Brazil’s Maria da Penha Domestic Violence Police Patrols: A Second-Response Innovation in Preventing Revictimization

Introduction

This article examines an innovation in police intervention in domestic violence in Brazil: the 300 or so patrols set up since 2015 by military police forces and municipal guards in cities around the country. The main function of these second-response units, generally known as “Maria da Penha” (hereafter MdP) patrols after the eponymous 2006 law on domestic violence, is to enforce court-issued protection orders by paying repeat visits to women at high risk of further violence, referring them to support services, and ensuring that the abusers stay away. The article analyzes their origins and modus operandi and evaluates their impacts on victims, abusers, the wider community, and internal police force culture. From a “what works” perspective, available evidence shows that victims enrolled in the MdP patrol programs are much less likely to suffer repeated assault or feminicide than those who are not. The article compares their approach with second-response police interventions developed elsewhere and examines the degree to which this intervention contributes to the protection networks provided by the state, the criminal justice system, and civil society.

Changing Police Interventions in Domestic Violence

As an offense, domestic violence has distinctive features that have shaped police approaches to intervention. It encompasses a wide spectrum of abusive behaviors that interfere with the autonomy, safety, and dignity of the victim. These include psychological, emotional, financial, and sexual abuse, as well as physical assault, which can occur over a very long period, creating victims who are unable, for many reasons, to escape their user. Women are revictimized, showing up in police data individually as “hot dots,” their households as “hot spots” (Farrell, 2015; Pease et al., 2018).

Like victims, offenders are a heterogeneous group (Piquero et al., 2006). Some are “escalators,” although intensification from verbal to physical violence may not be linear or predictable. In many cases of feminicide in Brazil and elsewhere, a verbal threat may be the only harbinger of a deadly assault. Improved understanding of domestic violence as a form of coercive control (Stark, 2009) or “patriarchal terrorism” (Johnson, 1995) helps to explain such behavior. In Rio de Janeiro, for example, 44% of men accused of feminicide in 2019 said they were motivated by rage at the victim ending their relationship (Mendes et al., 2020), consistent with data from other states and international patterns. Conversely, some abusers are “de-escalators”: An episode of violence may be followed by a “honeymoon” period of reconciliation. Victims recognize this lull in violence for what it is, a respite between outbursts of abuse. Some abusers maintain a stable low, or high, level of abuse over a prolonged period.

In response, victims become skilled managers of their own security (Monckton Smith & Williams, 2014). In seeking to maintain equilibrium, they may not seek help from the police, or disengage after the first contact, if the police intervention does not create a new, safer equilibrium but instead provokes the abuser to further violence. As a minimum, then, police interventions should do no harm and not leave the victim more vulnerable. Police response to domestic violence has also changed markedly since the 1970s. Reluctance to intervene in “private” family conflicts regarded as “not real crimes” has, to a large extent, been overridden by new laws that mandate the first-responder police attending an emergency call-out to arrest the aggressor (Buzawa, 2012). These laws were encouraged by a landmark experimental field study conducted in Minneapolis in the early 1980s that found that police arrest had a deterrent effect on reoffending (Sherman & Berk, 1984). However, further experiments revealed that arrests had no overall effect (Sherman, Smith, et al., 1992), a finding confirmed in a more recent meta-analysis of 11 studies (Hoppe et al., 2020). In fact, arrest had a negative effect on socially marginalized offenders with a nonconformist mindset, making them more likely to reoffend (Sherman, Schmidt, et al., 1992).

Police forces then began following up with victims as a protection strategy. But such second-response interventions, and literature detailing their impact, are much rarer. One pioneering project was the Domestic Violence Home-Visit Intervention, which started in New Haven in 2000. Within 72 hr of a domestic violence incident being reported to the police, a police-advocacy team comprising specially trained officers from the local community policing unit and clinicians or social workers specialized in child trauma and protection would conduct a home visit (Stover et al., 2009). Over several visits, information about legal and support services was given. The clearest positive impact was that victims were more likely to access child support services and to report further offenses to the police. However, analysis of this and similar second-response interventions showed that they still did not reduce the likelihood of repeat assault or encourage the victim to leave the abuser (Davis et al., 2010; Stover, 2012). A different approach is that taken by the Chula Vista, California, police department, which implemented a “graduated response protocol” for domestic incidents between intimate partners. This attempted to avoid direct criminal justice interventions and instead change the norms and behavior within the couple with educational messaging, written warnings, in-person follow-up visits to suspects and victims, and customized problem-solving. Findings showed a reduction in domestic violence crimes by 24% (Schmerler et al., 2020), indicating that it is the nature and detailed procedures of the police follow-through with victims and perpetrators that really matter, a point amply demonstrated by MdP patrols.

Research Design

This research was designed, first, to map the emergence and operational characteristics of MdP patrols and, second, to estimate their impact. Answering the first research objective is complicated by the highly decentralized nature of law enforcement in Brazil: Each of the 26 states and the Federal District operate both a military and a civil police force. One fifth of Brazil’s 5,570 municipalities also operate a municipal guard, largely deployed in a preventive, community policing role. There is no central database listing the location and active operational status of MdP patrols, and very little was published in the secondary literature, even though some patrols have attracted “best practice” prizes from government, civil society, media outlets, and international bodies. The number of MdP patrols in existence (as of June 2021) has been estimated by compiling data supplied by state-level military police forces and gleaned from the state appellate courts, which often sign agreements with municipal guards in the absence of military police initiatives. For the analysis of operational practices, the article draws on primary data, including local ordinances and internal standing orders setting out guidelines for local MdP patrols, data provided by police forces, and localized empirical studies, often produced by police “pracademics,” that is, officers enrolled in academic programs in which they carried out primary research. In addition, between 2019 and 2021, I conducted 11 formal interviews with pioneers/coordinators of MdP patrols in São Paulo city and in the states of Rio Grande do Sul, Minas Gerais, Maranhão, Paraíba, Bahia, Mato Grosso, and Espírito Santo and with officers working with victim protection networks in Ceará, Goiás, and Piauí states. Between 2016 and 2020, I also engaged in informal conversations with dozens of MdP patrol officers who were among the 450 participants in training sessions that I ran while developing a police training manual on gender-based violence with the Brazilian Forum on Public Safety (Fórum Brasileiro de Segurança Pública [FBSP], 2020b). The latter’s project of documenting “good practices” in policing gender-based violence also provided rich data from around the country.

With regard to impact, data must be treated with caution. Brazilian police recorded 3,730 female homicides in 2019 (FBSP, 2020a). Around half were likely committed by a current or former intimate partner, and should be classified as feminicides, which Brazilian law distinguishes from female homicides and defines as murders motivated by contempt for, or a discriminatory attitude toward, women or committed within the context of domestic violence (Macaulay, 2021). However, many local law enforcement agencies still systematically under-record feminicides. As this improves, the number of feminicides will appear to rise, whereas in reality this simply reflects changed recording practices.

There are similar problems in relation to nonlethal domestic violence. In the first half of 2020, women in Brazil reported 110,791 cases of bodily harm and 238,174 cases of threat to the police (FBSP, 2020a). As these offenses are persistently underreported, changes over time can be subject to confounding variables. For example, the COVID-19 context of local lockdowns in Brazil in 2020 likely explains a drop of around 10% and 18% in reports of bodily harm and threats in relation to 2019. Also, as the criminal justice system’s response to domestic abuse improves and victims feel more confident to report incidents, the number of reported incidents will rise. This, a lack of baseline victimization surveys, and uneven police recording practices are major caveats to interpreting either time-series or comparative data across different jurisdictions. Thus, like most international studies of police interventions in domestic violence, this article examines evidence from city-level studies.

Brazilian Police Interventions in Domestic Violence

Despite Brazil’s very high levels of social violence and the poor reputation of its police forces, the country has been a pioneer in offering new policing responses to gender-based violence. In 1985, the world’s first-ever ‘women’s police station,’ was installed in the center of São Paulo (C. M. Santos, 2005). Run by the civil police, they multiplied across the country and were emulated elsewhere, promising women a more sympathetic and effective environment in which to report domestic or sexual abuse than ordinary police stations. However, their role in potentially preventing revictimization has been limited by their relatively low number for the size of the country and spatial concentration in coastal, metropolitan areas. In 2019, only 417 municipalities reported having a women’s police station, a decrease from the 441 registered in 2014 (Instituto Brasileira de Estatística e Geografia, 2020). The stations’ remit also varies: In some states, women are expected to report domestic or sexual violence only at a women’s station; in others, any precinct will register a case. Some women’s stations will handle any crime committed against a woman, not just domestic or sexual abuse, and some extend their remit to other vulnerable groups such as children and the elderly, somewhat diluting their focus on gender-based offenses. Similarly, other support structures, such as refuges for battered women, are also unevenly distributed and relatively sparse. Thus, many victims of domestic violence will find themselves unable to receive dedicated support from a women’s police station or to leave their homes and find safe accommodation.

However, the military police is the one law enforcement institution with universal coverage across the national territory, with some 450,000 officers compared with the civil police’s 100,000. State military police forces also consistently report that domestic violence incidents constitute the single largest category of calls to their emergency hotlines, especially at night. In the first 6 months of 2019, the Rio de Janeiro police recorded 30,617 domestic violence–related calls (18.6% of the total volume). Beyond the clear demand for first-response police action, the need for effective second-response intervention to enforce court-issued emergency protection orders was also becoming clear (Azevedo et al., 2016). Some 12% of orders granted in the Federal District between 2006 and 2012 were breached, with no system for checking on victims or recording their reports of violations (Diniz & Gumieri, 2016). In the months leading up to the establishment of MdP patrols as a state-wide policy, Rio de Janeiro’s state law enforcement agency was seeing more than 5,000 military police units dispatched a month to emergency calls relating to domestic violence. Yet, only a very small percentage of those incidents—8.2%—ended up being officially reported as crimes against women, due to reluctance of victim to proceed (Tribunal de Justiça do Estado do Rio de Janeiro, 2019). Victims’ loss of faith in the police after the initial contact also reflected how they were treated. Qualitative interviews carried out in Porto Alegre with 21 women seeking assistance with domestic violence, and 25 service providers inside criminal justice, health, and social work agencies, revealed a system that was disjointed, slow and unresponsive, and often unsympathetic to the victim and her needs (Meneghel et al., 2011). The needs of victims, alongside a number of other factors both internal and external to the military police, meant that second-response police interventions emerged simultaneously in diverse states across Brazil. This occurred in an organic and decentralized way, eventually crystallizing into the MdP patrol model.

Internally, operational thinking began to change with the spread of community-oriented policing (COP) experiments that cropped up across Brazil in the 1980s and 1990s. The international literature shows that COP approaches encourage greater attention to preventive policing to protect vulnerable social groups. Thus, in the Federal District of Brasília, the Prevention of Family and Domestic Violence second-response units developed out of a COP literacy project run by the military police in a low-income neighborhood. It quickly shifted focus to domestic violence prevention, and in 2011, the first military police officers were trained for a dedicated patrol, with the project institutionalized in 2014 (Tribunal de Justiça do Distrito Federal e Territórios & Polícia Militar do Distrito Federal, 2019). Similarly, in nearby Goiás state, COP experiments in the early 2000s led the military police to establish a community policing center in 2013 (Pinheiro, 2020) to which MdP patrols would later be attached.

Another driver was the adoption of North American problem-solving, data-led, results-oriented, and preventive policing approaches, which relied on the quantitative and qualitative analyses of crime data to identify hot spots. MdP patrols frequently start life as pilot projects in hot spot areas. Some of the earliest, in Porto Alegre, were initiated in four high-violence neighborhoods where a community policing program already operated (Grossi & Spaniol, 2019). In Minas Gerais and Rio Grande do Sul, two of the states where, in the early 2000s, senior officers began to develop second-response policing projects on their own initiative, they were spurred by data and literature on repeat victimization to identify domestic violence as a policing priority (Gerhard, 2014; Moreira, 2006). In the former, the officer was influenced by studies urging immediate intervention so that the aggressor did not feel emboldened and victims were not left more vulnerable, processes that Pease (1998) termed “event boosts” and “flags.” He also took inspiration from first- and second-response protocols developed by British police forces in the 1990s (Moreira, 2006).

Hot spot/hot dot data analysis also underpinned many state-level homicide reduction initiatives developed in the last two decades. These brought together multiple stakeholders across local government to design tailored strategies aimed at reducing murder rates in particular localities or affecting specific social groups. Although the vast majority of domestic violence cases will not end in murder, prioritization of homicide reduction nonetheless provided justification and dedicated resources for domestic violence policing because of awareness that many feminicide cases were the culmination of domestic violence cases that had not received police attention. For example, in 2016, the law enforcement administration of Espírito Santo state established MdP patrols as part of the cross-government homicide reduction program, Estado Presente. At a local policing level, individual battalions similarly developed projects that became the precursors of MdP patrols. In 2014, in Rio de Janeiro state, the 10th battalion set up a tactical group called “Guardians of Life” to tackle homicides of children, the elderly, and women. The project soon focused on preventing intimate partner violence and feminicide when crime report analysis demonstrated that women constituted the bulk of vulnerable victims. The following year, the 38th battalion in Três Rios copied the initiative, creating the model project on which the subsequent state-wide MdP program, created in 2019, was based (Guimarães & Barros, 2017; Mendes et al., 2020). As a result of these factors, in many states, MdP patrols are institutionally embedded within local, preventive COP units and linked at headquarters to the state-wide strategic planning and homicide reduction programs. This enabled states such as Rio de Janeiro, Goiás, Mato Grosso, Rio Grande do Sul, and Minas Gerais to institutionalize MdP patrols and extend their coverage right across the state territory.

The main factor external to the police has been the 2006 Maria da Penha law on domestic violence and women’s movement pressure on the state to implement it. The law is named after a victim of attempted feminicide, whose husband remained at liberty years later due to the slowness of the Brazilian courts. The Inter-American Commission on Human Rights’ review of her case included a stinging criticism of the Brazilian government, which led to a broad-based feminist coalition developing a comprehensive law. This legislation gave impetus to early versions of MdP patrols by explicitly placing new responsibilities on the police forces tasked with crime prevention and first response (military police and municipal guards), not just on the police dedicated to the investigation (the civil police).

The law is novel in emphasizing protection and prevention (46 articles) over penal aspects such as mandatory arrest and punishment (four articles). Articles 10 and 11 further stipulate that the first-responder police must assist a victim who has been, or is about, to be assaulted; get her to a hospital, a safe place or a police station; and escort her back home to retrieve her belongings, if necessary. Article 22 empowers the courts to grant a wide range of measures intended to assist the victim, keep her and her children in their own home, where possible, and get her social welfare, health, and legal assistance. The “urgent protective measures” that the judge can decree include removal of the aggressor from the family home, and restraining orders prohibiting him from contacting or coming near the victim. The law’s emphasis on protective and preventive provisions gave police a reason to work with local victim support networks and legitimated new initiatives. In Rio Grande do Sul, in 2011, a senior military police commander set up “Operation Peaceful Family” to enforce the protection orders in her area. This attracted the attention of the state governor, who invited her to roll out the project in the capital. Thus, in July 2012, the very first patrols to bear the name of Maria da Penha were launched. They have since spread to all regions of Brazil through a process of horizontal policy mobility and transfer between police forces, with the patrol founders in Rio Grande do Sul and in Bahia, which copied the former’s model, becoming the key multipliers of the project.

How MdP Patrols Function

Broadly speaking, all the MdP patrols function in the same basic way because their primary duty is to provide the protection, enforcement, and prevention outlined in the law. A woman reports domestic violence, either through an emergency hotline or in person at a police station. After an assessment of her situation, a request for urgent protective measures is made to the court. The local MdP patrol receives the details of the case and visits the victim’s home. The patrol team, generally composed of a male and a female officer, speaks to the victim to inform her of her rights and refers her to other partners in the local support network. The perpetrator is separately informed that he is at risk of arrest if he breaks any of the court orders. Thereafter, the patrol visits the victim at home, or at work, periodically over a set period of time. The visits may be weekly, but could be less or more frequent, and can be prearranged or unscheduled. During the COVID-19 pandemic, the patrols have made much greater use of electronic communication to check on their clients’ welfare.

The operating practices of MdP patrols in different states and municipalities vary slightly, depending on the configuration of local institutional partnerships. For example, risk assessment and triaging of priority cases to pass on to the MdP patrol command may be carried out, variously, by the first response police, the officer in charge of the police precinct receiving the complaint, the women’s police station chief, the prosecutor’s office, or by the court (especially if this is a specialist Family and Domestic Violence Court), either by the judge alone or in conjunction with its multidisciplinary professional team. Different agencies use different risk assessment forms, although there have been attempts to standardize these (Macaulay, 2021).

Police have up to 48 hour to file a request for protective measures and the court has up to 48 hour to make a decision. Once granted, the courts either inform the aggressor and victims of the order, which they must then collect in person, or deliver the information through court officials or the police. However, domestic violence situations can be volatile, and this 4-day wait leaves the victim exposed. Thus, in some locations, MdP patrols will visit victims as soon as a report has been made or a request filed for protection. Some states now have fast-track protocols and will process a request for a restraining order within 4 hr.

Although local inter-agency arrangements are sui generis, there have been attempts, led by one of the key multipliers of MdP patrols, Major Denice Santiago, to standardize the service offered to victims. She provides to other states and municipalities setting up patrols Bahia’s model Technical Cooperation Agreement, which outlines the distinct and interlocking responsibilities of the various criminal justice system agencies and local government bodies. It specifies patrol operating practices in relation to initial risk assessment, data collection, and reporting on the victim’s status, providing standardized forms, so that information can be recorded properly and shared through the criminal justice system locally and with other states, via national databases of victims and offenders.

Impact of MdP Patrols

So, do MdPs work? Just as there is no national database of where the patrols are in operation, there is also no national-level study of their impact. Like the second-response experiments in the United States cited above, studies tend to be city-level, consist of time-restricted snapshots, and there are no quasi-experimental studies. The quality of the data provided by police departments and used in such studies is highly variable (Grossi & Spaniol, 2019), creating benchmarking and methodological problems. Yet, overall, there is a consistency of reported findings across Brazil. Available evidence suggests that MdP patrols have had a positive impact at several levels, first, and directly, on the victims and on the abusers; second, on the local communities where they operate, and third, on the police institutions and their attitudes toward community-oriented, preventive police work and domestic violence intervention.

For the victims, there are two metrics of success. One can be objectively measured: No further unwanted contact with the abuser and thus no need to report further abusive incidents to the police. There is also the subjective dimension of the victim’s personal sense of safety. Recognizing this aspect of their role, the patrols in Espírito Santo conduct visitas tranquilizadoras (“reassurance visits”). There is consistent evidence that enrolment in the MdP patrol program reduces revictimization. Police data from Manaus showed that of the 628 women in the program between September 2014 and February 2017, only 7% had suffered renewed violence (Silvestre, 2018). A study in Belém do Pará compared police records for 154 women served by the MdP patrol between 2016 and 2019. Prior to enrolment, 22% of women had reported three incidents to the police, and 16% had reported four. After enrolment, 60% reported no further incidents, and 22% reported only one further incident (Bernardo, 2019). Similarly, interviews with 37 MdP program clients in Vila Velha, Espírito Santo state found that 65% had not had to call the police again (Braga, 2017). This study replicated an earlier questionnaire employed by one of the pioneers of the patrols, Lt Col Nádia Gerhard whose extensive study surveyed the impact on 147 of the 1,468 women assisted by the MdP patrol between October 2012 and October 2013 in Porto Alegre (Gerhard, 2014). In terms of user satisfaction, 94% of respondents rated the service as good or excellent in Porto Alegre, as did 87% in Vila Velha. The overwhelming majority—88% and 95%, respectively—reported that the restraining order alone had not been sufficient to make them feel safe, and 91% and 78% agreed that they felt safer enrolled in the program.

The primary objective of the patrols is to enforce the stay-away order, that is, to influence the behavior of the abuser. Data provided by police forces typically give the number of arrests for violations. In the early days of MdP patrols, a high number of arrests were often presented as demonstrating police success in the task of enforcement. However, increasingly a low number of arrests are being heralded as success in the task of deterrence (Hanashiro & Sobral, 2017). In relation to the offender, the relevant metric is now prevention, not detention. The two studies above revealed that in 86% of cases in Porto Alegre, the order was being respected (65% in Vila Velha). Like the Chula Vista project cited above, the police also employ behavioral modification tools. Police forces are also increasingly using electronic tagging to monitor the movements of abusers, alerting the wearer, the victim, and police if they come within a certain range of the victim. As violation of a protection order became an imprisonable offense in 2018, abusers themselves will take evasive action to avoid a breach.

The close monitoring of the abuser has also inhibited lethal violence, as one of the features of feminicide is the tendency of aggressors to stalk the victim. Most MdP programs reported, in their data or interviews, that there had been no single feminicide, actual or attempted, of a woman enrolled in the program (Hanashiro & Schlittler, 2019; Hanashiro & Sobral, 2017). Overwhelmingly, the victims of feminicide are those women who have not reported abuse to the police and therefore have no restraining order. Recent state-level data confirm this: 93% of the 30 feminicide victims in 2019 in Mato Grosso do Sul had no protection order at the time of their murder (Roca, 2020). In Rio Grande do Sul, 94% of 79 feminicide victims in 2020 held no restraining order and 82% had not previously reported any domestic abuse to the police (Departamento Estadual de Proteção a Grupos Vulneráveis, Divisão de Proteção e Atendimento à Mulher, 2021).

Clearly, one of the most effective ways to prevent feminicide in the context of domestic abuse is to increase women’s confidence in reporting to the police. The visibility of the patrols, with their specially badged vehicles, as well as the word-of-mouth effect from satisfied service-users, tends to spur other women in the community to report domestic abuse to the police for the first time. In Porto Alegre, 70% of respondents said that at least one of their neighbors had done so (40% in Vila Velha). The two studies also showed that trust in the police rose markedly. Most official reports show a surge in reporting of domestic violence and a concomitant rise in the number of protection orders issued where patrols are active. In Mato Grosso, the courts issued an average of 61 protection orders a month in 2019, which tripled to 190 a month in 2020.

The accessibility and visibility of the patrols, which also conduct routine patrolling and policing work, have also led to greater trust in, and legitimacy for, a police presence in communities that are either neglected or overpoliced. This has the additional effect of validating community-oriented and preventive work within the military police, where there is still significant resistance. It furthermore validates law enforcement work on domestic violence, which is often seen as “not real policing,” a sinecure, and a waste of police resources. The patrols are also having some impact on gendered police cultures. Their proliferation has prompted the promotion of more women to command positions, while programs have been developed inside some forces to deal with police officers who are themselves domestic abusers (Schlittler, 2019). In Bahia and Maranhão states and in the Federal District, military police now also have units dedicated to supporting female officers around domestic violence, sexual harassment, and other forms of bullying and discrimination. That said, the military police’s highly militarized and macho culture remains embedded and reproduced in their training, hierarchy, and internal discipline. The fact that a great deal of routine police patrol work involves informal conflict resolution and social assistance, and that second-response policing is centered on care for victims is unlikely, in the absence of deeper structural reforms, to override police cultures built around masculinity and the use of force. It is more likely that MdP patrols will create a bounded subculture within police forces, much like the community policing initiatives, which exist in parallel with the dominant police ethos.

Conclusions and Recommendations

MdP patrols represent a significant innovation as a second-response intervention in situations of domestic abuse from which Northern countries could learn. There are similarities to the U.S. experiments mentioned earlier: The law mandates arrests for offenses involving physical injury, the patrols make visits to the household after offense, and they consciously seek to modify the conduct of the abuser. However, there are points of difference. The patrols’ focus on enforcement of the protection order means that they are not only just victim-focused but also offender-focused. With regard to the victim, the patrols’ follow-up is more prolonged than in U.S. models and tailored to individual circumstances. By deploying additional tools of focused deterrence policing such as electronic tagging, they shift the onus of compliance onto the offender, removing the burden of personal safety work from the victim. This shift had already been established, in principle, by the Maria da Penha law, which enabled the civil police and the courts to constrain the movements of the abuser, and the patrols have provided the necessary enforcement. Thus, MdP patrols do not just improve survivors’ trust in the police to report further offenses but also significantly reduce the likelihood of reoffending on the part of abusers.

The women’s police stations constituted an important earlier innovation in offering women an alternative first-response institutional space in which to report domestic violence and get help (Carrington et al., 2020). However, the evidence is weak that, by themselves, they prevent further victimization, including feminicide (Perova & Reynolds, 2017), given that women have to seek out the investigative police after the fact. MdP patrols were developed to fill the enforcement gap and form a necessary, although not sufficient, element of local municipal protection networks set up to assist survivors of domestic violence. These consist of the agencies of the criminal justice system (police stations, prosecutor’s and public defender’s offices, the local courts), state-provided social services, non-governmental organizations (NGOs), and voluntary groups. The patrols depend on these networks: Expansion of the scheme to more cities is always contingent on robust local partnerships. They also add a much-needed enforcement dimension to these preexisting services.

The patrols also constitute a vital form of tertiary prevention. They go to the victim, rather than vice versa, and they deploy secondary prevention tools such as risk assessment in shaping their service response to clients. To some extent, like the women’s police stations, they contribute to primary prevention strategies, with patrol members giving talks about domestic violence in schools, civil society organizations, and state bodies. In Bahia state, male MdP patrol officers run preventive awareness sessions for men in areas reporting high levels of domestic violence (Bueno & Brigagão, 2018). However, primary prevention requires a major cultural shift in Brazilian society around women’s status, rights, and bodily autonomy and is a task far beyond the remit of the criminal justice system (Pasinato et al., 2019). From a critical feminist criminological perspective, the patrols throw up paradoxes. On one hand, the focus on protection and prevention means that, after the first offense, they are arresting fewer aggressors for reoffending or violation of the court order and are able to divert these men into therapeutic and behavioral solutions, rather than jail. On the other hand, the protection framework also invites a paternalistic view of abused women as both “innocent” and vulnerable, which partly explains why socially conservative police officers continue to support and advocate for these patrols. But, does it matter if a good policy is pursued by some for the wrong reasons as long as it works for the victims?

User satisfaction surveys indicate that an immediate need is being met by the patrols, but this does not negate the poor, discriminatory, uncomprehending, or condescending treatment that domestic violence victims still encounter within the criminal justice system, even from those professionals who are supposed to be specialized in handling gender-based violence (Instituto Jones dos Santos Neves, 2019). Inadequate specialist training means that victims are frequently still subjected to sexist, racist, classist, and transphobic prejudices when trying to claim their right to social and physical protection. Police often have a very superficial understanding of domestic violence and of the Maria da Penha law. Specialist training programs are generally provided to officers who volunteer for MdP patrols but can vary from minimal to very comprehensive and regular. As the patrols become more institutionalized in their force, sessions on domestic violence are also now being incorporated into basic training for new recruits.

Although the existence of MdP patrols has improved local collaboration between the military or municipal police and the civil police and courts, police work is often still balkanized, with officers often unclear about the specific responsibilities of other partners and how to work with them to meet victim needs more seamlessly. Inter-agency coordination and coverage in a dispersed criminal justice system remain the biggest challenges to improving the support given to domestic violence victims. Just as the women’s police stations have limited geographical coverage, so MdP patrols based in the major cities are limited in their reach. To cover rural areas, therefore, all military police need training on domestic violence first-response intervention, while some states are engaging municipal guards in small towns to perform the second-response patrol task. Military police in the states of Minas Gerais and Bahia have been developing specific outreach tools to engage women in agrarian, indigenous, and black rural communities, in acknowledgment of their distinctive cultures (T. Santos, 2019).

The strongest finding and recommendation derived from the above account is that comprehensive domestic violence legislation that empowers police to enforce, protect, and prevent, rather than just mandates the arrest of offenders, is necessary to enable effective second-response interventions. The second recommendation is that equal attention should be given to both victim and abuser, given that the former’s primary concern is with the latter’s conduct. The third is that follow-up should be prolonged and repeated, if necessary. The fourth is that close working partnerships between different agencies, including those of the justice system, should encourage harmonized and standardized working practices that improve information-sharing and prevent victims from falling through the cracks. Taken together, these measures can improve women’s personal safety and increase their autonomy and wellbeing, which, ultimately, should be the aim of any criminal justice sector policy to tackle the persistent problem of domestic abuse.