Rickels 1985.

Methods	Randomised controlled trial, 6 weeks
Participants	504 outpatients with major depressive disorder, conducted in 3 treatment centres. Patients with a RDS score of at least 8 and an equal or less score on the CAS and at least 18 on HDRS Aged 18 to 70 years
Interventions	Alprazolam versus amitriptyline versus doxepin versus placebo (126/119/120/126) Placebo 'washout' period 4 to 7 days Maximum dose: alprazolam 4.5 mg, amitriptyline 225 mg, doxepin 225 mg, placebo 9 capsules Mean daily dose during last 2 weeks: alprazolam 3 mg, amitriptyline 148 mg, doxepin 143 mg
Outcomes	Primary outcome: Mean HDRS after 6 weeks: alprazolam 13.59 (initial 25.19), amitriptyline 14.77 (initial 25.48), doxepin 13.23 (initial 25.85), placebo 18.90 (initial 26.38) Secondary outcome: All‐cause withdrawals: alprazolam 24 (18.8%), amitriptyline 34 (27.4%), doxepin 23 (18.9%), placebo 59 (45.4%) Withdrawals due to side effects: alprazolam 3 (2.3%), amitriptyline 17 (13.7%), doxepin 12 (9.8%), placebo 10 (7.7%)
Notes	Endpoint analysis HDRS‐21
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"...randomly assigned..." No further information is provided.
Allocation concealment (selection bias)	Unclear risk	No information is provided
Blinding (performance bias and detection bias) Patients	Low risk	"Medication,..., was administered in identical capsules containing..."
Blinding (performance bias and detection bias) Physicians	Unclear risk	No information is provided
Blinding (performance bias and detection bias) Investigators	Unclear risk	Unclear if they used independent investigators for the assessments
Incomplete outcome data (attrition bias) HDRS	Unclear risk	"... persistent side effects and worsening of symptoms or lack of improvement permitted removal of the patient from the study." “... statistically significant differences among treatment groups with respect to dropout rate, with significantly more patients given placebo dropping out, end‐point analyses, including patients with at least one week of efficacy data, were also performed for all efficacy variables.” “... but the set of analyses based on actual patients numbers reached at each evaluation period provided rather similar results.”
Incomplete outcome data (attrition bias) Withdrawals/side effects	Unclear risk	Not all causes for withdrawal are described. Dropout rates are different between the groups due to ineffectiveness (alprazolam 19 patients, amitriptyline 11, doxepin 8 and placebo 48), but these last scores are taken in the analysis. There were also differences for dropouts due to side effects: alprazolam 3 patients, amitriptyline 17, doxepin 12 and placebo 10. It is not clear for which side effects patients were withdrawn from the study, so there can be a bias regarding patients who dropped out too soon before there was any effect of treatment.
Selective reporting (reporting bias)	Unclear risk	No information is provided
Other bias	High risk	The mean doses are given, but it is unclear how fast they increased the dosages Support from alprazolam manufacturer
Placebo washout	High risk	4 to 7 days