Prisoner death investigations: a means for improving safety in prisons and societies?

Unsafe prisons mean unsafe societies Prison safety matters and effective ombud institutions could stimulate improved prison safety (Tomczak [63]). In our case, the PPO seeks to effect change in prisons to ensure that the circumstances of prison deaths are brought to light, any relevant failing is exposed, examine whether any changes in operational methods, policy, practice or management arrangements would help prevent recurrence and 'ensure as far as possible that [...] lessons from the death are made clear' (PPO [49], p. 9). Keywords: Prison: Ombudsman;regulation;accountability;monitoring;oversight;serious case review;domestic homicide review EN Prison: Ombudsman regulation accountability monitoring oversight serious case review domestic homicide review 212 230 19 05/24/21 20210601 NES 210601 Introduction The investigation of deaths occurring in unnatural, surprising and unclear circumstances is an important area. [Extracted from the article] Copyright of Journal of Social Welfare & Family Law is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)


Introduction
The investigation of deaths occurring in unnatural, surprising and unclear circumstances is an important area. Death investigation matters for myriad reasons, including: (legal) accountability for risky practices, individuals and institutions; public health and safety; systemic institutional concerns and unaddressed risks; medical care quality; to address community concerns; and for bereaved individuals' understanding (Timmermans 2002, Freckelton 2007, Hanzlick 2016. If deaths are not effectively investigated and suitable remedial action is not taken to respond to findings and recommendation by coroners, 'the adverse consequences for the general community [. . .] can be disastrous' (Freckelton 2007, p. 1).
Prisoner mortality rates are up to 50% higher than in the wider community (UNOHCHR 2019, p. 9) and prisons are 'uniquely liable to abuses and distortion of power' (Liebling and Crewe 2013, p. 286). Prisoner deaths represent 'the extreme end of a continuum of near deaths and injuries', creating important learning which could avert further deaths, 1 'risks to custodial health and safety generally' (Coles and Shaw 2012, p. 2), and risks to societies (Link et al. 2019, Auty and Liebling 2020, McLeod et al. 2020). (Inter)national law imposes obligations to investigate prisoner deaths. These investigations deserve further attention in penal scholarship and practice globally. Every prisoner death investigation provides a window to identify, organise and apply learning that could safeguard prisons and societies. At present, an accountability deficit stretches across the criminal justice system and its overseers due to limited efficacy at preventing future deaths. There is an international need to develop best practice for investigating deaths in detention, which should consider how to stimulate penal change through death investigations.
To support the UK's compliance with Article 2 of the European Convention on Human Rights (ECHR), which protects the right to life, the Prisons and Probation Ombudsman (PPO) has been investigating prisoner deaths in England and Wales since 2004 (Owen and Macdonald 2015). Ombud institutions oversee prisons around the world and provide unelected accountability mechanisms, operating at the intersections of public administration and administrative law (Behan andKirkham 2016, Padfield 2018). Ombud institutions hold substantive potential to shape imprisonment, but have not received commensurate scholarly attention (Carl 2013). Not all Ombud institutions investigate deaths. As 'standard ombudsman territory' (Seneviratne 2010, p. 21) prisoner complaints have generated some limited scholarship (e.g. Sapers andZinger 2010, Calavita andJenness 2013).
This paper makes an important contribution, examining the Ombudsman's success at establishing feedback loops through prisoner death investigations. We identify potential to: i) extend self-conscious, confident communication work, early in the investigation, across PPO investigators; ii) communicate praise as well as deficits to prison staff throughout investigations; iii) adjust template PPO recommendations. An evidence base to inform recommendations is urgently required and holds potential to produce a step change in (inter)national prison oversight practices around the world.

Unsafe prisons mean unsafe societies
Prison safety matters and effective ombud institutions could stimulate improved prison safety (Tomczak 2021). Prison (un)safety is intrinsically concerning to anyone interested in their fellow citizens' social welfare and human rights (Liebling 2004, Rogan 2018. Unsafe prisons have implications far beyond prison walls. Prison (un)safety is instrumentally concerning as it suggests future increases in criminal reoffending and risks to public health. Poorer prisoner health and poorer quality of prison life both correlate with higher reoffending rates (Link et al. 2019, Auty andLiebling 2020). Moreover, prison health is public health (McLeod et al. 2020). Rates of disease, drug dependency and mental illness in prison populations are much higher than in general populations (UNOHCHR 2019, p. 9). Globally, around 30 million people are released from custody each year, so prisons are a vector for (community) transmission of infectious diseases, disproportionately impacting marginalised communities (Kinner et al. 2020). Mass imprisonment is a social problem, producing cumulative, intergenerational social inequality (Simon 2012). Prison safety, prisoner health and imprisonment rates matter, morally and for protecting the health and safety of prisoners, staff and societies. Prison health and safety cannot be the sole preserve of reformists: real people are already caught up in detention, so it cannot be overlooked by those prioritising an anti-carceral agenda (Carlton 2018). Addressing prison health and safety is now more urgent than ever before. Almost 11 million people are imprisoned globally, of whom 30% have not been convicted of a crime (Penal Reform International 2019). Amidst the global COVID-19 pandemic, these people are unable to leave environments that concentrate 'poverty, conflict, discrimination and disinterest' (World Health Organization 2000, p. 11).
In an adjacent area of inquiry, Serious Case Reviews in England were established under the Children Act 2004 to review cases where a child has died and abuse or neglect is known or suspected. Serious Case Reviews have been conceptualised as the tip of an iceberg: overall numbers of children dying through maltreatment are small, but many more suffer lower levels of abuse or neglect. In turn, every Serious Case Review provides a window 'allowing us to identify lessons to be learnt for safeguarding and promoting the welfare of children' (Brandon et al. 2012, pp. 1-2). Similarly, prisoner deaths represent the tip of an iceberg of prisoner ill health, neglect, poor quality of prison life and abuse, which are all morally problematic and correlate with increased reoffending and public health risks. (Inter)national law imposes obligations to investigate prisoner deaths (International Committee of the Red Cross/ICRC 2013, Rogan 2018). Prisoner death investigations create rich data which have been little utilised by academia, policy or practice (Tomczak 2018(Tomczak , 2021, but offer 'a starting point for much-needed analysis' into the quality of prison life, prison regimes (Liebling and Ludlow 2016, p. 238), prisoner (ill-)health and risks to societal health and safety. It is recognised that these investigations serve important functions in terms of providing an effective remedy for the next of kin, and holding those responsible accountable (ICRC 2013). Ryan (2019) argues that healthcare inquests and inquiries should seek to trigger demonstrable change, to reassure bereaved families that changes will be implemented as an outcome of their loved one's death. Every prisoner death investigation provides a window to identify, organise and apply learning that could not only reassure bereaved families, but safeguard and enhance safety in prisons and societies. Death investigations are potentially significant triggers for harm reduction, although we now demonstrate that this potential is yet unrealised.
Current prison conditions in England and Wales pose risks to the health and wellbeing of prisoners, staff, prisoners' families (Ismail 2020) and societies (Tomczak 2021). Prison safety dramatically deteriorated from 2012, with increased levels of suicide, violence, self-harm, radicalisation and gang formation (Ismail 2019, Prison Reform Trust/PRT 2019. Between 2012 and 2016, suicide rates more than doubled (Ministry of Justice/MoJ 2017), 'after a period of change for the prison workforce as a result of making efficiencies' (MoJ 2016, p. 41). Benchmarking policy generated the largest staff reductions in the service's history (Peacock et al. 2018). 2016's record prison suicide numbers drained hundreds of millions of pounds from public funds and harmed prisoners, prison staff and those bereaved (Tomczak 2018). Self-inflicted death rates remain elevated above pre-benchmarking rates (MoJ 2020, p. 2). Fiscal austerity has exacerbated the existing poor health of prisoners, impeding access to healthcare and productive activities (Ismail 2019). Statistics indicate i) declining prisoner health, which is crucial for behaviour change and has important implications for re-entry outcomes, reincarceration (Link et al. 2019) and public health (McLeod et al. 2020), and ii) declining prison safety and legitimate order, which is a precondition for prisoner growth and change (Auty and Liebling 2020). Criminal reoffending already costs £18.1 billion annually in England and Wales (Newton et al. 2019).

Prisoner death investigations
Deaths in coercive institutions involve legal obligations and legal responses (Rogan 2018) and threaten the fundamental human right to life, which forms the basic pre-condition of the enjoyment of other rights (Owen and Macdonald 2015). (Inter)national human rights law and humanitarian law impose obligations to respect and protect life in all circumstances, and to investigate suspected violations of the right to life (ICRC 2013). Nevertheless, in many prisons globally deaths are frequent, sometimes preventable and mostly considered 'natural' (Tomczak 2018). Prisoner death investigations remain ripe for development. Following an international survey, the Correctional Service Canada (CSC) Review on Non-natural Deaths in Custody (CSC 2018, pp. 58-9, emphasis added) concluded: 'there is a need to develop best practice' for investigating deaths in custody, rather than adopting best practice from other countries. We add that such best practice should consider how to identify, organise and apply learning from death investigations, in order to effect penal change through death investigations and in turn safeguard prisons and societies.
Applicable legal frameworks, and the agencies and techniques used to investigate deaths vary widely between countries and detention institutions. Legal frameworks include soft international law e.g. the United Nations Standard Minimum Rules for the Treatment of Prisoners/Mandela Rules: specifically Rules 71.1 (on investigating the circumstances and causes of deaths) and 71.2 (on dealing with the body), and international jurisprudence e.g. the ECHR. The ICRC sets out state obligations under international law and summarises guidelines for investigations in Annexe II (ICRC 2013).
In the Council of Europe's 46 member states, all deaths in compulsory state detention that are unexplained or related to violence and self-harm automatically engage Article 2 of the ECHR, which protects the right to life. Article 2 includes a procedural obligation, and duty to investigate potential violations of the i) positive obligation: to protect the life of a person involuntarily in state custody and ii) negative obligation: to prohibit intentional and unlawful taking of life by state agents. The form and nature of an Article 2 investigation varies across jurisdictions but must meet multiple criteria. The investigating authorities must act of their own motion. The investigation must be initiated promptly and proceed with reasonable expedition; be independent; be open to public scrutiny; involve the next of kin to the extent necessary to safeguard their legitimate interests; and be effective. Effective means that the investigation must: be conducted in a manner that does not undermine its ability to establish the relevant facts; comprise or obtain sufficient expertise; secure the relevant evidence, including witness evidence; identify those responsible for the death; reach conclusions in the central issues that are tenable and convincing and identify any shortcomings in the operation of the regulatory system; and ensure accountability of state agents and bodies (Owen and Macdonald 2015). These parameters are valuable and important, but primarily backwards facing. Whilst this legislative basis underpinning the crucial tasks of identifying, organising and applying learning from prisoner deaths is weak, the House of Lords to ensure so far as possible that the full facts are brought to light, that culpable and discreditable conduct is exposed and brought to public notice, that suspicion of deliberate wrongdoing if unjustified is allayed, that dangerous practices and procedures are rectified, and that those who have lost their relative may at least have the satisfaction of knowing that lessons learned from his death may save the lives of others.
The PPO has no statutory footing and operates using memoranda of understanding. Despite the primarily backwards-facing legislative basis for their investigations, the PPO assists the Coroner's inquest to fulfil the investigative obligation arising under Article 2 'by working together with coroners to ensure as far as possible that the full facts are brought to light and any relevant failing is exposed, any commendable action or practice is identified, and any lessons from the death are made clear' and to 'examine whether any change in operational methods, policy, practice or management arrangements would help prevent a recurrence' (PPO 2017, p. 9 2 ).
In the UK, a Coroner must investigate all deaths in compulsory state detention under the Coroners and Justice Act 2009 section 1(2)(c). Where the Coroner's inquest illustrates a concern that circumstances creating a risk of other deaths will occur or continue to exist in the future, the Coroner is obliged to send a Report to Prevent Future Deaths/ PFD report to organisations or persons with the power to take preventative action, even if the Coroner's concern is peripheral to the death. PPO investigations have barely been researched by scholars. By its own analysis, the PPO struggles to effect change: In 2018/19 we began 334 fatal incident investigations. [. . .] We saw a 23% increase in selfinflicted deaths this year with worryingly high numbers in some prisons. In many cases, we had to make the same recommendations as in previous years, where remedial action had been promised (PPO 2019, p. 11, emphasis added).
Related analyses of Coroners' inquests into custodial deaths highlight an accountability deficit across the criminal justice system and its overseers, because Coroner's interpretation of their power to make PFD reports is inconsistent across locales, nobody is responsible for judging whether responses to PFD reports are appropriate and effective, and there is no consistent mechanism for assessing implementation. Limitations are exacerbated by PFD reports being perceived as punishment rather than a learning opportunity, and a lack of clarity about who should engage stakeholders to create improvements (Coles andShaw 2012, Mendas 2012).
It is particularly concerning that death investigations' harm reduction potential is unrealised because England and Wales' range of prison oversight bodies form comparators for elsewhere (van Zyl Smit 2010, Tomczak 2021). Yet, the UK's 'blueprint' oversight apparatuses have neither challenged high imprisonment rates (with Scotland, then England and Wales leading Western Europe (PRT 2019, p. 56)), nor addressed recent dramatic declines in prison safety. The UK claims world-renowned detention monitoring methodologies and approaches, which its Foreign and Commonwealth Office actively promotes overseas. The UK actively participated in drafting and was amongst the first to ratify the Optional Protocol to the United Nations Convention against Torture and Inhuman or Degrading Treatment or Punishment (OPCAT), which is now being exported around the world (NPM 2016).

Theoretical underpinnings
Prison regulation is an enduring, essential 'counterweight to potential abuse of the special powers of the state' (Hood et al. 1999, p. 116), made only more urgent by neoliberal carceral expansionism in many jurisdictions, the globally expanding prison population and the COVID-19 pandemic (Tomczak 2021). Regulation is concerned with improving performance by steering the flow of events and behaviour (Braithwaite et al. 2007a), and holding key personnel responsible for it. Regulation encompasses sanctioning and supporting activities, most frequently involving education and persuasion but potentially escalating to litigation and prizes (Braithwaite et al. 2007b). Regulation can have transformative, emancipatory effects on individual people, public services and institutions (Braithwaite et al. 2007b, Smith 2009), as such: 'regulation matters, and therefore the development and empirical testing of theories about regulation also matter' (Braithwaite et al. 2007a, p. 4).
Regulation is essentially prospective (forward facing) but complements and overlaps with accountability processes which are essentially retrospective (backwards facing) (Black 2001). Regulation and accountability share common rationales of improving performance and holding actors responsible, and both seek to steer behaviour: establishing, monitoring and enforcing performance standards through rewards, awards and sanctions (Smith 2009). Prison oversight bodies differentially prioritise regulation and accountability. The PPO's work on death investigations and complaints most closely resembles accountability, but OPCAT is principally preventative and most closely resembles regulation. It is often unclear where regulation ends and accountability begins (Black 2001), but fair and effective social processes require them both (Smith 2009). Sapers andZinger (2010, p. 1515)  Pursuing individualised accountability for past failures and abuses for its own sake is a partial strategy that will not enable ombud institutions to fulfil their potential as agents of harm reduction, and also means that prevention efforts will not be adequately informed by the lessons of the past. As such, it is notable that the PPO in England and Wales do not sit on the UK's OPCAT National Preventative Mechanism. 3 Similar debates are also present in the broader scholarship on ombud institutions. Ombud institutions that focus on casework and investigating individual injustices miss opportunities to bring about good administration by acting as a 'system fixer' and 'learning agent', potentially by mobilising or establishing 'own-initiative' powers of investigation (Gill 2020). This is particularly important because the ability to perceive and name problems is structurally patterned (Calavita and Jenness 2013) and those with fewer resources and/or in vulnerable social locations are predisposed to self-blame and disentitlement (Michelson 2007). As such, it is not sufficient for ombud institutions to solely act responsively and pursue backwards-facing accountability efforts (Gill 2020).

Analytical approach: cybernetics
Applied to government and societies by Deutsch (1966), cybernetics is a generalised theory of governance which examines how order is created and maintained. The cybernetic paradigm valuably integrates existing governance models by exploring how diverse subsystems function simultaneously to create self-correcting institutions (Birnbaum 1989). Every ordered structure will in some form seek to control its environment, requiring it to perform actions through which it attempts to affect its environment to fulfil its purposes, monitor whether these actions had the intended effects and, if not, adjust its actions accordingly (Gadinger and Peters 2016). Cybernetics examines ways in which a system's output affects input into that system, and thus subsequent output. To claim control, a system must demonstrate cybernetic validity, i.e. having a feedback loop in which some output triggers a system change, being used as input for future operations (Green and Welsh 1988). Examining organisational control systems in terms of sensing mechanisms and feedback loops highlights institutional thermostats which monitor changes from acceptable levels of functioning and activate forces that return institutions to their previous stable state (Birnbaum 1989).
In our case, the PPO seeks to effect change in prisons to ensure that the circumstances of prison deaths are brought to light, any relevant failing is exposed, examine whether any changes in operational methods, policy, practice or management arrangements would help prevent recurrence and 'ensure as far as possible that [. . .] lessons from the death are made clear' (PPO 2017, p. 9). Following Gadinger and Peters' conceptualisation of feedback loops in foreign policy (2016), our analysis examines: i) how the PPO seeks to effect change in prisons following a prisoner death, ii) whether these actions had the intended effects and iii) if and how the PPO adjusts its actions.
Our approach is innovative, as cybernetics have been little used in criminal justice scholarship and/or practice. Our approach is valuable because analysing feedback (in) effects can transcend (sub)disciplinary boundaries, by combining analysis of structure and agency, over time, in a single framework (Gadinger and Peters 2016). Structuralist theories (e.g. Foucauldian analyses) tend to be static and obfuscate potential for agency and change. Cybernetics provides a non-deterministic perspective that links action (in our case, of the PPO) with the structures in which it is embedded (England and Wales' Prison Service), by which it is affected and affects itself. Neither agency nor structure is considered determinate, because no actor can fully control its environment and, by acknowledging fundamental elements of chance in the texture of the universe, cybernetics can acknowledge creativity and novelty in social relations (Deutsch 1966). Feedback happens over time, so forces analysts to think across longer intervals, which become easily obscured both in other types of scholarship and busy organisations. Feedback analysis can make long-term processes that (fail to) produce change visible. As feedback loops can amplify effects, cybernetics holds that 'small causes can have large consequences' (Gadinger and Peters 2016, p. 256). This approach is beneficial for this case, given i) the prevailing dystopias in criminological scholarship and tendency to produce teleological depictions of ever-proliferating governmentality which obfuscate any possibility for remoralising carceral regimes and deconstructing the carceral state (Carlen 2001, Zedner 2002, Bosworth 2011; and ii) the urgent need to reduce penal and concurrent social harms (Tomczak 2021).

Materials and methods
To investigate the extent to which the PPO meets its potential to improve safety in prisons and societies, a large research project was undertaken with multisectoral stakeholders. This article reports on one of the initial phases, which entailed 16 semistructured interviews with Ombudsman staff, spanning Senior Investigator to Senior Management roles. Volunteer participants were recruited using a purposive sample.
Interviews were carried out by Philippa Tomczak and research assistant Sara Hyde in December 2019, and were audio recorded with participants' consent. Ethical approval was granted by Her Majesty's Prison and Probation Service and the University of Nottingham.
This research resulted from the academic (Philippa Tomczak) and the Ombudsman (Sue McAllister) meeting in January 2019 at a Prison Reform Trust Prisoner Policy Network event. Tomczak had researched prisoner death investigations in England and Wales (Tomczak 2018) and remained concerned about death rates. McAllister was newly in post and seeking to increase the impact of Ombudsman investigations. Our partnership (Tomczak/McAllister) was initially funded by a University of Nottingham impact prize and ESRC impact accelerator grant, administered by Tomczak.
All interview transcripts were thematically coded and analysed in Word, using ethnographic content analysis (ECA) to answer the three research questions. Unlike positivist document analysis, ECA conceptualises document analysis as fieldwork. ECA entails discovery of analytical themes and sustained reflexivity about the research process and document production processes. Reflexive and recursive movement between concept development-sampling-data collection-data coding-data analysis -interpretation provides a systematic approach, whilst retaining flexibility to (re)develop analytical categories (Altheide and Schneider 2013). This flexibility can enable novel findings and stepchanges in analysis, facilitating important implications for practice.

Results
Results are now presented in four sections. The first provides context regarding PPO investigations to underpin the three following sections, which examine: i) how the PPO seeks to effect change in prisons following a prisoner death, ii) whether these actions had the intended effects and iii) if and how the PPO adjusts its actions.

Fatal incident investigations: context
The PPO investigate every prisoner death and produce good quality reports.  [. . .]. In more complex investigations, including self-inflicted deaths, we will nearly always go to the prison and interview staff and prisoners and medical staff.
These investigations therefore draw on a wide range of data sources, facilitating triangulation and strongly positioning the investigation to establish 'the full circumstances of prison deaths' (PPO 2017, p. 9) and inform the Coroner thereof. The investigation may also be guided by family input, adding both a further source of information and the potential to assuage family concerns: AVERY: A Family Liaison Officer, they always make contact at the very beginning with [. . .] the bereaved family, to ask if there's any questions they want looked at.
All of the available evidence is then compared to local and national prison policy: However, PPO staff were not constrained by these policies. Indeed, Beverley described two instances where the PPO's work had stimulated changes in national policy regarding prison operations. As such, there appears to be scope to pursue further systemic changes where required.
BEVERLEY: The restraints is a good example of our work [. . .] some fairly high-profile cases resulted in a change to the policy where healthcare considerations have to be given due weight [. . .]. Another example is about emergency response. So if you are a Prison Officer working on a landing and you go to a cell and you look through the hatch and you see somebody hanging, the policy, influenced in some part by our findings, is that subject to a dynamic risk assessment, if you think there is a threat to life and you assess that it is safe to do so you can go into the cell immediately.
Preventing prisoner deaths was expressed as an investigation aim by all 16 participants.

INTERVIEWER: investigations, what are the goals, what are they trying to achieve?
SIDNEY: Overall to make prisons safer and to stop prisoners committing suicide, or killing each other, or dying from drugs essentially.
HARPER: Our goal is to try and make things better, to try and make changes or to try and encourage the Prison Service to make changes, so [.] someone else in that same situation might not, and to try and [. . .] protect people.
PPO staff were therefore clearly not just investigating the circumstances of deaths but aiming to apply that knowledge to prevent further deaths. In order to protect prisoners and prevent deaths, Harper strived to produce their best investigation, identifying which problems happened, and where: HARPER: By doing as good as an investigation as you can, by working out what the problems are and where those, the sort of fatal flaws have happened.
Investigators were gathering extensive information and working hard to complete their investigations. Resultant findings are then fed back to prison staff at different points in time, but the main means through which the PPO seeks to control prisons is through recommendations, which are produced from the findings of their investigations.

i) How does the PPO seek to effect change in prisons following a prisoner death?
BEVERLEY: We find where things have gone wrong or not been done properly and we make recommendations as to what needs to be done to put those things right, to prevent recurrences in the future [. . .] We investigate what happened, we look at what happened, we look at what went wrong and we make recommendations to put those things right.
Staff discussed feeding back their findings to prison staff during the investigation's early stages, although this appeared to be inconsistent across investigators, dependent on individual initiative. For Finley, early feedback was rare, reserved for substantive issues requiring immediate attention.
FINLEY: If something particularly big comes out, i.e. where we feel that something needs to change straightaway, then we would inform the Governor 4 or Director 5 as soon as we can. That's relatively exceptional, I mean you would normally go in and [. . .]  Finley also identified that his interpersonal skills were less strong than his written work, but he was evaluated only on the latter. [.] We are always out there on our own with no real support, with no management oversight really.
For Ella, feeding back to staff at all levels during the early investigation was core practice that could address problems quickly. Ella identified this practice to be 'as important if not more important' than the report, and appeared confident approaching senior Governors.
ELLA: You have got six months to produce your investigation report, it will be daft not to flag something, [. . .] be foolish to leave that six months. So I would just go in and have a conversation with the Governor or Director and say 'look your staff don't understand what the medical emergency process is, you need to do something about that'. .] that it's clear, it's understandable, so that they can recognise that they need to make changes. [. . .] Getting them on side means that they will actually want to make changes [. . .]. It can be very hard but that's the way to do it, I think.
Ella and Harper displayed self-conscious, confident communication work, early in the investigation, which there is potential to extend across PPO investigators. However, participants overwhelmingly discussed communicating with prison staff about deficits. Nursing home regulation research has emphasised that inspectors' praise engendered the collective pride that is the stuff of high workplace morale and high performance. [. . .] Inspection teams who used praise a lot as a regulatory strategy improved compliance in the two years following an inspection significantly more than inspection teams that did not (Braithwaite et al. 2007b, pp. 116-117).
Praise improved compliance regardless of how deserving of praise the nursing homes were. When collectives are praised, all members want to share in the credit, and when individual members of a collective are praised, the collective claims a share of the praise. In contrast, when collectives are blamed or punished, 'each involved individual tends to believe it is someone other than them who is responsible; when individuals are blamed, the collective tends to disown or distance itself from the individual' (Braithwaite et al. 2007b, pp. 116-117). Given that informal praise is a more powerful form of social control than blaming, there is considerable scope for the PPO to communicate praise as well as deficits in order to better influence practice in prisons.

ii) Do PPO actions have the intended effects on prisons?
All participants attached importance to their work: including addressing power imbalances, making contributions to bereaved families, humanising deceased prisoners, supporting Coroners' inquests and supporting human rights litigation. BEVERLEY: I've been really exercised [.] by the question of is that enough, about could we, [. . .] should we be doing more to influence what actually happens in prisons and to reduce the number of people who take their own lives and to make prisons fairer and safer? SIDNEY: We really look into that death and see if there were any failings and if there were, find out why [. . .] and feed that back to the Governor and obviously write a report and make recommendations to the relevant people. [. . .] The problem comes when they don't, [. . .] they have to provide an action plan but it is whether they ever do anything to stop that happening again, that's the bit that frustrates me.
Seeing repeated failings, sometimes within the same prison, was an indication that the window of opportunity to identify, organise and apply learning that could safeguard prisons and societies was being squandered. Moreover, seeing repeated failings could be undesirable and difficult for PPO staff, even feeling soul-destroying.

INTERVIEWER: in a best case scenario, what would the work of the PPO achieve?
AVERY: That these same things wouldn't keep happening.
HARPER: Some prisons [. . .] it almost seems that [. . .] we could just copy and paste a previous report, change the names and the same things would apply. [. . .] just nothing changes, that they just pay, almost pay us, our reports and recommendations, almost lip service. [. . .] It can be fairly soul-destroying [. . .] when you have made recommendations and the prisons are pushing back or you hear that the same situation has happened and they are just not learning, so you can be quite despondent. [. . .] I'm almost saying I'm not sure it achieves a massive amount in some of the bigger prisons because we do keep saying the same things again and again and again.
Despite the primarily backwards-facing legislative basis for their investigations, PPO staff are invested in preventing prisoner deaths. Yet, consistently repeated recommendations indicate that PPO investigations do not reduce prisons' vulnerability to preventing future deaths and potentially establish a vicious cycle by demoralising PPO staff, which has implications for their wellbeing, productivity (Kalra et al. 2016), and motivation to invest in long-term, strategic thinking and/or attempts to adapt practice.

iii) How does the PPO adjust its actions?
As the interviews made clear, PPO staff sought to reduce prisoner deaths and could be greatly affected by repeated failings and continuing deaths, sometimes over long careers at the organisation. However, attempts to change PPO practice in response to this lack of impact appeared limited. Whilst these explanations are valid, such narratives divert attention from considering how the PPO could adjust its actions to attain control and make more of an impact on prison safety. Notably, none of the PPO staff considered the substance of their recommendations. Although recommendations neither flow directly from investigations nor have a clear relationship with changes to policy and practice (Stark 2019), they are within the PPO's control and have potential to achieve better organisational and policy learning (Hyland and Holme 2009).

INTERVIEWER
Drawing on regulatory theory, this partnership has identified potential to: i) extend self-conscious, confident communication work, early in the investigation, across PPO investigators; ii) communicate praise as well as deficits throughout investigations; iii) adjust template recommendations. These findings have implications for prisoner death investigations and prison oversight bodies globally.
Regarding recommendations, Serious Case Review critiques highlight that there is rarely 'a research evidence base cited for the recommendations', begging the question of 'whether there were clear rationales for making, or not making, recommendations' and 'the extent to which recommendations were thought to be likely to deliver change' (Brandon et al. 2012, p. 6). Moreover, the 'type of recommendations which are easiest to [. . .] implement may not be the ones which are most likely to foster safer, reflective practice' (Brandon et al. 2012, p. 6). Our next step involves designing a practical pilot to explore the potential of these three areas in PPO practice, including identifying good practice as well as systemic and individual failings throughout investigations. This could enhance the PPO's capacity to effect change: leading to safer prisons, fewer preventable deaths and increased PPO staff wellbeing.

Discussion
Extant (inter)national prison oversight apparatuses are extensive, and being implemented in an increasing number of countries globally through OPCAT. Prison oversight holds substantive, yet unrealised potential to shape imprisonment, for the benefit of those who work and live in prisons, their families, and the societies from which prisoners come and almost always return. But, there is a surprisingly limited research base to inform prison oversight (Hardwick and Murray 2019, Rogan 2019). We need a great deal more research to understand the impact of and means of improving all forms of prison oversight: on prisons (Padfield 2018) and societies. Beyond individual expertise, there is no guidance available to direct or facilitate evidence-based recommendations, despite the ubiquity of recommendations as a prison oversight tool.
Nobody has yet developed an evidence base to guide how those recommendations should flow from investigation and inspection findings. This is a significant gap globally. OPCAT's ratifying states must establish independent National Preventative Mechanisms (NPM) who regularly visit detention sites. NPM members are mandated to produce reports following their visits, which should, when appropriate 'contain recommendations addressed to the relevant authorities' (UN 2010, p. 3). Because the PPO is 'wholly independent' from the authorities in its remit, it is apparently equipped 'to execute fair and impartial investigations, making recommendations for change where necessary, without fear or favour' (PPO 2017, p. 1). Whilst independence is undeniably valuable, independence does not imply efficacy. Developing evidence-based, forward-facing recommendations for use by prison oversight bodies is an essential yet complicated task. The need for an evidence base for recommendations reflects a widespread issue in human rights practice: that 'indicators' of human rights progress are commonly formulated without evidence that they have a causal relationship to the intended outcome, hence we have little knowledge of what forms of torture prevention actually work (Carver and Handley 2020).
Our partnership (Tomczak/McAllister) will soon pilot a revised fatal incident investigation report structure and revised recommendations, based on our research findings and drawing on lesson learning strategies, which have improved safety in the airline industry, healthcare and policing. Lesson learning is based on the idea that operational failure provides an invaluable learning opportunity, whether caused by mismanagement, misfortune, mistake, or misconduct. Lesson learning strategies involve investigators and analysts identifying all of the factors that contributed to failure, rather than presuming human error, and then implementing the lessons learned (Smith 2009).