McFarlane 1996.

Methods	Allocation: randomised Design: multi centre Duration: 24 months Country: New York State, USA
Participants	N = 68 Diagnosis: DSM‐III‐R criteria for schizophrenia, schizoaffective disorder or schizophreniform disorder with also one or more additional complicating factors. Setting: the study was conducted at 3 mental health centres, 1 each in an urban, suburban, and rural area of New York State. Participants were selected during an admission to an inpatient service or an acute partial hospital or when they were receiving crisis services for an acute psychotic episode. In total 28% of the participants was in hospital when they entered the study. Age: 18‐45 years, mean 29.8 years Gender: 65% male Ethnicity: 78% white Substance abuse: 80% no history of abuse Living situation: 47% lives with family, 25% supervised without family, 26% unsupervised without family, Marital status: 84%never married, 9% divorced/separated Employment status: 91.2% unemployed, 5.9% sheltered work, 2.9% competitive employed Working history:‐ Motivation:‐ Education: 30% high school graduate, 29% some college, 22% some high school Disability benefit: ‐ Excluded: acutely violent or suicidal people and those with major medical illness or physical addiction requiring immediate medical hospitalisation were excluded until they were stabilised or detoxified.
Interventions	ACT + multifamily group (N = 37) After an initial psycho‐educational workshop for family members only, multifamily groups, each comprising 6 participants and their families, met with two ACT team members every other week for 2 years. The teams were guided by participants’ and family members’ preferences and intentions. One modification was made in the ACT approach: a more gradually paced recovery and rehabilitation concept was adopted from family psychoeducation to ensure the lowest risk of relapse and more consistent development of independent living skills ACT + crisis family intervention (N = 31) Co‐ordination between the team and family members occurred only during crises, without the input of other participants’ family members
Outcomes	Number of participants who obtained competitive employment Number of participants who obtained non‐competitive employment Mental health (PANSS)
Notes	We could not classify this comparison for the network meta‐analyses and direct comparison meta‐analyses No definition of competitive employment described, but they made a distinction between sheltered and competitive employment
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Participants and their families were randomly assigned
Allocation concealment (selection bias)	Unclear risk	"Post‐treatment interviews with the project staff failed to detect any bias in assigning participants to treatment conditions"
Blinding of participants and personnel (performance bias) All outcomes	High risk	Personnel and participants were not blinded. They could identify assignment by contents of programme
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"Raters were independent and blind to the treatment condition"
Incomplete outcome data (attrition bias) All outcomes	Low risk	No differences in attrition: three cases in each cohort (6/68)
Selective reporting (reporting bias)	Low risk	All listed outcomes were reported
Other bias	High risk	New York State Alliance for the Mentally Ill, which sponsored the project, and the New York State Office of Mental Health, which provided funding Baseline differences were considered chance effects and were entered as control variables in subsequent analyses. Participants in the intervention condition received care from the same treatment teams that provided services to participants in the control condition. The possibility of clinician bias existed.