Gowin 2012.

Study characteristics
Methods	Design: placebo‐controlled parallel trial
Participants	Participants: community living adults on probation or parole, in Houston area, USA Sex: (see note 1) whole sample (n = 12) = 10 male (83.3%), 2 female (16.7%); intervention (n = 6) = 5 male (83.3%), 1 female (16.7%); control (n = 6) = 5 male (83.3%), 1 female (16.7%) Age: (see note 1) intervention mean = 25.17 years (SD = 3.82), control mean = 32.00 years (SD = 5.02) Unit of allocation: individual Number randomised: 15 (see note 1); intervention = 6, control = 6. Three participants removed for positive urinalysis tests for prohibited substances (group membership not provided) Number completing: 12 (see note 1); intervention = 6 (100%), control = 100%) Setting: community clinic Inclusion criteria: free of illicit and prescription drugs during study period; participants on parole or probation; authors state that these participants were sought “because of high incidence of antisocial and aggressive behaviour associated with this population” (quote, p 983, column 2) Exclusion criteria: medical conditions (e.g. HIV, seizures, cardiovascular disease); pregnancy; any current or past psychiatric illness and axis I disorders (except past substance abuse/dependence) Ethnicity: (whole sample; see note 1) intervention = African‐American (n = 6, 100%), control = African‐American (n = 5, 83.33%) and Hispanic (n = 1, 16.67%) Baseline characteristics: Intervention group: high school education (n = 6, 100%); conduct disorder present (n = 3, 50%); ASPD present (n = 3, 50%); smoker (yes) (n = 4, 66.67%); number of cigarettes/day (mean = 4.33, SD = 4.59); on parole (n = 3, 50%); on probation (n = 2, 33.33%); Shipley Wechsler Adult Intelligence Scale (WAIS) (mean 105.33, SD = 12.66), Shipley WAIS‐R (mean 94.67, SD = 14.38). Control group; high school education (n = 5, 83.33%); conduct disorder present (n = 4, 66.67%); ASPD present (n = 3, 50%); smoker (yes) (n = 5, 83.33%); number of cigarettes/day (mean = 4.67, SD = 3.44); on parole (n = 1, 16.67%); on probation (n = 2, 33.33%); Shipley WAIS (mean 105.83, SD = 5.11); Shipley WAIS‐R (mean 96.50, SD = 6.25)
Interventions	Two conditions: tiagabine (n = 3 AsPD randomised); ascending dose escalation; week 1, non‐drug (baseline); week 2, placebo; week 3, 4 mg tiagabine; week 4, 8 mg tiagabine; week 5, 12 mg tiagabine; week 6, placebo (see note 2) placebo ‐ corn starch capsule (n = 3 AsPD randomised); week 1, non‐drug (baseline); week 2 to week 6, placebo corn starch capsule (see note 2) Duration of intervention: 6 weeks Duration of trial: 6 weeks Length of follow‐up: none Dose adjustment: increasing dose of tiagabine; 4 mg tiagabine (week 3); 8 mg tiagabine (week 4); 12 mg tiagabine (week 5)
Outcomes	Primary outcomes Aggression: Point Subtraction Aggression Paradigm (PSAP); Buss‐Perry Aggression Questionnaire (BPAQ); Lifetime History of Aggression Questionnaire (LHA); Retrospective Overt Aggression Scale (ROAS) Adverse events: medication side effects Secondary outcomes Leaving the study early; 3 randomised participants removed for positive urinalysis tests for prohibited substances (details of group membership not provided) Impulsivity; Eysenck Impulsivity Venturesomeness Questionnaire (EIVQ; Barratt Impulsiveness Scale (BIS‐II) Anger:State‐Trait Anger Expression Inventory (STAXI) Other outcomes Cognitive assessment; Shipley Institute of Living Scale (SILS) Timing of outcome assessments PSAP and cognitive assessment 2‐3 days per week; questionnaires at week 5
Notes	6/12 (50%) of total participants randomised satisfied for DSM‐IV TR criteria for AsPD (3 in intervention group, 3 in control group); 2/12 (16.7%) met DSM‐IV TR criteria for childhood conduct disorder; no data for AsPD sub‐sample All medications taken orally (via capsule) twice a day at 09:00 and 18:00 Study funding: National Institute on Drug Abuse (USA) Declaration of interests: none reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Comment: No information given on method of randomization, however 12 subjects randomised exactly to 6 control vs. 6 experimental, with each group having x1 female and 3 ASPD participants. This suggests that a truly random process was not used meaning that bias may have been introduced.
Allocation concealment (selection bias)	High risk	Comment: No information given on method of randomization, however 12 subjects randomised exactly to 6 control vs. 6 experimental, with each group having x1 female and 3 ASPD participants. This suggests that a truly random process was not used meaning that bias may have been introduced.
Blinding (performance bias and detection bias) of participants	Low risk	Comment: Placebo and tiagabine capsules were manufactured to look the same. There was however no assessment of whether participants were aware of which group they were in at the end of the study.
Blinding (performance bias and detection bias) of personnel	Low risk	Comment: Research assistants conducting drug administration and administering medication event monitoring system (MEMS) bottles were blind to allocation.
Blinding (performance bias and detection bias) of outcome assessors	Unclear risk	Comment: Use of computerized task may have reduced opportunity for bias from assessors however no information provided for potential impact on questionnaire‐based outcomes, or data analysis.
Incomplete outcome data (attrition bias) All outcomes	High risk	Comment: Did not use intention to treat (ITT). Appears that 3 subjects left study early due to current substance use but no further information or group membership information given.
Selective reporting (reporting bias)	Unclear risk	Comment: No protocol published prior to trial.
Other bias	High risk	Comment: Funded by National Institutes of Health grants, but apparent closeness of the authors of this paper and the developer of the PSAP who is acknowledged in the paper for “consultation, mentoring, and expertise of Don R Cherek, PhD without whom these experiments would not be possible” (p.989, col.1)