Abstract
Serious further offences are rare but represent catastrophic harm to the public by people under probation supervision. Where cases had previously been identified at high risk of harm and there are significant failings in their management, independent reviews are commissioned. Surprisingly, learning from these reports has received scant attention, with the last review 20 years ago. To address this, a pre-registered systematic search was carried out, identifying 37 reports of reviews of offences committed in the UK, published between 2005 and 2024. These reports were reviewed, drawing on the rapid evidence assessment approach, and antecedent behaviours were qualitatively synthesised.
Keywords
Serious further offences (SFOs) are specific violent and sexual offences committed by people who at the time of the offence are, or very recently were, under probation supervision. The probation service supervises approximately 240,000 people in the community at any one time and, in the year 2021–2022, there were 288 convictions for SFOs (Ministry of Justice [MoJ], 2024). SFO convictions consistently represent fewer than 0.5% of offenders under statutory supervision (MoJ, 2022).
HM prison and probation service has a system in place to formally review every case in order to identify any improvements which need to be made to the systems or to the practice of individual probation practitioners or managers. Cases for review are identified when serious violent and sexual offences are committed by a person under community supervision where that person has been assessed as posing a high risk of harm at some point during their sentence (HM Inspectorate of Probation [HMIP], 2020a). These cases are submitted to the Head Office but only progress to an ‘independent review’ where there are significant failings. The learning from these independent reviews is the subject of this report.
The independence of reviews is critical. When SFO reviews are completed internally by probation regions on themselves, a large proportion do not meet the required quality standard (HMIP, 2023a). In 2022/23, HMIP rated 47% of internal reviews ‘requires improvement’ or ‘inadequate’; a decline in the level of review quality compared to 31% receiving those ratings in 2021/22. A key practice deficit identified was practitioners not always seeking/using all available risk of harm information, including assessments not considering the breadth and nature of the risk posed and failing to recognise or respond to emerging risk issues. Underpinning this was practitioners not using reflection and having an optimism bias when interpreting behaviour – not taking a sufficiently inquisitive approach to risk (see HMIP, 2023a).
Current learning from independent reviews of serious further offences
HMIP uploads inquiry and review reports to its website (Justice Inspectorates, 2024). An HMIP (2020a) thematic inspection found that, at a national level, SFO reviews were too descriptive and not sufficiently analytical to identify underlying themes which could inform policy and support improvements in practice. Two issues were identified in a section of the website reflecting on the lessons from the Fishmonger's Hall terror attack in 2019. These two issues, namely the need for (a) sharper attention to ‘false compliance’ and (b) improved sharing of relevant information between agencies, were raised in the context of their resonance with the Joseph McCann case review (HMIP, 2020b). However, to our knowledge, there has not – for 20 years – been an integrative report of lessons learned across independent reviews of SFOs.
The previous ‘learning points derived from serious further offence full reviews’ report (National Probation Service, 2005) made four recommendations for improved case management. First, it recommended better, i.e. more timely and more complete, assessments of the risk of reoffending and serious harm. Notably, there was a need for improved risk of harm reviews in cases due to be transferred to a different probation area. Second, it advocated for an increased awareness of the significance of domestic abuse; including better sharing of information between probation and police services, more training in domestic abuse awareness and in the use of domestic abuse risk assessment tools, focussing especially on the protection of intimate partners. Third, it highlighted the need for better information sharing between agencies, suggesting improved interagency protocols and increased reflection and learning from SFOs, especially regarding cases under higher levels of supervision. Finally, the report suggested improved enforcement of licence conditions, including better recording of attendance, absence and breaches, and better prioritisation of limited resources to enforce orders during periods of understaffing.
The issue of enforcement is especially germane, as monitoring non-compliant and subversive institutional behaviour is recognised as a means of identifying people more likely to re-offend after release (Goodley and Pearson, 2024). Monitoring such behavioural warning signs, sometimes used as a component of or complement to resource-intensive structured professional judgement tools (Logan and Taylor, 2024), has demonstrated efficacy as an individualised and time-responsive means of identifying re-offending risk (Clark et al., 1993; McDougall et al., 2013; Nitsche et al., 2022). In England and Wales, monitoring and control – or strategies to detect the build-up to serious harm and prevent its occurrence – is identified as one of the ‘four pillars’ of risk management within multi-agency public protection arrangements (HM Prison and Probation Service, 2023). While providing a cross-situational signal regarding an individual's commitment to desist (Cochran and Mears, 2017), event monitoring also shines a light over the responsiveness of the clinical settings and services.
The current study
Although the previous ‘learning the lessons’ review was two decades ago, it underlined that failures are not simply reducible to the characteristics of the person offending. Failures may be better understood as an end-point that originates in human/clinical error and latent weaknesses in the organisational environment. In nature, hazards and human systems are known to interact (Gill and Malamud, 2016). Therefore, increasing the efficiency of the services around the person may mitigate the likelihood of an SFO. To this end, the present review drew upon ‘systems thinking’ to highlight errors at all points of the timeline leading to the SFO and highlight shortcomings within all services provided. This scheme looks beneath the ‘tip of the iceberg’ of major incidents and draws on a ‘Swiss Cheese Model’ approach to assessing risk (Reason, 2000). In Reason's model, any organisational system has in-built protective layers like slices of Swiss cheese where the holes represent weaknesses. A trajectory of opportunity for an adverse event (in this case, an SFO) is unsafeguarded when the holes come into alignment. The holes arise due to active failures and latent conditions, and, according to Reason (2000, p. 768), nearly all adverse events involve a combination of these two sets of factors.
Aims and research questions
This thematic review was completed in June 2024 and aimed to analyse all case inquiries and reviews of SFOs since 2005, including those in criminal justice and forensic mental health. We sought to identify any themes that could inform policy and support improvements in practice. The year 2005 was selected because the review aimed to provide, in its results, data that update and extend those of the previous report of ‘learning points derived from serious further offence full reviews’ (National Probation Service, 2005).
The clinical question guiding the current review was to identify what professional learning can be gleaned from independent reviews of SFOs. The present review therefore had the objective of identifying whether there are any patterns or themes across SFO case reviews and inquiries which highlight learning that can be applied to policy and/or prison and probation practice. To support this, we aimed to verify whether the findings of more recent independent reviews of SFOs (2015–2024) were dissimilar from the earlier (2005–2014) findings of independent reviews. Given our focus on organisational learning, we used the publication date of the review, rather than the date of the SFO incident (generally a full year earlier). For clarity, we used the mid-point for time split comparison; rather than filtering by one or more historical organisational events.1
Method
The current study was pre-registered with PROSPERO, Centre for Reviews and Dissemination, University of York (see https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023407686). As recorded there, the pre-registered methods were adapted shortly after first registration to ensure the search strategy included NHS England investigations following serious offences by mental healthcare users subject to probation supervision.
Eligibility
The current review targeted only independent inquiry and review reports published since 2005. These were further offences committed by people on probation in the UK and processed by HM inspectorate of probation and HM inspectorate of prisons, or committed by users of NHS forensic mental health services while subject to probation. Cases reviewed in academic journals, if found, were also eligible for inclusion.
Since SFO cases only progress to independent review by HM inspectorate where there are significant failings (HM Inspectorate of Probation, 2020a), these cases have already been selected from the overall pool of SFO cases as having significant shortcomings. Further to this, the current research aimed to ascertain patterns in the nature of these shortcomings and whether they had changed over the extended time period.
Search strategy
Data were collected via two search strategies: (a) a rapid evidence assessment (REA) search of health science databases; and (b) a hand search.
The REA search involved an online search of specific terms using an EBSCO web search across the following four databases: PubMed, MEDLINE, Criminal Justice Abstracts and PsycINFO. The following Boolean search terms were used across databases: ‘serious further offence OR SFO’, AND ‘offence OR offend’, AND ‘probation OR prison’, AND ‘independent AND (review OR investigation)’, and ‘UK OR UK OR England & Wales OR England OR Wales’. Filters were applied to identify reviews in English, between 2005 and 2024, and to include the terms in All Text, not just within the title.
The above terms were also searched in the Probation Journal and via Google, but the results of these specific resources were hand-searched for relevance. The Probation Journal was hand-searched, unlike the other scientific databases, due to its small size and the minimal number of search results that it yielded (n = 23 reports). It was therefore more efficient to hand-search Probation Journal results. Regarding Google, given its more numerous and repetitive results, hand-searching was most efficient as the few results that were not repeated were easier to spot by eye. The flow of studies from identification to inclusion is shown in Figure 1.
Data search PRISMA flowchart (Moher et al., 2009).
Study selection
A total of 37 reviews were extracted for inclusion. These 37 reports consisted of 11 cases within the prison/probation services and 26 forensic mental health patients. The majority of highly relevant and detailed data (n = 9 records) was collected from hand-searching HM inspectorate of probation inquiry and review reports (Justice Inspectorates, 2024) and from NHS England investigations (n = 26 records) following serious offences by mental healthcare users. This yielded some cases of patients who were on probation as well as being subject to a health care plan. As indicated above, to be included these cases needed to be on probation for a prior offence at the time of committing the serious further offence; whether on a court order or post-release licence.
As shown in Figure 1, 79 full reports were identified. Sixteen of those found in the hand search were excluded as duplicates of the same records that were less extensive than the official review. A further 26 reports were excluded as they were NHS cases that were not on probation at the time of the offence, leaving a final total of 37. Of these 37 reviews, 11 were published during the period 2005–2014, and 26 were published in 2015–2024. Two reviews covered the same case (NHS England & Niche Health and Social Care Consulting, 2017, 2018), but were published in different years and contained entirely different information. Both of these reviews were therefore included to contribute to a more complete data set. Of the final 37 reviews, 8 were HMIP reviews (9 records: Damien Hanson and Elliott White shared one HMIP review but this was split and analysed as two individual records). One of the remaining two reviews from the hand-search, regarding Jon Venables (Omand, 2010), was found within Google results. The final review report, concerning the Sonnex case (Cluley, 2010), was found via hand search of Probation Journal.
In this selection procedure, reviewer GC processed the publications under the supervision of reviewer DP, who also reviewed a small sample (20%) of publications to mitigate the risk of bias. There was no disagreement on inclusion/exclusion.
Data extraction
Relevant qualitative data were independently hand-selected from each of the reviewed publications and brought together using a data extraction form with pre-defined headings, e.g. Prior convictions, Index offence, Current SFO incident, Antecedent behaviours, Relationship with the victim. Data were processed in Microsoft Excel. The qualitative data in ‘Antecedent Behaviours’ were coded within QSR NVivo to identify themes in behaviours that related to the index offence and the SFO incident. Data were synthesised narratively, using the rapid evidence assessment approach (Barends et al., 2020).
Quality appraisal
No studies were excluded due to quality. All published reviews followed the same structure and quality checking during their completion by the official government bodies HM inspectorate of prisons, HM inspectorate of probation, and NHS England.
Results
Our aim was to identify patterns or themes across SFO case reviews and inquiries which highlight, or indicate a need for, organisational learning. To support this, first the data were compared and contrasted across the time period (2005–2024), using the mid-point (2014) to delineate two time periods. Then, the textual data extracted under ‘antecedent behaviours’ were synthesised. Each of the individual reports included in the review is *(asterisk) marked in the reference list.
Time split analysis
First, it was clear that the number of independent reviews increased over the time periods, from 11 to 26. Although offset by a reduction in release on temporary licence (RoTL) cases due to tightened RoTL eligibility criteria in 2015, this increase may be explained by the widened sample after the Offender Rehabilitation Act 2014. This introduced post-release licences for those, approximately 50,000, prisoners subject to prison sentences of less than 12 months.
Furthermore, the data for the two periods differed in a key aspect; the risk of harm (RoH) levels of the offenders on release from custody. In the earlier period (2005–2014), the greatest proportion of SFO perpetrators involved those assessed as lower RoH on release from custody (low/medium: 73%). Later (2015–2024), the greatest proportion of SFO perpetrators involved those assessed as high RoH on release from custody (high/very high: 62%). The reduction in the proportion of low and medium RoH cases may reflect the policy change (cited in HMIP, 2020a) where internal reviews were no longer triggered only by the severity of further offence, but the case also had to have been at high RoH during the sentence. The ongoing high prevalence of low/medium RoH offenders and associated mis-identification of current risk is an issue returned to below.
In both time periods, the most prevalent offender groups were those with prior convictions in assault and/or battery, closely followed by those with theft, burglary, robbery and GBH. Similarly, the most common index offences both pre- and post-2015 were burglary and assault. The least prevalent groups were the generally more targeted offences including criminal damage, arson and murder/attempted murder. Arguably, therefore, there is a sustained pattern across the time period of SFOs emerging from cases with a history of versatility in offending (violent and acquisitive). This is concordant with the Ministry of Justice (2023) findings for reoffending; that the highest reoffending rate is among those with index offences for theft and public order offences.
The majority of reviews (40%) from across time indicated that SFOs took place during the first 6 months following release from custody, especially among offenders on life licence. SFO offenders were also primarily (65% reviews) those who had served fewer than 5 years in custody. This should have led to enhanced supervision of those recently released from custody, especially those who had served less than 5 years; many of whom were ‘low RoH’ offenders at the time of the SFO.
Across time periods, a disproportionate number of SFOs took place in the neighbourhood into which the offender had been released from custody or where they were housed. In just over one-half (57%) of the cases the SFOs took place in the same location as the prior conviction, often related to being released back to their previous address. The exceptions to this included where the offender had been moved to a resettlement prison in a different region and was therefore released into the regional probation service and provided with accommodation in that area, as opposed to being released back to their previous address/area.
This is concordant with a further finding that the victim was often (58% reviews) a neighbour or person that the offender lived with, especially for forensic mental health patients. The majority of victims in the extracted cases were cohabiting with the offender; whether partners, mothers, friends, neighbours in the approved premises, or staff in inpatient units. Of these, multiple (23%) victims had been known survivors of domestic abuse by the offender. Conversely, 40% of victims were strangers who were not previously known to the offender, but many (27%) of these nevertheless took place in the offender's accommodation vicinity. This suggests that many SFOs are committed within the location of accommodation, and provides a potential implication for policy in safeguarding adults and not housing released offenders who have a history of high RoH with victims that they have previously domestically abused.
Finally, no doubt influenced by the majority (N = 26/37) being NHS patients, we found that 93% of the SFO offenders were either neurodivergent or suffered from one or more mental health disorders. Notably, approximately one-third (31%) suffered from schizophrenia / paranoid schizophrenia. Nevertheless, it is important to note that many of these cases were comorbid with substance abuse or other mental ill health. Similarly, nearly one-quarter (22%) of the offenders in this review were diagnosed with a personality disorder (PD). PD symptomatology varied slightly, leading to diagnoses including: antisocial PD; emotionally unstable PD; psychopathic PD; schizotypal PD; and dangerous and severe PD, a former personality diagnosis in those with a history of violent offending. These PDs made up a sizeable proportion (21%, 16/76) of the mental health disorders found in the offender cases in the present reviews.
Analysis of antecedent behaviours
Data extracted from ‘Antecedent Behaviour’ resulted in 309 references that could be categorised into three main themes: (a) missed behavioural risk factors; (b) poor management of clinical risk; and (c) organisational and management failures; each with subthemes explained below and illustrated in Figure 2. A notably large proportion (81%, 17/21) of identified factors were emulated in both the index offence and the subsequent SFO incident. This suggests that these factors may predict the SFO. Below we bring together these findings, with k referring to data references (data available from the authors on request).
Interrelation of thematic data.
Theme 1: Missed behavioural risk factors
This theme contained the largest number of subthemes (50%). These subthemes are offence-related behaviours that could have been identified and used to recognise an increased likelihood of reoffending. They include: assault (k = 15); threats to the lives of others (k = 12); inappropriate relationships or sexual advances (primarily with other vulnerable patients or staff) (k = 13); unhealthy and deteriorated relationships with family and friends (k = 7); failure to appreciate seriousness/impact of own behaviour (k = 7); and, gang activity (k = 5). The offender's decision to distance from society and prosocial groups and to frequent with antisocial peers was seemingly missed. In many instances, those who were assaulted and threatened (often the offender's parents) ended up as the victims of the SFO. There were missed opportunities for safeguarding following these assaults and threats to kill, and indeed these behaviours should have resulted in a risk of harm review. The failure to recognise the seriousness of one's own behaviour, e.g. the impact of the index offence or of a harm caused in a clinical setting or under community supervision, at best demonstrates limited insight; at worst an anti-social orientation.
A subtheme at the nexus of Themes 1 and 3 (i.e. both a ‘missed behavioural risk factor’ and an ‘organisational and management failure’) is the lack of engagement with services (k = 19). Many offenders had instances of missing probation appointments or not presenting for drug testing. This could be monitored more closely as, again, it may portray a lack of respect for the terms of one's licence and antisocial orientation.
Theme 2: Poor management of clinical risk
Themes that represented poor clinical management of offenders included lack of appropriate psychiatric assessment (k = 15) – primarily incorrect or frequently changing diagnoses leading to inappropriate care plans and medications; and, lack of communication between services (k = 18). Communication deficiencies were either directly, or, through insufficient completion of case records such that other services (e.g. probation, GP) did not have access to crucial details regarding past forensic and clinical histories. For example, in the McSweeney case (HM Inspectorate of Probation, 2023c), probation had no access to custodial behavioural records for the 1 year prior to release, making it difficult to complete a release risk assessment. This lack of access to records in some cases may have led to premature releases. There was also poor management of non-compliance with medication (k = 11) and of mental health events (k = 14). For the case to reach a mental health crisis point may suggest that the circumstances were not being sufficiently monitored and responded to.
Two substantial subthemes found at the nexus of Themes 2 and 1 (i.e. ‘poor management of clinical risk’ and ‘missed behavioural risk factors’) are aggression noted by professionals (k = 25), and non-compliance with licence conditions (k = 27). These two sub-themes highlight a lack of respect for authority: behaviour that was highlighted multiple times throughout the reviews, suggesting that the offenders had antisocial attitudes, and were not taking their circumstances seriously enough, nor appreciating the consequences of their actions sufficiently.
Theme 3: Organisational and management failures
The most prevalent subtheme identified under this theme was ‘issues with accommodation and homelessness’ (k = 18). References to this were found in both the index offence circumstances and the SFO. These were issues either with the location of the accommodation (e.g. falling within an exclusion zone per the licence conditions), accommodation arrangements not being made or being changed last minute, or issues arising at the accommodation (e.g. with other members of the probation premises). There were also missed opportunities to recall offenders (k = 8), which was of major importance in some cases as the SFO was committed while the offender should have been in prison following recall, but investigative delays by the police resulted in failure to recall.
There were also issues with probation officers (k = 6) and plans not being carried out (k = 3), though these were not seen in either the index offence or the subsequent SFO. That is, they were seen by the inspection/review team as issues in the cases but were not directly linked to the prior and subsequent offences. Issues with probation officers included: having excessively high caseloads and therefore not sufficiently monitoring the offender; changing probation officers regularly without sufficient handovers (usually due to the high caseloads); and, assigning inexperienced and newly qualified probation officers to high risk of harm offenders.
Two subthemes were identified at the nexus of Themes 3 and 2 (i.e. both ‘organisational and management failures’ and ‘poor management of clinical risk’). First, the subtheme with the largest representation (k = 38) regarded the risk of harm level. Primarily this included missed opportunities to update the risk assessment, inadequately completed risk of harm assessments, underestimation of risk of harm due to lack of information, and, lack of identification of risk of harm to previous victims of domestic abuse – even when the offender was going back to live with them. Only seen in the SFOs (not the index offence as well), there was a conspicuous incidence of a lack of appropriate care planning and supervision (k = 22). This was the post-release supervision plan, or care plan for health care patients.
The intersection of all three themes
Finally, two subthemes that fell under all three themes were: presentation at accident and emergency (A&E) in crisis (k = 5); and, response to evidence of drug and/or alcohol misuse (k = 21). Not only were these behavioural risk factors worthy of identification and close monitoring, but arguably they warrant better management. In some cases, offenders had presented at A&E following altercations where they had been stabbed and no action was taken. In other cases, mental health patients presented at A&E demanding hospital admission as they feared they may be a danger to themselves or others but were turned away due to a lack of beds. Generally also, there was inadequate enforcement of licences when offenders used drugs/alcohol in breach of the terms of their licence, with mental health patients smuggling in contraband to secure hospital facilities, and those on probation regularly drinking in pubs or heavily at probation-approved premises.
Notably, there was a significant overlap between references to behaviours seen in the index offence and behaviours in the subsequent SFO. As shown in Figure 3, the majority (62%) of the references seen in the SFO were also seen in the index offence. Furthermore, a staggering 93% of references to behaviour that was seen in the index offence also referenced behaviour that was seen in the ensuing SFO. This demonstrates a significant pattern in behaviour; that in the lead up to the SFO the offender repeated the same behaviours that led to the index offence. With careful monitoring, these behaviours could have been noted and measures put in place to manage the risk.
Diagram to show the overlap between references (k) in the antecedent data relating to the Index Offence (IO) and the serious further offence (SFO). Note. The intersection shows the extent of overlap, with k = 115 (93%) of 123 references corresponding to the index offence also corresponding to the serious further offence, and k = 115 (62%) of 186 references found in the serious further offence also corresponding to references in the index offence.
Certainty of evidence and risk of bias within reviews
Review findings may be consistent, contested, or mixed; with consistent evidence indicated where a range of different forms of evidence point to similar or identical conclusions (Barends et al., 2020). The evidence from the present reports was highly consistent, based on rich data collected by independent inquiry teams using very similar review templates and in comparable contexts. All of the reports addressed contexts directly relevant to the clinical question. This indicates the high efficacy of the results, where there is consistent evidence to show the observed effects.
Discussion
This article contributes by providing an integrative review of independent inquiry reports of serious further offences (SFOs) by people on probation. Surprisingly, there has been a lack of attention to learning from these serious incidents, with no single ‘learning the lessons’ report since National Probation Service (NPS) (2005). The present review's findings, explored below in the context of the literature, have expanded on those from that report. We also identify the present review's limitations and implications for policy and practice.
The NPS (2005) report called for more timely and complete risk of harm (RoH) assessments and better RoH procedures, especially for cases transferred between probation officers. This was not ameliorated across the present reviews with ongoing issues in the management of high RoH offenders. Risk assessment tools including those underpinning HMPPS's offender assessment system (OASys) have strong reliability and predict the risk of reoffending with approximately 70% accuracy (Yang et al., 2010). However, the reviews consistently noted issues such as missed opportunities to review/update the OASys, inadequately completed RoH sections, and underestimated RoH due to lack of information. While some of these errors may be attributable to problems with probation workload burden (Philips et al., 2016), it is also possible that high RoH offenders fall within the 30% of offenders that are not accurately predicted via the tools or are not adequately represented in the tool's calibration samples. Either way, monitoring behavioural factors such as those identified in this report may be helpful both in forecasting the risk presented and managing corresponding risk scenarios.
The NPS (2005) review also called for better awareness of the significance of domestic abuse in offenders at or previously at high RoH. In the current review survivors who had previously been victims of threats to kill, assault or domestic abuse by the offender were most commonly the victims of the later SFO. There was a link with the proximity of the victim to the offender in terms of accommodation. Spouses, parents and neighbours were the most frequent victims of SFOs if the offender was housed post-custody at their previous family/spousal address. In general, the location of post-release accommodation was most commonly the location of the SFO.
Also ongoing since the previous report, in a high proportion of SFO cases the offender breached the terms of their licence: whether through aggression, unchanged negative attitudes, refusing to engage with services, or not complying with the terms of their supervision including illicit use of alcohol or drugs. This was often inadequately managed by services with missed opportunities to recall and review the assigned RoH level.
These findings implicate the ‘central eight’ risk factors for criminal behaviour (Andrews and Bonta, 2010). The central eight includes: a history of antisocial behaviour; antisocial associates; antisocial attitudes; antisocial personality pattern; educational and/or employment issues; family or marital problems; substance abuse; and, leisure/recreation. The first four of these are described as the ‘Big Four’ risk factors for reoffending. In a recent meta-analysis these antisocial potential factors were found to be most predictive of post-release recidivism (Goodley et al., 2022). The dynamic risk factors among the central eight are considered treatment targets for those with an antisocial personality pattern (Andrews and Bonta, 2010) and positive changes, such as stabilised family/living arrangements, have been linked to reduced risk of harm (Barnes-Lee and Campbell, 2020; Eisenberg et al., 2019; Hilterman et al., 2016; Wooditch et al., 2014). Antagonistic behaviour and weak self-control manifesting in accommodation scenarios may indicate ongoing needs in managing anti-social personality. Given the superior reliability and validity of static behavioural markers in predicting post-release recidivism, Goodley et al. (2022) highlight the importance to risk management of monitoring changes in current behaviour related to indicators of antisocial potential.
There were multiple shortcomings in inter-agency communication highlighted in the present research, supporting the NPS (2005) review. This included poor communication of changes in psychiatric diagnosis and lack of access by community probation to information on the offender's behaviour in custody. The latter is notable given that in Goodley and Pearson (2024) behavioural warnings in the 6 months prior to full release were a risk factor for recall to higher security conditions, and nearly all of those reoffending in the community had received a behavioural warning in the 6 months prior to their release. Certainly, to ensure reliable information availability/sharing, there is a need for a rigorous flagging system for low RoH cases with a history of high RoH categorisation. Relatedly, in the present research SFOs were commonly associated with prior convictions for theft, burglary, assault/battery, robbery, or GBH. Offenders with convictions for these mid-tariff offences may not be readily seen as high RoH cases and therefore may receive lighter supervision following release from custody. Without sound inter-agency information sharing and in a context of inexperienced staff, high workloads, and weak management oversight, there is increased vulnerability to hazards breaching the defences or ‘slipping through the holes’ (cf. Reason, 2000).
Our finding that the majority of cases had mental health difficulties is not surprising given research indicating that 90% of prisoners have either substance abuse issues or at least one mental health problem (Durcan, 2023). The provision of prison mental health support notwithstanding, only one-in-seven of these prisoners receives psychological support for these issues. Additionally, there is no screening at reception by a mental health professional, meaning that individuals’ needs are missed at prison induction. In cases such as that of Jordan McSweeney, this proved to be critical as the lack of recognition of his ADHD was cited as a factor that may have contributed to the SFO (HMIP, 2023c). The report (p.10) states that ‘little analysis was undertaken of how this affected his day-to-day cognitive functioning and learning styles, and if there were links with offending behaviour’. Although most critically McSweeney's enduring pattern of violence associated with negative family dynamics was unrecognised and unsafeguarded, sufficient analysis of his mental health needs may have led to adaptations to his licence terms to support him and his supervisory team in preventing his SFO.
Conclusion, limitations, and implications
The current review highlights multiple failures that are arguably preventable with greater attention to offence-related behaviour and its implications for a more inquisitive approach to risk management. The similarity between the current findings and those of the previous ‘learning points derived from serious further offence full reviews’ report (NPS, 2005) highlights the lack of improvement and the persistence of shortcomings highlighted 20 years ago.
Specifically, we found similarities between the antecedents to the index offence and to the serious further incident; similarities that represented missed opportunities for action-related monitoring. Monitoring the correspondence between past and present behaviour may help combat the ‘optimism bias’ the inspectorate has observed within services (HMIP, 2023a). Preferably, risk assessment should draw on an accumulation of individualised evidence based on a behavioural monitoring approach that actively monitors those with a history of high RoH regarding any behaviour related to the index offence or seen in the lead up to the index offence (cf. Pearson and McDougall, 2017). The current review identifies particular concerns relating to: withdrawal from prosocial groups; substance misuse; or mental state including aggression noted by professional services; lack of compliance with licence terms; and, assaults or threats to life. These behaviours should result in an immediate review of risk and, where appropriate, removal from the spousal/family home.
Limitations
The present research was limited by the lack of complete data for all fields. This was due to various factors. The templates for the reviews were different depending on the region's governing body, meaning that some reviews included data that others omitted. To maintain anonymity the NHS reviews also omitted information that could identify the offender. Data from other inaccessible documents (e.g. OASys reports) were referred to but not quoted leading to an irregular dataset (accessible OASys reports N = 12/36).
Another limitation of the NHS England reviews (n = 26) was that their focus pursued the quality of the care and treatment provided to the patient, from a healthcare rather than criminal justice perspective. This means that the data overlooked or under-emphasised forensic assessments and prior/index offences and focussed instead on, for example, NHS Trusts’ provision of healthcare, making it hard to contrast NHS and HM inspectorate cases. Despite this, the findings were strikingly consistent.
A third limitation concerns the mental health data. In many cases, even after a full psychological assessment, the mental health teams struggled to diagnose offenders with one single disorder, often oscillating between different personality disorders and paranoid schizophrenia. Diagnoses seemed to change with each assessment. In the current review, the most recent diagnoses were extracted for analysis, though the inconsistency in this reduces the reliability of these data.
Finally, it is important to note that during the latter period (2015–2024) there were two disruptive probationary restructures. In 2014, under the government's ‘Transforming Rehabilitation’ (TR) reforms (MoJ, 2013), the probation service was fractured along risk lines, with lower-risk offenders being managed by private and non-profit organisations, and high-risk offenders remaining under the supervision of the national probation service (NPS). This involved the management of 50,000 more released prisoners and a greater concentration of high-risk cases among the NPS staff. There was further upheaval when the probation service was reunified in 2021 following the notable failing of TR. This disruption undoubtedly affected organisational and management factors, and surely impacted negatively on the quality of supervision of high-risk offenders. However, it is said that in probation indeterminate change is the norm (Robinson and Burnett, 2007) so probation must strive towards a practice that is robust to organisational upheaval.
Implications for policy and practice
This research has highlighted the importance of making timely and appropriate accommodation arrangements and devising supervision plans that account for previous behaviours and forensic history. Given that many of the current SFOs were committed within the location of release accommodation, this has implications for policy in safeguarding the constituents of the provided accommodations and not housing released offenders with victims that they have previously domestically abused. In many instances, those who were assaulted and threatened (often the offender's parents) ended up as the victims of the SFO. These assaults and threats to kill clearly must be taken more seriously, trigger a risk review and recall as appropriate, including updated victim safety planning.
Research suggests that good family relationships and access to a prosocial community of adults may significantly reduce the likelihood of reoffending and can dramatically change attitudes to authority (Barnes-Lee and Campbell, 2020). Therefore, housing with family where there is no previous evidence of domestic violence or abuse should be encouraged. Withdrawal from society and ruptures in prosocial groups such as family should also be sufficiently monitored by probation services. Beneficial future research would investigate what conditions are appropriate for effectively housing domestic abuse perpetrators back in the family/spousal home and the effectiveness of approved premises (probation hostels) as post-custody accommodation in preventing recidivism. This could be a systematic review of the many research outputs on this topic, bringing together their findings to inform policy.
As well as ruptures in relationships with family and friends, research and practice should examine the following as risk factors for SFOs: assault on others; threats to the life of others; inappropriate relationships or sexual advances; failure to recognise one's own risk of harm; and gang activity. These behaviours should be especially noted in the 6 months prior to release, and the transitional subsequent first 6 months. Supporting previous recommendations relating to long-term prisoners (Goodley and Pearson, 2024) behavioural monitoring should be a basis for updating assessments and triggering wraparound support – with recall action if the offender is not responsive to the intervention. Evidently, particular notice should be paid to those with a history of high RoH who have recently been released from custody having served a sentence of less than 5 years.
Most notably, the current review found that behaviour that was seen in the index offence, also referenced behaviour that was seen in the subsequent serious further incident. This demonstrates a significant pattern in behaviour; i.e. that the offender was likely to repeat in the lead up to the serious incident the same behaviours that led to the index offence. With a systematic behavioural monitoring approach and careful management, these behaviours could be noticed and measures put in place to help prevent a serious further offence. A natural recommendation therefore is for probation practitioners to be trained regarding the significance of behavioural information, both in the index offence and institutional offence behaviour, as a means of predicting and managing changes in the risk presented.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
