Alcohol consumption in homicide victims in the ... - Wiley Online Library

RESEARCH REPORT

doi:10.1111/j.1360-0443.2009.02716.x

Alcohol consumption in homicide victims in the city of São Paulo add_2716

1998..2006

Gabriel Andreuccetti1, Heráclito Barbosa de Carvalho1, Júlio de Carvalho Ponce2, Débora Gonçalves de Carvalho3, Túlio Kahn4, Daniel Romero Muñoz2 & Vilma Leyton2 University of São Paulo Medical School, Department of Preventive Medicine, São Paulo, Brazil,1 University of São Paulo Medical School, Department of Legal Medicine, São Paulo, Brazil,2 Institute of Legal Medicine, São Paulo, Brazil3 and Public Security Office, São Paulo, Brazil4

ABSTRACT Aims To assess the association between alcohol use and victimization by homicide in individuals autopsied at the Institute of Legal Medicine in São Paulo, Brazil. Design Cross-sectional study. Setting Excessive consumption of alcohol is a serious public health issue and a major factor in triggering violent situations, which suggests a strong association between alcohol ingestion and becoming a victim of homicide. Participants Data from 2042 victims of homicides in 2005 were obtained from medical examiner reports. Measurements The victim’s gender, age, ethnicity and blood alcohol concentration (BAC) were collected. The method of death and homicide circumstances, as well as the date, time and place of death were also studied. Findings Alcohol was detected in blood samples of 43% of the victims, and mean BAC levels were 1.55 ⫾ 0.86 g/l. The prevalence of positive BAC levels was higher among men (44.1%) than women (26.6%), P < 0.01. Firearms caused most of the deaths (78.6%), and alcohol consumption was greater among victims of homicide by sharp weapons (P < 0.01). A greater proportion of victims with positive BAC were killed at weekends compared to weekdays (56.4 and 38.5%, respectively; P < 0.01), and the correlation between homicide rates and the average BAC for the central area of the city was positive (rs = 0.90; P < 0.01). Conclusions These results highlight alcohol as a contributing factor for homicide victimization in the greatest urban center in South America, supporting public strategies and future research aiming to prevent homicides and violence related to alcohol consumption. Keywords

Alcohol, blood alcohol concentration, homicide, public health, South America, victimization, violence.

Correspondence to: Gabriel Andreuccetti, University of São Paulo Medical School, Department of Preventive Medicine, Av. Dr Arnaldo, 455 2° Andar, CEP 01246-903, São Paulo/SP, Brazil. E-mail: [email protected] Submitted 27 November 2008; initial review completed 3 March 2009; final version accepted 17 June 2009

INTRODUCTION Alcohol is a licit, easily available and low-cost drug, and is one of the most consumed psychoactive substances world-wide. A national survey conducted in Brazil in 2005 revealed that 74.6% are alcohol users (life-time use) and 12.3% are alcohol dependents, according to US National Household Surveys on Drug Abuse (NHSDA) criteria for alcohol dependence [1]. The World Health Organization (WHO) estimates that there are 2 billion alcohol consumers world-wide and that alcohol causes 1.8 million deaths each year [2]. It is also considered an important risk factor for traffic accidents, suicides, homicides and other causes of injuries [3–5].

Ethanol can cause a decrease in cognitive ability and increases aggressive reactions to a provocative stimulus [6]. Furthermore, an intoxicated individual has a diminished capacity of recognizing and acting upon dangerous circumstances, becoming more vulnerable in situations involving violence and accidents [7]. Wolfgang [8], in a classic US study about homicides, describes the ‘victim-precipitated homicide’, which characterizes a situation where the alcohol-intoxicated victim first provokes the aggressor. Alternately, intoxicated individuals can become easy targets of thieves and other predatory criminals, which can often culminate in a homicide [9]. Rates and patterns of violent death vary by country and region. Poorer countries, especially those with large

© 2009 The Authors. Journal compilation © 2009 Society for the Study of Addiction

Addiction, 104, 1998–2006

Alcohol consumption in homicide victims

gaps between the rich and the poor, tend to have higher rates of homicide than wealthier countries, where suicides are more prevalent [10]. An estimated 1.6 million deaths were caused by violence around the world in 2000, a third of which were homicide victims [11]. During the same year, homicides were the third leading cause of death in Brazil (following cerobrovascular diseases and ischemic heart disease), with a rate of 30.5 deaths per 100 000 inhabitants, and the leading cause for people aged 15–44 years [12]. In São Paulo, the largest and most populated city in Brazil, with more than 10 million inhabitants and a territory extension of 1509 km2, homicide rates have declined since 2000. The homicide rate was 57.3 per 100 000 inhabitants in 2000 and declined to 26 deaths per 100 000 inhabitants in 2005 [13], probably because of police enforcement and the 2004 restrictive gun law implementation [14]. However, these rates have remained high compared to the average rate in high-income countries for all violent deaths (14.4 per 100 000) [11], and they are even more remarkable in some poor districts of São Paulo, where the homicide rate reached 74.7 deaths per 100 000 inhabitants in 2005. Alcohol is present in post-mortem samples in a great number of medico–legally investigated deaths [15], making the determination of blood alcohol concentration (BAC) a valuable tool for establishing an association between alcohol consumption and violence. A meta-analysis including 65 English-language medical examiner studies of non-traffic injury fatalities found the highest aggregate percentage of deceased people positive for alcohol among homicides (47.1%), followed by unintentional injury deaths (38.5%) and suicide cases (29%) [16]. Despite an alarming number of homicides and other violent deaths in Brazil, there is a lack of studies that allow epidemiological approaches to support governmental decisions regarding public health policies on this issue. There is also little information about the association between alcohol and homicides [17], which should be included in preventive strategies aiming to reduce the number of violent deaths and their related financial costs. In this sense, the city of São Paulo can be a useful model for other Brazilian regions and urban centers in developing countries, as the influence of alcohol on violent deaths and great numbers of homicides have been investigated thoroughly in the city. In addition, São Paulo represents a huge challenge to the control of the harmful use of alcohol as it lies in the region with the greatest life-time use of alcohol in Brazil (80.4%) [1], and has a high density of alcohol retail outlets (an estimated proportion of one to every 16 people in a São Paulo suburb) [18].

1999

The purpose of the present study was to examine how blood alcohol in homicide victims killed in São Paulo during 2005 are related to demographic characteristics of these victims and to homicide circumstances, information that is essential for formulating policies to prevent alcohol-related violence.

METHODS A cross-sectional study was conducted including information from homicide victims autopsied at the Institute of Legal Medicine in the city of São Paulo between January and December 2005. Homicides were defined as injuries inflicted by another person with intent to injure or kill, by any means [19]. Demographic characteristics of victims and homicide data were obtained from medical examiner reports and linked to data from the results of toxicological tests. Our study comprised only homicide victims for whom the determination of BAC was requested. Measurements of blood alcohol levels in homicide victims can be requested by the police authority or medical examiners but they are not mandatory in Brazil for violent deaths, except for victims of traffic accidents. Thus, alcohol blood tests are not performed in all homicide victims autopsied because medical examiners may deem it unnecessary to perform an internal examination to establish the cause of death or for any other reasons (e.g. advanced decomposition of the corpse). Data collected included information on the victims’ gender, age, ethnicity, method of death and BAC levels. Based on medical examiner reports, ethnic identification was classified into five groups: white (Caucasian), black African American, mulatto, Asian and unknown (for those victims whose skin color could not be identified). Mulatto is the term used commonly in Brazil to designate the offspring resulting from the union of white and black people. The method of death was defined based on the weapon or method utilized to inflict the injury responsible for the homicide, and was classified as: injury by firearms, injury by sharp weapons and other methods (hanging, beating, drowning and others). Blood samples were analyzed by headspace gas chromatography for measuring alcohol levels. A BAC level greater than 0.2 g/l was used as the main threshold value (positive confirmation of alcohol in the victim’s blood), as in Brazil it is illegal to drive a motor vehicle with BAC higher than this value. Above this limit, BAC was divided into four categories as follows: 0.3–0.5 g/l; 0.6– 1.5 g/l; 1.6–2.5 g/l; >2.5 g/l. With the purpose of analyzing other potential circumstances influencing the occurrence of homicides, date, time and place of death were obtained from a


Addiction, 104, 1998–2006

2000

Gabriel Andreuccetti et al.

confidential database of the Public Security Office of São Paulo. All data were collected using a standardized form and victims with inconsistent records of BAC levels (no information available or levels higher than 7 g/l) were excluded from the study (five cases), resulting in a total of 2042 victims analyzed. Information on age and circumstances of death were not available in some cases, with an exclusion percentage of 1.3% (26 cases) for age and 22.5% (459 cases) for circumstances of death, resulting in a total of 2016 and 1583 victims analyzed for these variables, respectively. Information on the circumstances of death was missing, due most probably to incomplete integration of data obtained from police and medical examiner reports, as the combination of these two databases did not provide these data for all victims analyzed in this study. A geospatial analysis was conducted to study the place of death and the influence of alcohol consumption in those deaths. A map of the city of São Paulo divided into 96 districts was constructed based on homicide rates calculated by dividing the numbers of victims included in this study in a given district by the respective estimated population in 2005. This map was then compared to a similar map including, rather than homicide rates, the average positive BAC of homicide victims per district. A color gradient was applied to the maps to indicate higher (red) and lower (yellow) homicide rates and mean BAC using the software MapInfo Professional version 8.0. Databases were checked by at least two different researchers to avoid potential misinterpretation. Descriptive statistics were used to generate measures of frequency, means and standard deviations. All mean BAC levels (unless noted otherwise) refer to that found in victims with a positive confirmation for alcohol. Statistical differences between groups for nominal variables were analyzed using Pearson’s c2 test, while for interval variables with normal distribution, Student’s t-test was applied. Spearman’s correlation coefficient was used to test the association between homicide rates and average BAC among different districts in the map (Stata/SE version 9.0 for Windows). For the assessment of this correlation, districts were grouped into five regions: central, western, northern, eastern and southern, according to the city’s administrative division. Differences with P < 0.01 were considered statistically significant. This study was based on confidential data and all victims’ identity was preserved. The study protocol was approved by the Research Ethics Committee of University of São Paulo Medical School.

RESULTS Homicide victims were predominantly males (93%; 1899 cases). Their mean age was 30.4 ⫾ 11.5 years, and females were slightly older (32.2 ⫾ 13.5 years) than males (30.3 ⫾ 11.3 years) (t = 1.89; P = 0.03). Table 1 presents demographic information, BAC levels and methods of death. In the sample studied, 43% (876) of the victims had a positive BAC. The minimum positive BAC found was 0.3 g/l and the maximum BAC was 6.3 g/l, with mean BAC levels of 1.55 ⫾ 0.86 g/l. Mean BAC levels in males (1.56 ⫾ 0.86 g/l) were greater than those seen in females (1.21 ⫾ 0.65 g/l), and the difference was statistically significant (t = 2.48; P < 0.01). The prevalence of positive BAC was also greater among men (44.1%) than women (26.6%), c2 = 16.73; df = 1; P < 0.01. Most intoxicated victims were aged 25–34 years, representing 37.6% (325 victims) with positive BAC, followed by those who were aged 15–24 years (229). Otherwise, the proportion of positive BAC was greater

Table 1 Description of homicide victims by demographic characteristics, blood alcohol concentration levels and method of death.

Category Gender Male Female Total Age (years) 64 Total Ethnicity White Mulatto Black Asian Unknowna Total Method of death Firearms Sharp weapons Othersb Total

All n (%)

Positive n (%)

Negative n (%)

1899 (93.0) 838 (44.1)* 1061 (55.9) 143 (7.0) 38 (26.6)* 105 (73.4) 2042 876 1166 17 704 672 389 154 47 33 2016

(0.8) 3 (17.7) (34.9) 229 (32.5) (33.3) 325 (48.4) (19.3) 211 (54.2) (7.6) 78 (50.7) (2.3) 13 (27.7) (1.6) 4 (12.1) 863

14 475 347 178 76 34 29 1153

(82.4) (67.5) (51.6) (45.8) (49.4) (72.3) (87.9)

1053 743 212 18 16 2042

(51.6) 439 (41.7) (36.4) 336 (45.2) (10.4) 92 (43.4) (0.9) 1 (5.6) (0.8) 8 (50.0) 876

614 407 120 17 8 1166

(58.3) (54.8) (56.6) (94.4) (50.0)

1605 (78.6) 644 (40.1)* 961 (59.9) 193 (9.5) 117 (60.6)* 76 (39.4) 244 (11.9) 115 (47.1)* 129 (52.9) 2042 876 1166

a Victims whose identification of skin color was not possible. bOther methods such as hanging, beating, drowning, etc. *The differences between groups were statistically significant (P < 0.01).


Addiction, 104, 1998–2006


among victims aged 35–44 years (54.2%; n = 211). It is worth mentioning that 16.9% (23 of 136 victims) of those younger than 18 were found to be BAC positive. The majority of the victims (51.6%) were Caucasian, followed by mulattoes (36.4%) and black (10.4%). Data analysis did not show any statistical difference (c2 = 2.22; df = 2; P = 0.33) in alcohol consumption among whites, mulattoes and black (Asians and unknown groups were excluded from the analysis because of the low number of victims in each group: 18 and 16, respectively). With regard to method of death, of all homicide victims, 78.6% (1605 victims) were injured by firearms, 9.5% (193) by sharp weapons and 11.9% (244) by other methods. Alcohol consumption by victims was different according to the method of death. Among those injured by firearms, 40.1% (644) had consumed alcohol while a higher proportion (47.1%; 115 victims) was found among those injured by other methods. An even higher proportion of alcohol consumption was seen among homicides by sharp weapons, reaching 60.6% (117). These differences were found to be statistically significant (c2 = 31.57; df = 2; P < 0.01). Table 2 highlights differences by gender according to BAC results. More than half the victims were male

2001

teenagers and young adults, as males aged 15–34 years represented 63.7% (1285 victims) of total homicides for which age was assessed. Despite a much lower number of victims in all age groups, female victims were also most often young. In all victims studied, no differences by gender were found for alcohol consumption among ethnic groups. A greater proportion of homicides by firearms was found in both males (80.5%) and females (53.1%). The percentage of homicides by a sharp weapon and other methods was greater among women (18.9 and 28%, respectively) than men (8.7% and 10.7%, respectively) (c2 = 60.18; df = 2; P < 0.01). Table 3 shows the distribution of BAC levels by age groups in males and females. The predominant level for positive BAC was 0.6–1.5 g/l for both males and females, with no significant distribution differences of positive BAC levels (c2 = 7.06; df = 3; P = 0.07). However, the distribution of positive BAC levels was found to be different in males by age (c2 = 39.30; df = 18; P < 0.01), with a predominance of levels higher than 0.5 g/l in those aged 25–54 years, but the same was not seen among females (c2 = 13.80; df = 12; P = 0.31).

Table 2 Comparison of demographic characteristics and method of death by gender according to blood alcohol concentration levels of homicide victims. Male

Female

All n (%) Age (years) 64 Total Ethnicity White Mulatto Black Asian Unknowna Total Method of death Firearms Sharp weapons Othersb Total

Positive n (%)

Negative n (%)

All n (%)

Positive n (%)

Negative n (%)

12 665 620 365 140 43 29 1874

(0.6) (35.5) (33.1) (19.5) (7.5) (2.3) (1.5)

2 218 313 201 74 13 4 825

(16.7) (32.8) (50.5) (55.1) (52.9) (30.2) (13.8)

10 447 307 164 66 30 25 1049

(83.3) (67.2) (49.5) (44.9) (47.1) (69.8) (86.2)

5 39 52 24 14 4 4 142

(3.5) (27.5) (36.6) (16.9) (9.9) (2.8) (2.8)

1 11 12 10 4 0 0 38

(20.0) (28.2) (23.1) (41.7) (28.6)

4 28 40 14 10 4 4 104

(80.0) (71.8) (76.9) (58.3) (71.4) (100) (100)

985 690 196 14 14 1899

(51.9) (36.3) (10.3) (0.7) (0.7)

424 319 87 1 7 838

(43.0) (46.2) (44.4) (7.1) (50.0)

561 371 109 13 7 1061

(57.0) (53.8) (55.6) (92.9) (50.0)

68 53 16 4 2 143

(47.6) (37.1) (11.2) (2.8) (1.4)

15 17 5 0 1 38

(22.1) (32.1) (31.3)

53 36 11 4 1 105

(77.9) (67.9) (68.8) (100) (50.0)

1529 (80.5)* 166 (8.7)* 204 (10.7)* 1899

622 (40.7) 107 (64.5) 109 (53.4) 838

907 (59.3) 59 (35.5) 95 (46.6) 1061

76 (53.1)* 27 (18.9)* 40 (28.0)* 143

(50.0)

22 (28.9) 10 (37.0) 6 (15.0) 38

54 (71.1) 17 (63.0) 34 (85.0) 105

a Victims whose identification of skin color was not possible. bOther methods such as hanging, beating, drowning, etc. *The differences between groups were statistically significant (P < 0.01).


Addiction, 104, 1998–2006


20.0 2.6 1.9 8.3 7.1 0 0 4.2

0 23.1 13.5 20.8 7.1 0 0 15.5

0 2.6 7.7 8.3 14.3 0 0 6.3

0 0 0 4.2 0 0 0 0.7

The majority of homicides (67.2%) occurred between 6 p.m. and 6 a.m. Forty-two per cent of homicides (667 victims) took place at weekends (Saturdays and Sundays), mainly in the evenings and at night. The day of the week with the highest number of homicide victims (134) was Sunday, between midnight and 6 a.m. The greater proportion of victims with positive BAC was also found during weekend evenings and nights. Of the victims of homicides, 56.4% occurring during weekends were positive for alcohol, contrasting with 38.5% on weekdays (c2 = 49.41; df = 1; P < 0.01). The city areas with higher homicide rates matched with those areas with higher mean BAC of homicide victims, as seen in the geospatial analysis presented in Fig. 1. It can be noted that homicide victims are clustered in the central and peripheral areas (northern, southern and eastern) of the city, the same regions with, on average, higher BAC levels. However, Spearman’s correlation coefficient between homicide rates and average BAC for the 96 districts was not significant (rs = 0.13; P = 0.19), but a strong positive correlation was found between these two variables for the eight districts comprising the central area (rs = 0.90; P < 0.01), while this correlation was not significant for the remaining areas.

12 665 620 365 140 43 29 1874 Age (years) 64 Total

*The distribution of positive BAC levels was different in males of different age groups (c2 = 39.30; df = 18; P < 0.01).

80.0 71.8 76.9 58.3 71.4 100 100 73.2 5 39 52 24 14 4 4 142 0 2.3 6.1 10.7 9.3 4.7 3.4 5.8

n

83.3 67.2 49.5 44.9 47.1 69.8 86.2 56.0

0 3.9 4.0 2.7 6.4 2.3 0 3.8

16.7 18.3 23.4 22.2 19.3 7.0 6.9 20.4

0 8.3 16.9 19.5 17.9 16.3 3.4 14.1

n

Negatives 0–0.2 % >2.5 % 1.6–2.5 % 0.6–1.5 % 0.3–0.5 % Negatives 0–0.2 % Blood alcohol concentration levels (g/l)

Male*

Table 3 Distribution of blood alcohol concentration (BAC) levels by age groups in both male and female victims.

Female

0.3–0.5 %

0.6–1.5 %

1.6–2.5 %

>2.5 %

2002

DISCUSSION The results show that alcohol consumption was common among homicide victims in the city of São Paulo during 2005. According to the Public Security Office of the State of São Paulo, 2684 homicide victims were reported in the city of São Paulo in 2005. Of these, 2042 victims (76%) were analyzed. Although we do not intend to consider the study sample as representative of homicide victims in São Paulo, there is no reason to suspect that they are any different in respect to alcohol consumption. Also, there is no indication that blood alcohol tests were performed because there was an assumption that the victim was under the influence of alcohol. Furthermore, because the time elapsed between injury (or alcohol consumption) and death was not an exclusion criterion in our study, the results are likely to underestimate the proportion of cases in which alcohol was present in victims’ blood at their death. People with high BAC levels, as found in the present study, can have emotional instability, decreased inhibition and loss of critical judgement [20], which supports the hypothesis that pharmacological effects of alcohol can induce violent behavior and diminish awareness of potential harmful and life-threatening situations, making an intoxicated individual a potential victim of crimes such as homicides.


Addiction, 104, 1998–2006


2003

Figure 1 Maps of the city of São Paulo divided by districts. (a) Homicide rate per 100 000 inhabitants, calculated by dividing the number of victims undergoing blood alcohol tests in a given district by the respective estimated population in 2005* (n = 1583). (b) Average blood alcohol concentration (grams of ethanol/l of blood) of homicide victims per district (n = 1583). *Source: estimated population in the districts of Sao Paulo (total population for the city of São Paulo: 10 895 521 inhabitants) according to the Brazilian Institute of Geography and Statistics and the Municipal Secretary of Planning of São Paulo (IBGE/SEMPLA), 2005

Female victims showed a proportion of positive cases for ethanol and mean BAC levels lower than males, suggesting that women are less prone to abuse and to be dependent on alcohol [21]. This may also contribute to fewer female homicide victims. Individuals aged 25–34 years comprised the largest number of intoxicated victims both in males and females, which is consistent with the current high prevalence of alcohol consumption in this age group in Brazil [1], and corroborates the existence of an association between alcohol consumption and victimization by homicides. Additionally, the study found victims younger than 18 with positive BAC levels, which is a cause of concern as they are below the legal drinking age in Brazil. Also noteworthy is the magnitude of alcohol consumption revealed in the analysis of the most frequent BAC levels found in males aged 25–54 years, who presented the highest prevalence of alcohol levels greater than 0.5 g/l. Nordrum et al. [15] conducted a study in Norway investigating BAC levels in 1113 violent death victims and found a predominant ethanol level similar to our study, ranging between 1.0 and 2.9 g/l. The authors also suggest that BAC greater than 0.5 g/l in post-mortem samples should be considered a potential contributing factor to violent deaths.

Statistical analysis of the data did not reveal any significant differences in the proportion of victims with positive BAC levels by ethnicity. This association, however, has already been reported by previous studies [9], which reported that alcohol was more common among victims of Latin origin than blacks, whites and others in Los Angeles. It is important to emphasize that the distribution of victims by age and ethnicity in this study is consistent with the demographic statistics of the metropolitan area of São Paulo, where the majority of the population is comprised of individuals aged 15–49 years (56.2%) and the ethnic group with the greatest representation is white (60.3%), followed by mulatto (30.3%) and black (7.2%) [22], demonstrating that the sample selection for the present study was not biased. A firearm was the most commonly used instrument in homicides, especially among men. Compared to men, female homicides were perpetrated more often using sharp weapons and other instruments, which points to a gender-related difference in the social dynamic involved in this type of death. The literature suggests that violence against women is usually perpetrated by family members or intimate partners and, hence, results from ‘private’ conflicts. Among males, cases of aggression are more likely to involve


Addiction, 104, 1998–2006

2004


strangers in a public area and be related to urban criminality [23]. The high proportion of positive BAC levels in cases involving a sharp weapon reported in the present study and in other studies [8,16] may indicate greater impulsiveness in crimes perpetrated with this type of instrument, where the people involved look for an object that can be used as a weapon (for example, a knife or a pair of scissors) while intoxicated. The findings regarding alcohol consumption according to the day of the week and time of the occurrence of homicides revealed a higher proportion of alcoholrelated homicides on Saturday and Sunday evenings and nights, which is probably related to high-risk consumption of alcohol in bars and at parties [21]. The association between daily distribution of alcohol intoxication and homicide deaths has already been demonstrated in Udmurt Republica, Russia [24], where high social tolerance for heavy drinking may increase the risk of violent outcomes. Furthermore, the geospatial analysis used in this study provided important data on the geographical distribution of homicide victims according to alcohol use, as demonstrated by the positive correlation between homicide rates and average BAC for the central area. Even though further studies using additional geostatistical methodology are recommended, the results presented here are consistent with previous studies which indicated that peripheral and central areas are hotspots for homicides in the city of São Paulo [25]. The central area has a constant traffic of people at all hours of the day, and includes a vast variety of commercial buildings, restaurants and bars. Peripheral regions are typically residential areas with high levels of social deprivation. It has already been demonstrated that they have a high density of alcohol outlets [18], although no correlation was found in this study. Policies aiming to reduce high-risk drinking can reduce violent behaviors that often culminate in alcoholrelated injuries, such as motor vehicle accidents and assaults [26]. Restrictions such as limited hours or days of alcohol sale have been shown to be effective in reducing injury associated with violence in Brazil [27]. This strategy may result in a decrease of homicides if applied in conjunction with an approach targeting risk areas and tackling the association between alcohol and homicide. Similarly, age restrictions for alcohol purchase must be addressed because teenagers have easy access to alcoholic beverages in Brazil [28]. In this sense, the findings of our study have significant health implications because they suggest the need for specific alcohol policies based on demographic characteristics, regions and conditions that influence alcohol-related victimization by homicide in the city of São Paulo.

A traffic legislation recently adopted in Brazil established one of the lowest BAC limits (0.2 g/l) for drivers in the world, aiming to reduce the number of alcoholrelated traffic accidents, which seems to be effective when accompanied by increased law enforcement [29]. Policies for alcohol measurement and control in victims and offenders of violence cases such as homicides, assaults and domestic violence could also play an important role in the prevention and reduction of harmful health and social effects related to alcohol use. A number of limitations should be considered in the present study, as well as other studies, of the association between alcohol consumption and victimization by homicides. First, it is difficult to obtain alcohol levels in ‘control’ populations or non-fatal assault victims, not allowing estimates of relative risk [9]. A similar approach has already been used with injured patients screened for alcohol consumption in emergency rooms, and the estimated relative risk of injury within an hour after alcohol use, compared with no alcohol consumption, was 4.33 [30]. Secondly, the present study could not assess different physiological and behavioral responses produced by alcohol consumption in different individuals and BAC levels for perpetrators, even though it is believed that the influence of alcohol tends to be similar for both offenders and their victims at the moment of crime [31,32]. Other methodological limitations include the exclusion of 459 cases from the analyses of circumstances of death, which could, possibly, have affected the results for these variables. However, exclusion was shown to be arbitrary, as the cases excluded had demographic characteristics similar to the rest of the sample. Finally, the high rate of homicides in Brazil contrasts with that seen in high-income countries [10]. There are different alcohol consumption patterns and social differences among countries, which should be considered in the generalizability of this study’s findings across international borders. Although homicide rates have been falling in São Paulo, this reduction is not due to increased control on alcohol availability, and efforts should be directed towards limiting alcohol availability in high-risk areas, as suggested in the present study. Even though a causal relationship could not be established between alcohol use and homicide occurrence, the association between alcohol consumption and victimization by homicide has been reinforced, further improving the understanding of alcohol as a contributing factor for violence and providing substantial input for the promotion of alcohol policies. Moreover, the integration of information from police and medical examiner reports can improve current knowledge about the causes of alcohol-attributable death


Addiction, 104, 1998–2006


and provide important data for further investigation of preventive strategies aiming to reduce harmful alcoholrelated consequences. Declarations of interest None. Acknowledgements CNPq-Conselho Nacional de Desenvolvimento Científico e Tecnológico (Brazilian National Council for Scientific and Technological Development) provided financial support. The authors also wish to thank Dr Hideaki Kawata and Dr Neide S.F. Oliveira from the Institute of Legal Medicine, and Sueli A. Moraes and Miriam Regina de Souza for their valuable contributions. References 1. Carlini E. A., Galduroz J. C. F. II. Levantamento Domiciliar Sobre o Uso de Drogas Psicotrópicas no Brasil: Estudo Envolvendo as 108 Maiores Cidades do País [II Domestic Inquiry about the Use of Psychoactive Drugs in Brazil: Study Involving the 108 Major Cities of the Country]. São Paulo, Brazil: CEBRID/ UNIFESP; 2005. 2. World Health Organization. Global Status Report on Alcohol 2004. Geneva: World Health Organization; 2004. 3. Bedford D., O’Farrell A., Howell F. Blood alcohol levels in persons who died from accidents and suicide. Ir Med J 2006; 99: 80–3. 4. Rosen M., Haglund B. Trends in alcohol-related mortality in Sweden 1969–2002: an age-period-cohort analysis. Addiction 2006; 101: 835–40. 5. Skibin L., Bilban M., Balazic J. Harmful alcohol use of those who died a violent death (the extended region of Ljubljana 1995–1999). Forensic Sci Int 2005; 147: S49–52. 6. Exum M. L. The application and robustness of the rational choice perspective in the study of intoxicated and angry intentions to aggress. Criminology 2002; 40: 933–66. 7. Peterson J. B., Rothfleisch J., Zelazo P. D., Pihl R. O. Acute alcohol intoxication and cognitive functioning. J Stud Alcohol 1990; 51: 114–22. 8. Wolfgang M. E. Patterns in Criminal Homicide. Philadelphia, PA: University of Pennsylvania Press; 1958. 9. Goodman R. A., Mercy J. A., Loya F., Rosenberg M. L., Smith J. C., Allen N. H. et al. Alcohol use and interpersonal violence: alcohol detected in homicide victims. Am J Public Health 1986; 76: 144–9. 10. Rosenberg M., Butchart A., Mercy J., Narasimhan V., Waters H., Marshall M. S. Interpersonal violence. In: Jamison D. T., Breman J. G., Measham A. R., Alleyne G., Claeson M., Evans D. B. et al., editors. Disease Control Priorities in Developing Countries, 2nd edn. New York, NY: Oxford University Press and the World Bank; 2006, chapt. 40, p. 755–70. 11. Krug E. G., Dahlberg L. L., Mercy J. A., Zwi A. B., Lozano R., editors. World Report on Violence and Health. Geneva: World Health Organization; 2002. 12. Pan American Health Organization. Health Statistics from the Americas-2006 Edition. Washington, DC: Pan American Health Organization; 2006. Available at: http://www. paho.org/English/DD/AIS/HSA2006.htm (accessed 7 May 2009).

2005

13. Centre for the Study of Violence, University of São Paulo. Homicide Rates for the City of São Paulo Divided by Districts. São Paulo, Brazil: Centre for the Study of Violence, University of São Paulo; 2005. Available at: http://www. nevusp.org/portugues/index.php?option=com_content& task=view&id=1372&Itemid=71 (accessed 13 November 2008). 14. Marinho de Souza Mde F., Macinko J., Alencar A. P., Malta D. C., de Morais Neto O. L. Reductions in firearm-related mortality and hospitalizations in Brazil after gun control. Health Aff (Millwood) 2007; 26: 575–84. 15. Nordrum I., Eide T. J., JŁrgensen L. Alcohol in a series of medico-legally autopsied deaths in northern Norway 1973– 1992. Forensic Sci Int 2000; 110: 127–37. 16. Smith G. S., Branas C. C., Miller T. R. Fatal nontraffic injuries involving alcohol: a metaanalysis. Ann Emerg Med 1999; 33: 659–68. 17. Gawryszewski V. P., Kahn T., de Mello Jorge M. H. [Linking of information from health and security databases on homicides]. Rev Saude Publica 2005; 39: 627–33. 18. Laranjeira R., Hinkly D. Evaluation of alcohol outlet density and its relation with violence. Rev Saude Publica 2002; 36: 455–61. 19. World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death, vol. 1, 10th revision. Geneva: World Health Organization; 1992. 20. Dubowski K. M. Alcohol determination in the clinical laboratory. Am J Clin Pathol 1980; 74: 747–50. 21. Laranjeira R., Pinsky I., Zaleski M., Caetano R. I Levantamento Nacional Sobre os Padrões de Consumo de Álcool na População Brasileira [I. National Inquiry about Alcohol Consumption Patterns in Brazilian Population]. Brasília, Brazil: Brazilian National Antidrug Secretariat (SENAD); 2007. 22. Brazilian Institute of Geography and Statistics. Síntese de Indicadores Sociais [Summary of Social Indicators]. Rio de Janeiro: IBGE; 2007. Available at: http://www.ibge.gov.br/ home/estatistica/populacao/condicaodevida/ indicadoresminimos/sinteseindicsociais2007/default.shtm (accessed 15 March 2009). 23. Kellermann A. L., Mercy J. A. Men, women, and murder: gender-specific differences in rates of fatal violence and victimization. J Trauma 1992; 33: 1–5. 24. Pridemore W. A. Weekend effects on binge drinking and homicide: the social connection between alcohol and violence in Russia. Addiction 2004; 99: 1034–41. 25. Camargo E. C., Druck S., Monteiro A. M., Freitas C. C., Camara G. [Mapping of homicide risk with binomial cokriging and simulation: a case study in Sao Paulo, Brazil]. Cad Saude Publica 2008; 24: 1493–508. 26. Holder H. D., Gruenewald P. J., Ponicki W. R., Treno A. J., Grube J. W., Saltz R. F. et al. Effect of community-based interventions on high-risk drinking and alcohol-related injuries. JAMA 2000; 284: 2341–7. 27. Duailibi S., Ponicki W., Grube J., Pinsky I., Laranjeira R., Raw M. The effect of restricting opening hours on alcohol-related violence. Am J Public Health 2007; 97: 2276–80. 28. Romano M., Duailibi S., Pinsky I., Laranjeira R. [Alcohol purchase survey by adolescents in two cities of State of Sao Paulo, Southeastern Brazil]. Rev Saude Publica 2007; 41: 495–501. 29. Mann R. E., Macdonald S., Stoduto L. G., Bondy S., Jonah B., Shaikh A. The effects of introducing or lowering legal per se


Addiction, 104, 1998–2006

2006


blood alcohol limits for driving: an international review. Accid Anal Prev 2001; 33: 569–83. 30. Borges G., Cherpitel C., Mittleman M. Risk of injury after alcohol consumption: a case–crossover study in the emergency department. Soc Sci Med 2004; 58: 1191–200.

31. Gillies H. Homicide in the West of Scotland. Br J Psychiatry 1976; 128: 105–27. 32. Chervyakov V. V., Shkolnikov V. M., Pridemore W. A., McKee M. The changing nature of murder in Russia. Soc Sci Med 2002; 55: 1713–24.


Addiction, 104, 1998–2006