Beutler 2003.

Methods	Design: randomised controlled trial Follow‐up: 20 weeks post‐treatment and 6 months post‐treatment follow‐up. Setting: outpatients recruited from community and advertising Country: USA
Participants	Participants: 40 patients with comorbid DSM‐IV depression disorder and stimulant (cocaine or methamphetamine) dependence on the Structured Clinical Interview (SCID) Mean age (years): 33.06 (SD = 8.67) Sex: 57% male Ethnicity: 25% minority group
Interventions	Description: Cognitive Therapy (CT) for drug abuse (Beck 1993), Narrative Therapy (NT; Scogin 1987) versus Prescriptive Therapy (PT). CT was therapist‐guided, symptom‐focused and emotionally supportive. NT was patient‐led, non‐confrontational and insight‐focused; outside self‐help groups (e.g. AA, NA) were recommended as part of this treatment. PT targeted the patient‐treatment fit between four patient qualities (functional impairment, coping style, resistance traits, subjective distress) and four treatment variables (intensity, focus, directiveness, affective regulation). Format:individual therapy Duration: CT: 14‐18 weeks (16‐20 sessions); NT: 14‐18 weeks (patient‐selected frequency; weekly recommended); PT: 14‐18 weeks (16‐20 sessions). Allocation: CT n = 15; NT n = 12; PT n = 13
Outcomes	Treatment retention: dropout defined as missing 3 consecutive sessions; recoded as treatment retention for review Self‐report: BDI mean (SD) total score (past 7 days); TLFB mean days/week (past 30 days; modified self‐report version) Interviewer‐rated: HDRS mean (SD) total score (past 7 days), ASI mean (SD) alcohol and drug index scores (past 30 days) Attrition: proportion with missing data at post‐treatment and 6‐month follow‐up HDRS data (overall attrition rate of each treatment group not reported)
Notes	Therapists carefully selected and trained Funding: National Institute of Drug Abuse (NIDA) RO1DA09294 (Beutler) Conflict of interest: no statement provided.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (via author email): "Assignment one of the three groups was accomplished by drawing poker chips representing the three groups from a hat. The drawing for therapists within groups was accomplished by poker chips on which were written a therapist number that had been pre‐assigned." Comment: participants were randomly assigned to a treatment group and therapist.
Allocation concealment (selection bias)	Low risk	Quote (via author email): "Each participants group assignment was logged into a computer file and stored within a password protected file." Comment: investigators enrolling participants could not foresee assignment because group assignment was concealed in a password‐protected file.
Blinding of participants and personnel (performance bias) subjective outcomes	High risk	Quote (via author email): "Participants ‐ We cautioned therapists and staff to avoid telling patients which group to which they were assigned. Groups were referred to by numbers indicating group and therapist (e.g., group #2, therapist #5). At the end of treatment we checked to see if patient's knew which group they had been assigned to." Comment: blinding of study participants was attempted, the success of this was checked, but not reported and blinding could have been broken and influenced outcomes. Three separate sets of therapists were used to deliver each treatment to reduce treatment allegiance/contamination effects. Quote: "Therapist belief in and satisfaction with the treatment model selected was routinely and formally assessed (p. 72); all had high levels of therapist commitment; PT therapists thought it would more acceptable to their peers. Nevertheless, the authors were unable to blind therapists to the psychological treatment they were delivering, which could have influenced outcomes."
Blinding of participants and personnel (performance bias) objective outcomes	High risk	Participant and providers knowledge of group allocation may influence treatment engagement/participation/ attendance
Blinding of outcome assessment (detection bias) self‐report	Unclear risk	Blinding of study participants was attempted, but success of blinding was not reported. Comment: The authors judge risk of bias to be unclear. While participants were not blind to treatment allocation and self‐report measures may be impacted by self‐presentation bias or client insight, it is unlikely that any such risk of bias will vary by treatment condition.
Blinding of outcome assessment (detection bias) interviewer‐rated outcomes	Low risk	Quote (via author email): "Blind ratings were made by clinicians who were unfamiliar with the particular patients."
Blinding of outcome assessment (detection bias) treatment retention/attendance	Low risk	Blinding unclear, but outcome measure is unlikely to be influenced by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	High risk	Treatment retention: 48% of participants; Quote: "Treatments differed significantly in rate of drop out (Chi2 = 9.70; P < 0.05), favouring lower dropout in the NT group (p. 77) Comment: low treatment retention rate (< 50%), between group differences in treatment dropout rate Outcome assessment: attrition rate in each treatment group using HDRS data Attrition post‐treatment: Total 50%; CT 47%; NT 42%; PT 62% Attrition at the 6 month follow‐up: Total 53%; CT 47%; NT 58%; PT 54% No reasons for missing data reported. Comment: high level of attrition (> 50%) at post‐treatment and 6‐month follow‐up unlikely to be related to true outcome, even though the attrition rates across the three treatment groups were similar.
Selective reporting (reporting bias)	Unclear risk	Quote: "While the study was constructed as a RCT in which the three groups were compared, the usual methods of analysing RCT designs using end‐point or intent‐to‐treat analysis are inappropriate for detecting multiple contributors to change. (p. 70) ....The current study afforded the opportunity of comparing the relative yield of traditional intent‐to‐treat comparisons with the more flexible and inclusive Hierarchical Liner Methods (p. 70)" Comment: risk is unclear as no study protocol is available. All expected outcomes specified in the method are published in the paper, but end‐point analysis results are only reported at post‐treatment, and not at 6‐month follow‐up.
Other bias	High risk	Fidelity: fidelity ratings of videotapes of early and late sessions conducted by supervisors not independent raters. Other treatments: unclear whether participants were able to receive concurrent out‐of‐study psychotherapy. Reported that antidepressants were prescribed by project psychiatrist; however, the type of medication, dose, and consistency of medication adherence between group conditions is not reported.