Jabre 2009.

Methods	RCT comparing etomidate (n = 234) and ketamine (n = 235)
Participants	Adult pre‐hospital, emergency department, or ICU patients requiring intubation for being comatose, shock, acute respiratory failure, or 'other'
Interventions	Etomidate 0.3 mg/kg versus ketamine 2 mg/kg
Outcomes	Primary: maximum Sequential Organ Failure Assessment (SOFA) score. Secondary: change in SOFA score, 28 day mortality, ICU LOS2, ventilation duration, vasopressor use, ACTH stimulation test performed 0 to 48 h after intervention
Notes	ICU length of stay reported as "ICU free days at day 28", Vasopressor use reported as "Catecholamine free days until day 28", ventilation duration reported as "Mechanical ventilation free days at day 28". These variables were transformed by subtracting these results from 28 days to estimate the duration of each of these variables. This analysis assumes a single ICU visit, ventilation treatment, vasopressor use; 69% of patients in both arms were intubated for being comatose, while the minority were intubated for shock, acute respiratory failure, or 'other' reasons
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Utilized a computerized random number generator
Allocation concealment (selection bias)	Low risk	Study drugs were sealed in sequentially numbered, identical boxes containing the entire treatment for each patient. "The emergency physician enrolling patients was aware of study group assignment. However nurses and intensivists in the intensive care unit were masked to the treatment assigned" (Jabre 2009. p294). While this single blind methodology raises concern, the allocation concealment was adequate to prevent enrolling physicians from altering the allocation sequence
Incomplete outcome data (attrition bias) Mortality	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) Adrenal Gland Dysfunction	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) Hospital LOS	Low risk	N/A
Incomplete outcome data (attrition bias) ICU LOS	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) Duration of mechanical Ventillation	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) Vasopressor requirements	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) Organ Dysfunction	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing data on 2 patients from each group, and one patient in the etomidate withdrew consent. Unlikely to affect results, given the small proportion of the sample size. Effect on risk estimates likely to be negligible
Selective reporting (reporting bias)	Low risk	All relevant data reported
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Emergency physicians giving the drugs were not blinded, but nurses and ICU physicians providing care for the duration of the study were blinded. Unlikely to affect the performance of participants and personnel, hence the effects of the intervention
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Emergency physicians giving the drugs were not blinded, but nurses and ICU physicians providing care for the duration of the study were blinded. Unlikely to affect the outcome assessment, and hence opportunity for information bias minimized