Gerra 1995.

Methods	Randomised, placebo‐controlled, double‐blind trial
Participants	Setting: hospital outpatient clinic, Parma, Italy. Participants: 152, drug abuse disorder by DSM‐III‐R, heroin users. Group sizes: (1) n = 33, (2) n = 42, (3) n = 58, (4) n = 19. Similarity of groups not reported. Age: 18 to 32 years. 82% men
Interventions	(1) Clonidine, 0.15 mg iv 3 times a day. (2) Clonidine, 0.15 mg iv 3 times a day + naltrexone, 12.5 mg day 2 then 50 mg/day for 3 months. (3) Clonidine, 0.15 mg iv 3 times a day + naloxone, 0.2 mg iv day 2, 0.4 mg 2 times a day on days 3 and 4, then naltrexone 50 mg/day from day 5. (4) iv saline + oral placebo. Daily clinic attendance with 4 hours iv therapy in morning, 3 hours in afternoon. (Groups 2 and 3 not considered for this review.) Scheduled duration of treatment unclear
Outcomes	Mean total withdrawal score at 48 and 72 hours; bar graphs for days 1, 2, and 3 showing ratings of individual items of withdrawal scale; morphine metabolites in urine; Hamilton Rating Scale for Depression on day 1, day 8, and 6 months
Notes	Withdrawal assessed by observer only using 9‐item scale, mainly of objective signs. Source of funding not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "All the patients were randomly divided into four groups ..." Comment: Group sizes differed, and similarity of the characteristics of the groups was not discussed. The adequacy of sequence generation is doubtful
Allocation concealment (selection bias)	Unclear risk	Method not reported
Blinding (performance bias and detection bias) Subjective outcomes ‐ intensity of withdrawal, adverse effects	Unclear risk	Double‐blind stated, but given the differences in group sizes, it is doubtful whether the blind was maintained for treating personnel, participants, and observers
Blinding (performance bias and detection bias) Objective outcomes ‐ duration of treatment, completion of treatment	Low risk	Double‐blind stated. Although it is doubtful whether the blind was maintained, these outcomes are considered unlikely to be affected by knowledge of treatment allocation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Drop‐out in first week higher in placebo compared with other groups. Given the marked difference in withdrawal severity between clonidine and placebo groups, the differential drop‐out is unlikely to have a clinically significant impact on withdrawal scores (the main outcome reported)
Selective reporting (reporting bias)	Low risk	None apparent
Other bias	Low risk	None apparent