Cohn 1986.

Methods	1. Randomized 2. Double blind 3. Three parallel groups: ‐Buspirone 15‐50mg/d ‐Alprazolam 1.5‐5mg/d ‐Lorazepam 1‐10mg/d 4. Duration: 4 weeks with 4‐7 placebo washout period beforehand 5. Analysis: Cochran‐Mantel‐Haenszel test used endpoint analysis of categoric responses
Participants	1. Diagnosis: GAD (DSM‐III) 2. N=60 3. Age (mean): Buspirone = 40.5 Alprazolam = 35.9 Lorazepam = 38.8 Sex: 55% females Setting: outpatients
Interventions	1. Buspirone (N = 20) 2. Alprazolam (N = 20) 3. Lorazepam (N = 20)
Outcomes	1. HAM‐A 2. Raskin Depression Scale 3. Covi Anxiety Scale 4. physician's questionnaire measuring global psychopathology
Notes	(a)Statistical analysis support from Mead Johnson Pharmaceutical Group (b) Alprazolam used in the analysis using HAM‐A
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	B ‐ Unclear