Parallels in Practice: Applying Principles of Research Integrity and Ethics in Domestic Violence Fatality Review (DVFR)

Within the context of the big data society, new systems of data collection on domestic violence and abuse (DVA) have emerged. One such system is Domestic Violence Fatality Review (DVFR) which captures the various dimensions of gender, violence, and abuse required to form an evidence base for prevention. However, to date, there has been limited dialogue between practitioners and researchers about the ‘doing’ of DVFRs. As DVFR systems vary by jurisdiction, we conducted a case study of Domestic Homicide Reviews (DHRs) in England and Wales. Applying the Research Integrity Framework (RIF) developed by the four Women’s Aid Federations in the United Kingdom (UK), this article examines both the practice of DHR and how it is utilised as data in research. Informed by our situated perspectives as researchers and/or practitioners working in the field, our analysis demonstrates how undertaking DHR as a practitioner parallels collecting, accessing, and analysing data from DHRs as a researcher. Guiding principles are identified to help practitioners and researchers navigate the parallel challenges they confront and, critically, inform dialogue between practice and research. Implications for both professional practice and research are presented. To increase transparency and confidence, we argue that more attention should be afforded to the methodological and ethical issues inherent in both the practice of DHRs, and their utilisation as a source of data in research. While DHRs have differences to DVFRs in other jurisdictions, these findings also have implications for these other systems which will also be discussed.


Introduction
Domestic violence and abuse (DVA) are recognised internationally as pervasive problems, requiring multi-sectoral and -systems interventions (García-Moreno et al., 2015). In response, new systems of administrative and civil society data collection on DVA have emerged regarding prevention and advocacy (Walklate et al., 2020). While these modes of knowledge production have prompted concern regarding data-driven surveillance programmes and carceral data regimes (Dencik et al., 2016), these systems contain a wealth of data that could provide new insights into the needs and experiences of victim-survivors, as well as for services looking to understand the complex barriers to help-seeking and access (Storer et al., 2021).
Domestic Violence Fatality Review (DVFR) systems are one way of collating data to respond to DVA, and can capture various dimensions of gender, violence, and abuse required to form an evidence base for prevention. DVFRs began in the United States (US) in the 1990s and have since developed in several high-income countries, including Australia, Canada, New Zealand, and some parts of the United Kingdom (UK). While varying by jurisdiction -including in terms of context, scope, and process -DVFR systems share a common approach (Dawson, 2017a): they draw on a range of different sources and stakeholders to retrospectively examine deaths that have occurred because of DVA by an intimate partner and/or family member, and sometimes other types of relationship (e.g., deaths by suicide, or the deaths of bystanders). The purpose of this examination is to produce findings, or data, about DVA-related deaths and, thereby, improve practice, policy, and system responses to prevent future deaths (Dawson, 2021;Websdale, 2020).
Challenges associated with localised systems of DVFRs (e.g., funding, legal considerations) have been found to influence the production and use of findings . Nonetheless, the findings of DVFRs are increasingly used in research and to inform policy and practice. Globally, DVFRs have formed the basis for large-scale reviews on victim/ perpetrator demographics, relationships, and event circumstances (e.g., ADFVDRN & ANROWS (2022) in Australia; Cheng and Jaffe (2021) in Canada); the identification of risk factors in domestic homicides (e.g., Dawson & Piscitelli, 2021); and considerations for specific populations (e.g., Buxton-Namisnyk, 2021). Findings from DVFR systems are routinely used as data to inform our understanding of DVA and domestic homicide, including who is at risk, under what circumstances, with what effect, and how (or if) agencies have responded. Yet, there has been relatively little examination of how DVFRs are conducted in practice and how these findings later come to inform research. Moreover, the utilisation of DVFRs -both as a narrative and aggregate record of DVA and domestic homicide -as a source of data has been neglected. This is because the current literature about DVFRs is "limited, fragmented, and primarily descriptive" (Dawson, 2017b, p. vii). Viewing DVFRs through a lens of data means problematizing DVFR findings as data, that is, identifying and questioning implicit and explicit assumptions about how these systems operate. This also requires an interrogation of how and what researchers choose to extrapolate from DVFR reports as well as how they come to inform their own research findings. Thus, what are the methodological and ethical challenges faced by practitioners and researchers respectively in their engagement with DVFRs? To what extent are the challenges comparable? If so, can a practitioner-researcher dialogue enable better navigation of these challenges, including with respect to the ethical principles and frameworks that might guide their work? This paper addresses these questions by viewing DVFR systems as a way of producing data, both in respect of DVFR practices (which produce accounts of victims' lives and identify recommendations for preventing similar deaths) and in DVFR research (where these accounts and recommendations are utilised by researchers in further analysis).
The paper proceeds in four parts. In the first section we outline an analytical framework through which we seek to problematize DVFRs and interrogate the methodological and ethical challenges for practitioners and researchers. This established, the second section introduces a case study methodology and rationale -focused on the specific jurisdictional form of DVFR as found in England and Wales -before providing detail on how we applied the analytical framework. The third section presents results from our analysis according to five analytical themes, highlighting the comparable and somewhat parallel nature of challenges encountered by both practitioners and researchers. The implications of this analysis, and recommendations for a shared dialogue between practice and research, are considered in the final section. While our focus is on DHRs as a case study, in discussing the literature we refer to DHRs and/or DVFRs as appropriate.

Analytical Framework
Across DVFR systems, there are no standardised ethical frameworks governing review processes, and only limited engagement with the ethical implications of using DVFR findings. While some research has sought to address ethics in DVFR (see Albright et al., 2013;Dale et al., 2017), this scholarship targets DVFR practice, rather than the ethical implications of using DVFRs in subsequent research or the methodological challenges of doing so.
Addressing this gap, and to frame this paper, we applied the Research Integrity Framework (RIF) (Women's Aid et al., 2020) to a case study in England and Wales, where the DVFR system is known as Domestic Homicide Review (DHR). The RIF was developed by the English, Northern Irish, Scottish, and Welsh Women's Aid Federations in partnership with academics to promote best practice in DVA research. The RIF sets out an ethical framework made up of five "pillars": safety and well-being; transparency and accountability; equality, human rights, and social justice; engagement; and research ethics. The pillar of safety (physical and emotional) and wellbeing is based on the ethical principle of do not harm and is fundamental to both DVA practice and research. Transparency and accountability mean identifying who is doing the research, how, where, with whom and for what purpose. Building on these pillars, equality, human rights and social justice includes awareness of individual, social, and cultural factors and intersecting structural inequalities that underpin DVA. Engagement involves participation, collaboration or co-production with key stakeholders (including survivors). The fifth pillar, research ethics, cut across these four pillars, but also includes formal review of research and evaluation.
The RIF was applied to examine both the doing of DHR as a way of producing data about DVA-related deaths and the subsequent use of such data in research. The RIF was selected for this analysis because of the role of the four Women's Aid federations, all of which are closely tied to "each country's women's liberation movement" and therefore specifically attuned to the issues facing the DVA sector and research in the UK context (Women's Aid et al., 2020, p. 2). Whilst this is not the only framework that could be applied, these principles provide a starting point to navigate the range of methodological and ethical challenges associated with DVFR generally and specifically within DHRs in our chosen jurisdiction.

Selecting a Case Study
As DVFR systems vary, we undertook a case study of the system in England and Wales which, known as DHRs, has been in place since 2011. 1 DHRs were established under the Domestic Violence, Crime and Victims Act 2004 and implemented in 2011 (Payton et al., 2017). DHRs examine killings by (former) intimate partners, family members or members of the same household, as well as deaths by suicide. The DHR system has several features that distinguish it from other DVFRs including scope, timing, stakeholder involvement, outputs, and oversight. In theory, all DVA-related deaths are examined, and DHRs commence in parallel with the criminal justice process. This examination is stand-alone, with a DHR being conducted for each case. Alongside a multi-agency review team (hereafter: the review panel), the family should be equal stakeholders and others, like friends, neighbours, and colleagues may also be involved (hereafter: testimonial networks). As a product, DHR reports are usually published. Finally, although DHRs are delivered locally, there is national oversight.
For context, this section sets out the DHR process, summarising, without comment, the requirements detailed in the statutory guidance (Home Office, 2016b). A DHR begins with a notification to the relevant local Community Safety Partnership (CSP), 2 usually from the police, that a DVArelated death has occurred. The CSP, in consultation with local agencies, assesses the case to see if it meets the criteria for a DHR to be commissioned. If a case meets the criteria, an independent chair is appointed to lead the DHR (and is usually also the author of the report). A review panel is also established. This panel includes agencies that had contact with the victim, perpetrator, and/or any children, with these agencies documenting and analysing their contact for evidence of good practice or the identification of improvements. Other agencies may be invited for their expertise, for example about the needs and experiences of minoritized communities. Testimonial networks (such as family or friends) should be invited to contribute, potentially providing both additional information (some of which may be unknown to agencies) and sometimes an alternative perspective (this is usually facilitated via the independent chair). A victim's family have specific rights during a DHR (our discussion in this paper focuses on family involvement), including meeting the review panel if they wish. Upon conclusion of a DHR, a report is (usually) published, describing the case circumstances and any learning, as well as developing an action plan and making recommendation(s). The commissioning CSP is responsible for overseeing the implementation of an action plan and its recommendations.
Notably, the DHR system mixes local delivery via a CSP with national oversight through the Home Office (the government department responsible for crime and justice). The Home Office issues the aforementioned statutory guidance and convenes a national Quality Assurance (QA) panel that must approve all DHR reports before publication. For a summary of the DHR system, see Fig. 1.
Although the absence of a national repository has impeded learning (Benbow et al., 2019), DHR findings have been brought together to produce case profiles (Montique, 2019; Home Office, 2016a), including, for example, about risk and need (Chantler et al., 2020), process evaluation for family (Rowlands & Cook, 2022) and children (Stanley et al., 2019). DHRs have also been considered from the perspective of participant experience and satisfaction (Sharp-Jeffs & Kelly, 2016), and studies have begun to explore how DHRs operate in practice (Boughton, 2022;. However, in the same way that DVFR is generally underexamined, the in situ doing of DHR remains a "black box" (Rowlands & Bracewell, 2022). In referring to the in situ doing, we mean the operational, discursive, and symbolic interactions that comprise DHRs, the practices that constitute the tasks considered routine within it, and how and what shape outputs take as a result. Unpacking DHRs in this way, as Rowlands (2020b) argues, reinforces that they are a contingent process of meaning-making, reflecting their multi-agency nature and therefore, so too, are the findings that they generate.

Analysing the Case Study
Applying the RIF to a case study of the DHR system in England and Wales, we drew upon our collective experiences as researchers utilising DHRs as data and, for JR, leading reviews as an independent chair. We have all previously been, or currently are, involved in the practice or research of DHR, with disciplinary backgrounds including sociology, social work, and/or criminology. We came together to participate in a process of peer production and collective writing (Peters et al., 2019). This process involved a series of collaborative, critical and reflective exchanges, including sharing experiences or knowledge of best practices, concerns, and strategies for negotiating challenges in the field. These exchanges took place over a series of meetings, with each person driving a different analytical strand while supporting each other to develop individual contributions, and move beyond fixed positions, anecdotes, or single claims to knowledge. Individuals discussed and reviewed each strand to give their own understanding and worked together to develop a shared knowledge. This collaborative process of writing and knowledge exchange holds strength in that it represents different layers of interpretation, ideologies, and experiences based on our respective engagement with fatality review systems. While this form of collaborative working has value in a substantive sense (i.e., pooling best practices), it is also valuable as a source of peer support in a sensitive and difficult area of research and practice (Williamson, et al., 2020).

Results
Across the five pillars of the RIF, our analysis demonstrates how the work of doing DHRs parallels research practices which utilise DHR findings. This framework allowed us to interrogate the parallel processes of i) being, as a practitioner, part of a system that generates data about DVA and domestic homicide and ii) collecting and analysing DHR findings as a researcher. These parallels encompass both methodological issues such as case selection and access, but also ethical issues relating to power and dialogue between participants, confidentiality, and risk of harm. Based on this analysis, we offer guiding principles for practitioners and researchers to navigate the parallel challenges they confront and, critically, inform an ongoing dialogue between practice and research. As noted previously, in addition to our exploration of the extant DHR literature, we also draw on the wider DVFR literature.

Safety and Wellbeing
Safety and wellbeing reflect general concerns about social research, and the specific challenges when researching DVA (Bender, 2017). These tensions arise because of the need to balance hoped-for benefits (for participants, as much as researchers) with potential risks (such as the harm arising from participation). Operationalising this pillar, the RIF distinguishes between physical and emotional safety, and within DHRs, tensions are evident for participants and researchers.
Before exploring this further, it is useful to recognise that, compared to other DVFR systems, DHRs present a specific risk. As noted above, a DHR is a stand-alone case examination and is usually published, so any DHR subject(s) are potentially identifiable to those with either case knowledge or via media reports Websdale, 2020). Consequently, it is difficult to make claims for internal or external confidentiality (Tolich, 2004). This raises ethical Fig. 1 The DHR process in England and Wales. Reproduced from  issues regarding the deceased victim (discussed later). There are also considerations for living subjects -including the perpetrator, any children, and wider testimonial networks -as they are potentially identifiable, regardless of participation. As a result, the practices of anonymisation required by the statutory guidance, such as using pseudonyms (Home Office, 2016b, p. 24), are perhaps no more than "smoke and mirrors" (Rowlands, 2020a, p. 24). In addition, review panellists are identifiable, given the requirement to name those involved (Home Office, 2016b, p. 11). The individual and public nature of DHRs is also relevant to researchers. Given researchers are interpreting the contributions and representations of numerous participants -some or all of whom are identifiable -this raises questions about how, for example, published DHRs are used in research. Thus, DHRs should be treated as complex not inert documents (Rowlands & Bracewell, 2022).
DHRs may be a source of risk for various stakeholders. In terms of physical safety, for participants, tragically, the most immediate safety consideration -a perpetrator's risk to the victim -is removed. Yet, physical risks could still arise not least because perpetrators' behaviours and emotions are often at the centre of these deaths (Websdale, 2020). For example, in a DHR conducted into a death by suicide, the (alleged) perpetrator is unlikely to have been convicted and, if there were children from the relationship, there may be an ongoing risk to a family (Monckton-Smith et al., 2022). This may affect family willingness to participate. For practitioners, this same potential for risk arises because review panellists are named. Reflecting these concerns, this may affect decisions about publication by CSPs (H. Candee, personal communication, November 8, 2021), for example, either meaning they make the decision not to publish DHRs into deaths by suicide, or only publish an executive summary. Given these potential risks, the RIF usefully prompts consideration of the location of any interactions. While there has been a recognition of the potential sensitivities of testimonial network engagement in DHRs/DVFRs, participant safety has not been considered specifically, either for family members or professionals (Albright et al., 2013;Dale et al., 2017;Rowlands & Cook, 2022). This is a significant omission, given in DVA research, physical safety for participants is key (Bender, 2017). For researchers, physical safety is less of an issue, although emotional safety may have physical implications (considered below). Nonetheless, during fieldwork, there should routinely be an assessment of safety and the presumed absence of the perpetrator (either because they are deceased or in prison) should not be taken as a proxy for safety.
Concerning emotional safety, the potential emotional impact of DHRs has been recognised (Mullane, 2017;Websdale, 2012). Given participants in a DHR discuss the circumstances leading to a death, there is a risk of secondary trauma (Williamson et al., 2020). Illustratively, testimonial networks may face secondary trauma in recounting their loved one's experiences, being reminded of loss, or learning about what agencies did or did not do that might have averted a death. Meanwhile, practitioners must critically reflect on their and other agencies' practice and seek to identify learning, while researchers are often closely reading DHRs/DVFRs (Cullen et al., 2021).
Risks to emotional safety are posed not only by the circumstances of the death, but the impact of DHRs as a process. For testimonial networks, there is a presumption that DHRs/DVFRs may bring therapeutic or cathartic benefits (Jaffe et al., 2013;Rowlands & Cook, 2022). Yet, DHRs can have a significant, detrimental impact. First, a DHR may be just one of several state mechanisms that begin after a death and affect family wellbeing (Tomczak & Cook, 2022). Second, family members may be re-traumatized, particularly if untrained individuals interview them, which has the subsequent effect of limiting the scope of information provided (Mullane, 2017). In England and Wales, advocacy services only support a small proportion of families through the DHR process (Montique, 2019). Despite DHRs highlighting the need for services to respond to children's voices, they rarely involve children, reporting concerns of compounding children's trauma (Stanley et al., 2019).
For practitioners, a DHR may have an emotional impact. DHRs may be an unsatisfactory experience, with many professionals, including independent chairs, lacking the training to participate, or expressing frustrations with the process . As with any DVFR system, these frustrations can arise because of the need to navigate professional conflict (Albright et al., 2013;Dale et al., 2017). These dynamics might generate added pressure for DVA services, given the emphasis on their role as experts (Home Office, 2016b, p. 11), but also in terms of perhaps being in a minority in arguing for a feminist DVA analysis (Sheehy, 2017).
In summary, DHRs have the potential to be a risk to safety and wellbeing. Using the RIF as a framework helps draw attention to the tensions inherent in the DHR process, which can affect testimonial networks, practitioners, and researchers.

Transparency and Accountability
Transparency and accountability are critical when assessing the rigour of research and is evidenced by rationales for, and accounts of, methods used to collect and analyse data (Macy, 2018). While the statutory guidance sets out process expectations (Home Office, 2016b), questions have emerged regarding both the transparency of how individual DHRs are conducted, and how DHRs are administered and function within a wider system. For both research and practice, transparency and accountability raise many questions, most notably those centring on if and when to commission a DHR, decision-making during the DHR, and the delivery and implementation of any recommendations that are produced.
Whether a DHR is commissioned has implications for issues of transparency and accountability. In terms of practice, as noted above, DHRs begin with a notification to a CSP (Home Office, 2016b, p. 9). However, CSPs are not required to report on how many cases are referred or their commissioning decisions. Nor are CSPs required to provide basic information about these cases (e.g., on sociodemographic or relationship type). This makes it difficult to identify and address the consistency of decision-making, and some deaths may be excluded (Benbow et al., 2019;Condry & Miles, 2022). While family should be informed of this decision, it is not clear if, when and how this happens (Rowlands & Cook, 2022). This is relevant in cases where a decision is made not to undertake a DHR, given the potential impact on a family .
For researchers, this lack of transparency means it is not clear which subsets of domestic homicides (and, for DHRs, domestic abuse-related suicides) are reviewed and therefore whether critical information on improving systems and practice may be missing (Dawson, 2021). In addition, DHR reports can vary in terms of quality and content Chantler et al., 2020;Chopra et al., 2022) and, if published, are often difficult to locate, removed after short periods or only executive summaries are made available (Bridger et al., 2017). This means researchers using DHRs as data do not know to what extent their sample is representative.
Concerning decision-making, in practice, it can be very difficult to identify how this operates within DHRs. For example, although DHR reports should include a statement about the independent chair's independence and skills, and review panel members should be named (Home Office, 2016b, pp. 12, 11), issues of equality and inclusion are not always transparent . This can result in reviews being undertaken without a comprehensive understanding of the complexities of DVA and the intersecting oppressions that can structure risk (Dawson, 2021). Moreover, despite calls for the inclusion of cultural and community skills and knowledge in DVFRs/DHRs (Bent-Goodley, 2013;Jones et al., 2022), services with knowledge of minoritized communities or communities with specific needs may not always be included on a review panel (Montique, 2019).
Furthermore, the positionality of the independent chair and the review panel in relation to DVA is rarely noted. Existing evidence demonstrates that domestic homicide is gendered in both its dynamics and characteristics (Chantler et al., 2020;Chopra et al., 2022), underscoring the importance of a gendered approach when undertaking reviews (Sheehy, 2017). A gender-neutral approach may create a narrow lens that focuses on the homicide event and minimises the importance of contextual, relational, and power dynamics (Hester, 2013). These considerations are particularly important for cases in which a man is murdered by his (ex) partner following his ongoing abusive and repeatedly violent behaviour towards her. One consequence of a narrow lens may be that decision-making in a DHR (and therefore the findings) is affected, including the risk of victim blame (Rowlands, 2020b) and/or that there may be a focus on perpetrators to the exclusion of the victim .
For researchers, these challenges mean that it is important to recognise that DHR reports are the result of layers of interpretation, beginning with victim contact with an agency, through to the decisions made in a DHR itself, and ultimately by a researcher (Rowlands & Bracewell, 2022).
Concerning learning, in particular the delivery and implementation of any recommendations, the statutory guidance requires DHRs to make recommendations in response to any learning and include an action plan to identify the changes or improvements that should be made. CSPs should have appropriate governance mechanisms in place to disseminate learning and monitor the action plan (Home Office, 2016b, p. 22). However, the extent to which implementation is monitored is unclear, as there are no statutory requirements to report this, raising questions over exactly who is accountable for ensuring change following a DHR and the effectiveness of any changes .
There are unfortunately comparable issues in the context of DVA research, where findings can either lack application to real-world practices, or fail to translate well to policy arenas. How research evidence comes to inform policy and practice is a problem that is interminable across all disciplines, but is particularly pertinent in the context of DVA where research can pose ongoing risks to participant safety (Bender, 2017).
In summary, issues of transparency and accountability are critical when assessing the ethical and methodological rigour of research. It prompts researchers and practitioners to consider how and why participants are selected for a review or study, the processes of decision-making undertaken about equality, representativeness and diversity of cases, and the delivery and implementation of any findings/recommendations that are produced.

Equality, Human Rights, and Social Justice
The equality, human rights and social justice pillar emphasises how inequalities and power imbalances can impact research. There are many aspects of conducting DHR that are potentially informed by structural inequality and that pose specific risks to, for example, transparency and accountability in decision-making (above) and stakeholder engagement (below). These issues speak directly to what has been described by Fricker (2007) as a form of "epistemic injustice", which can include testimonial injustice and hermeneutical injustice, both of which are relevant to DHR practices and research.
Addressing the former, testimonial injustice occurs when someone's statements are given more or less credibility based on prejudices instead of other concerns, such as expertise or experience. In practice, within DHRs/DVFRs, victims cannot speak about their experience, so testimony from others who can speak on their behalf is crucial, particularly as testimonial networks may have access to information not known to agencies (Dawson, 2021). According to statutory guidance (Home Office, 2016b, p. 17), aside from the emotional impacts of involvement, in undertaking a DHR, the review panel should "recognise that the quality and accuracy of the review is likely to be significantly enhanced by family, friends and wider community involvement". However, there are tensions in the DHR process which mean testimonial networks may be seen as less credible than practitioners. Rowlands and Cook (2022) explore this in their analysis of bereaved family involvement in DHR processes, demonstrating that review panels comprise multiple stakeholders with sometimes competing, sometimes complementary interests which must be managed by an independent chair.
Existing DHR literature and guidance recognises the risk of a "hierarchy of testimony" (Home Office, 2016b, p. 17) that favours particular stakeholders, namely professionals, rather than achieving a balance, with this influencing the shape of DHR recommendations . Mullane (2017) states that even though families provide valuable information that might be unknown to professionals, this may not be given the same weight as other (professional) contributors. Stanley et al. (2019) demonstrated that children are not prominent figures in the DHR process, potentially due to the risk of secondary trauma and anonymity. However, children who live at home are likely to be exposed to, and victims of, DVA perpetrated before the homicide. As the DHR report gives very little space to showing how testimony is elicited, selected, or presented in the final output, it is difficult to ascertain whether stakeholder contributions are attributed equal weight.
Similarly, much academic research, whether qualitative or quantitative, requires that researchers work with participants to elicit testimony, experience, or knowledge, and then represent it to a different audience. However, when researchers use DHR findings, their presentation of testimony is contingent on the meaning-making processes of DHRs, including decisions about what to include or exclude, and interpretation, which may not be transparent. As already noted, information included in DHR reports is inconsistent and often missing. Such absences risk testimonial injustice that researchers could compound. For example, separate studies Chopra et al., 2022) report that ethnicity was missing from over 40% of the DHRs examined and tended to be reported only when victims and perpetrators were "visibly different" (Chantler et al., 2020, p. 487). Without critically engaging with DHRs, researchers risk reproducing inequalities and reinforcing a very particular image of domestic homicide, victims, and perpetrators. Testimonial injustices, therefore, risk creating barriers to who can contribute to collective knowledge and marginalise the experiences of minorities.
This lack of knowledge creates a significant barrier to making sense of experiences, constituting what has been termed hermeneutical injustice. Hermeneutical injustice occurs when a knower's experience cannot be understood by listeners, and even the knower themselves, as they do not have the concepts to make sense of their experience (Fricker, 2007). In practice, there is a risk to hermeneutical justice within DHRs, either because they do not recognise the contribution of testimonial networks or a victim's experiences cannot be understood within the interpretative framework being used. As noted above, there may not be adequate representation on a review panel in terms of cultural competence. This means that a DHR about a victim from a minoritized community may not engage with the structural context of their death (e.g., where there may be obstacles to help-seeking and accessing support services) . A lack of language and knowledge erases the opportunity to receive support that may reduce the risk of domestic homicide (Sabri et al., 2018), both in relation to practitioners who must generate recommendations and for researchers who seek to utilise and analyse these recommendations (e.g., standardizing terminology, breaking down stereotypes about ageing, health, and abuse) (Benbow et al., 2019).
Analysing DHR practice and research, according to tenets of equality, human rights, and social justice, highlights both opportunities and challenges for responding to epistemic injustices in the field of DVA. While DHR/DVFR practitioners and researchers risk compounding such injustices by discrediting certain types of knowledge or failing to diversify panel membership (Bent-Goodley, 2013;Chantler et al., 2022), they can redress those injustices experienced by victims and their families, in part, by engaging with the structural contexts within which domestic homicide occur.

Engagement
Engagement is closely tied to collaboration, involvement, coproduction, and co-design. This pillar is not exclusive to one stage of the DHR or research process. For example, engagement can be considered at the point of review/research design and terms of reference, delivery in the form of peer research, consultation, or advisory capacities, or as part of collaborative policymaking and dissemination (McAra, 1 3 2017). Central to each of these is stakeholder involvement and the contribution of expertise.
The first issue of stakeholder involvement has received increasing attention across health, justice, and social care settings (Crépault, 2016). In practice, a range of stakeholders are involved in DHRs both directly (e.g., practitioners/ professionals, representing a range of agencies participating on a review panel, and testimonial networks including family), and indirectly (e.g., local, CSPs, and nationally the Home Office and QA panel). While different aspects of family involvement have been addressed in the previous pillars -including safety and wellbeing, transparency and accountability, and the risk of testimonial injustice -challenges also arise in terms of the procedural aspects of doing DHRs. The statutory guidance suggests a range of different opportunities for family involvement (Home Office, 2016b, pp. 17-19). Additionally, the centrality of family involvement in DHRs is closely tied to the work of advocacy efforts by the charity AAFDA, whose 7-step model identifies practical guidance to ensure they are integral to the process. 3 The model recommends that families be involved in setting terms of reference, have opportunities to meet the panel, receive a draft report with time to comment, and be aware of how/if recommendations have been implemented. However, practices of involvement have been little elaborated on, and a lack of clarity remains as to the exact purpose, process, and outcomes of DHR for families (Rowlands & Cook, 2022). The practice of participating in a DHR from the perspective of practitioners is similarly under-elaborated, with a current lack of a national training programme for independent chairs and review panellists. Balancing stakeholder perspectives, particularly when practices seem relatively opaque (as discussed above for transparency and decision-making), is, therefore, a complex process, particularly if stakeholders are under-prepared and unsupported .
Within the context of research, the issue of stakeholder involvement raises comparable questions. Exploring this in relation to health research, Tembo et al. (2021) identified involvement in setting scope and priorities, delivering, managing, and monitoring the research and its outcomes, as well as having (free) access to outputs, as particularly important for effective engagement of community stakeholders. Coproduction is increasingly recognised as a complex practice that presents both ethical and governance issues, not least because the meaning of a problem, such as DVA, can vary across stakeholder groups and be reflected in imbalanced power relationships (Crawford, 2020). Much of the extant research into DHRs has focused on thematic analyses of recommendations, risk factors and key findings rather than the substantive process of doing reviews. Consequently, very little is known about what works for professionals, agencies or families or their experiences of the review process (Boughton, 2022;Haines-Delmont et al., 2022;Sharp-Jeffs & Kelly, 2016).
The second issue, therefore, following closely to that of stakeholder involvement is the contribution of expertise, as it raises questions regarding value judgements made about expertise, what is considered as knowledge, and whose experience this represents.
Within the context of practice, who has a seat at the table, and what experiences they represent, are particularly pertinent issues when it comes to the role of Equality, Diversity and Inclusion (EDI) in understanding DVA and homicide. As previously addressed, specific expertise may be required to understand how victims from minoritized communities might experience help-seeking differently, or how discriminatory attitudes might impact their access to support and intervention (Bent-Goodley, 2013;Chantler et al., 2022). Other representation might include specialists with experience in working with victims with disabilities, lesbian, gay, bisexual, trans and queer (LGBTQ) victims, and men. As women account for most victims in DHRs, panels should account for (and credit) feminist knowledge by including DVA services (Dawson, 2021;Sheehy, 2017). These issues are not only relevant for analysing how domestic homicide and its causes are understood, but how DVFRs inform community-level change (Websdale, 2012). However, testimony is not always credited with equal weight, particularly when it comes to that provided by statutory versus non-statutory agencies .
In research, similar concerns arise. While co-production and the involvement of stakeholders implies, as Crawford (2020) states, a solution to how knowledge is constituted, mobilized, and used, co-production presents challenges in the way that we solicit knowledge, translate and apply it in policy. Such hierarchies of knowledge pose epistemological questions regarding how we come to know the extent, nature, and dynamics of DVA and where (or who) this knowledge is gathered from, and the power relations imbued within these processes. Although stakeholder involvement often appears in a more consultatory form in research (e.g., advisory boards), the use of collaborative inquiry highlights the potential for sharing lived experience knowledge (Jury, 2022). Involving bereaved families in DHRs present numerous ethical challenges around safety and wellbeing. However, such methods are perhaps key to ensuring accountability in research as demonstrated above.
In summary, the fourth pillar of the RIF, engagement raises two key issues that cut across DHRs as a process and in using DHRs in research: who is involved in the review/ research process, and how much weight their expertise is attributed. These issues speak directly to ethical tensions around accessibility, safety and wellbeing, in particular, how to safeguard participants who are potentially vulnerable in DHR practice and research, whilst also generating risks to equality and careful inclusion of the range of individuals affected by DVA.

Research Ethics
Research ethics is the final pillar of the framework, promoting an explicit engagement with ethical considerations and the procedural aspects of DVA research, including ethical approval. To some extent, this pillar encompasses many of the issues already raised, with these featuring in both professional and academic ethical frameworks (see, for example, the British Sociological Association's Statement of Ethical Practice, (2017) or the British Association of Social Workers' Code of Ethics (2021)). While the application of ethical guidelines to DHR practice and research has been little explored, both ethical frameworks and institutional review boards (IRBs) provide key oversight mechanisms.
In relation to DHR practice in England and Wales, and DVFRs elsewhere, there has been very limited research into how they are conducted. Yet, Albright et al. (2013) discuss, with reference to DVFRs more generally, how the process speaks to ethical frameworks and principles. While procedural ethics are important, DVA research arguably requires a "widened ethical lens" that moves beyond the requirements of IRBs and considers the risks of harm or exploitation at every step of the research process (Clark & Walker, 2011, p. 1503. DHRs, like any form of DVFR, involve, as Albright et al., (2013, p. 451) suggest, a series of decision-making moments and "the applicability and credibility of fatality review findings and recommendations are affected by decisions made at every stage of the process". Therefore, it is surprising that there has been relatively little investigation of the method behind DHRs/DVFRs, and guidance for practitioners as to how to navigate ethical challenges that arise. Two exceptions to this include Albright et al. (2013) and Dale et al. (2017). Applying the Guiding Principles of Evaluation recommended by the American Evaluation Association, Albright et al. (2013) provide an analysis of potential ethical issues using a program evaluation framework, utilising five guiding principles of ethics in evaluation, including: systematic inquiry, competence, integrity and honesty, respect for people, and responsibilities for general and public welfare. Further developments have been offered by Dale et al. (2017) who identified eight further issues which present ethical dilemmas for practitioners, including confidentiality, review approaches, community interactions, case selection, waivers, family participation, the dissemination of information, and recommendations. However, these issues warrant further investigation in DVFR systems, including DHRs specifically. Given their specific jurisdictional form, DHRs raise new ethical conundrums surrounding consent, anonymity, and harm. Drawing from archival research, although a DHR is in the public domain, "the person…about whom the data were produced" did not, indeed cannot, give consent (Tesar, 2015, p. 103). This may also be the case for bereaved families whose loved ones' life and death have been described and explained. These ethical conundrums unfold in different ways. As stated previously, no ethical framework for DHRs exists in England and Wales, and therefore ensuring and regulating consistent ethical compliance across the system is challenging. Despite being a national system with a framework for delivery via the statutory guidance and a national QA panel, DHRs are directed by localised, regional practices (as each is commissioned by the relevant local CSP). In the absence of an ethical framework, how should practitioners foreground ethical issues in generating findings and address these in DHR reports?
While ethical issues saturate the entire RIF and our analysis, the research ethics pillar is centrally concerned with ethical review and approval. Interestingly, the question of whether DHR research itself requires ethical approval has rarely been asked. Often, studies using DHRs as data state that ethical approval is not required as reports are publicly accessible (e.g., Benbow et al., 2019;Chantler et al., 2020), while in others ethical issues are unaddressed (e.g., Chopra et al., 2022). IRBs primarily review research that involves direct contact with human participants, yet DHRs are secondary data that is usually in the public domain. Therefore, the question arises as to whether there should be more explicit engagement with DHRs as data. When using DHR findings as data, researchers should routinely consider the issues identified across the five pillars. For example, is it ethical for researchers to be using materials from a DHR as a source of data if relatives have been disappointed with, or have not endorsed, the report or its findings? And how should researchers engage with evidence of testimonial injustice?
In summary, practice and research require consideration of both the procedural aspects of ethical approval and review and an iterative process of reflecting upon ethical guidance through the review/research process. The DHR system in England and Wales provokes specific ethical dilemmas for practitioners and researchers which warrant further investigation and can assist both in improving the review/research process.

Implications for Practice and Research
Thus far, we have demonstrated how viewing DHRs as data -both in terms of practices (where findings are generated) and in research (where findings are utilised by researchers) -is a powerful framework through which to analyse the methodological and ethical challenges involved. While there are different challenges for practitioners and researchers across the five pillars, there are commonalities and lessons to be learned by keeping research and practice in dialogue with one another. In DHRs, challenges for both research and practice stem from the nature of the system itself, including the fact that DHRs are stand-alone case examinations, which struggle with tensions between local implementation and national oversight. Analysing the parallels between DHR practices and research practices holds promise as a means of assisting those involved in the doing of, and research into, DHRs. Identifying the parallels in this way presents an opportunity for dialogue between those with a stake in the DHR system, given practice seeks to feed into research and research seeks to influence practice (Rowlands & Bracewell, 2022). Consequently, given the limited empirical research into the doing of DHRs (and DVFRs), this analysis represents an original contribution to the field. In Table 1, we summarise the implications for practice and research across the five pillars, identifying focal points and then considerations for both practitioners and researchers respectively.

Limitations
This paper is a contribution to the limited literature on the ethics and practice of 'doing' DVFRs, building on earlier contributions (Albright et al., 2013;Dale et al., 2017) using a case study of DHRs specifically. However, it is not without limitations. First, we are all white scholars from the Global North. Whilst we bring a range of research and practice perspectives, we do not claim to represent the diversity of voices of those involved in DHRs/DVFRs, including practitioners, testimonial networks, and researchers. Nonetheless, we hope that our open, critical, reflective dialogue stimulates and contributes to an ongoing conversation about the method and ethics of DHRs/DVFRs.
Second, while we have applied the RIF as an analytical framework, we have not evaluated whether it achieves its stated goal of promoting best practices in DVA research. Despite potential challenges, partnerships between advocates, activists and researchers are valuable, not least for ethics and safety (Zimmerman et al., 2016). As demonstrated here, as a product of such collaboration, the RIF can enable, at the very least, an analysis of best practices. However, research integrity and ethics are intrinsically Table 1 Parallels for practice and research a Testimonial injustice occurs when someone's statements are given more or less credibility. Hermeneutical injustice occurs when a knower's experience cannot be understood by listeners, and even the knower her/himself, as they do not have the concepts to make sense of their experience (Fricker, 2007)

Conclusion
DVFRs emerged as a key data collection system on DVA, generating findings which are becoming routine material for researchers. However, there has been little consideration of how DVFRs are conducted in practice, and how findings are subsequently utilised in research. Addressing this gap, we applied the RIF to a case study of DHRs, the DVFR system in England and Wales, to identify the parallels between practices and research. The RIF allowed us to foreground the methodological and ethical issues in research and practice across five key pillars: safety and wellbeing, transparency and accountability, equality, human rights and social justice, engagement, and research ethics. Viewing DHRs/DVFRs as data provides an opportunity to identify parallels in research and practice, and to facilitate dialogue between the two. For practitioners, this analysis shows that to increase transparency, they must interrogate how their findings are generated and how they come to inform system change. For researchers, this analysis demonstrates the value of viewing DHRs/DVFRs critically as data which is to some extent malleable and fallible as data so often are. If policymakers and practitioners are to learn from DHRs/DVFRs, and researchers are to continue to utilise them as data, these systems and their research findings must therefore be problematized as data. Learning from our analysis, the next step might involve developing practice guidance about the wider application of the RIF for DHR practitioners and researchers, with this approach potentially usefully applied to other review systems. DVFR are not the only responses to violence and abuse where the production and use of data needs to be interrogated. Responses to other forms of violence and abuse, such as elder abuse, partner abuse, child abuse, need similar analyses. The RIF provides a framework to untangle the complex relationships between system responses, practitioners, policy makers and researchers.