Absconsion from forensic psychiatric institutions: a review of the literature

Background: Absconding from mental health units is referred to as a patient leaving without permission and can have significant consequences for the patient, family, community, and institution. The varying definitions of absconsion involve breaching security of an inpatient unit, accessing grounds or community without permission, gaining liberty during escorted leave or being absent for longer than permitted from authorized or trial leave. While considerable literature exists on absconsion from acute psychiatric units, there is a paucity of literature specific to forensic absconsions, despite inherent differences between patients and systems. Forensic patients are offenders who are found unfit to stand trial, or not criminally responsible on account of mental disorder. The literature indicates the absconding rate within the forensic population is expected to be low, based on the fact that the level of security in forensic units is higher than general psychiatric units. Despite the rates being considered low, the outcomes of absconding in this population can potentially be serious, thus the exploration of factors surrounding these incidents is essential. Purpose: To review the literature regarding absconsion from forensic psychiatric institutions. This review will identify potential risk factors and motivations of forensic patients that have absconded. Methods: Electronic database and hand searches were conducted to locate articles pertaining to absconding specific to forensic psychiatric institutions published from 1969-present. Search terms included “abscond”, “escape”, “AWOL”, “runaway”, “psychiatric inpatient”, “forensic institution”, & variants. All full-text articles meeting inclusion & exclusion criteria were appraised for qualitative themes, limitations, and assessed for risk of bias using appropriate CASP Checklists. The review is structured following the PRISMA checklist and framework. Results: A total of 19 articles meeting literature review criteria were identified. The majority of the articles were of retrospective case-control design (n=12). Three systematic reviews were found on absconsion that included analyses from both forensic and general psychiatric populations. Definitions for absconding were omitted or varied making comparisons between studies difficult. Much research compared demographic, static and dynamic factors. History of previous absconsion, scores on validated risk-of-violence assessment tools, substance-use disorder, acute mental state, and socio-environmental factors were consistently noted as risk-factors. Four distinct motivations for absconding emerged: goaldirected, frustration/boredom, symptomatic, and accidental. Overall, the literature suggested forensic absconsion was a rare event of short duration with low risk to the public and few reoffending incidents. Conclusions: There is a paucity of literature on forensic absconsions. A consistent definition of absconsion and use of standardized reporting protocols across forensic programs would be beneficial in order to be able to compare data on absconding events. Also, prospective studies should be undertaken to better understand the motivations and dynamic risk factors of forensic patients who have absconded and would help inform a forensic absconsion risk assessment protocol.

Background: Absconding from mental health units is referred to as a patient leaving without permission and can have significant consequences for the patient, family, community, and institution. The varying definitions of absconsion involve breaching security of an inpatient unit, accessing grounds or community without permission, gaining liberty during escorted leave or being absent for longer than permitted from authorized or trial leave. While considerable literature exists on absconsion from acute psychiatric units, there is a paucity of literature specific to forensic absconsions, despite inherent differences between patients and systems. Forensic patients are offenders who are found unfit to stand trial, or not criminally responsible on account of mental disorder. The literature indicates the absconding rate within the forensic population is expected to be low, based on the fact that the level of security in forensic units is higher than general psychiatric units. Despite the rates being considered low, the outcomes of absconding in this population can potentially be serious, thus the exploration of factors surrounding these incidents is essential. Purpose: To review the literature regarding absconsion from forensic psychiatric institutions. This review will identify potential risk factors and motivations of forensic patients that have absconded. Methods: Electronic database and hand searches were conducted to locate articles pertaining to absconding specific to forensic psychiatric institutions published from 1969-present. Search terms included "abscond", "escape", "AWOL", "runaway", "psychiatric inpatient", "forensic institution", & variants. All

Introduction
Absconsion of patients from forensic mental health units can have significant consequences for the patient, hospital, and greater community [1][2][3][4][5][6][7][8][9][10][11]. Absconsion can be defined as an unauthorized leave of absence from mental health inpatient services. Within the literature, however there is no standard definition of teams realize the risk of absconsion exists when granting this leave [23]. However, lack of evidence on absconsion specific to the forensic population, makes clinical decision-making and identifying patients at high risk of forensic absconding more difficult. The purpose of this literature review is to evaluate the state of literature in relation to absconsions from forensic mental health units. It will aim to identify risk factors and motivations highlighted in the literature that can help inform decisionmaking in granting leave.

Methods
Searches of electronic databases were conducted to locate articles pertaining to absconding specific to forensic mental health patients published from 1969present. The following databases were searched: Ovid, MEDLINE, PsychINFO, CINAHL, PubMed and Web of Science. A hand search was also conducted of Google Scholar, the Web, and relevant publications' reference lists. The following search terms were used with variants in parentheses: "Abscond" (absconding, absconder, absconded), "Escape", "Elope", "AWOL", "AWOP", "Runaway" (at large), "Psychiatric Inpatient" (Patient, Resident, Absconder, Offender, Offender-Patient), "Forensic Institution" (Special Hospital, High-Security Psychiatric Hospital, Forensic Hospital/Ward, Secure Hospital). Included publications were in English. Initially, the search timeframe was set from 2007 to present with November 20th, 2018 being the last date searched. However, due to the low quantity of articles (n=10), the timeframe was adjusted to include relevant publications from 1969. The search excluded theses or other grey literature and only literature published in peer-reviewed journals was included. Full-text articles meeting criteria were appraised for qualitative themes, limitations, and assessed for bias using CASP Checklists [25]. The review was guided by the PRISMA checklist and Explanation and Elaboration document [26]. Figure 1 illustrates the different phases of the literature review. absconsion used. Definitions can include instances of breaching the security of an inpatient unit, accessing hospital grounds or the community without permission, fleeing from staff while on community outings, being absent for longer than permitted, or failure to return from an authorized or trial leave [1][2][3][4][5][6]8,9,[11][12][13][14]. Regardless of the circumstances surrounding the event, the potential for serious outcomes exists.
Absconsion from forensic mental health units can compromise a patient's safety and result in suicide or serious self-harm [1,5,8,15] with the risk of suicide being elevated immediately upon return of the patient to the unit [2,8,16]. Substance and alcohol use are also commonly reported during absconding with over 50% of absconders in one study reporting use on leave [2,4,[7][8][9]11,14]. There is risk of exposure to the environmental elements since many patients do not have a home to seek shelter [1,15]. Absconsions can lead to distrust in psychiatric services by families who expect the hospital to keep patients safe. The reputation of the forensic hospital and its processes can be put under scrutiny by a community expecting adequate management of patients who may pose a danger within a community or are at risk for violence [1][2][3][4][5][6][7][8][9][10][11]17]. In absconsions from forensic institutions, police are notified which may precipitate media attention, exacerbating both the situation and stigma [2,5,18,19].
Within forensic psychiatric institutions, patients are under much higher security due to potential risk for violence [1,13,17]. This level of security may decrease overall rates of absconsion, however also results in a lack of autonomy related to treatment, heightened perceived stigma, and isolation of patients from the hospital and community [1,13,[20][21][22]24]. Lengthy durations of stay in hospital in comparison to non-forensic patients is a unique source of stress, making forensic rehabilitation and motivations for absconsion more complex [4,6,9,11,[21][22][23][24].
However, for nearly all forensic patients some form of leave is a fundamental part of their rehabilitation [17,20,21,23]. Clinical
Three systematic reviews were found on the subject of absconsion; all included articles from both forensic and general psychiatric institutions [1,12,13]. Remaining studies involved qualitative observation of clinician decision making, and a description of an absconsion protocol [19,23].

Definitions of Absconsion
There was no consistent definition of absconding used within the literature on forensic mental health units (Table 1). particular subgroups of absconders: opportunity makers and opportunity takers [2,4]. Opportunity takers make up the majority of absconders, meaning those who make use of circumstances or chance events where impulsivity, fear of transfer, revoking of parole, amount of money in possession, and difficulties in accepting detention are associated with increased risk [2,4]. Opportunity makers, by contrast, engineer situations in which they can abscond, and absconsion was often goaloriented and planned, especially with those from higher security wards [2,[4][5][6].
Researchers identified four distinct motivations of absconding behavior: goaldirected, frustration/boredom, symptomatic/disorganized, and accidental [4,9,11]. Frustration/boredom and goaldirected behaviours account for the vast majority of incidents [4,11]. These patients often exhibited higher HCR-20 scores, difficult behaviors, absconding ideation and voicing of discontent to staff, and upcoming or recently unfavourable Review Board hearings in the weeks prior to absconding [4,9,11]. When individuals returned, it was at their leisure, they minimized the situation, externalized blame, or defended behavior [4]. Goal-directed individuals are those with a desire to abscond in order to complete a specific goal which they were likely to have voiced to staff [4,9]. Symptomatic/disorganized absconders appeared to act in response to auditory hallucinations or delusional beliefs [4,9,11]. Active symptoms of illness, notable instability, medication changes, missed medication and stating of psychotic beliefs with absconding ideation preceded incidents [4,9]. Accidental absconders were those who lost track of time or met situations beyond their control resulting in a report of absconsion [4,9,11].

Characteristics of Absconding Incidents
Incidents of absconding from secure hospitals were found to be rare and acts of violence during absconsion were infrequent [1][2][3][4]6,8,11,12]. Incidents occurring in lowsecurity units appeared to be higher compared to medium or maximum-security units [3,5,7,13]. Most absconding incidents occurred from individual or community trips outside the hospital [3][4][5]8,13]. The number Amongst studies that provided a definition of absconding, there were variations in time-periods a patient could be absent before being deemed an absconder [1,3,4,6,7,13]. In addition, most studies classified both failures to return from leave and escape as absconsion [6,10]. One study differentiated between escape and absconsion [7] and only one clearly differentiated between absconding, failure to return from leave, escape, and attempts [1,12,13]. Three studies did not include definitions at all [18,23,27].

Static Risk Factors
History of previous absconsion or attempts was found to be significantly associated with future absconding incidents in many studies [2,[4][5][6][7][8][9]14]. Those who abscond are likely to be young, Caucasian, male, and diagnosed with a psychotic disorder, compared to non-absconders [1,5,13,18]. A history of alcoholism, unemployment, and being an older sibling was characteristic of some absconders [10]. Some studies looked at scores on commonly used violence risk assessment tools. Higher risk scores on the HCR-20 were noted in three studies as being a good predictor of absconding behaviour [4,9,11]. Higher PCL-R scores of absconders and escapes were also associated with a higher risk of absconding [4,11,27].

Dynamic Risk Factors
Absconding was also found to be significantly associated with dynamic factors such as patients' acute mental state or recent stressful events such as death/loss or transitions in care (usually from lower to higher security) [4,11,14,25]. Furthermore, in the year prior to absconding, absconders were more likely to be involved in property damage, verbal aggression, self-harm, substance abuse and to be non-compliant with treatment [3][4][5]9]. Absconders were more likely to have a comorbid substance use disorder and problematic personality traits/disorder than personality disorder alone [9,11].

Motivations to Abscond
Only four studies to date have included explicit analysis of specific motivational factors [4,6,9,11]. Two commented on or resentful [7,10,28]. Because of conflicting data about timings of absconsion, focusing instead on motivations behind observed trends may be more useful in assessing risk than the static factor of length of stay [4,6,7,9,10,14,28].

Outcomes of Absconsion
Forensic absconsion is rare, with locked doors and high secure units correlated to the lowest rates, but with overall rates increasing [1][2][3][4][5][6][7][8]. Studies focused on the duration of absconding events and circumstances of the return, (see Table 1) and in most cases the duration was short, the majority of patients were returned within 24 hours, and all absconders were eventually found [1][2][3][4][5][6]8,11,12]. A recent study noted the median absconding duration of only 4 hours [11], and one noted most patients were caught within minutes by hospital staff [5]. Between suburban and urban hospital settings, it may be easier for absconders to evade detection and gain access to the wider community in a metropolitan city [4,11,26]. Police were involved in the patients' recapture usually without incident, and in some cases, the patients returned voluntarily, although this was less likely in escapees [5,7]. The ultimate level of public endangerment posed by those who absconded is low, with few patients reoffending during absence [2][3][4]. An Ontario forensic institution found that despite higher HRC-20 scores and substance-use disorders among the 57 patients responsible for 102 absconsions over 2 years, only one incident of minor violence and very few other illegal behaviours occurred [4]. Three English maximum-security hospitals had very low rates of absconding from thousands of rehabilitation outings over a five-year period, all with minimal risk to the public [2]. Huws et al., reported 11 cases of offending from 36 AWOPs over 13 years, and with only two serious offences (rape and manslaughter) [3]. Several studies noted negative outcomes for absconders: selfharm during leave, elevated risk of suicide upon return, and substance abuse during leave [1][2][3][4]8,15]. Beyond public safety risks, re-offense, and patient harm, other outcomes included slowed patient rehabilitation, the reputational risk to the hospital, affects on the legal status of the patient, and perpetuation of stigma towards patients [1][2][3][4][5][6][7][8][9][10][11]18].

Timing of Absconding
Some studies observed that most incidents occurred during warm weather from May to September and in holiday months such as December when patients wished to be close to loved ones [4,10]. One study noted absconsion took place primarily during the afternoon/evening, weekends or times with lower staff level [6]. Two studies indicated that absconsions tended to occur in clusters suggesting a "copycat" or "contagion" effect during times of ward stress [5,6]. There is disagreement over whether length of stay is related to absconsion risk. Three studies confirmed significant time lapses between the date of admission and first absconsion episode [4,9,28]. One study noted that the longer a patient has been known to staff and not absconded; the less likely staff may be to expect absconsion, increasing the risk [28]. This is inconsistent with previous research, which hypothesized no relationship, or that risk of absconsion is highest earlier in inpatient stay when patients are unsettled

B. Completed CASP case-control risk of bias checklist tool
Did the study address a clearly focused issue?
Appropriate method to answer their question?
Were cases recruited in an acceptable way?
Were controls selected in an acceptable way?
Was exposure accurately measured to minimise bias?
If the results of the review have been combined, was it reasonable to do so?
Can the results be applied to the local population?
Were all important outcomes considered?
Are the benefits worth the harms and costs?
ability to assess risk factors [5,28]. Attempts were made to minimize confounding factors and bias in some studies by having a control group, matched comparison group of non-absconders, or other comparison group, though appropriate matching criteria were difficult to discern and inconsistent across studies [2,3,[5][6][7][8]10,11,18,27,28]. The past 10 years have seen an increase in forensic absconsion literature, possibly attributable to negative media responses to increasing rates or recent high profile absconsions prompting interest in better risk management and increased pressure for understanding of outcome risks and motivations as mentioned in some studies [2][3][4].

Most
research conducted was retrospective in nature. Due to incomplete EMRs and biased clinician recall, many retrospective studies were operating with incomplete data and particularly limited in exploring dynamic risks [2,3,5,7]. Selection bias was evident when outcomes were not available for several absconders, or for attempts, especially considering history of absconding being a stable predictor [2,[4][5][6][7][8][9]14,18]. Limitations of retrospective analysis based on EMRs is further compounded by the fact that there exists no evidence of systematic documentation for absconding episodes across forensic institutions over time. While retrospective studies attempted to discern absconder profiles, characteristics, and motivations many noted these were based off incomplete records [2,3,[5][6][7][8]10,11,18,27,28]. Refinement of record-keeping of incidents may assist in the better extrapolation of risk factors [5,19]. A standard debriefing protocol for patients who have been returned could further inform motivations and strengthen therapeutic relationships [4,5,9].
Given that risks and factors have been established from retrospective studies, more prospective studies should be undertaken. These should incorporate a standardized definition of absconding from forensic institutions and conduct in-depth analysis of motives and characteristics on

Discussion
Within this review, the use of an a priori protocol, defined research question, objectives, and scope reduces bias in selection [29]. The major limitation in this review was the scarcity of data on absconsion within the forensic population, prompting date range adjustment to include as many relevant studies as possible. Inconsistent definitions of absconsion made it difficult to draw comparisons between studies, as well as inconsistent reporting of rates, comparison groups, and settings (see Table 1). CASP is seen generally seen as an appropriate tool for qualitative risk of bias assessment, however, it appears to be inferior to others in terms of sensitivity [25,30]. Studies were all found to be of qualitative value to include in a review of the literature, individual article limitations were examined, and all were found to be of low risk of bias as per CASP guidelines (see Table 4 A-D).
The deficiency in volume and quality of literature on the topic of absconsion from forensic institutions may reflect the rarity of such incidences, resulting in smaller sample sizes and difficulties in producing comparisons. The small sample sizes or discrepancy between sizes of control and absconding groups in individual studies affects the significance of results measured in case-control and cohort studies and the the HCR-20 and PCL-R that showed utility in predicting absconsion, so further analysis of these scales' absconsion predictivity would be valuable [4,9,12].The LARA is based on a review of violent risk assessment and absconsion literature, but the tool has not yet been tested for validity/reliability and is not specific to forensic settings [12]. The LARA could be refined for forensic use, although it would need to be tested prospectively [12]. Prospective analysis of an empiricallyderived Leave Application Form to assist clinical teams in leave decisions showed a 33% decrease in absconding following implementation in one population [4,9]. Future cross-referencing of this form's validity in different settings would aid in confirming its reliability and shows promise for evolving into a standardized forensic tool [9]. In addition to EMR review, interviewing patients about past absconsions to further discern motivational risks to be incorporated into a tool would be valuable [9,12]. Ultimately literature is starting to emerge but any proposed risk assessment tools are in very early stages.
Statistical risk prediction of absconsion is inherently problematic due to the low volume of forensic absconding events, and risk assessment based on static and demographic factors may not be appropriate [12,28]. While compared in nearly all retrospective chart reviews, demographic factors may be time and context-specific, or the offender patient population at one institution may be skewed depending on the type of surrounding community from which patients originate [12,28]. History of absconsion, recent PCL-R, and HCR-20 remaining steadily predictive across studies may be related to consistent motivations [1,4,6,9,11,27]. Clinical interventions may affect observed findings for factors such as definite psychopathy and record of physical violence being negatively associated with absconsion, as clinical teams may impose additional restrictions on those who have high-risk diagnoses or violent behaviour [13,14]. These patients may thus present as lower risk since they have fewer opportunities to abscond, rather than these larger forensic sample sizes to replicate the significance and clinical utility of risk factors previously identified in retrospective literature.
Additionally, literature reviews pooled both forensic and general psychiatric populations in their analyses [1,12,13]. It is imperative to study forensic populations separately; there are differences between forensic patients and general psychiatric patients and the systems through which they receive rehabilitative treatment, that impacts how we analyze absconsion from these settings and development of appropriate risk tools [16,23,29]. General psychiatric patients may be admitted voluntarily or involuntarily depending on clinical presentation, while criminal history and serious mental illness are inherent for admittance to a forensic psychiatric facility [12,[23][24][25]27]. The forensic patient also experiences the unique stresses associated with the forensic system, such as decreased autonomy, increased length of stay, Review Board hearings, revocation of liberty, and heightened security and stigma due to their criminal offenses [11,[22][23][24]27]. Ultimately these clear differences between the two settings that warrant separate investigations into factors associated with absconsion from these two distinct populations, and a review of forensic-specific literature is warranted [1,12,13].
One of the biggest gaps is in risk assessment. At present only one qualitative study has contributed to the understanding of clinician decision making in granting forensic leave however the researcher's own influence needed to be reviewed for observation bias and reporting bias being the sole observer [23]. These leave decisions tend to be unstructured, with absconding risk and current mental state rarely referenced explicitly, and time constraints predisposing conversations to brevity [23]. Despite attempts, overall, there still exists no "thorough, well designed, rigorously carried-out trials of interventions to reduce absconding" valid for forensic populations [9,12,15]. Two studies that were analyzed explored tools incorporating validated and widely used violence risk and psychopathy assessment scales such as potentially targeted with appropriate preventative interventions [4,7]. For instance, if substance abuse is both a motivator and risk factor for absconsion, an inpatient program for patients targeting substance use disorders could prove effective [4,7]. Although the most straight forward method of reducing absconding would seem to be to increase hospital wide security, locking ward doors and reducing leave, such measures do not stop absconding altogether and need to be balanced against potential negative consequences such as a volatile ward environment, depression associated with lack of freedom, and negative acute mental states [4,11,13,14,25].
Finally, several studies suggested ensuring adequate staffing levels and relational security, increasing positive relationships with the clinical team, and improving the education of staff themselves on risks and precipitators to absconsion in order to better detect their presence, would aid in creating a cohesive, transparent, rehabilitative environment for patients [4,6,9,23].

Conclusion
Ultimately, there is a deficit in forensicsetting literature and inconsistency in many areas of interest on absconsion. While a rare event of ultimately low risk to the public, there is a need to develop a consistent definition of absconsion and use standardized reporting for incidents and outcomes across forensic settings. Prospective studies with patient interviews centered on absconder motivations and acute mental state should be undertaken so as to rely less on static risk factors and retrospective data. Development of a validated forensic absconsion risk assessment tool based on evidence, and preventative measures targeting motivations will aid in reducing risk.
Conflict of Interest: none characteristics actually representing lowrisk.
Retrospective studies have thoroughly established static factors, and several acknowledge the potential weight of motivational factors, yet very little has been done to actually evaluate motivational factors as this data is difficult to obtain retrospectively [4,11,12,14,25]. Recent stressful life events and clinical events should be carefully monitored in patients in the context of their relationship to potential motivators since they may increase frustration, be indicative of boredom, precede goal-oriented behaviour, or aggravate symptoms of mental illness prompting accidental or symptomatic absconsion [3][4][5]9,11,12,14,25,27]. Clinicians acknowledge the time spent without leave on a ward is frustrating and contributes to a negative rehabilitative environment, boredom, and despair for patients facing long hospitalizations alongside the revocation of their autonomy compared to other psychiatric settings [1,4,6,7,9,11].
Substance use is unsurprisingly a predominant subtype of goal-directed motivation [4,7]. Future prospective analysis should also consider incorporating patient and staff interviews alongside medical record review following absconsion to better discern patient motives.
Behaviours that are strong predictors for absconsion in forensic settings could be