11 Forensic psychiatry11.3 Practising psychiatry in a secure setting
11.3.4 Forensic risk management planning
7 min read
Written by Resident Doctor Committee · Published
Predictors of reoffending
A documented history of violence or aggression is the strongest single predictor of future violence, and NICE (the National Institute for Health and Care Excellence) says risk assessment must always take it into account.
Relapse into substance misuse after release is a dynamic criminogenic risk factor that raises the likelihood of reoffending.
Deviant sexual interest and an antisocial or psychopathic orientation are the two strongest predictors of sexual reoffending.
Measured deviant sexual arousal to children, on plethysmography, is the single strongest predictor of sexual reoffending in the pooled follow-up evidence of Hanson and Bussiere, later Hanson and Morton-Bourgon.
Prior offending history is the strongest predictor of sexual reoffending, ahead of age, education, employment and relationship status.
Denial, mental illness and childhood abuse do not predict sexual reoffending.
A general lack of empathy for others is a psychopathic trait that predicts sexual reoffending, whereas an empathy deficit confined to the offender's own victims does not.
Actuarial and structured professional judgement approaches
Feature | Actuarial | Structured professional judgement (SPJ) |
|---|---|---|
Items | Fixed, mainly historical (static) | Static and changeable (dynamic) |
Scoring | A set rule; the total converts to a reference group's reoffending rate | No fixed total score or numerical probability |
Clinical judgement | None, with no override after coding | Ends in a formulation, risk scenarios and a management plan |
SPJ is preferred in UK practice because its dynamic factors support an individual formulation and a management plan that treatment and supervision can act on, not a bare group-based probability.
Instruments differ in scope: only some, such as the HCR-20, cover violence of any kind by adults, while others are limited to sexual offending, partner violence or young people.
Some actuarial tools are built and validated specifically for sexual reoffending, unlike tools for general violence, any reoffending, psychopathy or risk to self.
Psychopathy and inpatient screening
The Psychopathy Checklist-Revised (PCL-R) is the reference standard measure of psychopathy, with 20 personality traits and antisocial behaviours each scored 0 to 2 from interview plus collateral file review.
The PCL-R rates the construct rather than predicting risk directly; its score feeds actuarial tools, and high scorers are a narrower group than people with antisocial personality disorder.
The Broset Violence Checklist screens for imminent inpatient violence over about 24 hours from observed ward behaviour alone, with no history or interview; it is quick, needs minimal training and is repeated each shift.
Actuarial instruments
Instrument | Origin | Items | Predicts |
|---|---|---|---|
RRASOR (Rapid Risk Assessment for Sexual Offence Recidivism) | Hanson | Four static: prior sexual offences (heaviest weight, 0 to 3), age under 25 at release, any unrelated victim, any male victim | Sexual reoffending |
Static-99 | Merged RRASOR with Thornton's SACJ-Min, built for the English prison service | Ten static, scored from records without interview or psychopathy rating | Sexual reconviction in adult men |
Risk Matrix 2000 | Thornton, for UK police, prison and probation | A base score, then a second step of aggravating factors | Separate sexual, violent and combined scales |
VRAG (Violence Risk Appraisal Guide) | Harris, Rice and Quinsey, on men in a Canadian maximum-security hospital in Ontario | Twelve weighted static historical items | Violent reoffending, as a risk bin with an expected rate |
SORAG (Sex Offender Risk Appraisal Guide) | An adapted VRAG | Fourteen, adding sexual history and phallometric items | Violent reoffending of any kind in men convicted of sexual offences |
Static-99R changed only the age weights of Static-99, and Static-2002 has fourteen items.
One VRAG item is separation from a parent before age 16.
The original VRAG used DSM-III (Diagnostic and Statistical Manual, third edition) personality disorder and schizophrenia items. These are historical instrument definitions, not current diagnostic guidance; record clinical diagnoses using ICD-11 and follow the exact instrument manual when scoring.
The PCL-R total carries the largest weight in the VRAG and is also weighted in the SORAG, but in no other common risk instrument.
The HCR-20
The HCR-20 (Historical, Clinical, Risk Management-20) is an SPJ tool for adult violence with 20 items, leading to a formulation, risk scenarios, a management plan and a summary rating, not a summed score.
Scale | Items | What it rates |
|---|---|---|
Historical | 10 | Static lifetime factors, such as problems with employment |
Clinical | 5 | Recent factors that can change with treatment |
Risk management | 5 | Forward-looking changeable factors in the expected future setting: personal support, living situation, plans, treatment response, stress or coping |
The HCR-20 has no dedicated items on animals and property, suicide or protective factors.
Violence in the HCR-20 is unwanted harm to another person, so threats, attempts, fear-inducing conduct and illness-driven assaults count, but harm in a licensed consensual combat sport does not.
Other structured professional judgement instruments
Tool | Harm or group | Structure |
|---|---|---|
SARA (Spousal Assault Risk Assessment) | Violence to a current or former partner | 20 items, general violence factors plus partner-specific ones (escalating assaults, breaches of court or no-contact orders, attitudes excusing abuse); supports a safety plan for a named partner |
SAM (Stalking Assessment and Management) | Repeated unwanted pursuit of one person | 30 items under the nature of the stalking, the perpetrator's risks and the victim's vulnerability |
SAVRY (Structured Assessment of Violence Risk in Youth) | Ages about 12 to 18 | 24 risk items in historical, social or contextual, and individual or clinical domains, plus six protective factors |
SAPROF (Structured Assessment of Protective Factors for violence risk) | Protective factors | 17 items in internal, motivational and external groups, balancing an HCR-20 formulation, without scenario or management steps |
START (Short-Term Assessment of Risk and Treatability) | Repeated short-term review | 20 dynamic items, each rated as a strength and a vulnerability, over weeks to about three months |
SVR-20 (Sexual Violence Risk-20) | Sexual violence | The SPJ tool for sexual violence risk |
START covers violence, self-harm, suicide, self-neglect, unauthorised leave, substance use and victimisation.
ODARA (Ontario Domestic Assault Risk Assessment) is also partner-specific but, unlike SARA, is actuarial.
Security and outcome measures
The DUNDRUM-1 triage security scale anchors every item to a level of security and is used at admission.
The DUNDRUM-4 recovery scale is rated later, to judge readiness to move to a less secure setting.
HoNOS-secure adds security ratings to the twelve standard clinician-rated Health of the Nation Outcome Scales and is repeated to measure change, so it is an outcome measure, not a risk prediction.
Formulation and scenarios
A formulation joins historical patterns, stable characteristics, current symptoms, circumstances and protective factors such as supportive relationships, meaningful activity and self-control.
A scenario names who might be harmed, the sequence, triggers, likely timing, severity and recurrence, keeping severity separate from imminence.
Scenario | What it describes |
|---|---|
Repeat | The index behaviour happens again |
Escalation | The same pathway with greater severity, such as weapon access making an assault fatal |
Change in nature | A different type of harm, such as fire-setting, if history supports it |
Twist | A different victim or circumstances, such as a support worker becoming the focus |
The twist is the scenario clinicians most often omit and inquiries most often find.
The management plan
Each action needs an owner, a frequency and a response to failure: "named nurse contacts the patient the same day after a missed depot" beats "monitor compliance".
Early warning signs are agreed with the patient in their own words, and the response is graded from more contact to urgent review, hospital assessment or a lawful recall request.
Victim safety is planned separately with victim liaison services, and every condition must be necessary, proportionate and reviewable.
Treatment of sexual offending and fire-setting
ATSA (the Association for the Treatment of Sexual Abusers) recommends cognitive behavioural treatment as the core intervention for adults who have sexually offended, with medication only as an adjunct, because drugs reduce drive but leave offence-supporting cognitions untouched.
Drug treatment for problematic sexual arousal follows a risk-graded hierarchy of selective serotonin reuptake inhibitors, anti-androgens, gonadotrophin-releasing hormone agonists and their combinations.
A phosphodiesterase-5 inhibitor improves erection and has no place in this hierarchy.
Cyproterone acetate lowers androgen activity, and its most commonly reported psychiatric adverse effect is depression.
Deliberate fire-setting has one evidence-based treatment, a structured cognitive behavioural group programme: FIPP in prisons and FIP-MO in forensic mental health services.
Multi-agency public protection arrangements (MAPPA)
Feature | Detail |
|---|---|
Statutory basis | Criminal Justice Act 2003, England and Wales |
Responsible authority | Police, prison and probation services jointly |
Duty to co-operate | Health, local authority, housing and other relevant bodies |
Levels 1, 2 and 3 | Ordinary agency management, active multi-agency coordination, and senior coordination with exceptional resources |
The four categories are registered sexual offenders, violent and other sexual offenders, other dangerous offenders, and terrorist or terrorism-risk offenders; category does not set the management level.
MAPPA coordinates information and management, but the lead agency keeps its statutory duties, and MAPPA cannot grant restricted patient leave.
Information sharing is limited to what manages the identified risk, with the lawful basis, purpose and recipients recorded; disclosure to employers, partners or victims needs its own assessment.
Reviewing the plan
Review on a set date and after any material clinical, social or legal change, lifting each restriction that no longer serves a necessary, proportionate purpose.