Chang 1981.

Study characteristics
Methods	Randomised, double‐blind, 3‐period cross‐over trial
Participants	8 people (6/8 (75%) men/2/8 (25%) women) aged 17‐58 years (median = 41 years), 7/8 (88%) participants were cannabis naive Tumour types: resected soft tissue sarcoma Chemotherapy regimen: adjuvant doxorubicin and cyclophosphamide every 4 weeks until a total cumulative doxorubicin dose of 500‐550 mg/m2 Doxorubicin (70 mg/m2)and cyclophosphamide (700 mg/m2) were given at constant doses for all participants Chemotherapy emetogenicity: high
Interventions	Dronabinol 10 mg/m2 orally every 3 hours for total 5 doses, if vomited then participant given marijuana cigarettes 900 mg, containing THC 1.93% (approximately 17.4 mg), n = 8 Placebo, n = 8
Outcomes	Episodes of nausea and vomiting on day of therapy
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Order of THC‐placebo administration was randomized into paired trials"
Allocation concealment (selection bias)	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	High risk	17/27 (63%) participants received THC, 16/27 (59%) participants received placebo
Selective reporting (reporting bias)	Low risk	Data reported for primary outcome
Other bias	Unclear risk	Assumed washout period sufficient. Paired analysis was performed. Unclear if groups balanced at baseline
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Identical gelatin capsules with sesame oil. Identical cigarettes, the odour and taste of a lit placebo cigarette were identical to those of cannabis cigarette"
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Blinding assumed as study reported as "double‐blind". "Neither patients nor nursing staff was [sic] informed which drug was administered"