www.ipeadata.gov.br/ (Acessado em 30 de outubro de 2011).
» Disponível em http://www.ipeadata.gov.br/ -
15Bailey TC, Gatrell AC. Interactive spatial data analysis. Essex: Longman Scientific; 1995.
-
16R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. Disponível em http://www.R-project.org.
» Disponível em http://www.R-project.org -
17Câmara G, Druck S, Monteiro AM, Carvalho MS. Análise Espacial de Dados Geográficos. Brasília: EMBRAPA, 2004. Disponível em http://www.dpi.inpe.br/gilberto/livro/analise/cap5-areas.pdf (Acessado em 31 de janeiro de 2011).
» Disponível em http://www.dpi.inpe.br/gilberto/livro/analise/cap5-areas.pdf -
18Sá Carvalho M, Bivand R. Exploring and analysing rates of Hansen's Disease in Olinda, Brazil. 2003. Disponível em: http://leg.ufpr.br/lib/exe/fetch.php/pessoais:marilia:bivandmarilia.pdf (Acessado em 19 de novembro de 2011).
» Disponível em: http://leg.ufpr.br/lib/exe/fetch.php/pessoais:marilia:bivandmarilia.pdf -
19Barata RB, Ribeiro MCSA, de Sordi M. Desigualdades sociais e homicídios na cidade de São Paulo, 1998. Rev Bras Epidemiol 2008; 11(1): 3-13.
-
20Gawryszewski VP, Costa LS. Homicídios e desigualdades sociais no Município de São Paulo. Rev Saúde Pública 2005; 39(2): 191-7.
-
21Briceño-León R. Urban violence and public health in Latin America: a sociological explanatory framework. Cad Saúde Pública 2005; 21(6): 1629-64.
-
22Santana FS. Homicídios em Salvador, 1998-2001: Efeitos das desigualdades socioeconômicas na mortalidade por violência [dissertação de mestrado]. Rio de Janeiro: Escola Nacional de Saúde Pública FIOCRUZ; 2004.
-
23Nadanovsky P, Celeste RK, Wilson M, Daly M. Homicide and impunity: an ecological analysis at state level in Brazil. Rev Saúde Pública 2009; 43(5): 733-42.
-
24World report on violence and health: a summary [Relatório mundial sobre violência e saúde: um resumo]. Genebra, Organização Mundial da Saúde, 2002.
-
25Messner SF, Anselin L, Baller RD, Hawkins DF, Deane G, Tolnay SE. The spatial patterning of county homicide rates: an application of exploratory spatial data analysis. Journal of Quantitative Criminology 1999; 15(4): 423-50.
-
26Breetzke GD. A socio-structural analysis of crime in the city of Tshwane, South Africa. S Afr J Sci 2012; 106(11/12): 4-7.
-
27Rothman KJ, Greenland S, & Lash TL. Modern Epidemiology. Philadelphia: Lippincott, Williams & Wilkins; 2008.
-
28Minayo MCS. A violência social sob a perspectiva da saúde pública. Cad Saúde Pública 1994; 10 (Suppl 1): 7-18.
-
Financial support: Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro.
Publication Dates
-
Publication in this collection
Mar 2014
History
-
Received
27 Nov 2012 -
Reviewed
23 Sept 2013 -
Accepted
13 Nov 2013