Interventions for drug‐using offenders with co‐occurring mental health problems

Summary of findings for the main comparison. Therapeutic community compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Therapeutic community compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with therapeutic community
Re‐arrests assessed with official records Follow‐up: 12 months	266 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	RR 0.67 (0.53 to 0.84)	Study population
				98 per 100	32 fewer per 100 (46 fewer to 16 fewer)
Re‐incarceration assessed with official records Follow‐up: 12 months	266 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	RR 0.40 (0.24 to 0.67)	Study population
				59 per 100	36 fewer per 100 (45 fewer to 20 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (blinding and selective reporting).

Summary of findings 2. Therapeutic community and aftercare compared to cognitive behavioural skills for drug using women offenders with co‐occurring mental illness

Therapeutic community and aftercare compared to cognitive‐behavioural skills for drug‐using women offenders with co‐occurring mental health problems
Patient or population: drug‐using women offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community and aftercare Comparison: cognitive‐behavioural skills
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with cognitive‐behavioural skills	Risk difference with therapeutic community and aftercare
Self‐reported drug use Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.78 (0.46 to 1.32)	Study population
				17 per 100	4 fewer per 100 (9 fewer to 6 more)
Re‐arrest for any type of crime assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.69 (0.44 to 1.09)	Study population
				33 per 100	10 fewer per 100 (19 fewer to 3 more)
Criminal Activity assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.74 (0.52 to 1.05)	Study population
				33 per 100	9 fewer per 100 (16 fewer to 2 more)
Drug‐related crime assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.87 (0.56 to 1.36)	Study population
				21 per 100	3 fewer per 100 (9 fewer to 8 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear reporting in the paper raises concerns about the potential high risk of bias with regards to blinding and methods used in the randomisation procedure; we downgraded by one. ^bOne study with 95% confidence intervals through the line of no effect.

Summary of findings 3. Therapeutic community compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Therapeutic community compared to waiting list control for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community Comparison: waiting list control
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with waiting list control	Risk difference with therapeutic community
Return to prison (recidivism) post parole assessed with California Department of Correction's computerised Offender Based Information System Follow‐up: 36 months	478 (1 RCT)	⊕⊕⊕⊝ MODERATE^a	RR 0.60 (0.46 to 0.79)	Study population
				40 per 100	16 fewer per 100 (21 fewer to 8 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (randomisation process, concealment, and selective reporting).

Summary of findings 4. Mental health treatment court with assertive case management model compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Mental health treatment court with assertive case management model compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: court Intervention: mental health treatment court with assertive case management model Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with mental health treatment court with assertive case management model
Conviction for a new crime assessed with data from probation office Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	RR 1.05 (0.90 to 1.22)	Study population
				72 per 100	4 more per 100 (7 fewer to 16 more)
Re‐incarceration to jail assessed with data from probation office Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	RR 0.79 (0.62 to 1.01)	Study population
				71 per 100	15 fewer per 100 (27 fewer to 1 more)
Self‐reported drug use assessed with Addiction Severity Index (ASI) Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	‐	Mean self‐reported drug use was 0.08	MD 0.00 (‐0.03 lower to 0.03 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (allocation concealment and blinding of assessors) and by one for imprecision.

Summary of findings 5. Motivational interviewing/mindfulness and cognitive skills compared to relaxation training for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to relaxation training for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: motivational interviewing and cognitive skills Comparison: relaxation training
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Impact
Self‐reported marijuana use continuous	181 (1 RCT)	MODERATE^a	This study compared cognitive skills to a relaxation training intervention for adolescents in prison with depressed mood. Researchers measured marijuana use at 3‐months follow‐up assessment using the Risks and Consequences Questionnaire (RCQ). They report a main effect < .007, with participants in the motivational interviewing group showing fewer problems than participants in the relaxation training group.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for unclear risk of bias (random allocation and blinding).

Summary of findings 6. Motivational interviewing/mindfulness and cognitive skills compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to waiting list control for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: motivational interviewing and cognitive skills Comparison: waiting list control
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with waiting list control	Risk difference with motivational interviewing and cognitive skills
Self‐reported drug use assessed with Addiction Severity Index (ASI) composite drug score across 13 items of drug use in the last 30 days Follow‐up: 6 months	31 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean self‐reported drug use was 0.44	MD ‐0.04 lower (‐0.37 lower to 0.29 higher)
Abstinence from drug use Follow‐up: 6 months	31 (1 RCT)	⊕⊕⊝⊝ LOW^a	RR 2.89 (0.73 to 11.43)	Study population
				15 per 100	29 more per 100 (4 fewer to 160 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal information size not met.

Summary of findings 7. Motivational interviewing/mindfulness and cognitive skills compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: medium secure hospital and jail Intervention: motivational interviewing and cognitive skills Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Risk with treatment as usual	Risk difference with motivational interviewing and cognitive skills
Self‐reported frequency of marijuana use assessed with TCU‐CRTF (Texas Christian University: Correctional Residential Treatment Form) Scale from 0 to 32 Follow‐up: 3 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean self‐reported frequency of marijuana use was 1.50	MD ‐1.05 lower (‐2.39 lower to 0.29 higher)
Arrest frequency post release assessed with official police records Follow‐up: 36 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean arrest frequency post release was 1.47	MD ‐0.66 lower (‐1.31 lower to ‐0.01 lower)
Time to first arrest or offence assessed with official police records Follow‐up: 36 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean time to first arrest or offence was 1.6	MD 0.87 higher (‐0.12 lower to 1.86 higher)
Positive drug screen or refusal to provide a urine sample assessed with urine sample Scale from negative to positive Follow‐up: 12 months	84 (1 RCT)	⊕⊝⊝⊝ VERY LOW^a,b	‐	Mean positive drug screen or refusal to provide a urine sample was 3.25	MD ‐0.7 lower (‐3.5 lower to 2.1 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded by one for risk of bias (incomplete outcome measures).

Summary of findings 8. Multi‐systemic therapy involving family and juveniles compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Multi‐systemic therapy involving family compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: community Intervention: multi‐systemic therapy involving family Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with multi‐systemic therapy involving family
Drug dependence assessed with DUDIT questionnaire Scale from 0 to 44 Follow‐up: 7 months	156 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean drug dependence was 3.55	MD ‐0.22 lower (‐2.51 lower to 2.07 higher)
Arrested assessed by corroborating with police data Follow‐up: 7 months	158 (1 RCT)	⊕⊕⊝⊝ LOW^a	RR 0.97 (0.70 to 1.36)	Study population
				47 per 100	1 fewer per 100 (14 fewer to 17 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (blinding measures) and downgraded by one for imprecision.

Summary of findings 9. Multi‐systemic therapy involving family compared to group substance abuse therapy for drug‐using adolescents with co‐occurring mental illness

Multi‐systemic therapy involving family compared to group substance abuse therapy for drug‐using adolescents with co‐occurring mental health problems
Patient or population: drug‐using adolescents with co‐occurring mental health problems Setting: court Intervention: multi‐systemic therapy involving family Comparison: group substance abuse therapy
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with group substance abuse therapy	Risk difference with multi‐systemic therapy involving family
Arrests Follow‐up: range 6 months to 24 months	112 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean arrests were 1.19 SD	MD ‐0.24 SD lower (‐0.76 lower to 0.28 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; SD: standard deviation.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (selective reporting of outcomes) and by one for imprecision.

Summary of findings 10. Interpersonal psychotherapy compared to a psychoeducational intervention for drug‐using offenders with co‐occurring mental illness

Interpersonal psychotherapy compared to a psychoeducational intervention for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: interpersonal psychotherapy Comparison: psychoeducational intervention
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with a psychoeducational intervention	Risk difference with interpersonal psychotherapy
Substance abuse relapse post release Follow‐up: 3 months	38 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	RR 0.67 (0.30 to 1.50)	Study population
				47 per 100	16 fewer per 100 (33 fewer to 24 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded by one for risk of bias (selective reporting outcomes).

Summary of findings 11. Legal defence service and wrap‐around social work services compared to legal defence service only for drug‐using offenders with co‐occurring mental illness

Legal defence service and wrap‐around social work services compared to legal defence service only for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: court Intervention: legal defence service and wrap‐around social work services Comparison: legal defence service only
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with legal defence services only	Risk difference with legal defence services and wrap‐around social work services
Committing new offences Follow‐up: 12 months	29 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	RR 0.64 (0.07 to 6.01)	Study population
				1 per 100	2 fewer per 100 (0 fewer to 2 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded for risk of bias (incomplete outcome data).

Antecedentes

available in

Esta revisión forma parte de una familia de tres revisiones que proporcionan un examen detallado de lo que funciona para reducir el consumo de drogas y la actividad delictiva entre los delincuentes que consumen drogas. Estas tres revisiones informan sobre ensayos que generan varias publicaciones y numerosas comparaciones (Perry forthcominga; Perryforthcomingc). Dos de las tres revisiones representan un interés específico en las intervenciones farmacológicas y las intervenciones para las mujeres delincuentes. Las tres revisiones provienen de una revisión sistemática Cochrane anterior (Perry 2006). Se considera la efectividad de las intervenciones en base a dos resultados clave: el consumo de drogas y la actividad delictiva. Aquí se ha presentado el método revisado para esta revisión individual, que se centra en el impacto de las intervenciones para delincuentes que consumen drogas con trastornos mentales concomitantes.

Descripción de la afección

Las personas implicadas en problemas con el sistema de justicia penal son más propensas a presentar trastornos mentales. Muchos estudios informan cifras diferentes de prevalencia en dependencia de los métodos utilizados para calcular la prevalencia (Fazel 2016). Algunos estudios informan cifras genéricas que representan todos los trastornos mentales graves ‐ p.ej., más de la mitad (64%) de los reclusos en las cárceles de los Estados Unidos informan trastornos mentales graves (Glase 2006) ‐ y otros tratan de desglosar diferentes tipos de diagnósticos de salud mental (p.ej., psicosis vs. depresión mayor). En una revisión sistemática de 33 000 reclusos, uno de cada siete presentaba depresión mayor o psicosis, con pocos cambios en las tasas de diagnóstico en las últimas tres décadas (Fazel 2012).

Las diferencias en la prevalencia de los trastornos mentales se diferencian entre hombres y mujeres y por edad. Un estudio de los trastornos mentales en las cárceles encontró que más mujeres que hombres (31% y 14,5%, respectivamente) presentan un trastorno mental grave (Steadman 2009), y una estimación indica que dos tercios de los menores detenidos tienen un trastorno mental lo suficientemente grave como para limitar su capacidad para actuar (Shufelt 2006). Además, se descubrió que las mujeres delincuentes violentas tienen cinco veces más probabilidades que los hombres de presentar trastornos de ansiedad (Waserman 2005). Otros estudios han informado que una mayor proporción de las personas que presentan trastornos mentales tienen más probabilidades de ser arrestadas en comparación con la población general (Lamb 1998).

También se sabe que las tasas de comorbilidad entre los trastornos mentales y el consumo de sustancias son altas (Butler 2011). Esta comorbilidad empeora el pronóstico del trastorno psiquiátrico individual y aumenta la probabilidad de reincidencia y de mortalidad prematura después de la liberación (Chang 2015). A pesar de estas dificultades, se desconoce cuán bien las intervenciones concebidas para tratar esta comorbilidad abordan estos problemas (Fazel 2002).

Descripción de la intervención

Se han adoptado muchos tratamientos diferentes para el consumo indebido de sustancias (p.ej., desintoxicación, comunidades terapéuticas) para su uso en el sistema de justicia penal. Esta revisión incluye cualquier intervención diseñada para reducir, eliminar o prevenir la reincidencia en el consumo de drogas o la actividad delictiva, o ambas. Este objetivo ha dado lugar a la inclusión de una amplia variedad de tratamientos, que incluyen tribunales de tratamiento de salud mental con un modelo de gestión asertiva de los casos, comunidades terapéuticas, entrevistas motivacionales (EM) con habilidades cognitivas, uso de terapia multisistémica/multidimensional que involucra a las familias y entrenamiento de conciencia plena, servicio de defensa legal con servicios sociales relacionados y psicoterapia interpersonal .

La gestión de casos ha evolucionado tradicionalmente para abordar las necesidades de los programas de reinserción de los reclusos, que incluyen el empleo, la educación, la salud, la vivienda y el apoyo familiar mediante la evaluación y la conexión de los clientes con los servicios apropiados (Austin, 1994). La gestión de casos en los Estados Unidos se ha aplicado en los programas Treatment Accountability for Safer Communities (Marlowe 2003b); ha demostrado una eficacia inicial, pero sin evidencia sistemática que apoye el proceso. En el Reino Unido, en los años ochenta se desarrolló una prestación de servicios relacionados similar en un intento por proporcionar servicios más completos utilizando un enfoque "conjunto" (Synder 2012). La prestación de servicios relacionados requiere un enfoque basado en equipo que incluya al joven, la familia y los proveedores de servicios en el desarrollo, la implementación y la evaluación de cada parte de cualquier plan de apoyo (Wilson 2008).

Los tribunales de tratamiento de salud mental ayudan a vincular a los delincuentes que normalmente estarían en prisión con un tratamiento comunitario a largo plazo. Se basan en evaluaciones de salud mental, planes de tratamiento individualizado y monitorización judicial continua para abordar las necesidades de salud mental de los delincuentes y las preocupaciones de seguridad pública de las comunidades. Al igual que otros tribunales de resolución de problemas, como los tribunales de drogas, los tribunales de violencia doméstica y los tribunales comunitarios, los tribunales de salud mental tratan de abordar los problemas subyacentes que contribuyen a la conducta delictiva. Los tribunales de salud mental comparten características con los equipos de intervención en situaciones de crisis, los programas de desviación carcelaria, los programas especializados de libertad condicional y libertad bajo palabra, y una serie de otras iniciativas de colaboración destinadas a abordar el problema de la sobrerrepresentación significativa de personas con enfermedades mentales en el sistema de justicia penal.

A partir de la década del 60, en los Estados Unidos se han utilizado intervenciones terapéuticas comunitarias junto con programas de trabajo en libertad condicional para rehabilitar a los delincuentes en un entorno de apoyo durante un período relativamente largo. Las intervenciones terapéuticas comunitarias, específicamente las que proporcionan atención posterior, tienen efectos modestos sobre la reducción de la reincidencia y el consumo de drogas (Mitchell 2012a; Pearson 1999), pero se sabe menos sobre el impacto del uso de dichos esquemas en las personas que tienen trastornos mentales y problemas de consumo de drogas que ocurren de forma conjunta (p.ej., Sacks 2008).

Los enfoques cognitivo‐conductuales, incluyendo el autocontrol, el establecimiento de metas, el entrenamiento en autocontrol, el entrenamiento en habilidades interpersonales, la prevención de la reincidencia, el trabajo en grupo y la modificación del estilo de vida, han mostrado signos de éxito (Lipsey 2007). Los estudios de investigación anteriores basados en revisiones sistemáticas han excluido las evaluaciones que se centran específicamente en las necesidades de los delincuentes que consumen drogas o de los delincuentes con trastornos mentales, pero no en las personas con trastornos mentales y problemas de consumo de drogas concomitantes. Las técnicas de entrevistas motivacionales se emplean a menudo para promover la retención en el tratamiento y están dirigidas a mejorar el cambio motivacional y reducir la reincidencia posterior (McMurran 2009; Smedslund 2011).

La terapia multisistémica/multidimensional (MST/MDST) consiste en un tratamiento intensivo basado en la familia y la comunidad que se proporciona a adolescentes con dificultades clínicas, sociales y emocionales importantes. Los estudios de investigación sobre la eficacia de la MST no han logrado determinar que el MST sea más eficaz que otros servicios para evitar las disposiciones restrictivas en la vida fuera del hogar, reducir los arrestos o las condenas, o mejorar la vida y el funcionamiento de la familia (Littell 2005). La transferibilidad de estos sistemas ha sido cuestionada con resultados variables cuando se emplean en diferentes países y contextos (Bogt 2006). La MDST también se ha empleado a través del modelo de tribunal de drogas para menores, que está diseñado para abordar la relación entre el consumo de sustancias y la actividad delictiva y en el trabajo actual se compara con la terapia ocupacional grupal para el consumo de sustancias (tratamiento grupal para adolescentes [TGA]) (Dakof 2015).

A pesar de los conocimientos cada vez mayores sobre la eficacia de los programas de tratamiento para los delincuentes, al parecer ninguna evidencia reciente proveniente de una revisión sistemática se ha centrado en la eficacia del tratamiento en los delincuentes que consumen drogas y presentan trastornos mentales concomitantes.

De qué manera podría funcionar la intervención

Las intervenciones que se ofrecen a los delincuentes que consumen drogas bajo la custodia del sistema de justicia penal han variado con el tiempo. La gestión de casos se utiliza para describir lo que equivale a una serie de prácticas diversas y modelos de supervisión que abarcan varios servicios diferentes, incluida la libertad condicional. Se han utilizado ejemplos de gestión de casos para coordinar e integrar todos los aspectos de la supervisión comunitaria, desde la evaluación inicial de las necesidades de los delincuentes hasta la realización del programa y la finalización prevista de la orden o la obligación de la sentencia (Partridge 2004). De manera similar, la atención integral tiene varias fortalezas en su enfoque, que incluye la adaptación, centrada en la familia y culturalmente sensible, de cada plan de servicio, a las necesidades, valores y talentos de cada persona (Synder 2012).

Los tribunales de tratamiento de salud mental tienen como objetivo identificar a los clientes en las primeras etapas del proceso penal, ya sea en la cárcel o por parte del personal del tribunal, como los funcionarios de los servicios previos al juicio o los trabajadores sociales de la oficina del defensor público. La mayoría de los tribunales tienen criterios relacionados con los tipos de cargos, antecedentes penales y diagnósticos que se aceptarán. Por ejemplo, un tribunal puede aceptar solo a los acusados de delitos menores que no tengan antecedentes de delitos violentos y que tengan un diagnóstico de Eje I según criterios diagnósticos reconocidos. A los acusados que cumplen con los criterios sobre la base de la evaluación inicial, por lo general se les realiza una evaluación más completa para determinar su interés en participar y sus necesidades de tratamiento en la comunidad. Los acusados que aceptan participar reciben un plan de tratamiento y otras condiciones de supervisión en la comunidad. Los casos son desestimados o la sentencia se reduce en gran medida en el caso de los que cumplan su plan de tratamiento durante el tiempo acordado, por lo general de entre seis meses y dos años.

A partir de la década de los 60, en los Estados Unidos se han utilizado intervenciones terapéuticas comunitarias junto con programas de trabajo en libertad condicional para rehabilitar a los delincuentes en un entorno de apoyo durante un período relativamente largo. Por lo general, esto abarca la transición entre estar en prisión y trabajar dentro de la comunidad (Prendergast 2011). La filosofía de una intervención terapéutica comunitaria es enfocarse en el tratamiento de todo el ser (no en el consumo de drogas per se) y en los problemas sintomáticos subyacentes, con residentes fundamentales en el funcionamiento de la comunidad terapéutica (Mitchell 2012a). Estas intervenciones generalmente se basan en actividades grupales proporcionadas para tratar la enfermedad mental a largo plazo, los trastornos de personalidad y la adicción a las drogas. El enfoque suele ser residencial, con clientes y terapeutas viviendo juntos.

Los enfoques de la terapia cognitivo‐conductual (TCC) que utilizan programas basados en la teoría psicológica se han empleado para tratar de ayudar a las personas a abordar su comportamiento delictivo y, en general, han recibido un buen apoyo de la bibliografía en su reducción de la reincidencia. Esta terapia a menudo se describe como una intervención psicosocial cuyo objetivo es mejorar la salud mental. La TCC se centra en desafiar y cambiar distorsiones cognitivas (p.ej., pensamientos, creencias, actitudes) y comportamientos poco útiles, mejorar la regulación emocional y desarrollar estrategias personales de afrontamiento para resolver problemas actuales. Originalmente, se diseñó para tratar la depresión, pero sus usos se han expandido para incluir el tratamiento de varios trastornos mentales, incluyendo la ansiedad.

La psicoterapia interpersonal (PIP) es una psicoterapia breve y enfocada en las relaciones que se centra en resolver problemas interpersonales y lograr la recuperación sintomática. Se trata de un tratamiento sustentado de forma empírica que sigue un enfoque altamente estructurado y limitado en el tiempo y que está previsto que se complete en un plazo de 12 a 16 semanas. La PIP se basa en el principio de que las relaciones y los eventos de la vida impactan en el estado de ánimo, y que lo contrario también es cierto.

Miller y Rollnick desarrollaron entrevistas motivacionales como un proceso para motivar el cambio en consumidores de sustancias (Miller 1991). Esta técnica utiliza diferentes estrategias, como expresar empatía, evitar argumentar a favor del cambio y trabajar en la ambivalencia para fortalecer el compromiso con el cambio. Los metanálisis apoyan el uso de las entrevistas motivacionales como tratamiento independiente y en combinación con programas más intensivos (Vasilaki 2006). A esta idea de compromiso con el cambio está vinculada la idea del autocontrol, que ha establecido vínculos entre el consumo de sustancias y el comportamiento antisocial (Malouf 2014). La teoría indica que el uso de la conciencia plena implica una mayor conciencia de sí mismo, lo que puede promover una respuesta reflexiva en lugar de reactiva y puede ayudar a mejorar el estado de ánimo y la conducta problemática (Shonin 2013).

Por qué es importante realizar esta revisión

Muchas personas que están bajo la custodia del sistema de justicia penal presentan trastornos mentales y problemas de consumo de drogas concomitantes. Aunque los estudios de investigación anteriores han evaluado ampliamente los programas de tratamiento para delincuentes, se sabe poco acerca de los desafíos, tratamientos y oportunidades de rehabilitación para los delincuentes con trastornos mentales y consumo de drogas concomitantes. Por lo tanto, se considera que la evaluación de la evidencia existente sobre el impacto de las intervenciones en los delincuentes que consumen drogas con trastornos mentales concomitantes podría ser útil para identificar los tratamientos para reducir el consumo de drogas y la actividad delictiva en esta población vulnerable.

Objetivos

available in

La revisión aborda las siguientes cuestiones.

¿Hay algún tratamiento para los delincuentes que consumen drogas con trastornos mentales concomitantes que reduzca el consumo de drogas?
¿Hay algún tratamiento para los delincuentes que consumen drogas con trastornos mentales concomitantes que reduzca la actividad delictiva?
¿El contexto del tratamiento (tribunal, comunidad, prisión/establecimiento seguro) afecta el resultado de la intervención?
¿El tipo de tratamiento afecta el/los resultado/s del tratamiento?

Métodos

available in

Criterios de inclusión de estudios para esta revisión

Tipos de estudios

Ensayos controlados aleatorizados (ECA).

Tipos de participantes

Se incluyeron las personas implicadas en problemas con el sistema de justicia penal con trastornos mentales y problemas de consumo de drogas concomitantes sin importar el género, la edad o el origen étnico. El consumo de drogas incluyó cualquier estudio que se refiriera a participantes que consumían de manera ocasional, eran dependientes o se sabía que abusaban de las drogas. Los delincuentes se definieron como personas implicadas en problemas con el sistema de justicia penal. Las personas pueden residir en hospitales especiales, prisiones o en la comunidad, o ser derivadas de los tribunales o sometidas a planes de remisión de casos de arresto para recibir tratamiento. El contexto del estudio podría cambiar durante el proceso del estudio. Por ejemplo, las personas implicadas en problemas con el sistema de justicia penal podrían comenzar en la cárcel pero progresar a través de un proyecto de trabajo en libertad condicional en un contexto comunitario. Se consideró que los delincuentes presentaban trastornos mentales concomitantes cuando el documento lo señaló explícitamente. Se utilizaron varios mecanismos diferentes para identificar las muestras de estudio con trastornos mentales, que incluyeron:

pruebas diagnósticas de referencia como los criterios del Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM‐IV), o la International Statistical Classification of Diseases and Related Health Problems Tenth Revision (ICD‐10);
la naturaleza de la intervención (p.ej., un tribunal de salud mental); o
descripciones de los autores de los estudios de los participantes que presentan "antecedentes de problemas psiquiátricos" o un "trastorno mental grave" con consumo de sustancias concomitante.

Tipos de intervenciones

Las intervenciones incluidas se diseñaron, total o parcialmente, para eliminar o prevenir entre los participantes la recaída en el consumo de drogas o la actividad delictiva, o ambas. En la revisión se incluyeron varias intervenciones.

Intervenciones experimentales incluidas en la revisión

Cualquier intervención farmacológica (p.ej., buprenorfina, metadona)
Cualquier intervención psicosocial (p.ej., comunidad terapéutica, manejo de casos, terapia cognitivo‐conductual, psicoterapia interpersonal, entrevistas motivacionales)

Intervenciones control incluidas en la revisión

Ningún tratamiento o control en lista de espera
Tratamiento mínimo o alternativo (p.ej., el informe del uso de una intervención similar pero menos intensa, utilizando un enfoque teórico diferente con los mismos componentes o una intervención alternativa diferente)
Tratamiento habitual (incluido cualquier estudio que informara una combinación o componente de (1) una intervención psicológica (p.ej., manejo de la ira, entrevistas motivacionales, consejería, reemplazo de la agresividad, terapia familiar), (2) un programa educativo (p.ej., salud, educación relacionada con el consumo de sustancias sobre las conductas de riesgo), o (3) destrezas para la vida (p.ej., planificación financiera, habilidades laborales, habilidades informáticas, aptitudes interpersonales en las entrevistas)

Tipos de medida de resultado

Resultados primarios

Cuando los documentos informaron varios períodos de seguimiento diferentes, se informó el período más largo, ya que se considera que esta medida proporciona la estimación más conservadora de la efectividad. Se proporcionaron:

medidas de consumo de drogas informadas como:
- consumo de drogas autoinformado (drogas no específicas, uso de drogas específicas sin incluir el alcohol, puntuaciones compuestas del Addiction Severity Index); o
- uso biológico de drogas (medido por medio de pruebas de orina o análisis del cabello); y
actividad delictiva medida por:
- actividades delictivas autoinformadas u oficialmente denunciadas (incluido el arresto por cualquier delito, delitos de drogas o reincidencia).

Métodos de búsqueda para la identificación de los estudios

Búsquedas electrónicas

Updated searches identified records from 2014 to 6 February 2019.

Cochrane Central Register of Controlled Trials (CENTRAL; issues to February 2019).
MEDLINE (1966 to February 2019).
Embase (1980 to February 2019).
PsycINFO (1978 to February 2019).
SciSearch (Science Citation Index) (1974 to February 2019).
Social SciSearch (Social Science Citation Index) (1972 to February 2019).
Applied Social Sciences Index and Abstracts (ASSIA; 1987 to February 2019).
National Technical Information Service (NTIS; 1964 to March 2014).^a
Sociological Abstracts (1963 to March 2014).^b
Healthcare Management Information Consortium (HMIC; to February 2019).
Public Affairs Information Service (PAIS; 1972 to February 2019).
Criminal Justice Abstracts (1968 to February 2019).
Latin American Caribbean Health Sciences Literature (LILACS; 2004 to February 2019).
Current Controlled Trials (December 2009).^c
SPECTR (March 2004).^d
Cumulative Index to Nursing and Allied Health Literature (CINHAL)plus (up until February 2019).

^aPaid access only ‐ insufficient resources to search.

^bNot available to search through York University.

^cNo longer available to search.
^dNo public access through Campbell Collaboration website, which previously hosted the database.

To update the review, we restricted the search to studies that were published since the end date of the previous search (May 2014). We did not search several original databases indicated by the key at the end of the database list. One database (NTIS) was fee charging, and the other three databases (Sociological Abstracts, Current Controlled Trials, and SPECTR) were not available for searching due to changes in the provision of databases through the University of York.

We developed search strategies for each database to exploit the search engine most effectively and to make use of any controlled vocabulary. We included methodological search filters designed to identify RCTs. Whenever possible, we used filters retrieved from the InterTASC Information Specialists' Sub‐Group (ISSG) Search Filter Resource site (www.york.ac.uk/inst/crd/intertasc/). If filters were unavailable from this site, we substituted search terms based on existing versions. We did not place any language restrictions on identification and inclusion of studies in the review.

We have listed details of the update search strategies and results and the websites searched in Appendix 1, Appendix 2, Appendix 3, Appendix 4, Appendix 5, Appendix 6, Appendix 7, Appendix 8, Appendix 9, Appendix 10, and Appendix 11.

Búsqueda de otros recursos

Reference checking

We scrutinised the reference lists of all retrieved articles for additional references and searched the catalogues of relevant organisations.

Personal communication

We sought out experts for their knowledge of other published or unpublished studies relevant to the review.

Obtención y análisis de los datos

Selección de los estudios

A team of review authors independently inspected the search hits by reading the titles and abstracts. Each potentially relevant study was obtained as a full‐text article. Each article was independently assessed for inclusion. In the case of discordance, a third independent review author arbitrated. One review author undertook translation of articles not written in the English language.

We divided the screening process into two key phases. Phase one used eight key questions as reported in the original review.

Prescreening criteria: phase one

Is the document an empirical study? If not, exclude the document
Does the study evaluate an intervention, a component of which is designed to reduce, eliminate, or prevent relapse with drug‐using offenders?
Are participants referred by the criminal justice system at baseline?
Does the study report pre‐ and post‐programme measures of drug use?
Does the study report pre‐ and post‐programme measures of criminal behaviour?
Is the study an RCT?
Do the outcome measures refer to the same length of follow‐up for the two groups?

Papers included after phase one screening were then scrutinised for further assessment.

Prescreening criteria: phase two

Does the study population comprise wholly participants with diagnosed mental health problems using DSM‐IV or ICD‐10 diagnostic criteria? if yes, include the document
Does the study population comprise wholly participants identified on screening to have a mental health problem(s) based on intervention eligibility (e.g. mental health court)? if yes, include the document
When the full study population does not comprise offenders with diagnosed or presumed mental health problems, are separate results given for those participants with mental health problems? if no, exclude the document

Extracción y manejo de los datos

We used data extraction forms to standardise the reporting of data from all studies obtained as potentially relevant. Two review authors independently extracted data and subsequently checked them for agreement. The narrative tables presented study details (e.g. author, year of publication, country of study origin), study methods (e.g. random assignment), participants (e.g. number in sample, age, gender, ethnicity, age, mental health status), interventions (e.g. description, duration, intensity, setting), outcomes (e.g. description, follow‐up period, reporting mechanism), and notes (e.g. country, funding).

Evaluación del riesgo de sesgo de los estudios incluidos

The team of review authors independently assessed risk of bias of all included studies using the 'Risk of bias' assessment criteria recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

The recommended approach for assessing risk of bias in studies included in Cochrane Reviews involves a two‐part tool that addresses four specific domains, namely, sequence generation and allocation concealment (selection bias), blinding of outcome assessors (detection bias), incomplete outcome data (attrition bias), and selective outcome reporting (reporting bias). The first portion of the tool involves describing what was reported to have happened in the study. The second portion of the tool involves assigning a judgement related to the risk of bias for that entry, in terms of low, high, or unclear risk. To make these judgements, we used the criteria indicated by the Cochrane Handbook for Systematic Reviews of Interventions, as adapted to the addiction field. See Appendix 12 for details.

The domains of sequence generation and allocation concealment (avoidance of selection bias) were addressed in the tool by a single entry for each study.

Participants and personnel cannot be blinded to the type of intervention; moreover, we think that being aware of receiving a psychosocial treatment is in itself part of the therapeutic effect; for these reasons, we did not assess risk of performance bias.

Detection bias was considered separately for objective outcomes (e.g. dropout, use of substance of abuse measured by urine analysis, participants relapsed at end of follow‐up, participants engaged in further treatments) and subjective outcomes (e.g. duration and severity of signs and symptoms of withdrawal, participant self‐reported use of substance, side effects, social functioning as integration at school or at work, family relationship).

Incomplete outcome data (avoidance of attrition bias) was considered for all outcomes except for dropout from treatment, which is very often the primary outcome measure in trials on addiction.

For studies identified in the search, the review authors attempted to contact study authors to establish whether a study protocol was available.

Medidas del efecto del tratamiento

The mean differences (MD) with 95% confidence intervals (CIs) was used for continuous outcomes measured on the same scale, and the standardised mean difference (SMD) was used for continuous outcomes measured on different scales. Higher scores for continuous measures are representative of greater harm. We presented dichotomous outcomes as risk ratios (RRs), with 95% CIs.

Cuestiones relativas a la unidad de análisis

To avoid double‐counting of outcome measures (e.g. arrest, parole violation) and follow‐up periods (e.g. 12 months, 18 months), we checked all trials to ensure that multiple studies reporting the same evaluation did not contribute towards multiple estimates of programme effectiveness. We followed Cochrane guidance, and where appropriate, we combined intervention and control groups to create a single pair‐wise comparison. When this was not appropriate, we selected one treatment arm and excluded the others.

Manejo de los datos faltantes

We attempted to contact study authors via email when we noted missing data in the original publication.

Evaluación de la heterogeneidad

We assessed heterogeneity using the I² statistic and the Chi² statistic (Higgins 2011). We regarded heterogeneity as substantial if I² was greater than 50% or if the P value was lower than 0.10 for the Chi² test for heterogeneity (Deeks 2017). In keeping with the guidance provided in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017), we distinguished the following values to denote no important heterogeneity and moderate, substantial, and considerable heterogeneity, respectively: 0% to 40%, 30% to 60%, 50% to 90%, and 75% to 100%.

Síntesis de los datos

We used the RevMan software package to perform a series of meta‐analyses for continuous and dichotomous outcome measures (RevMan 2012). We used a random‐effects model to account for the fact that participants did not come from a single underlying population. We combined two studies of the therapeutic community and aftercare in comparison to treatment as usual.

Análisis de subgrupos e investigación de la heterogeneidad

We had planned to conduct sensitivity analyses to assess the impact of studies at high risk of bias compared with those at low or unclear risk of bias. Because of the overall high risk of bias of the included studies, this analysis was not possible.

Grading of evidence and 'Summary of findings' tables

We assessed the overall quality of the evidence for the following primary outcomes using the GRADE system: relapse, frequency of use, extent of use, any adverse events, and dropout from treatment. The GRADE Working Group developed a system for grading the quality of evidence (Schunemann 2013); this system takes into account issues related not only to internal validity but also to external validity, such as directness of results.

We have presented the main findings of the review in a 'Summary of findings' table. This transparent and simple tabular format provides key information concerning quality of evidence, magnitude of effect of the interventions examined, and sums of available data for the main outcomes.

The GRADE system uses the following criteria for assigning grades of evidence.

High: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

Grading is decreased for the following reasons.

Serious (‐1) or very serious (‐2) study limitations for risk of bias.
Serious (‐1) or very serious (‐2) inconsistency between study results.
Some (‐1) or major (‐2) uncertainty about directness (correspondence between the population, the intervention, or the outcomes measured in the studies actually found and those under consideration in our systematic review).
Serious (‐1) or very serious (‐2) imprecision of the pooled estimate.
Publication bias strongly suspected (‐1).

Results

Description of studies

Results of the search

As shown in Figure 1, our update searches identified 9653 records. We screened out 9424 references based on titles and abstracts. We examined the remaining 229 records in full text and excluded 224 of them (see Characteristics of excluded studies). We included five new trials,(Cullen 2012; Dakof 2015; Malouf 2017; McCarter 2016; Sundell 2008), along with one follow‐up study to an existing trial within the review (Lanza 2014), and we included three ongoing trials (Baldus 2011; Tinland 2013; VanDorn 2017), along with eight studies from the previous review. The total number of included studies is 13 (see Characteristics of included studies).

Figure 1

Study flow diagram.

Included studies

Population

The 13 included trials randomised a total of 2606 participants and were published between 1999 and 2017. Seven of the 13 trials included adult drug‐using offenders. Three studies investigated the impact on interventions with adolescents and/or youth (Dakof 2015; McCarter 2016; Stein 2011). Two studies included females only (Johnson 2012; Sacks 2008). Three studies reported on juveniles or youth involved in the criminal justice system (Dakof 2015; McCarter 2016; Stein 2011). Adult male offenders were the focus of study populations in the remaining studies, with a mean age of 30 years. In all but two studies (Cullen 2012; McCarter 2016), most participants were of white ethnic origin.

Mental health diagnoses varied across studies (see Table 1.

See summary of findings Table for the main comparison.

Table 1. Mental health diagnoses

Study, year	Criteria used for diagnoses	Description of mental health problem
Cosden 2003	Determined by a psychiatrist/psychologist on the basis of a clinical interview and observations	Mood disorder Schizophrenia Bipolar disorder Other Dual diagnosis
Cullen 2012	Primary clinical diagnosis of a psychotic disorder. Diagnosis mechanism not reported	Schizophrenia Schizoaffective disorder Bipolar disorder Other psychotic disorder
Dakof 2015	Diagnostic Interview Schedule for Children (DISC‐2) ‐ identifying presence of mental disorders according to the DSM‐III Youth Self‐Report	Presence of mental health disorders Externalising subscales
Johnson 2012	Hamilton Rating Scale for Depression Median duration of index episode in months Number of depressive episodes Number of previous suicide attempts DSM‐IV Axis I disorders using the SCID‐I/II	Criteria for a major depressive disorder at least 4 weeks after substance abuse treatment Minimum score of 18 on the Hamilton Rating Scale for Depression
Lanza 2014	DSM‐IV Mini International Neuropsychiatric Interview Anxiety Sensitivity Index	Anxiety Mental health disorders Antisocial personality disorder Major depressive disorder Generalised anxiety disorder
Malouf 2017	Borderline Personality Disorder Features assessed with the Personality Assessment Inventory	Affective instability Identity problems Negative relationships Impulsivity
McCarter 2016	Youth Self‐Report that contain scales orientated to the DSM‐IV	Somatic complaints Anxiety and depression Social problems Internalising and externalising (thought and attention problems)
Sacks 2004	DIS	Diagnosis of lifetime Axis I or Axis II mental disorder Antisocial personality disorder
Sacks 2008	Global Severity Index Beck Depression Inventory Lifetime of mental health PTSD Symptom Scale ‐ Interview Posttraumatic Stress Diagnostic Scale	Depression PTSD Lifetime of mental health
Sacks 2011	DSM‐IV diagnostic criteria Beck Depression Inventory Post Traumatic Stress Disorder Symptom Scale Brief Symptom Inventory Global Severity Index	Depression PTSD Psychological distress
Stein 2011	CES‐D Scale	Scores > 16 indicate presence of significant depression; 69.8% had significant depressive symptoms
Sundell 2008	DSM‐IV diagnostic criteria Youth Self‐Report	Conduct disorder Internalising and externalising Total behaviour problems
Wexler 1999; Prendergast 2003; Prendergast 2004	Not specified	Antisocial personality disorder Phobias PTSD Depression Dysthymia Attention deficit hyperactivity disorder

CES‐D: Center for Epidemiological Studies ‐ Depression; DIS: Diagnostic Interview Schedule; DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; PTSD: post‐traumatic stress disorder; SCID: Structured Clinical Interview for DSM Disorders.

Settings

Eight studies were conducted in a secure setting (Johnson 2012; Lanza 2014; Malouf 2017; Sacks 2004; Sacks 2008; Sacks 2011; Stein 2011; Wexler 1999), two studies were conducted in community settings (Cosden 2003; Sundell 2008), and two studies were conducted in court settings (Dakof 2015; McCarter 2016). One study was conducted with a medium forensic secure hospital population in the United Kingdom (Cullen 2012). Studies were published in the United States (n = 10/13; 76%), Spain (n = 1/13; 7.6%), the United Kingdom (n = 1/13; 7.6%), and Sweden (n = 1/13; 7.6%).

Duration of trials

Trial duration varied between three‐month follow‐up in Johnson 2012, Lanza 2014, Stein 2011, and Sundell 2008, and five‐year follow‐up in Wexler 1999. Six‐month follow‐up was reported in Cosden 2003,Dakof 2015, and Sacks 2008. The remaining studies reported on outcomes at 12, 24, and 36 months (Cosden 2003; Cullen 2012; Dakof 2015; Malouf 2017; McCarter 2016; Sacks 2011; Sacks 2004). Treatment duration was most intensive (e.g. lasting between three and seven days per week) when a therapeutic community model was employed for periods of up to 12, 18, and 24 months (e.g. Sacks 2004; Sacks 2008; Sacks 2011); typically all other treatment interventions lasted between four and six months (e.g. Cullen 2012; Lanza 2014). The shortest treatment intervention was delivered in a 90‐minute session followed by a 60‐minute booster session upon release (Stein 2011).

Outcome measures

A total of 5 of 13 (38%) trials reported drug outcomes (Cullen 2012; Johnson 2012; Lanza 2014; Stein 2011; Sundell 2008), 5 of 13 (38%) reported crime outcomes (Dakof 2015; McCarter 2016; Sacks 2004; Sacks 2011; Wexler 1999), and 3 of 13 (23%) reported both drug and crime outcomes (Cosden 2003; Malouf 2017; Sacks 2008).

Interventions

Therapuetic interventions and aftercare

Four studies compared a therapeutic community (TC) intervention with aftercare versus treatment as usual (Sacks 2004; Sacks 2011), another intervention (Sacks 2008), or no intervention (Wexler 1999). Sacks 2004 compared a modified TC residential treatment programme using a cognitive‐behavioural curriculum to change attitudes and lifestyles versus a programme of intensive psychiatric services with medication, weekly individual therapy and counselling, and specialised groups of cognitive‐behavioural work, anger management, therapy and education, domestic violence, parenting, and weekly drug/alcohol therapy sessions.

Sacks 2008 evaluated a modified TC group with programme activities supplemented by peer‐led activities on weekends in comparison to an intensive outpatient programme that consisted of an educational programme on substance abuse treatment.

Sacks 2011 consisted of a re‐entry residential TC programme where participants worked in the community and saved money for independent living. This was compared to participants who were released to a community corrections facility during the day; they left the facility to go to work, receive treatment, and report to parole officers. This group engaged with brokering community‐based services and directly received support and counselling services. A weekly relapse prevention group and daily medication monitoring were provided. Psychiatric and substance abuse services were provided by outside agencies (community parole officers helped clients choose). The Wexler study compared a TC treatment programme with aftercare in the community versus a waiting list control.

Mental health court

One study compared use of a mental health court and case management to treatment as usual (Cosden 2003). The mental health treatment court (MHTC) consisted of case management and assertive community treatment (ACT) provided via a case management model. This model included weekly or bi‐weekly court supervision and frequent contact with case managers, followed by treatment as usual (if required), and compared this to treatment as usual, which included traditional court proceedings and county mental health services (Cosden 2003).

Motivational interviewing, mindfulness, and cognitive skills

Four studies compared motivational interviewing, mindfulness, and cognitive skills to no intervention (Lanza 2014), another intervention (Stein 2011), or treatment as usual (Cullen 2012; Malouf 2017). Stein 2011 was a manualised motivational intervention focused on empathy ‐ not arguing and developing discrepancy; self‐efficacy; and personal choice, and compared this approach to a relaxation intervention that included progressive muscle relaxation, use of guided imagery, and feedback on use of techniques.

Malouf 2017 used a manualised group intervention for jail inmates nearing release into the community. The intervention incorporated and adapted elements from several mindfulness‐based interventions (MBIs), including acceptance and commitment therapy, mindfulness‐based relapse prevention (MBRP), and dialectical behavioural therapy (DBT), and was compared to programmes that were normally available within the prison (e.g. anger management financial planning, health education).

Lanza 2014 used cognitive‐behavioural therapy (CBT) to change behaviour through cognitive restructuring and compared to ACT, which aimed to construct an alternative context in which behaviour aligned with one’s values is more likely to occur.

Multi‐systemic therapy including families

Two studies compared multi‐systemic therapy including families versus treatment as usual (in Sundell 2008) and another intervention (in Dakof 2015). Sundell 2008 compared an intensive family‐ and community‐based treatment to support prosocial development versus individual counselling, family therapy, addiction treatment, and special education services.

Dakof 2015 compared an intervention that involved therapists who worked individually with each family in four areas of treatment (adolescent, parent, family, and community) versus adolescent group therapy based on cognitive‐behavioural therapy and motivational interviewing.

Legal defence and social work

One study compared legal defence and wrap‐around social work to legal defence service only (McCarter 2016). The wrap‐around approach provides a collaborative and co‐ordinated response of service providers that organises and streamlines service delivery. This includes attending any team meeting with or on behalf of youth, providing service referrals, and connecting families and guardians to local providers for appropriate mental health, substance abuse, and educational services and support. This was compared to provision of only legal defence service.

Interpersonal psychotherapy

One study compared interpersonal psychotherapy versus another intervention (Johnson 2012). Study participants in the intervention group received manualised group and individual sessions in prison for treatment of substance misuse and mental health problems. These approaches were compared to an attention‐matched manualised in‐prison and post‐release psychoeducation course on mental health and drug problems.

Excluded studies

We excluded 224 full‐text studies. (See Characteristics of excluded studies for further details.) Reasons for exclusion were lack of criminal justice involvement in referral to the intervention; lack of reporting of relevant drug or crime outcome measures, or both, at pre‐ and post‐intervention periods; and allocation of participants to study groups that were not strictly randomised or did not contain original trial data.

Risk of bias in included studies

Allocation

Randomisation

All 13 studies were described as randomised. Nine of the included studies reported on how the randomisation sequence was generated and were judged as having low risk of bias (Cosden 2003; Dakof 2015; Johnson 2012; Lanza 2014; Malouf 2017; McCarter 2016; Sacks 2011; Stein 2011; Sundell 2008). The remaining four studies did not report how the randomisation sequence of participants was generated (Cullen 2012; Sacks 2004; Sacks 2008; Wexler 1999).

Characteristics at baseline

Eight of the 13 studies were similar in terms of drug use at baseline (Cullen 2012; Dakof 2015; Johnson 2012; McCarter 2016; Sacks 2008; Sacks 2011; Stein 2011; Wexler 1999); four studies were rated unclear (Cosden 2003; Lanza 2014; Malouf 2017; Sundell 2008); and one study showed comparable baseline differences (Sacks 2004). For similarity on criminal justice measures, nine studies were rated as similar (Cosden 2003; Cullen 2012; Dakof 2015; Johnson 2012; McCarter 2016; Sacks 2008; Sacks 2011; Sacks 2004; Wexler 1999), and four were rated as unclear (Lanza 2014; Malouf 2017; Stein 2011; Sundell 2008).

Allocation concealment

Of the 13 studies, only two adequately reported that the allocation process was concealed (Johnson 2012; Sundell 2008 ). The remaining 11 (85%) studies were rated as unclear.

Blinding

We assessed risk of detection bias across subjective and objective measures (see Appendix 12). We rated eight studies as having unclear risk (Cosden 2003; McCarter 2016; Sacks 2004; Sacks 2008; Sacks 2011; Stein 2011; Sundell 2008; Wexler 1999); two studies as having low risk (Cullen 2012; Lanza 2014); and the remaining three studies as having high risk of bias.

Incomplete outcome data

Loss to follow‐up was reported to a differing extent in the included studies. We rated six studies as having low risk with limited attrition noted (Johnson 2012; Lanza 2014; Sacks 2004; Stein 2011; Sundell 2008; Wexler 1999); three studies as having unclear risk (Dakof 2015; Sacks 2008; Sacks 2011); and four studies as having high risk of bias (Cosden 2003; Cullen 2012; Malouf 2017; McCarter 2016).

Selective reporting

We rated five of the thirteen trials as having unclear risk of bias (McCarter 2016; Sacks 2004; Stein 2011; Sundell 2008; Wexler 1999); six studies as having low risk (Cosden 2003; Cullen 2012; Lanza 2014; Malouf 2017; Sacks 2008; Sacks 2011), and two studies as having high risk of bias (Dakof 2015; Johnson 2012).

See Figure 2 and Figure 3 for details.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Effects of interventions

1. Therapeutic community and aftercare versus treatment as usual

Impact on self‐reported drug use

This was not reported.

Impact on self‐reported criminal activity

Sacks 2011 and Sacks 2004 were combined to show a significant reduction in the number of re‐arrests (risk ratio (RR) 0.67, 95% confidence interval (CI) 0.53 to 0.84) and re‐incarcerations (RR 0.40, 95% CI 0.24 to 0.67), with moderate‐certainty evidence at 12‐month follow‐up (266 participants; see Analysis 1.1).

2. Therapeutic community and aftercare versus cognitive‐behavioural skills for drug‐using women

See summary of findings Table 2.

Impact on self‐reported drug use

Sacks 2008 showed no significant reduction in self‐reported drug use (RR 0.78, 95% CI 0.46 to 1.32), with low‐certainty evidence at six‐month follow‐up (314 participants; see Analysis 2.1).

Impact on self‐reported criminal activity

Sacks 2008 showed no significant reduction in re‐arrest for any type of crime (RR 0.69, 95% CI 0.44 to 1.09), criminal activity (RR 0.74, 95% CI 0.52 to 1.05), or drug‐related crime (RR 0.87, 95% CI 0.56 to 1.36), with low‐certainty evidence at six‐month follow‐up (314 participants; see Analysis 2.2,Analysis 2.3, and Analysis 2.4).

3. Therapeutic community versus waiting list control

See summary of findings Table 3.

Impact on self‐reported drug use

This was not reported.

Impact on self‐reported criminal activity

Wexler 1999 showed a significant reduction (but a trend towards favouring) return to prison in favour of the therapeutic community intervention (RR 0.60, 95% CI 0.46 to 0.79), with moderate‐certainty evidence at 36‐month follow‐up (478 participants; see Analysis 3.1).

4. Mental health treatment court with assertive case management model versus treatment as usual

See summary of findings Table 4.

Impact on self‐reported drug use

Cosden 2003 showed no significant reduction in Addiction Severity Index (ASI)‐self‐reported drug use (mean difference (MD) 0.00, 95% CI ‐0.03 to 0.03), with low‐certainty evidence at 12‐month follow‐up (235 participants; see Analysis 4.3).

Impact on self‐reported criminal activity

Cosden 2003 showed no significant reduction in conviction for a new crime (RR 1.05, 95% CI 0.90 to 1.22) or re‐incarceration to jail (RR 0.79, 95% CI 0.62 to 1.01), with low‐certainty evidence at 12‐month follow‐up (235 participants; see Analysis 4.1 and Analysis 4.2).

5. Motivational interviewing/mindfulness and cognitive skills versus relaxation therapy

See summary of findings Table 5.

Impact on self‐reported drug use

Stein 2011 compared cognitive skills to a relaxation training intervention for adolescents in prison with depressed mood. This study measured marijuana use at three‐month follow‐up assessment using the Risks and Consequences Questionnaire (RCQ). Researchers reported a main effect < .007, with participants in the motivational interviewing group showing fewer problems than participants in the relaxation training group. No further numerical information is available (moderate‐certainty of evidence; 181 participants).

Impact on self‐reported criminal activity

This was not reported.

6. Motivational interviewing/mindfulness and cognitive skills versus waiting list control

See summary of findings Table 6.

Impact on self‐reported drug use

Lanza 2014 reported no significant reduction in self‐reported drug use based on the ASI (MD ‐0.04, 95% CI ‐0.37 to 0.29) and abstinence from drug use (RR 2.89, 95% CI 0.73 to 11.43), with low‐certainty evidence at six months (31 participants; see Analysis 5.1 and Analysis 5.2).

Impact on self‐reported criminal activity

Studies did not assess this outcome.

7. Motivational interviewing/mindfulness and cognitive skills versus treatment as usual

See summary of findings Table 7.

Impact on self‐reported drug use

Malouf 2017 found no significant reduction in frequency of marijuana use (MD ‐1.05, 95% CI ‐2.39 to 0.29), with very low‐certainty evidence at three months post release (40 participants; see Analysis 6.1).

Cullen 2012 found no significant reduction in positive drug screens (MD ‐0.7, 95% CI ‐3.5 to 2.1), with very low‐certainty evidence at 12 months (84 participants; see Analysis 6.4).

Impact on self‐reported criminal activity

Malouf 2017 found a significant reduction in frequency of re‐arrest (MD ‐0.66, 95% CI ‐1.31 to ‐0.01) but not in time to first arrest (MD 0.87, 95% CI ‐0.12 to 1.86), with very low‐certainty evidence up to 36 months (40 participants; see Analysis 6.2 and Analysis 6.3).

8. Multi‐systemic therapy (involving family) and juveniles versus treatment as usual

See summary of findings Table 8.

Impact on self‐reported drug use

Sundell 2008 found no significant reduction in drug dependence on the Drug Use Disorders Identification Test (DUDIT) score (MD ‐0.22, 95% CI ‐2.51 to 2.07), with low‐certainty evidence up to seven months (156 participants; see Analysis 7.2).

Impact on self‐reported criminal activity

Sundell 2008 found no significant reduction in arrests (RR 0.97, 95% CI 0.70 to 1.36), with low‐certainty evidence up to seven months (158 participants; see Analysis 7.1).

9. Multi‐systemic therapy (involving family) versus adolescent group substance abuse therapy

See summary of findings Table 9.

Impact on self‐reported drug use

This was not reported.

Impact on self‐reported criminal activity

Dakof 2015 reported no significant reduction in re‐arrests (MD ‐0.24, 95% CI ‐0.76 to 0.28), with low‐certainty evidence up to 24 months (112 participants; see Analysis 8.1).

10. Interpersonal psychotherapy versus a psychoeducational intervention

See summary of findings Table 10.

Impact on self‐reported drug use

Johnson 2012 reported no significant reduction in self‐reported drug use (RR 0.67, 95% CI 0.30 to 1.50), with very low‐certainty evidence up to three months (38 participants; see Analysis 9.1).

Impact on self‐reported criminal activity

This was not reported.

11. Legal defence service and wrap‐around social work services versus legal defence service only

See summary of findings Table 11.

Impact on self‐reported drug use

This was not reported.

Impact on self‐reported criminal activity

McCarter 2016 reported no significant reduction in the number of new offences committed (RR 0.64, 95% CI 0.07 to 6.01), with very low‐certainty evidence up to 12 months (29 participants; Analysis 10.1).

Discusión

available in

Resumen de los resultados principales

Esta revisión sistemática proporciona evidencia de 13 ensayos con 2606 participantes que evalúan 11 comparaciones diferentes y fue posible realizar un análisis agrupado. Por lo general, la certeza de la evidencia fue baja. La mayoría de las intervenciones se realizaron en centros penitenciarios (ocho estudios; 61%), en tribunales (dos estudios; 15%), en la comunidad (dos estudios; 15%) o en hospitales de seguridad media (un estudio; 8%). La mayoría de los estudios compararon una intervención versus tratamiento habitual u otra intervención (11/13 estudios; 84%).

Las 11 comparaciones de tratamientos diferentes se dividieron en:

comunidad terapéutica y atención posterior versus tratamiento habitual (Sacks 2004; Sacks 2011);
comunidad terapéutica y atención posterior versus un curso de habilidades cognitivo‐conductuales (Sacks 2008); y
comunidad terapéutica y atención posterior versus un control en lista de espera (Wexler 1999).

Dos estudios que compararon intervenciones terapéuticas comunitarias informaron una reducción significativa de la reincidencia y la actividad delictiva posterior en comparación con el tratamiento habitual (Sacks 2004; Sacks 2011), con evidencia de certeza moderada. Sacks 2008 adaptó un tratamiento terapéutico comunitario para mujeres delincuentes en comparación con un curso de habilidades cognitivo‐conductuales. Este estudio comparó a mujeres asignadas a tratamiento terapéutico comunitario o a tratamiento estándar versus un programa de recuperación cognitivo‐conductual y prevención de la recaída, conocido en el sistema como el "programa intensivo ambulatorio" (Sacks 2008), con evidencia de certeza baja. A los seis meses, los investigadores encontraron que ambos grupos mejoraron significativamente en variables de salud mental, consumo de sustancias, comportamiento criminal y riesgo de VIH. Los autores del estudio señalaron que se requiere una mayor exploración de cada modelo en los diferentes grupos de delincuentes a fin de permitir una utilidad más precisa de cada modelo. Concluyeron que estos hallazgos preliminares indican la importancia de proporcionar enfoques integrales y sensibles según el género dentro del sistema penitenciario para responder a las complejas necesidades de consumo de sustancias de las mujeres que delinquen (Sacks 2008). Se encontró que el tratamiento terapéutico comunitario es más beneficioso que la terapia cognitivo‐conductual, ya que prolonga el tiempo pasado en la comunidad antes de la reincidencia (Sacks 2008). Este hallazgo apoya parcialmente investigaciones anteriores que indican que la combinación del tratamiento terapéutico comunitario y la atención posterior a la liberación parece producir los resultados más consistentes y exitosos entre los delincuentes que no presentan trastornos mentales concomitantes (Mitchell 2012a). Aunque esto no se aborda en esta revisión, los clientes que permanecieron en tratamiento durante el período más prolongado parecieron ser los más beneficiados (Sacks 2004). Estas diferencias parecen confirmarse hasta por 36 meses en comparación con las personas que no recibieron nada, lo que indica que con el tiempo, el impacto de la intervención finalmente se disipó (Wexler 1999), con evidencia de certeza moderada. Solo uno de los cuatro estudios informó sobre los resultados del consumo de drogas (en mujeres) y no encontró reducciones después de la intervención en comparación con la asistencia a un curso de habilidades cognitivas (Sacks 2008). No se sabe si el consumo de drogas se reduce en los hombres con trastornos mentales concomitantes.

Tribunales de tratamiento de salud mental (TTSM) y uso de un modelo de gestión asertiva de los casos versus tratamiento habitual

Las personas bajo la custodia del sistema de justicia penal de ambos grupos mostraron mejoría en una variedad de resultados en la satisfacción con la vida, una disminución en los niveles de angustia y una mejoría en la vida independiente. En general, el patrón de actividad delictiva indicó que ambos grupos pasaron tiempo en la cárcel, pero por razones diferentes. Los individuos con TTSM tuvieron más probabilidades de ser "fichados" por un delito y no ser condenados y de tener más condenas por violación de la libertad condicional en comparación con los individuos que solo habían recibido tratamiento habitual. Las personas que recibieron tratamiento habitual fueron más propensas a ser condenadas por un nuevo delito (Cosden 2003), con evidencia de certeza baja.

Entrevistas motivacionales/conciencia plena y habilidades cognitivas versus control en lista de espera; entrevistas motivacionales/conciencia plena y habilidades cognitivas versus entrenamiento de relajación; entrevistas motivacionales/conciencia plena y habilidades cognitivas versus tratamiento habitual

Ver Cullen 2012, Lanza 2014, Malouf 2017 y Stein 2011.

Cuatro estudios de entrevistas motivacionales/conciencia plena y habilidades cognitivas comparadas con un control en lista de espera, entrenamiento de relajación y tratamiento habitual informaron evidencia de certeza moderada a muy baja. No se observaron diferencias significativas entre estos estudios, lo que indica que el uso de dichas habilidades puede no reducir el consumo posterior de drogas o la actividad delictiva en comparación con cualquiera de las alternativas. Además, uno de los cuatro estudios fue un ensayo controlado aleatorizado (ECA) piloto de entrevistas motivacionales versus tratamiento habitual, lo que indica que se necesitan estudios más grandes para apoyar cualquier hallazgo futuro. El uso de medidas autoinformadas, a menudo contaminadas por el sesgo de conveniencia social, significa que la confianza en estos resultados puede ser limitada (evidencia de certeza moderada; Malouf 2017).

Terapia multisistémica (TMS) que involucra a las familias versus tratamiento habitual; TMS que involucra a las familias versus terapia grupal para el consumo de sustancias

Dos estudios de terapia multisistémica para los jóvenes incluyeron las familias y la compararon con tratamiento habitual o un tratamiento alternativo grupal para el consumo de sustancias (Dakof 2015; Sundell 2008). Los hallazgos muestran que la TMS no apoyó la efectividad a corto plazo en relación con los servicios generalmente disponibles para los jóvenes con trastornos de conducta en Suecia (Sundell 2008). Este resultado es contrario a otros trabajos realizados en los Estados Unidos y Noruega, pero similar al trabajo realizado en Canadá (Cunningham 2002). Sundell 2008 destacó la importancia de medir y monitorizar la fidelidad durante la traspolación y la administración de las intervenciones a otros contextos y a diferentes países en todo el mundo; la importancia del impacto del contexto social no se debe subestimar (evidencia de certeza baja).

Psicoterapia interpersonal versus intervención psicoeducativa

Un estudio piloto de psicoterapia interpersonal en comparación con una intervención psicoeducativa no mostró una reducción significativa en el consumo posterior de drogas. Sin embargo, estos resultados se deben interpretar con cautela, debido a la pequeña muestra y el corto período de seguimiento (evidencia de certeza muy baja; Johnson 2012).

Trabajo de defensa legal y servicios sociales relacionados versus trabajo de defensa legal solo

Un estudio piloto del trabajo de defensa legal y de los servicios sociales en comparación con el trabajo de defensa legal con menores no mostró una reducción significativa en el retorno posterior a la cárcel en el período de seguimiento de 12 meses. Los autores del estudio argumentan que los servicios de representación global pueden proporcionar factores de protección que podrían afianzar los riesgos y las necesidades subyacentes de los jóvenes, lo que podría contribuir a una mayor intervención del tribunal o a la reincidencia en el futuro (evidencia de certeza muy baja; McCarter 2016).

Elementos exitosos del tratamiento y control de problemas complejos y concomitantes

En cuanto a abordar algunos de los problemas complejos de los individuos con trastornos mentales y consumo de sustancias concomitantes, la evidencia de esta revisión sistemática proporciona poca información.

A lo largo de estos ensayos se informaron varios elementos exitosos del tratamiento, y se pueden identificar varios temas clave.

En primer lugar, se observó que la cuestión de la participación en el tratamiento era importante. En el tribunal de salud mental, el apoyo informal de familiares y amigos estimuló la participación de los clientes dentro de la comunidad para obtener beneficios a largo plazo, pero se necesitan más estudios de investigación para evaluar si las intervenciones que empoderan a las familias pueden mejorar y mantener los resultados por más tiempo que las intervenciones que no se basan en la familia (Cosden 2003; Dakof 2015).

En segundo lugar, los programas que se adaptaban específicamente a las necesidades de los clientes de salud mental tendían a incluir un plan de estudios cognitivo‐conductual que hacía hincapié en el pensamiento y la conducta delictivos junto con las clases psicoeducativas. El propósito de combinar estos dos tipos de mecanismos es mejorar la capacidad del individuo de reconocer y comprender con mayor detalle su problema de consumo de sustancias y trastornos mentales (Sacks 2004).

Tercero, cuanto más tiempo un individuo se involucra en el tratamiento, mejor/es es/son el/los resultado/s (Wexler 1999).

Compleción y aplicabilidad general de las pruebas

Aplicabilidad general

La aplicabilidad de la evidencia se ve obstaculizada en general por la variedad de ensayos que tratan varias opciones de tratamiento diferentes, lo que hace inapropiado agrupar los resultados de los estudios. La mayoría de los juicios se realizaron dentro del sistema judicial de los Estados Unidos; por lo tanto, su generalización a los sistemas de justicia penal fuera de los Estados Unidos es limitada. Tres ensayos realizados en España (Lanza 2014), el Reino Unido (Cullen 2012) y Suecia proporcionan una perspectiva europea pero con evidencia de certeza moderada (Sundell 2008). Es por eso que los hallazgos de los estudios se deben interpretar con precaución.

Información sobre la salud mental

Aunque esta revisión buscó específicamente identificar los estudios que incluyeran participantes con trastornos mentales concomitantes, las descripciones de los estudios de la salud mental variaron (ver Tabla 1). Cosden 2003 utilizó a un psiquiatra o a un psicólogo para realizar una entrevista clínica para determinar un diagnóstico de salud mental junto con el consumo de sustancias. Esto resultó en una muestra del tribunal de salud mental de individuos diagnosticados con varios trastornos mentales, incluyendo trastorno del estado de ánimo, esquizofrenia, trastorno bipolar y diagnóstico dual. Otros documentos se refirieron al uso de los criterios diagnósticos del Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM‐IV), similares a la medida Youth Self‐Report (Dakof 2015; McCarter 2016; Sacks 2011; Sundell 2008), pero posteriormente proporcionaron poca información con respecto a las necesidades individuales de salud mental. La información demográfica de Sacks 2004 reveló otros aspectos del pronóstico de salud mental, incluyendo el tratamiento de por vida de la salud mental, la atención vitalicia al paciente y los medicamentos recetados.

La serie de estudios de Wexler de 1999 informó una variedad de diagnósticos, incluyendo trastorno de personalidad antisocial, fobias, trastorno de estrés postraumático, depresión, distimia y trastorno de déficit de atención, pero no describió cómo estos diagnósticos fueron confirmados o evaluados dentro de la población.

Seis de 13 ensayos informaron sobre cambios en el bienestar de la salud mental. Tres ensayos informaron el uso del Beck Depression Inventory, el Global Severity Index y la Posttraumatic Diagnostic Scale (Sacks 2004; Sacks 2008; Sacks 2011). Otro estudio informó sobre la depresión, pero utilizó la Hamilton Rating Scale for Depression (Johnson 2012). Cuatro estudios informaron la presencia de trastornos del estado de ánimo junto con esquizofrenia, trastorno general de ansiedad o trastorno de personalidad antisocial (Cosden 2003; Cullen 2012; Lanza 2014; Malouf 2017). Cuatro estudios analizaron los efectos diferenciales del tratamiento sobre la gravedad de la depresión (Cosden 2003; Johnson 2012; McCarter 2016; Stein 2011). Cosden 2003 señaló que es necesario comprender mejor cómo ayudar a los clientes con trastornos mentales a través de diferentes niveles de tratamiento. Johnson 2012 señaló que en los participantes que recibieron psicoterapia interpersonal se redujeron de manera significativa los niveles de depresión y de consumo de sustancias sobre los controles apareados por la atención. Los autores del estudio anotaron que la intensidad del tratamiento administrado una vez que el individuo es liberado en la comunidad es clave para mantener buenos resultados. Sin embargo, señalan que las personas bajo la custodia del sistema de justicia penal a menudo tienen retrasos en la provisión del tratamiento y los servicios en el momento de la puesta en libertad, e indican que los servicios alternativos, como el tratamiento telefónico, podrían ser útiles para proporcionar un tratamiento más intensivo y útil después de la puesta en libertad en tiempos de crisis.

Calidad de la evidencia

Ocho de 13 (62%) estudios se consideraron con riesgo de sesgo incierto en más de cuatro de ocho dominios. El principal factor limitante fue la falta de información de la evidencia, lo que impidió a los autores de la revisión emitir un juicio claro sobre el sesgo. Debido a que la imprecisión del informe disminuye la calidad de la evidencia, es muy probable que los estudios de investigación adicionales tengan una marcada repercusión sobre la confianza en la estimación del efecto y es probable que esta cambie. Además, se describieron varias limitaciones específicas relacionadas con el diseño de los estudios (que dan lugar a problemas de sesgo de selección), y los tamaños de las muestras fueron pequeños. Se observó que Stein 2011 y Cullen 2012 tuvieron relativamente poco poder estadístico. Es necesaria la replicación de estos estudios para mejorar la generalizabilidad y la validez externa de los hallazgos de los estudios.

Se informaron tamaños de muestra modestos similares, y algunos ensayos se definieron como estudios piloto (p.ej., Malouf 2017; McCarter 2016). Sacks 2011 and Cosden 2003 sugieren que se debe utilizar un tamaño de muestra más grande para proporcionar estimaciones del efecto más precisas. Los tamaños de muestra pequeños limitan la generalizabilidad de la población de la muestra a otros contextos, y pocos estudios recopilaron datos longitudinales suficientes para apoyar el uso continuo de tales esquemas sin que se encargaran ensayos adicionales más grandes (Cullen 2012; Dakof 2015; McCarter 2016). Cosden 2003 también informó sobre la posibilidad de sesgo de resultado, ya que el entrevistador no estaba ciego a la condición de resultado del participante ni a la pérdida durante el seguimiento (el 25% de la muestra del estudio se perdió durante el seguimiento) a los 12 meses.

Otra posible preocupación con respecto al sesgo de selección en la serie de estudios de Wexler fue que los participantes fueron asignados al azar a tratamiento terapéutico comunitario en prisión y a condiciones habituales de prisión, pero no a atención posterior (Prendergast 2003; Prendergast 2004; Wexler 1999). Los autores del estudio señalaron que las posibles diferencias en la motivación personal podrían explicar algunos de los resultados positivos asociados con el apoyo continuado a los participantes por los servicios de atención posterior. Posteriormente, se observó que estos participantes tenían las "puntuaciones de disposición" más altas, lo que indica que la motivación es una consideración importante para la selección de los clientes (Wexler 1999). Cullen 2012 informó sobre el uso de la asignación al azar dentro de los sitios, lo que puede haber dado lugar a la contaminación entre los grupos de tratamiento, y a la probabilidad de que haya surgido un sesgo de selección adicional debido al hecho de que los pacientes que declinaron participar estaban más enfermos o eran más antisociales, y que estos factores podrían influir en los resultados del tratamiento (Cullen 2012).

En general, se consideró que la certeza de la evidencia fue moderada a muy baja para todas las intervenciones incluidas.

Sesgos potenciales en el proceso de revisión

Además de las limitaciones ya analizadas, el método de búsqueda se limitó a las bases de datos a las que se podía acceder a través de la Universidad de York, y no se realizaron búsquedas exhaustivas en sitios web. No se realizaron búsquedas en registros de ensayos específicos. Como resultado, es posible que en esta versión actualizada se haya omitido parte de la bibliografía.

Figure 1

Study flow diagram.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Analysis 1.1

Comparison 1 Therapeutic community and aftercare vs treatment as usual, Outcome 1 Criminal activity.

Analysis 2.1

Comparison 2 Therapeutic community and aftercare vs cognitive‐behavioural therapy, Outcome 1 Self‐reported drug use at 6 months.

Analysis 2.2

Comparison 2 Therapeutic community and aftercare vs cognitive‐behavioural therapy, Outcome 2 Arrested any for 6 months.

Analysis 2.3

Comparison 2 Therapeutic community and aftercare vs cognitive‐behavioural therapy, Outcome 3 Criminal activity at 6 months.

Analysis 2.4

Comparison 2 Therapeutic community and aftercare vs cognitive‐behavioural therapy, Outcome 4 Drug‐related crime.

Analysis 3.1

Comparison 3 Therapeutic community vs waiting list control, Outcome 1 Re‐incarceration at 36 months.

Analysis 4.1

Comparison 4 Mental health treatment court with assertive case management vs treatment as usual, Outcome 1 Committing a new crime.

Analysis 4.2

Comparison 4 Mental health treatment court with assertive case management vs treatment as usual, Outcome 2 Re‐incarceration to jail at 12 months.

Analysis 4.3

Comparison 4 Mental health treatment court with assertive case management vs treatment as usual, Outcome 3 ASI drug use at 12 months.

Analysis 5.1

Comparison 5 Motivational interviewing and cognitive skills vs waiting list control, Outcome 1 Abstinence from drug use at 6 months.

Analysis 5.2

Comparison 5 Motivational interviewing and cognitive skills vs waiting list control, Outcome 2 ASI drug score at 6 months.

Analysis 6.1

Comparison 6 Motivational interviewing and cognitive skills vs treatment as usual, Outcome 1 Marjuana frequency at 3 months.

Analysis 6.2

Comparison 6 Motivational interviewing and cognitive skills vs treatment as usual, Outcome 2 Arrest frequency 3 years post release.

Analysis 6.3

Comparison 6 Motivational interviewing and cognitive skills vs treatment as usual, Outcome 3 Time to first arrest or offence 36 months post.

Analysis 6.4

Comparison 6 Motivational interviewing and cognitive skills vs treatment as usual, Outcome 4 Positive drug screen at 12 months.

Analysis 7.1

Comparison 7 Multi‐systemic therapy vs treatment as usual, Outcome 1 Arrest by police.

Analysis 7.2

Comparison 7 Multi‐systemic therapy vs treatment as usual, Outcome 2 DUDIT scores.

Analysis 8.1

Comparison 8 Multi‐systemic therapy vs adolescent substance treatment group, Outcome 1 Arrests between 6 and 24 months.

Analysis 9.1

Comparison 9 Interpersonal psychotherapy vs psychoeducational controls, Outcome 1 Substance abuse relapse at 3 months.

Analysis 10.1

Comparison 10 Legal defence services and wrap‐around social work services vs legal defence work only, Outcome 1 Number of new offences committed at 12 months.

Summary of findings for the main comparison. Therapeutic community compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Therapeutic community compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with therapeutic community
Re‐arrests assessed with official records Follow‐up: 12 months	266 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	RR 0.67 (0.53 to 0.84)	Study population
				98 per 100	32 fewer per 100 (46 fewer to 16 fewer)
Re‐incarceration assessed with official records Follow‐up: 12 months	266 (2 RCTs)	⊕⊕⊕⊝ MODERATE^a	RR 0.40 (0.24 to 0.67)	Study population
				59 per 100	36 fewer per 100 (45 fewer to 20 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (blinding and selective reporting).

Summary of findings for the main comparison. Therapeutic community compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Summary of findings 2. Therapeutic community and aftercare compared to cognitive behavioural skills for drug using women offenders with co‐occurring mental illness

Therapeutic community and aftercare compared to cognitive‐behavioural skills for drug‐using women offenders with co‐occurring mental health problems
Patient or population: drug‐using women offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community and aftercare Comparison: cognitive‐behavioural skills
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with cognitive‐behavioural skills	Risk difference with therapeutic community and aftercare
Self‐reported drug use Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.78 (0.46 to 1.32)	Study population
				17 per 100	4 fewer per 100 (9 fewer to 6 more)
Re‐arrest for any type of crime assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.69 (0.44 to 1.09)	Study population
				33 per 100	10 fewer per 100 (19 fewer to 3 more)
Criminal Activity assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.74 (0.52 to 1.05)	Study population
				33 per 100	9 fewer per 100 (16 fewer to 2 more)
Drug‐related crime assessed with Colorado Department of Corrections Record Information System (CDOC‐RIS) Follow‐up: 6 months	314 (1 RCT)	⊕⊝⊝⊝ LOW^a,b	RR 0.87 (0.56 to 1.36)	Study population
				21 per 100	3 fewer per 100 (9 fewer to 8 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aUnclear reporting in the paper raises concerns about the potential high risk of bias with regards to blinding and methods used in the randomisation procedure; we downgraded by one. ^bOne study with 95% confidence intervals through the line of no effect.

Summary of findings 2. Therapeutic community and aftercare compared to cognitive behavioural skills for drug using women offenders with co‐occurring mental illness

Summary of findings 3. Therapeutic community compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Therapeutic community compared to waiting list control for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: therapeutic community Comparison: waiting list control
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with waiting list control	Risk difference with therapeutic community
Return to prison (recidivism) post parole assessed with California Department of Correction's computerised Offender Based Information System Follow‐up: 36 months	478 (1 RCT)	⊕⊕⊕⊝ MODERATE^a	RR 0.60 (0.46 to 0.79)	Study population
				40 per 100	16 fewer per 100 (21 fewer to 8 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (randomisation process, concealment, and selective reporting).

Summary of findings 3. Therapeutic community compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Summary of findings 4. Mental health treatment court with assertive case management model compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Mental health treatment court with assertive case management model compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: court Intervention: mental health treatment court with assertive case management model Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with mental health treatment court with assertive case management model
Conviction for a new crime assessed with data from probation office Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	RR 1.05 (0.90 to 1.22)	Study population
				72 per 100	4 more per 100 (7 fewer to 16 more)
Re‐incarceration to jail assessed with data from probation office Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	RR 0.79 (0.62 to 1.01)	Study population
				71 per 100	15 fewer per 100 (27 fewer to 1 more)
Self‐reported drug use assessed with Addiction Severity Index (ASI) Follow‐up: 12 months	235 (1 RCT)	⊕⊝⊝⊝ LOW^a	‐	Mean self‐reported drug use was 0.08	MD 0.00 (‐0.03 lower to 0.03 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (allocation concealment and blinding of assessors) and by one for imprecision.

Summary of findings 4. Mental health treatment court with assertive case management model compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Summary of findings 5. Motivational interviewing/mindfulness and cognitive skills compared to relaxation training for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to relaxation training for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: motivational interviewing and cognitive skills Comparison: relaxation training
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Impact
Self‐reported marijuana use continuous	181 (1 RCT)	MODERATE^a	This study compared cognitive skills to a relaxation training intervention for adolescents in prison with depressed mood. Researchers measured marijuana use at 3‐months follow‐up assessment using the Risks and Consequences Questionnaire (RCQ). They report a main effect < .007, with participants in the motivational interviewing group showing fewer problems than participants in the relaxation training group.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for unclear risk of bias (random allocation and blinding).

Summary of findings 5. Motivational interviewing/mindfulness and cognitive skills compared to relaxation training for drug‐using offenders with co‐occurring mental illness

Summary of findings 6. Motivational interviewing/mindfulness and cognitive skills compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to waiting list control for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: motivational interviewing and cognitive skills Comparison: waiting list control
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with waiting list control	Risk difference with motivational interviewing and cognitive skills
Self‐reported drug use assessed with Addiction Severity Index (ASI) composite drug score across 13 items of drug use in the last 30 days Follow‐up: 6 months	31 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean self‐reported drug use was 0.44	MD ‐0.04 lower (‐0.37 lower to 0.29 higher)
Abstinence from drug use Follow‐up: 6 months	31 (1 RCT)	⊕⊕⊝⊝ LOW^a	RR 2.89 (0.73 to 11.43)	Study population
				15 per 100	29 more per 100 (4 fewer to 160 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal information size not met.

Summary of findings 6. Motivational interviewing/mindfulness and cognitive skills compared to waiting list control for drug‐using offenders with co‐occurring mental illness

Summary of findings 7. Motivational interviewing/mindfulness and cognitive skills compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Motivational interviewing/mindfulness and cognitive skills compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: medium secure hospital and jail Intervention: motivational interviewing and cognitive skills Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	Risk with treatment as usual	Risk difference with motivational interviewing and cognitive skills
Self‐reported frequency of marijuana use assessed with TCU‐CRTF (Texas Christian University: Correctional Residential Treatment Form) Scale from 0 to 32 Follow‐up: 3 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean self‐reported frequency of marijuana use was 1.50	MD ‐1.05 lower (‐2.39 lower to 0.29 higher)
Arrest frequency post release assessed with official police records Follow‐up: 36 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean arrest frequency post release was 1.47	MD ‐0.66 lower (‐1.31 lower to ‐0.01 lower)
Time to first arrest or offence assessed with official police records Follow‐up: 36 months	40 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	‐	Mean time to first arrest or offence was 1.6	MD 0.87 higher (‐0.12 lower to 1.86 higher)
Positive drug screen or refusal to provide a urine sample assessed with urine sample Scale from negative to positive Follow‐up: 12 months	84 (1 RCT)	⊕⊝⊝⊝ VERY LOW^a,b	‐	Mean positive drug screen or refusal to provide a urine sample was 3.25	MD ‐0.7 lower (‐3.5 lower to 2.1 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded by one for risk of bias (incomplete outcome measures).

Summary of findings 7. Motivational interviewing/mindfulness and cognitive skills compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Summary of findings 8. Multi‐systemic therapy involving family and juveniles compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Multi‐systemic therapy involving family compared to treatment as usual for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: community Intervention: multi‐systemic therapy involving family Comparison: treatment as usual
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with treatment as usual	Risk difference with multi‐systemic therapy involving family
Drug dependence assessed with DUDIT questionnaire Scale from 0 to 44 Follow‐up: 7 months	156 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean drug dependence was 3.55	MD ‐0.22 lower (‐2.51 lower to 2.07 higher)
Arrested assessed by corroborating with police data Follow‐up: 7 months	158 (1 RCT)	⊕⊕⊝⊝ LOW^a	RR 0.97 (0.70 to 1.36)	Study population
				47 per 100	1 fewer per 100 (14 fewer to 17 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (blinding measures) and downgraded by one for imprecision.

Summary of findings 8. Multi‐systemic therapy involving family and juveniles compared to treatment as usual for drug‐using offenders with co‐occurring mental illness

Summary of findings 9. Multi‐systemic therapy involving family compared to group substance abuse therapy for drug‐using adolescents with co‐occurring mental illness

Multi‐systemic therapy involving family compared to group substance abuse therapy for drug‐using adolescents with co‐occurring mental health problems
Patient or population: drug‐using adolescents with co‐occurring mental health problems Setting: court Intervention: multi‐systemic therapy involving family Comparison: group substance abuse therapy
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with group substance abuse therapy	Risk difference with multi‐systemic therapy involving family
Arrests Follow‐up: range 6 months to 24 months	112 (1 RCT)	⊕⊕⊝⊝ LOW^a	‐	Mean arrests were 1.19 SD	MD ‐0.24 SD lower (‐0.76 lower to 0.28 higher)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; SD: standard deviation.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by one for risk of bias (selective reporting of outcomes) and by one for imprecision.

Summary of findings 9. Multi‐systemic therapy involving family compared to group substance abuse therapy for drug‐using adolescents with co‐occurring mental illness

Summary of findings 10. Interpersonal psychotherapy compared to a psychoeducational intervention for drug‐using offenders with co‐occurring mental illness

Interpersonal psychotherapy compared to a psychoeducational intervention for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: prison Intervention: interpersonal psychotherapy Comparison: psychoeducational intervention
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with a psychoeducational intervention	Risk difference with interpersonal psychotherapy
Substance abuse relapse post release Follow‐up: 3 months	38 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	RR 0.67 (0.30 to 1.50)	Study population
				47 per 100	16 fewer per 100 (33 fewer to 24 more)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded by one for risk of bias (selective reporting outcomes).

Summary of findings 10. Interpersonal psychotherapy compared to a psychoeducational intervention for drug‐using offenders with co‐occurring mental illness

Summary of findings 11. Legal defence service and wrap‐around social work services compared to legal defence service only for drug‐using offenders with co‐occurring mental illness

Legal defence service and wrap‐around social work services compared to legal defence service only for drug‐using offenders with co‐occurring mental health problems
Patient or population: drug‐using offenders with co‐occurring mental health problems Setting: court Intervention: legal defence service and wrap‐around social work services Comparison: legal defence service only
Outcomes	№ of participants (studies) Follow‐up	Certainty of the evidence (GRADE)	Relative effect (95% CI)	*Anticipated absolute effects (95% CI)**
				Risk with legal defence services only	Risk difference with legal defence services and wrap‐around social work services
Committing new offences Follow‐up: 12 months	29 (1 RCT)	⊕⊕⊝⊝ VERY LOW^a,b	RR 0.64 (0.07 to 6.01)	Study population
				1 per 100	2 fewer per 100 (0 fewer to 2 fewer)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded by two for optimal size not met. ^bDowngraded for risk of bias (incomplete outcome data).

Summary of findings 11. Legal defence service and wrap‐around social work services compared to legal defence service only for drug‐using offenders with co‐occurring mental illness

Table 1. Mental health diagnoses

Study, year	Criteria used for diagnoses	Description of mental health problem
Cosden 2003	Determined by a psychiatrist/psychologist on the basis of a clinical interview and observations	Mood disorder Schizophrenia Bipolar disorder Other Dual diagnosis
Cullen 2012	Primary clinical diagnosis of a psychotic disorder. Diagnosis mechanism not reported	Schizophrenia Schizoaffective disorder Bipolar disorder Other psychotic disorder
Dakof 2015	Diagnostic Interview Schedule for Children (DISC‐2) ‐ identifying presence of mental disorders according to the DSM‐III Youth Self‐Report	Presence of mental health disorders Externalising subscales
Johnson 2012	Hamilton Rating Scale for Depression Median duration of index episode in months Number of depressive episodes Number of previous suicide attempts DSM‐IV Axis I disorders using the SCID‐I/II	Criteria for a major depressive disorder at least 4 weeks after substance abuse treatment Minimum score of 18 on the Hamilton Rating Scale for Depression
Lanza 2014	DSM‐IV Mini International Neuropsychiatric Interview Anxiety Sensitivity Index	Anxiety Mental health disorders Antisocial personality disorder Major depressive disorder Generalised anxiety disorder
Malouf 2017	Borderline Personality Disorder Features assessed with the Personality Assessment Inventory	Affective instability Identity problems Negative relationships Impulsivity
McCarter 2016	Youth Self‐Report that contain scales orientated to the DSM‐IV	Somatic complaints Anxiety and depression Social problems Internalising and externalising (thought and attention problems)
Sacks 2004	DIS	Diagnosis of lifetime Axis I or Axis II mental disorder Antisocial personality disorder
Sacks 2008	Global Severity Index Beck Depression Inventory Lifetime of mental health PTSD Symptom Scale ‐ Interview Posttraumatic Stress Diagnostic Scale	Depression PTSD Lifetime of mental health
Sacks 2011	DSM‐IV diagnostic criteria Beck Depression Inventory Post Traumatic Stress Disorder Symptom Scale Brief Symptom Inventory Global Severity Index	Depression PTSD Psychological distress
Stein 2011	CES‐D Scale	Scores > 16 indicate presence of significant depression; 69.8% had significant depressive symptoms
Sundell 2008	DSM‐IV diagnostic criteria Youth Self‐Report	Conduct disorder Internalising and externalising Total behaviour problems
Wexler 1999; Prendergast 2003; Prendergast 2004	Not specified	Antisocial personality disorder Phobias PTSD Depression Dysthymia Attention deficit hyperactivity disorder
CES‐D: Center for Epidemiological Studies ‐ Depression; DIS: Diagnostic Interview Schedule; DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; PTSD: post‐traumatic stress disorder; SCID: Structured Clinical Interview for DSM Disorders.

Table 1. Mental health diagnoses

Comparison 1. Therapeutic community and aftercare vs treatment as usual

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Criminal activity Show forest plot	2	532	Risk Ratio (M‐H, Random, 95% CI)	0.57 [0.42, 0.77]

1.1 Any criminal activity	2	266	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.53, 0.84]
1.2 Re‐incarceration	2	266	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.24, 0.67]

Comparison 1. Therapeutic community and aftercare vs treatment as usual

Comparison 2. Therapeutic community and aftercare vs cognitive‐behavioural therapy

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Self‐reported drug use at 6 months Show forest plot	1	314	Risk Ratio (M‐H, Random, 95% CI)	0.78 [0.46, 1.32]

2 Arrested any for 6 months Show forest plot	1	314	Risk Ratio (M‐H, Random, 95% CI)	0.69 [0.44, 1.09]

3 Criminal activity at 6 months Show forest plot	1	314	Risk Ratio (M‐H, Random, 95% CI)	0.74 [0.52, 1.05]

4 Drug‐related crime Show forest plot	1	314	Risk Ratio (M‐H, Random, 95% CI)	0.87 [0.56, 1.36]

Comparison 2. Therapeutic community and aftercare vs cognitive‐behavioural therapy

Comparison 3. Therapeutic community vs waiting list control

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Re‐incarceration at 36 months Show forest plot	1	478	Risk Ratio (M‐H, Fixed, 95% CI)	0.60 [0.46, 0.79]

Comparison 3. Therapeutic community vs waiting list control

Comparison 4. Mental health treatment court with assertive case management vs treatment as usual

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Committing a new crime Show forest plot	1	235	Risk Ratio (M‐H, Fixed, 95% CI)	0.79 [0.62, 1.01]

2 Re‐incarceration to jail at 12 months Show forest plot	1	235	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.91, 1.24]

3 ASI drug use at 12 months Show forest plot	1	235	Mean Difference (IV, Fixed, 95% CI)	0.0 [‐0.03, 0.03]

Comparison 4. Mental health treatment court with assertive case management vs treatment as usual

Comparison 5. Motivational interviewing and cognitive skills vs waiting list control

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Abstinence from drug use at 6 months Show forest plot	1	31	Risk Ratio (M‐H, Random, 95% CI)	2.89 [0.73, 11.43]

2 ASI drug score at 6 months Show forest plot	1	31	Mean Difference (IV, Random, 95% CI)	‐0.04 [‐0.37, 0.29]

Comparison 5. Motivational interviewing and cognitive skills vs waiting list control

Comparison 6. Motivational interviewing and cognitive skills vs treatment as usual

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Marjuana frequency at 3 months Show forest plot	1	40	Mean Difference (IV, Random, 95% CI)	‐1.05 [‐2.39, 0.29]

2 Arrest frequency 3 years post release Show forest plot	1	31	Mean Difference (IV, Random, 95% CI)	‐0.66 [‐1.31, ‐0.01]

3 Time to first arrest or offence 36 months post Show forest plot	1	31	Mean Difference (IV, Fixed, 95% CI)	0.87 [‐0.12, 1.86]

4 Positive drug screen at 12 months Show forest plot	1	84	Mean Difference (IV, Fixed, 95% CI)	‐0.70 [‐3.50, 2.10]

Comparison 6. Motivational interviewing and cognitive skills vs treatment as usual

Comparison 7. Multi‐systemic therapy vs treatment as usual

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Arrest by police Show forest plot	1	156	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.70, 1.36]

2 DUDIT scores Show forest plot	1	156	Mean Difference (IV, Random, 95% CI)	‐0.22 [‐2.51, 2.07]

Comparison 7. Multi‐systemic therapy vs treatment as usual

Comparison 8. Multi‐systemic therapy vs adolescent substance treatment group

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Arrests between 6 and 24 months Show forest plot	1	112	Mean Difference (IV, Fixed, 95% CI)	‐0.24 [‐0.76, 0.28]

Comparison 8. Multi‐systemic therapy vs adolescent substance treatment group

Comparison 9. Interpersonal psychotherapy vs psychoeducational controls

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Substance abuse relapse at 3 months Show forest plot	1	38	Risk Ratio (M‐H, Fixed, 95% CI)	0.67 [0.30, 1.50]

Comparison 9. Interpersonal psychotherapy vs psychoeducational controls

Comparison 10. Legal defence services and wrap‐around social work services vs legal defence work only

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of new offences committed at 12 months Show forest plot	1	29	Risk Ratio (M‐H, Random, 95% CI)	0.64 [0.07, 6.01]

Comparison 10. Legal defence services and wrap‐around social work services vs legal defence work only