Miklowitz‐USA.

Methods	Allocation: randomised Blinding: single (assessors) Setting: Emory University, Harvard University, University of Calgary, University of California Los Angeles, University of California San Diego, University of North Carolina, Yale University and Zucker Hillside Hospital, USA Inclusion criteria: age 12‐35 years, speaking and writing English, meet SIPS/SOPS criteria Exclusion criteria: diagnosis of schizophrenia or schizoaffective disorder (DSM‐ IV‐TR), pervasive developmental disorders, current substance or alcohol dependence, neurological disorders Duration: 18 months (6 months of treatment, 12 months of follow‐up)
Participants	Diagnosis: high risk for developing psychosis N = 129 Sex: men and women, ˜60:40% M:F Age: 12‐35 years, average 17 SD 4 History: no details
Interventions	1. FFT: 18 sessions of psychoeducation, communication enhancement training and problem‐solving skills training in 6 months, average 11 sessions SD 7: N = 66 2. Enhanced care: 3‐session family psychoeducational therapy, average 2.4 sessions SD 1.2: N = 63 Drug treatment not requirement of study. When participants were taking medications, their pharmacotherapy was managed by a study psychiatrist, unless they wished to consult a community provider. 27 (20.9%) were taking antipsychotic medications at randomisation
Outcomes	Transition to psychosis Leaving the study early Mental state: SOPS (positive), at 6 months post‐treatment Prescription of antipsychotics, by 6 months Unable to use: Mental state: SOPS (negative symptoms) (no usable data), SOPS – at 1 year (no data) Functioning: GAF, GFR, GFS (no usable data) Additional outcomes: Family interactions (e.g. perceived criticism): PCPW, CBQ‐ mother report, 10‐min problem‐solving family interaction task
Notes	Funding: National Institute of Mental Health (NIMH) grants 1RC1MH088546 (TDC, DJM), and R01MH093676 (DJM), and a grant from the Stanley Family Foundation (TDC). Power, sample size calculation: quote: "Power for the study’s repeated measure design, calculated prior to the study based on an expectation of 120 participants and 20% attrition, was 80% to detect a medium‐sized (0.50 SD) group difference in symptoms (alpha = 0.05, two‐tailed). Our study design had 95% power to detect a three‐way interaction between treatment, age group, and time with a medium effect size (f = 0.25) (p<0.05)." Adherence: see Table 14
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Lead study investigator who was neither involved in the provision of treatments nor the follow‐up evaluations conducted the random assignments to groups, with 50% of participants allocated to each condition, allocations, performed using Efron’s biased coin toss were stratified by site and whether or not the participant was prescribed an antipsychotic medication at baseline, allocation results were sent by email to each site’s principal investigator.
Allocation concealment (selection bias)	Unclear risk	See above
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants not blinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Although this was an RCT with ‘blind’ evaluations of clinical outcome, the clinical supervisors knew whether they were rating FFT or enhanced care sessions
Incomplete outcome data (attrition bias) All outcomes	Low risk	Out of 129 randomised participants, 27 discontinued the study before 6‐month assessment (attrition 21%). In the FFT group attrition rate was 17% (N = 11) and in the enhanced care group 25% (N = 16). The reasons for dropping out of the study were classified as “withdrew or missed assessment“ in both groups. In the FFT group, 11 participants withdrew prior to first session and in the enhanced care group 10 withdrew prior to first session.
Selective reporting (reporting bias)	Unclear risk	All outcomes mentioned in registered protocol (NCT01907282) and publications' methods reported for post‐treatment point (6 months), but some of the data were only partially reported, unusable figures or reported per age groups, and not per randomised groups. Data for 1‐year follow‐up not reported
Other bias	Low risk	We did not identify any other sources of bias.