Table 1

Characteristics of included studies

Authors	Country	Setting	Inclusion criteria	Number	Withdraw/dropout	Intervention	Control	Other
Aho-Mustonen et al 15	Finland	High secure	Schizophrenia and schizoaffective.	39 total: 35 men, 4 women.	1 Int. 2 TAU. ITT: unsure.	Psychoeducation programme eight weekly sessions. Therapists: two psychologists who completed 2-day training programme. Fidelity reported.	TAU.
Bernstein et al 16	The Netherlands	High secure	Personality disorder.	35 All men.	5 not sure which arm. ITT: yes.	Three years of schema therapy usually delivered twice a week. Therapists completed 8-day training programme and 2× monthly supervision groups. Fidelity reported.	TAU. Clinics free to choose. Typically a form of cognitive–behavioural, psychodynamic or humanistic psychotherapy.	Part of a long-term study of 102 patients.
Cullen et al.18 Some outcomes reported in Cullen et al.40	UK	Medium secure	Schizophrenia, schizoaffective disorder, bipolar disorder or other psychotic disorder; a history of violence; no prior participation in Reasoning and Rehabilitation (R&R) programme previously.	84 All men.	31 Int. 4 TAU. 23 out of 44 (52.3%) did not complete treatment. ITT: yes.	R&R programme structured, manualised programme. Minimum of 36 2-hour sessions (2 or 3 per week). Therapists -experienced staff who received 3–5 days’ training from programme authors. Fidelity reported	TAU.	Small sample size and low base rates of violent behaviour reducing likelihood of effect sizes. Randomisation occurred within sites and may have led to contamination across treatment. Possibility of bias as it was not possible to blind researchers to allocation status. 23% of referred patients refused to participate in the study.
Doyle et al.19 Some outcomes reported in Tarrier et al.41	UK	High secure	Personality disorder	63 total. All men.	At 36 months: 19 Int. 14 TAU. At 24 months 14 Int. ITT : yes.	SFT. Weekly 1-hour sessions for at least 18 months. Therapists were two experienced cognitive therapists who received specialist SFT training with ongoing supervised practice and supervision. Fidelity reported – uncertain therapist competence.	TAU. >14 reported. Group-based enhanced thinking skills and sex offender treatment were the most frequently provided therapies recorded on the TAU logs.	Problems recruiting participants and refusing to be interviewed or filling in forms incorrectly (37 out of initial 136 patients considered – 29.4%). High attrition; poor statistical power; insufficient frequency of ST; and the provision of ST by only two therapists.
Haddock et al 21	UK	High, medium and low secure 48 (62.3%). The other patients were living in the community.	Schizophrenia, schizoaffective and history of violent behaviour.	77 total: 66 men (85.7%), 11 women (14.3%).	9 in total. Unclear regarding which groups. ITT: Yes.	CBT. 25 sessions. Therapists experienced in CBT for people with psychosis, received training in the protocol and supervision. Fidelity assessed.	Social activity therapy. 25 sessions. Fidelity assessed.	108 of whom were identified as meeting initial inclusion criteria. Thirty-one refused to be assessed to determine eligibility (28.7%).
Hakvoort et al 22	The Netherlands	High secure	Personality disorder. Addictions. No previous TBS admission.	14 total. All men.	2 ITT: unsure.	Cognitive–behavioural music therapy and anger management 20×1 hours weekly sessions. Five therapists experienced in CBT music therapy in forensic psychiatry. Trained on a standard protocol. Fidelity assessed.	TAU. Most also received anger management sessions. 3 hours per week.	Six out of 21 identified refused to participate (28.6%).
MacInnes et al 23	UK	Medium secure	Schizophrenia, schizoaffective, bipolar, depression psychosis and personality disorder.	112 total: 91 men (85%), 16 women (15%). Five data missing.	Int: 7 (13%) 6 months. 8 (15%) 12 months. TAU: 15 (26%) 6 months. 15 (26%) 12 months. ITT: yes	Computer-aided solution focused on brief therapy. 6×1 hour monthly session. Staff trained in SFBT techniques. Fidelity assessed.	TAU.	Significantly more women withdrew from the study.
Tomlinson and Hoaken25	Canada	Medium secure	Pts experiencing emotional dysregulation. Schizophrenia, schizoaffective, bipolar, depression psychosis and personality disorder.	18 total: 14 men (78%), 4 women (22%).	Int: 3 (33%). TAU: 2 (22%). ITT: unsure.	DBT skills training sessions provided weekly for 1.5 hours for 6 months (24 total sessions). Followed training manual. 1 hour weekly staff consultation groups. Fidelity not reported.	TAU.
Walker et al 27	UK	High, medium and low secure.	Schizophrenia, schizoaffective, bipolar, depression and psychosis.	81 total: 79 men, 2 women.	16. Int. 0 TAU. ITT: unsure.	Psychoeducation programme. Two sessions per week for 11 weeks. Therapists: consultant psychiatrist and clinical nurse specialist. Fidelity not reported.	TAU. No psychological interventions but able to attend social and occupational activities.	Recruitment problems 26 out of 107 (24.3%) eligible participants not included. Recorded as not randomised.

CBT, cognitive–behavioural therapy; DBT, dialectical behaviour therapy; Int, Intervention; ITT, intention to treat; SFBT, solution-focused brief therapy; SFT, schema-focused therapy; TAU, treatment as usual.