Appendix.

Support for Bias Assessment Judgment for Each Domain by Study Presented in Table 3

Bias Domain	Dos Santos et al, 202170	Glue et al, 201762	Glue et al, 202064	Glue et al, 201863	Taylor et al, 201868	Gasser et al, 201467	Gasser et al, 201566	Wolfson et al, 202065	Grob et al, 201169
Random sequence generation (selection bias)	Simple randomization performed by researcher who did not participate directly in the experimental sessions or have access to the raw data	Nonrandomized	Active placebo randomly injected into ketamine vials	Nonrandomized	Randomization conducted by Yale Investigational Drug Service and kept separate from investigators	Randomization methodology not mentioned	Randomization methodology not mentioned	Randomized using web-based randomization system with unique container numbers; randomization monitored by individuals without communicating with site staff, study monitors, or analysts	Order in which subjects received the treatments randomized and known only by the research pharmacist; methodology not mentioned
Allocation concealment (selection bias)	Pharmacist-prepared solution intended to closely mimic color, smell, and taste of ayahuasca; all participants naive to ayahuasca	Open label	Psychoactive-controlled, identical vials	Open label	Placebo-controlled with normal saline	Active placebo of LSD; capsules of identical size, color, and shape and bottled in sequentially numbered containers	Active placebo of LSD; capsules of identical size, color, and shape and bottled in sequentially numbered containers	Placebo was 125 mg lactose prepared by pharmacist in identical gelatin capsules	Placebo was 250 mg niacin in identical clear capsules
Blinding of participants and researchers (performance bias)	Volunteers and researchers were right about what had been ingested in 100% of the sessions	Not blinded	After ketamine dosing, all subjects reported dissociative symptoms; minimal effects for midazolam	Not blinded	17 of 18 patients correctly identified receiving ketamine	All participants correctly guessed the LSD dose administered	All participants correctly guessed the LSD dose administered	Not reported	Drug order almost always apparent to subjects and investigators
Blinding of outcome assessment (detection bias)	Participants and researchers correctly guessed what they received	Participant rated	Participant rated	Participant rated	Participant and clinician rated	Participants and independent clinician raters correctly guessed what they received	Participants and independent clinician raters correctly guessed what they received	Participant and independent rater blinded	Participant and investigator rated; almost always apparent whether psilocybin or placebo
Incomplete outcome data (attrition bias)	100% completion rate; primary outcome analysis missing data of 2 in ayahuasca group and 1 in placebo group	100% completion rate without missing data	92% completion rate without missing data	90% completion rate without missing data	94% completion rate without missing data	75% completion rate without missing data	83% completion rate; 1 audio recording and 1 questionnaire missing	94% completion rate without missing data	66% completion rate
Selective reporting (reporting bias)	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported	All prespecified outcomes reported
Other bias	Nonrepresentative population (mostly undergraduate female students); "unnatural" context of the speech test	Almost all patients had prior MDD; none depressed at time of testing, but MDD history could represent a clinical phenotype responsive to ketamine	Most patients had prior MDD; none depressed at time of testing, but MDD history could represent a clinical phenotype responsive to ketamine	Four patients requested twice-weekly dosing for early recurrence of anxiety (usually within 4 days of dosing), and 16 received once-weekly dosing	Not mentioned	Participants had grave somatic diseases, and the course of somatic illness (worsening or improving) can impact psychological parameters independent of the intervention and contribute to missing data	No control group for the follow-up because of the crossover design	Degree of group differences impacted by an outlier in the placebo group who responded exceptionally well to psychotherapy alone compared to other participants in the placebo group during the blinded segment; potential placebo effect because 2 of 3 placebo participants believed they were assigned MDMA during the blinded segment	Variability in the extent of contact with subjects after treatment; ad hoc communication depended on the needs and wishes of subjects, some of whom were near death while others were more functional

LSD, lysergic acid diethylamide; MDD, major depressive disorder; MDMA, 3,4-methylenedioxy-methamphetamine.