Correa Filho 2013.

Study characteristics
Methods	Study design: randomized controlled trial Country: Brazil Setting: 0utpatient alcohol treatment center Eligibility criteria:iIncluded were males ages 18‐60 years with an ICD‐10 diagnosis of alcohol dependence and enrolled in the outpatient alcohol treatment program. Excluded were individuals with significant medical disease that might have interfered with the evaluation of the study medication or presence of a safety concern, positive screen for other types of substance misuse except nicotine, clinically significant psychiatric illness, previous treatment with ondansetron within 6 months of randomization, current use of disulfiram, naltrexone or acamprosate, current use of any psychotropic medication, inability to give full informed consent, and clinical history of developmental disorder Duration of follow‐up: 3‐months Informed consent: all participants provided written informed consent Ethical approvals: The Ethics Committee of the Clinical Hospital of the University of Sao Paulo
Participants	Sample size: 52 (50 Odansetron, 52 Placebo) Description of the target population: Males in outpatient treatment for alcohol dependence Age: 42.9 years Sex: 100% male Race/Ethnicity: 33.3% White, 21.6% Black, 45.1% Mixed Marital status: 60.0% married, 15.8% single, 24.2% separated/widowed Harmful alcohol use (baseline): mean AUDIT Score = 29.1 Co‐occurring disorders: mean Hamilton Depression Rating Scale Score = 9.0
Interventions	Type: pharmacologic + non‐pharmacologic Description: Odansetron, brief cognitive behavioral interventions, encouragement to attend Alcoholics Anonymous Duration and frequency: Two 8 mg capsules Odansetron per day for 12 weeks + 12 weekly appointments with cognitive behavioral intervention and encouragement to attend Alcoholics Anonymous Delivery and provider: trained research assistants Comparison group: two placebo capsules per day for 12 weeks + 12 weekly brief cognitive behavioral intervention sessions including encouragement to attend Alcoholics Anonymous
Outcomes	Primary outcome(s): abstinence, heavy drinking Primary outcome measurement tool(s): Timeline followback, Hepatic lab indices, Collateral (family) reporting Secondary outcome(s): mean drinks per day, side effects, retention, depression Secondary outcome measurement tool(s): Timeline followback, UKU Side Effect Rating Scale, Obsessive‐Compulsive Drinking Scale Time points: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12‐weeks (Depression, Obsessive‐Compulsive Drinking assessed at weeks 1, 6, and 12)
Notes	Study funding and conflicts of interest: The State of Sao Paulo Research Foudation (FAPESP) Linked study records: None
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The participants were randomly divided into 2 groups through computer‐generated random numbers" Pg. 2046
Allocation concealment (selection bias)	Low risk	Quote: "Once a week the patients received an envelope with 2 packages containing 7 capsules. One package was designated for morning dosing and the other for night‐time." Pg. 2046 "All capsules in each treatment group were identical in appearance and size and had been manufactured and distributed by the Pharmace Sector" Pg. 2046
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "The codes referent to the medications used were revealed to the researchers only after all patients had completed the study. Only 2 pharmacists from the pharmacy sector at the Psychiatric Institute of the Clinical Hospital of the University of Sao Paulo knew what medication corresponded to which specific code. The packages containing the capsules were distributed to patients by 2 trained research assistants, who had also been blinded to the study and who assessed the outcome of each patient throughout the study period." Pg. 2046
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "The codes referent to the medications used were revealed to the researchers only after all patients had completed the study. Only 2 pharmacists from the pharmacy sector at the Psychiatric Institute of the Clinical Hospital of the University of Sao Paulo knew what medication corresponded to which specific code. The packages containing the capsules were distributed to patients by 2 trained research assistants, who had also been blinded to the study and who assessed the outcome of each patient throughout the study period." Pg. 2046
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "Reasons for dropout were classified as refusal to continue (2%), protocol violation (26%) or lost to follow‐up (72%). Dropout rates among participants randomized to placebo were 57.7% and 42% among those randomized to ondansetron. The overall difference between both groups was not significant" Pg. 2047 "As half of our sample discontinued the treatment, we first used GEE to analyze the data of those who completed it. Little's test was used to verify the missing‐data mechanism. Subsequently, we performed a multiple‐imputation analysis with Markov Chain Monte Carlo approaches and then used GEE in SPSS‐18. Maintaining the original variability of the missing data was achieved by creating imputed values based on variables associated with the causes of missing data." Pg. 2046.
Selective reporting (reporting bias)	Unclear risk	Protocol not available