Physical violence inside female prisons in Brazil: prevalence and related factors Violência física dentro das prisões femininas no Brasil: prevalência

Resumo O presente artigo objetiva identificar a prevalência de violência física no interior das prisões femininas brasileiras, bem como fatores associados. Estudo transversal de abrangência nacional, realizado em 15 unidades prisionais femininas nas cinco regiões do Brasil em amostra selecionada por múltiplos estágios. Realizouse análise univariada; análise estratificada em relação ao desfecho (sofrer violência física dentro da prisão) e as variáveis preditoras através do teste de Chi-quadrado de Pearson; cálculo da Odds Ratio; e regressão logística múltipla. Utilizou-se o teste Hosmer-Lemeshow para análise de qualidade de ajuste e adequação do modelo. A prevalência de violência física dentro das prisões femininas brasileiras foi de 37.4%. Houve correlação entre a vitimização da violência física na prisão e as seguintes variáveis: vitimização física prévia (p = 0.013), isolamento (p = 0.000), sofrimento mental (p = 0.003), ingestão abusiva e/ ou dependência alcoólica (p = 0.011), uso atual ou prévio de cocaína injetável (p = 0.002) e ócio prisional (p = 0.003). A violência física tornouse inerente ao sistema prisional feminino brasileiro. Estudos futuros são necessários no intuito de fornecer propostas para intervenções efetivas de modo a prevenir a violência física dentro das instalações prisionais. Palavras-chave Violência física, Prisões, Instituições penais, Prisões femininas Abstract The present study aims to identify the prevalence of physical violence against female prisoners in Brazil, as well as related factors. This is a cross-sectional national survey conducted in 15 female prisons in five regions of Brazil selected in multiple stages. The following types of analysis were performed: univariate analysis; stratified analysis relating the outcome (suffer physical violence inside prison) to predictor variables, using the Pearson chi-square test; calculation of the Odds Ratio (O.R.); and multiple logistic regression. The Hosmer-Lemeshow test was used for analysis of goodness of fit and adequacy of the model. The prevalence of physical violence inside female prisons was 37.4%. There was a correlation between physical violence victimization in prison and the following variables: physical victimization prior to arrest (p = 0.013), solitary confinement (p = 0.000), mental suffering (p = 0.003), current or previous abusive intake of alcohol (p = 0.011), current or previous injection of cocaine (p = 0.002) and not performing prison labor (p = 0.003). Physical violence has become inherent in the Brazilian female prison system. Continued studies are needed to monitor the situation and to develop interventions to prevent physical violence inside the facilities.

introduction Physical violence inside prisons is a prevalent phenomenon worldwide [1][2][3] . Nevertheless, frequent reports uncover severe violations against female inmates' emotional and physical integrity. Violence against women is defined as "any act of gender-based violence that results in, or is likely to result in, physical, sexual or mental harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or in private life" 4 .
More than 625,000 women and girls are held in penal institutions around the world for preventive reasons or for conviction 5 , and this specific population experiences more frequent and severe situations of interpersonal violence than the general female population [6][7][8][9][10] .
Nearly one-third of them are incarcerated in the United States, China, Russia and Brazil 5 . The Brazilian female prison population grew by 567% between 2000 and 2014, while the general growth rate of the prison population was 119% in the same period 11 . Among the specificities of the Brazilian prison system, the institutionalization of gender violence is a classic problem, arising from the imposition of a penal and prison system that is culturally androcentric 6,7,9,14,16 .
Research indicates a significant association between a woman experiencing episodes of physical violence and her violent behavior reproduced in prison. These data shed light on the question of female imprisonment and report an aspect of these women's lives, revealed in a universe of suffered and perpetrated violence previously described in a few studies 6,7,9,15 .
Therefore, it seems that inmates who suffered physical violence are more vulnerable to repeat these experiences in adult life, as victims or authors 9,10 . Nonetheless, there are few studies on institutional violence among female inmates 6,8,14,[17][18][19][20] and research that examines the epidemiology and context of physical violence inside prisons is almost nonexistent 18 . Subsequently, this study aims to identify the physical violence suffered by women in female prisons in Brazil, as well as its associated factors.

materials and methods
A Cross-sectional analytical study of the project "National survey on the health of female prison population and female prison officers" was developed in 15

study population and sample
The population investigated was comprised of incarcerated women in Brazilian prison system (N = 22,496) in 259 female prison facilities in 2012 according to INFOPEN database (Statistical Information of the Brazilian Penitentiary System). Sampling was performed in multiple stages: 1) Initially, the two states that had the largest female prison populations were intentionally selected by region; 2) Prisons were stratified considering the number of inmates by state, capital, metropolitan region and countryside; 3) Prisons with more than 75 residents were included in the sample plan, due to their obligation of having health services 21 , in addition 4/5 of female prisons in Brazil are located in prisons with more than 75 inmates; 4) On this stage 24 prisons were selected involving 10,668 female inmates separated by five Brazilian regions; 5) Then 30% of violence inside the prison was estimated 2 considering a level of significance 95%, power 80%.
The initial sample size comprised 1,861 female inmates.
Then, sample size composed was randomly selected using the Intemodino Group random number generator 22 , and the identification was done through the nominal listing provided in each prison unit. Women who were unable to attend the research for any reason such as administrative custody, isolated or sick to participate on the day the survey was conducted, or lack of fluency in the Portuguese language were excluded from the sample.
There was an 8.3% loss in the initial selection (n = 154), due to the actual number of female inmates found in prison at the time of data collection, 10.5% withdrawal or refusal (n = 195) and impossibility of reaching some prisons facilities and another obstacles resulted in a lack of 9.9% missing data (n = 185). In the end, 1327 female inmates were approached. Therefore, a Complex Survey Data (CSD) was performed based on adjustment the weights in each stratum previous selected for determinate population correction where the clusters are sampled without replacement and also to correct the population number of female prisoners in

Data collection
The Data were collected through self-applied questionnaires, using the Audio Computer-Assisted Self-Interviewing (ACASI) technology using a 10'' tablet. The choice of ACASI was due to its applicability in studies which need sensitive approach, private and health risk behaviors information 24 . The option "I do not know or I do not want to answering" was added in all questions due to no obligation to answer.
The definition of the outcome variable (physical violence) was characterized in Law no. 11,340/06 (Law that provides mechanisms to curb domestic and family violence against women in Brazil.) and in the ICD-10 (International Statistical Classification of Diseases and Related Problems, chapter "Assaults" (X85-Y09).) classification 25 . Questions related to physical violence victimization experienced before and during the imprisonment were expressed in four questions, later grouped for analysis.
"Has anyone ever slapped you, pushed, pinched or pulled your hair to hurt you?" (Y04: Assault through physical strength.) "Has anyone ever slapped, beaten, burned or tried to hang you?" (Y04: Assault through physical strength; X91: Assault by means of hanging, strangulation and suffocation.) "Has anyone ever injured you with a knife, another piercing object (Stanley knife, glass shard, etc.) to hurt you?" (X99: aggression by means of sharp or penetrating object; Y00: aggression through a blunt instrument) "Has anyone ever injured you with a revolver or another firearm, besides other objects that caused injury to hurt you?" (X93 : aggression by means of hand firearm shooting; X94: aggression by shotgun, carbine or larger caliber firearm; X95: aggression by shooting from another firearm or unspecified weapon.) Predictive variables comprised: profile, prison history, alcohol consumption, tobacco, prescribed drugs, illicit drugs and mental suffering. Profile was accessed by age, education level, race/ skin color, religion and marital status (self-reported following criteria set forth in Brazil). Prison history reported number of arrests, reason for current arrest, number of people in the same cell, solitary confinement and prison labor in the facility.
The Alcohol Use Disorders Identification Test (AUDIT) was used to identify different alcohol consumption patterns among female prisoners. The AUDIT is comprised of 10 items and evaluates recent use and alcohol consumption-related problems, in addition to dependence symptoms 26 . The classification of alcohol consumption was stratified into low risk and risk/high risk. Low risk consumption is related to prisoners who obtained 0-7 scores and risk either high risk is related to those who obtained 8 or above scores.
Tobacco consumption was measured through following questions: Have you ever-smoked cigarettes?; How many cigarettes do you smoke a day?; How often you smoke?; How long time you take to smoke since you wake up in the morning?; Does it difficult for you not smoking in not allowed places?; Do you smoke even when you are sick? ; Is there smokers sharing the cell with you? However, only smoking practice was considered as a predicted variable on this study.
Prescribed drugs include benzodiazepines and tranquilizers and illicit drugs comprised marijuana, cocaine (aspirated and/or injectable), ecstasy, LSD, crack, oxidato or oxi (Oxidato or oxi is a highly addictive and hallucinogenic blend of cocaine paste, gasoline, kerosene and quicklime (calcium oxide) 27 ), opioids and solvents.
Self-Reporting Questionnaire-20 (SRQ-20) was used to measure the levels of suspicion of mental suffering. The instrument is comprised of 20 questions, the first four of which are related to physical symptoms, and the other 16 to psycho-emotional ones. The application of SRQ-20 allow the early detection of signs and symptom of mental health compromise, including fatigue, insomnia, irritability, as well as subclinical aspects. It is highly recommended, for population-based studies, especially in groups of worker, in view of the association of signs and symptoms with the reduction of working and social roles. Since it is a tracking instrument, the determination of cutoff points for the detection of cases is fundamental to guarantee sensitivity and specificity. In this study, we adopted the minimum score of 7 affirmative responses to SRQ-20 for CO's, in conformity with the guidelines set forth for women 28 .

Data Analysis
The data were analyzed using Stata ® version 11. The Complex Survey Data (CSD) was applied to adjust the sample loss. The CSD is composed of survey data from the survey design, variables to identify information about the strata, clusters, sampling weights, and population correction. Sampling weights was carried out for precise adjustment estimates. Weighting, clustering, and stratification are considered on survey design to minimize standard errors 23 .
Descriptive statistic analyses were calculated for outcome physical violence victimization inside prison and the predictable variables, which comprises: profile, prison history, alcohol consumption, tobacco, prescribed drugs, illicit drugs and mental suffering. Crude and adjusted logistic regression analyses were performed to assess association the violence outcome and the predictive variables using Pearson's chi-squared test. Predictive variables within increased the probability for suffering physical violence inside the prison up to 20% were taken to the final Logistic Regression model In order to control confounding factors, the odds ratios and 95% confidence intervals were estimated. The Hosmer-Lemeshow test was used for analysis of goodness of fit and adequacy of the model.
The research was approved by the Committee of Ethics in Research under protocol 188.211. Since this research involves human beings, all participants agreed with and signed the Free and Informed Consent Form.

Results
The sample collected consisted of 1,327 women held in the prison units of the following investi- Women participants were mostly younger being under 34 years of age (62.3%). The sociodemographic profile is presented as follows: The majority non-white (71.1%) and had received less than secondary education (78.9%); more than a half (59.5%) arrested for drug trafficking and 16.7% theft or robbery; almost half (47.5%) also reported previous incarceration; share a cell with 6 people or more (67.7%); 41.9% were christians and 37.2% protestant .The majority was mothers (79.9%), having paid legal occupation prior prison (75.6%) ( Table 1).
The prevalence of physical violence inside Brazilian female prisons was 37.4%. In relation to the perpetrators, the prisoners themselves were the most accused (67%), followed by prison officers (21.9%). Among these victims, 67.5% reported having received health care inside the prison unit. As for first aid, in case of more severe violence, the prison officers were the first ones to  Table 2).
The multivariate logistic regression analysis by adjusted models demonstrated a possible association between physical violence victimization inside the prison and the following variables: physical victimization prior to imprisonment (p = 0,013), being taken to solitary confinement (p = 0.000), a positive diagnosis for mental suffering (p = 0.003), current or abusive intake of alcohol / alcohol dependence (p = 0.011), current or previous injecting cocaine use (p = 0.002) and not performing labor in the facility (p = 0.003) ( Table 4).
The Hosmer-Lemeshow chi-square (goodness-of-fit) test was not statistically significant (p = 0.9893), thus, the proposed model is consistent and adequate to explain the observed outcome.

Discussion
It was found that the Brazilian female prisoners have experienced physical violence not only during (37.4%) but also prior incarceration (81.9%). Those experiences were associated to multiple forms of violence, poverty, alcohol / another drugs abuse, and mental suffering. The rates of violence inside prison are high since the prisoners must be protected by the State under human rights law 29 . This fact is supportive to another studies in literature, which asserts that violence and aggression belong to prison defined as a territory of psychological and physical torture 1,9,30 . The finds were consistent with another study also reporting violence inside female prison in United States which found a prevalence of 39.5% 1 . Concerning to physical violence aggressors, inmate-on-inmate violence were the most mentioned (67%) followed by officer-on-inmate (21.9%). Although inmate-on-inmate violence was more frequent in this study, surveys carried out in Brazilian prison units showed that aggressions are committed, in large part, by prison officers 13,19 .
According to Scherer et al. 14 , a study carried out in a female prison in the countryside of São Paulo, the prison is scenery of multiple forms of violence caused by prison officers, of both sexes, who acts severely causing fear and intimidation to the female inmates. Officer-on-inmate violence is used instead of rehabilitation which violence goes against the duties of the National Penitentiary Department (DEPEN) 31 . Serra 32 explains that corporal punishment, torture, and other punitive practices are historical in Brazilian society and have persisted for centuries, which shows that the Brazilian society has not yet renounced rigorous and severe punishment. The author establishes that as punitive practices are not aligned with the existing criminal codes, therefore, a "long-term interdisciplinary theoretical-methodological approach" is needed to try to assimilate the punitive logic in the Brazilian social formation.
Regarding violence prior incarceration, the currently study showed that 81.9% of the interviewees suffered some type of physical violence, either as children, adolescents and/or adults. This percentage is high compared to physical violence among Brazilian women, which is 30% according to Schraiber et al. 33 emphasizing a lifetime victimization of the female prison population.
The multivariate analysis considered a possible factors associated to the woman who suffer physical violence inside the prisons: physical violence at some point in life before going to prison, mental suffering, history of alcohol and other drugs abuse, and no labor activity inside the prison. Researches indicate a significant association between a woman experiencing episodes of physical violence and her violent behavior reproduced in prison. Therefore, it seems that inmates who suffered physical violence are more vulnerable to repeat these experiences in adult life, as victims or authors 9,10 .
In terms of alcohol abuse and to physical violence victimization inside prisons, the current study observed a prevalence of 37.3%. Such numbers are higher than the average of abusive consumption of alcohol by the Brazilian population (13.7%) 34 . Mullings et al. 35 using the same classification to detect alcohol abuse (AUDIT) among female prisoners in the state of Texas, found the prevalence of 40%. There is a greater chance that these alcohol-dependent women have grown up in a dysfunctional family context, with problems related to their family members' use of alcohol and / or other drugs, as well as neglect, physical and sexual abuse in childhood.
Mental suffering was also related to physical violence among inmates. Pereira et al. 36 , found the prevalence of 80% of women diagnosed with moderate to severe depression in a prison unit in São Paulo. Also, Logan and Blackburn 37 detected that women incarcerated by violent act were four times more likely to be diagnosed with mental disorder than those who committed less severe violence. Therefore, it is suggested that this population is more likely to develop mental suffering among other psychic disorders. That is, such a cyclical mechanism that results in the imprisonment of women seems to be related to violence in childhood, alcohol, drugs, violence in adulthood, aggression and other victimizing conditions. Drug consumption was strongly associated with suffering violence inside prison. Dias 38 , inside São Paulo prisons in Brazil, reaffirms the link between drug-related debts and the outbreak of  39 and drug trafficking has become, in recent years, the activity that most destines women to imprisonment in Brazil. Studies aimed at drug trafficking policy and drug use inside prisons are scarce, probably because they constitute illegal acts. However, Soares and Ilgenfritz 12 , in an interview with a prison officer of female unit, revealed the existence of a mentality rooted in the prison system that the drug would be necessary to improve the psychological state, at least for a period, and to calm the inmates.
Overall, the study found that the general profile of women incarcerated in Brazil corroborates other comprehensive national and international studies 6-10,13-20 which comprehend young women between 18 and 30 years of age, non-white, low education, convicted for drug trafficking, and unmarried (although the marital status indication is quite variable and not precise by definition).
In terms of education, 64.9% of the interviewed women had less than nine years of study, while the overall of Brazilian women with less than 9 years of study was 48.7%, according to PNAD 2009 40 . Studies show that the female prison population comes essentially from the less favored strata of society, with the lowest levels of education 41,42 . Due to the low level of education of most of the interviewees, it was difficult to handle the technology of the tablet during the application of the self-interview instrument. Therefore, the interviewers assisted them by reading the questions and answers and, at other times, also by marking the answers according to the interviewee's testimony. However, this fact may have led to limitations in this study.
In addition, the data presented may be underestimated in relation to the concrete reality experienced by these women, both before and during the imprisonment. The constant presence of prison officers at interview sites may have inhibited women's speech, either for fear repression or retaliation from prison officers, as well as intimidation and / or prejudice from other inmates. Because of that, there was also a need to restructure the questions of the violence component so that the answers did not lead to a possible report or confession, offering risk to the interviewee inside the prison unit.
Finally, these data expose the issue of female imprisonment in Brazil by unveiling a universe of suffered and perpetrated violence. Such patterns of violence were experienced before, during and after imprisonment, when another stigma is added to the trajectory of these women. Thus, the prison system diverges from the real meaning of imprisonment, failing to fight crime and violence in society by spreading the violence within institutions that, in fact, are supposed to extinguish it. This fact often reiterates the violent course and the continuity of women in the world of crime.
Several laws and policies aimed at the re-socialization of women were cited throughout the research. Such laws are designed to ensure the physical and mental integrity of female prisoners. However, such policies and laws are not put in practice. When the outcome is incarceration, the state should be responsible for providing all the health care that is inherent to that individual, including the promotion of physical and mental integrity. Nonetheless, the data show the non-fulfillment of the duties of the State. It gets clearer and clearer that the violence inside prison in the life of these women is being reaffirmed.
By identifying, in this research, the cycle of violence and revictimization suffered by women in penal institutions, it is suggested that collective actions targeting these women be carried out in order to identify their vulnerabilities and take steps to prevent violence inside the institutions. However, such actions will only be effective with the participation and technical and humanized training of the whole female institutional penitentiary sector in order to break the popular punitive logic and the trend of institutionalized punitive rigidity.

Conclusion
The vulnerability of incarcerated women with a history of alcoholism, the use of psychoactive substances, mental suffering and a history of violence throughout their lives are evident. The lack of support for alcohol and drug addicts inside many prisons opens up gaps for the spread of violence. In addition, the absence of such health supports constitutes institutional violence.
It is necessary to expand the discussion and promote studies on this subject with emphasis on interdisciplinary approaches in the Brazilian prison units. This way, concrete subsidies will be provided in the constitution of a theoretical body that will guide proposals for effective intervention in the Brazilian female prison units.