Nathan 2019.

Study characteristics
Methods	Design: multisite, parallel, randomised control trial
Participants	Participants: male prisoners with personality disorder (whole sample, see notea) Sex: all male Age: (see notea) intervention group ('Resettle') mean age = 35.8 years (SD = 11.3); control group (treatment‐as‐usual; TAU) mean age = 32.6 years (SD = 11.6); total sample mean age = 34.3 years (SD = 11.5) Unit of allocation: individual participant Number randomised: (see notea) intervention (n = 38); control (n = 34) Number completing: (see note a) intervention group (Resettle: primary outcome ‐ intention‐to‐treat (ITT), n = 28; secondary outcome at y1, n = 25; secondary outcome at y2, n = 12); control group (TAU: primary outcome ITT, n = 29; secondary outcome at y1, n = 24; secondary outcome at y2, n = 11) Setting: prison and community in the North West of England Inclusion criteria: male prisoners over 18 years of age, likely to have personality disorder and identified as 'high risk' and in need of multi‐agency risk management arrangements (MAPPA) Exclusion criteria: severe intellectual impairment or psychotic mental illness identified from a review of the records or from the initial baseline assessment Ethnicity: (see note a) intervention (Resettle) group (white British ‐ 34 (89.5%); white Irish ‐ 1 (2.6%); white & black Caribbean ‐ 0; African ‐ 1 (2.6%); other mixed backgrounds ‐ 2 (5.3)); control (TAU) group (white British ‐ 33 (97.1%); white Irish ‐ 0; white & black Caribbean ‐ 1 (2.9%); African ‐ 0; other mixed backgrounds ‐ 0) Baseline characteristics: (see note a) Intervention (Resettle) group Age at first conviction: 'before age 15': n = 13 (34.2%); 'age 15‐17': n = 11 (28.9%); 'age 18 +': n = 14 (36.8%) Number of previous convictions: 13.3 (SD = 9.6) Number of previously convicted offences: 32.5 (SD = 25.4) Index offence: violent (n = 25, 65.8%), sexual (n = 10, 26.3%), burglary (n = 1, 2.6%), robbery (n = 7, 18.4%), other (n = 7, 18.4%) Psychopathy Checklist‐Screening Version (PCL‐SV): total mean score = 16.2 (SD = 4.6), facet 1 mean = 7.1 (SD = 3.0), facet 2 mean = 9.0 (SD = 3.3) Number of days from release to follow‐up: mean = 882.5 (SD = 187.1) SCL‐90 Global Severity Index: mean score = 0.85 (SD = 0.66) DSM‐IV personality disorders definite diagnosis: paranoid (n = 5, 12.2%); schizoid (n = 0); schizotypal (n = 0); antisocial (n = 34, 89.5%); borderline (n = 9, 23.7%); narcissistic (n = 1, 2.6%); avoidant (n = 1, 2.6%); obsessive compulsive (n = 0); not otherwise specified (n = 3, 7.9%) DSM‐IV personality disorders probable diagnosis: paranoid (n = 3, 7.9%); schizoid (n = 3, 7.9%); schizotypal (n = 0); antisocial (n = 2, 5.3%); borderline (n = 7, 18.4%); narcissistic (n = 0); avoidant (n = 3, 7.9%); obsessive compulsive (n = 1, 2.6%); not otherwise specified (n = 0) Control (TAU) group Age at first conviction: 'before age 15': n = 13 (38.2%); 'age 15‐17': n = 12 (35.5%); 'age 18 +': n = 9 (26.5%) Number of previous convictions: n = 14.2 (SD = 10.9) Number of previously convicted offences: n = 30.2 (SD = 27.2) Index offence: violent (n = 25, 73.5%), sexual (n = 5, 14.7%), burglary (n = 3, 8.8%), robbery (n = 7, 20.6%), other (n = 5, 14.7%) Psychopathy Checklist‐Screening Version (PCL‐SV): total mean score = 15.7 (SD = 4.5), facet 1 mean = 6.5 (SD = 3.5), facet 2 mean = 9.2 (SD = 2.4) Number of days from release to follow‐up: mean = 832.6 (SD = 144.1) SCL‐90 Global Severity Index: mean score = 0.82 (SD = 0.78) DSM‐IV personality disorders definite diagnosis: paranoid (n = 1, 2.9%); schizoid (n = 0); schizotypal (n = 1, 2.9%); antisocial (n = 31, 91.2%); borderline (n = 4, 11.8%); narcissistic (n = 2, 5.9%); avoidant (n = 1, 2.9%); obsessive compulsive (n = 2, 5.9%); not otherwise specified (n = 3, 8.8%) DSM‐IV personality disorders probable diagnosis: paranoid (n = 1, 2.9%); schizoid (n = 2, 5.9%); schizotypal (n = 0); antisocial (n = 2, 5.9%); borderline (n = 5, 14.7%); narcissistic (n = 0); avoidant (n = 1, 2.9%); obsessive compulsive (n = 1, 2.9%); not otherwise specified (n = 3, 8.8%)
Interventions	Two conditions: Resettle programme; TAU Intervention (n = 34 definite AsPD participants randomised): Resettle programme Control group (n = 31 definite AsPD participants randomised): TAU Details of conditions: Resettle intervention programme is an individual and group‐based psychosocial intervention consisting of 3 levels; (i) the therapeutic milieu generated by appropriate and prosocial relationships with a focus on enhancing social learning within a safe and boundaried environment; (ii) regular group work aimed at developing enhanced capacities for self‐reflection and understanding of others; and (iii) individually‐tailored psychosocial interventions, with a focus on risk management, well‐being and social integration. All Resettle participants are subject to detailed case formulations which form the basis of individual risk management and intervention plans. Participants initially attend the programme 6 days per week and there are 2 key‐worker sessions a week. Following a period of familiarisation, less frequent weekly attendance is negotiated on the basis of individualised assessments of need and risk. The control condition is TAU; standard probation supervision following release from prison. TAU comprises regular meetings (weekly initially) with the offender manager and engagement with other services where specified in the licence conditions. Duration of intervention: at least 2 years 6 months (6 months prior to release, then 2 years after release) Duration of trial: variable but approximately 30‐36 months Length of follow‐up: 2 years following discharge from prison; outcomes measured at 1 year and 2 years after release
Outcomes	Primary outcomes Recidivism: number and type of officially recorded offending according to the Police National Computer (PNC) (data were obtained for every offence recorded on the PNC between the point of initial release until the completion of the study) (see notec) Recidivism: non‐convicted offences identified by self‐report or incident reporting of antisocial behaviour using the Self‐Reported Delinquency (SRD) scaleb over the previous year (see notec) Adverse events: death (reported incidentally) Secondary outcomes Leaving the study early: number of participants not included in ITT analysis of primary outcome (reported incidentally) Other outcomes None
Notes	aIn the intervention group (n = 38), 34 participants had a definite AsPD diagnosis and 2 had a probable AsPD diagnosis. In the control group (n = 34), 31 participants had a definite AsPD diagnosis and 2 had a probable AsPD diagnosis. Reported participant demographic data are for the whole intervention group (n = 38) and control group (n = 34) bThe SRD scale is a 32‐item self‐report measure that asks respondents to indicate the frequency with which they have engaged in a wide range of antisocial behaviours (from theft to sexual or violent offending) over the previous year (Huizinga 1986), amended for use in an adult group (Palmer 2000). cRaw study data was provided by the study authors, allowing data extraction for the primary outcome to be undertaken for a 100% AsPD subsample
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomisation was carried out by an administrator unconnected to the study using random numbers generated by the study statistician. The Minim stratified randomisation programme was utilized to minimise the imbalance between the two groups for the type of index offence (violent versus sexual offence), SCID I diagnosis of drug and alcohol abuse (presence versus absence), and the designated probation office." (p 3, column 1)
Allocation concealment (selection bias)	Low risk	Quote: "The administrator informed the researcher, who had undertaken the baseline assessment, of the group allocation. In turn, the researcher informed the offender manager of allocation." (p 3, column 1)
Blinding (performance bias and detection bias) of participants	Unclear risk	Comment: Due to the nature of the intervention, it is not possible to blind participants to their allocation; the impact of this on risk of bias is unclear. In the intervention arm: "Participants initially attended Resettle for 6 days each week. In this phase, there were two key‐worker sessions a week...." (quote, p 3, column 1) In the control arm: "Usual care involved standard probation supervision following release from prison custody. This entailed regular meetings (weekly initially) with the offender manager and engagement with other services where specified in the licence conditions. Although the offender manager may have visited the participant in prison prior to release and if he was returned, this was very limited contact in comparison to the contact between the Resettle practitioners and the participants randomised to the intervention group." (quote, p 3, column 1)
Blinding (performance bias and detection bias) of personnel	High risk	Comment: This is an open‐label study: the offender manager was aware of group allocation. Resettle practitioners only worked with the participants randomised to the intervention. Quote: "The administrator informed the researcher, who had undertaken the baseline assessment, of the group allocation. In turn, the researcher informed the offender manager of allocation. For control group allocation, the offender manager made usual arrangements. In the event of allocation to the intervention group, there was liaison between the offender manager and the Resettle service." (p 3, column 1)
Blinding (performance bias and detection bias) of outcome assessors	High risk	Quote: "Follow‐up was not conducted blindly because assignment to the treatment and control groups was evident from the contact process." (p 4)
Incomplete outcome data (attrition bias) All outcomes	High risk	Comment: The attrition rate was high for the intervention and control groups. ITT was utilised in the analysis but numbers were still smaller than would be expected in the ITT and reasons for the missing numbers were unclear. E.g. Resettle: n = 38 randomised, ITT primary outcome reported as n = 28; TAU: n = 34 randomised, ITT primary outcome reported as n = 29
Selective reporting (reporting bias)	High risk	Comment: primary outcome measures reported in protocol as of 13 June 2013 were significantly different from those reported in the paper. The trial register indicated that several outcome measures were administered (e.g. Inventory of Interpersonal Problems (IIP); Barratt Impulsivity Scale (BIS)) but the study only reported data for reoffending and self‐reported antisocial behaviour. Protocol primary outcomes of 13 June 2013: 1. Reoffending data (Records from Police National Computer (PNC); Probation records) Nathan 2019 primary outcome: "The primary outcome was number and type of officially recorded offending according to the Police National Computer (PNC)." (quote, p 3‐4). Secondary outcomes reported in the protocol as of 13 June 2013: 1. Antisocial behaviour (Self‐Report Delinquency (SRD) scale). 2. Personality functioning (Inventory of Interpersonal Problems (IIP); Barratt Impulsivity Scale (BIS)). 3. Psychiatric illness/symptom and substance abuse/dependency (Structured Clinical Interview for DSM‐IV (SCID I); Symptom Check List‐90‐Revised (SCL‐90‐R)) Nathan (2019) secondary outcome: "The secondary outcome measure was self‐reported antisocial behaviour. This was recorded at 1 and 2 year follow‐up assessments using the Self‐reported Delinquency Scale (SRD) (Huizinga & Elliott, 1986) amended for use in an adult group (Palmer & Hollin, 2000)." (quote, p 4 column 1)
Other bias	High risk	Comment: Allegiance bias: high risk of bias. The Resettle programme (formerly known as the Community Risk Assessment and Case Management Service ‘CRACMS’) was jointly funded by the Ministry of Justice and the Department of Health (England and Wales) as part of the Dangerous and Severe Personality Disorder (DSPD) services. Author V Baker is described as ’Associate Director, Resettle Project, Speke, Liverpool, UK’ in the Miller 2010 publication. This allegiance to the Resettle programme was not declared in Nathan 2019. Vested interest bias (funding and/or author affiliations): high risk of bias. This work was supported by the 'Dangerous and Severe Personality Disorder’ (DSPD) programme established by the UK Home Office and the Department of Health (DoH) 1999. The Resettle programme (formerly known as the Community Risk Assessment and Case Management Service) was jointly funded by the Ministry of Justice and the DoH as part of the DSPD services. The intervention was therefore developed by DoH, and this study was funded in part by the DoH. Publication bias: unclear risk of bias. Though the study was published in a peer reviewed journal, there has been a long period of time between study commencement (2008) and publication of results (2019).

A + E = accident and emergency
AIDS = acquired immune deficiency syndrome
APQ = Antisocial Personality Questionnaire
ASI = Addiction Severity Index
AsPD = antisocial personality disorder
AUDIT = Alcohol Use Disorders Identification Test
BCSS = Brief Core Schema Scales
BDI = Beck Depression Inventory
BIS = Barratt Impulsivity Scale
BPAQ‐SF = Buss‐Perry Aggression Questionnaire‐12‐Item Short‐Form
BPD = borderline personality disorder
BPRS = Brief Psychiatric Rating Scale
BSI = Brief Symptom Inventory
CBT = cognitive behavioural therapy
CIRCLE = Chart of Interpersonal Reactions in Close Living Environments
CM = contingency management
CORE‐OM = Clinical Outcomes in Routine Evaluation‐Outcome Measure
CRACMS = community risk assessment and case management service
CSRI = Client Service Receipt Inventory
DAST = Drug and Alcohol Screening Test
DBT = dialectical behaviour therapy
DC = drug counseling
DES = Dissociative Experiences Scale
DFST = dual‐focus schema therapy
DoH = Department of Health (England and Wales)
DSM (III, IV) = Diagnostic and Statistical Manual of Mental Disorders (Third Edition, Fourth Edition)
DrinC = Drinker Inventory of Consequences 
DSPD = dangerous and severe personality disorder
DWI = driving while intoxicated
EQ‐5D = European Quality of Life‐5 dimensions
ESS = Experience of Shame Scale
EUROQOL = EuroQol Research Foundation
GAF = Global Assessment of Functioning
GCSE = General Certificate of Secondary Education
HADS = Hospital Anxiety and Depression Scale
HCR‐20 = Historical Clinical Risk Management‐20
HIV = human immunodeficiency virus
HTA = Health Technology Assessment
IBRS = Institutional Behaviour Rating Scale
ICC = intra‐class correlations
ICD‐10 = International Classification of Diseases‐Tenth Revision 
IIP = Inventory of Interpersonal Problems
ILC = impulsive lifestyle counselling
IOP = intensive outpatient programme
IPDE = International Personality Disorder Examination
IQ = intelligence quotient
IRP = individualised relapse prevention
ITT = intention‐to‐treat analysis
MACT= manual‐assisted cognitive behaviour therapy
MANSA = Manchester Short Assessment of Quality of Life
MAPPA = multi‐agency public protection arrangements
MCVSI = MacArthur Community Violence Screening Instrument
MINI = Mini International Neuropsychiatric Interview
MINIM = randomisation program for running minimisation in clinical trials
MOAS = Modified Overt Aggression Scale
n/a =  not applicable
NAS = Novaco Anger Scale
NCTU = Nottingham Clinical Trials Unit
NEP = needle exchange programme
NHS = National Health Service
OAS = Overt Aggression Scale
PAS‐Q = Quick Personality Assessment Schedule
PCL‐R = Psychopathy Checklist‐Revised
PCL‐SV = Psychopathy Checklist‐Screening Version
PD = personality disorder
PD‐NOS = personality disorder‐not otherwise specified
PDQ = Personality Diagnostic Questionnaire
PEPS = psychoeducation and problem‐solving
PNC = Police National Computer
PTSD = post‐traumatic stress disorder
QALY = quality‐adjusted life year
QOL = quality of life
REBT = rational emotional behaviour therapy
Resettle = programme of psychosocial interventions for high risk personality disordered offenders
SASII = Suicide Attempt Self‐Injury Interview
SBCM = strengths‐based case management
SCID = Structured Clinical Interview for DSM [Diagnostic and Statistical Manual of Mental Disorders]
SCL‐90‐R = Symptom Checklist 90
SD = standard deviation
SE = standard error
SFQ = Social Functioning Questionnaire
SIDP = Structured Interview for DSM‐IV Personality Disorders
SM = standard maintenance
SMI = schema mode inventory
SMT = schema modal therapy
SNAP = Schedule for Nonadaptive and Adaptive Personality 
SPSI‐R = Social Problem Solving Inventory‐Revised
SRASBM = Self‐Report of Aggression and Social Behaviour Measure
SRD = Self‐Reported Delinquency Scale
ST = schema therapy
START = Short‐Term Assessment of Risk and Treatability
STAXI = State‐Trait Anger Expression Inventory
SVR‐20 = Sexual Violence Risk‐20
TAU = treatment‐as‐usual
TBS = Terbeschikkingstelling
TER = Treatment Engagement Rating Scale for Forensic Outpatient Treatment
USD = United States Dollar
VRS = Violence Risk Scale
WAI = Working Alliance Inventory
YSQ = Young Schema Questionnaire‐Short Version
ZAN‐BPD = Zanarini Rating Scale for Borderline Personality Disorder