NAOMI 2009.

Methods	Allocation: "A computer‐generated randomisation list of permuted blocks of two, four, and six was used. Patients were assigned to receive diacetylmorphine, methadone, or hydromorphone in a 45:45:10 ratio. Randomization was stratified according to centre and according to the number of previous methadone treatments (two or fewer vs. three or more)." Blindness of patients and or care providers in respect to treatment:"The investigators and participants were aware of whether the assigned study drug was oral methadone or one of the injectable drugs, but diacetylmorphine and hydromorphone were administered in a double‐blind fashion". "Evaluations were performed at a separate research office that operated independently from the treatment clinic in each city" there is no mention whether the assessors were blinded to treatment allocation. Duration of treatment within the study: 12 months
Participants	Opiate‐dependent using injected heroin on regular basis, not responding in the past or currently in MMT. Diagnosis: opioid dependence (meeting three or more of seven criteria listed in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition,14 including tolerance or withdrawal) N= 251 Age=25 years or older Sex= Male154 (61.4) History= duration of injecting drug use yr 16.5±9.8
Interventions	Injected heroin + oral methadone (N = 115) oral methadone (N = 111) injected Dilaudid + oral methadone (N = 25) Psychosocial interventions: All patients were offered a comprehensive range of psychosocial and primary care services in keeping with Health Canada best practices.
Outcomes	retention in addiction treatment reduction in illicit‐drug use or other illegal activity
Notes	Country: Canada 2005‐2008
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The authors describe "a computer‐generated randomisation list of permuted blocks of two, four, and six". "Randomization was stratified according to center and according to the number of previous methadone treatments (two or fewer vs. three or more)."
Allocation concealment (selection bias)	Low risk	central allocation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Almost no missing outcome data: "We obtained 12‐month retention data on 245 of 251participants (97.6%) and response data on 240 of 251 participants (95.6%). The baseline characteristics of the groups were similar"
Selective reporting (reporting bias)	Low risk	Details from the published protocol on table 3.
Blinding (objective outcomes: drop out, use of substances measured by urine analysis)	Low risk	No blinding but the objective outcome measurement were not likely to be influenced by lack of blinding: "Retention was assessed with the use of detailed data on daily prescription‐drug use and, when possible, with the use of administrative data and pharmacy and physician records."
Blinding (subjective outcomes: use of substances as measured by self report, side effects)	Low risk	Self‐reported nonuse of illicit heroin was confirmed by means of urine testing at 100% of 46 visits in the group of patients randomly assigned to receive hydromorphone (the double‐blind portion of the study). (information obtained by the authors on request)