Likar 2008.

Methods	Randomised, double‐blind study
Participants	13 patients: semi‐conscious, terminal cancer with predicted life expectancy of up to 3 days. All had metastatic disease. Primary cancer: bronchial (10), rectal (2), bladder (1). Age: Intervention A = 71.3 + 3.8 years Intervention B = 71.8 + 5.4 years Gender: Intervention A = 5 males, 2 females Intervention B = 5 males, 1 female.
Interventions	Intervention A = Hyoscine hydrobromide 0.5 mg every 6 hours intravenously Intervention B = Glycopyrronium bromide 0.4 mg every 6 hours intravenously. Discontinued if no abatement of death rattle after third injection.
Outcomes	Death rattle assessed using scale of one to five: 1 = noisy breathing; 2 = minimal rattle; 3 = moderate rattle; 4 = severe rattle; 5 = very severe rattle Assessment carried out two‐hourly from zero hours till 12 hours. Stronger decrease in death rattle at various time points in those who had Intervention B (i.e. glycopyrronium) compared to those who had Intervention A: statistically significant difference. Mean time from first treatment to death (SD): Intervention A = 19.5 + 5.4 hours Intervention B = 12.8 + 5.0 hours; not statistically significant
Notes	Restlessness and expressions of pain also assessed on scale of one to three: 1 = slight; 2 = moderate; 3 = severe Outcome: Restlessness: appeared to be greater with Intervention B at 2 hours, then no statistical difference. Expressions of pain: no difference
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Lack of detailed description: 'envelope method' used
Allocation concealment (selection bias)	Unclear risk	Lack of detailed description: 'envelope method' used
Selective reporting (reporting bias)	Unclear risk	Lack of detailed data for secondary outcomes
Incomplete outcome data addressed	Low risk	All participants accounted for.
Blinding	Low risk	Injection solutions blinded by hospital pharmacy