Eisenberg 1988.

Methods	Multi‐centre randomised controlled trial.
Participants	People with severe head injury (GCS 4 to 7). Aged 15‐50 years. Elevated ICP refractory to conventional management. 73 people took part in the study: 37 received barbiturate treatment in addition to standard care and 36 received standard care only.
Interventions	Pentobarbital: Loading dose 10 mg/kg over 30 minutes, 5 mg/kg every 1hr for three hours. Maintenance dose 1 mg/kg/hr with serum level monitoring. Control: standard care.
Outcomes	Control of raised intracranial pressure. With regard to the primary outcome criteria, there were only two possibilities: (1) Treatment success ‐ declared when a person's ICP fell below 20 mmHg (or 15 mmHg for those classified as "skull opened"); (2) Treatment failure ‐ declared when ICP became uncontrollable, or the patient developed a unilateral dilated pupil, cardiovascular collapse, or died. Mortality and morbidity according to GOS score at 30 days and 6 months.
Notes	The study took place at Albert Einstein College of Medicine, New York; Baylor College of Medicine, Houston; University of California at San Diego; University of Texas Health Sciences Center, Houston; and the University of Texas Medical Branch, Galveston. People were recruited between December 1982 and December 1985.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients were randomised (fixed block size) by centre, according to the last available GCS score (4 to 7) using sealed opaque envelopes." pp.16‐17
Allocation concealment (selection bias)	Low risk	Quote: "...using sealed opaque envelopes." p.17
Blinding (performance bias and detection bias) Participants	Low risk	Review authors' judgement: Low risk as people had a Glasgow Coma Scale score of less than 7 and therefore had reduced cognitive function.
Blinding (performance bias and detection bias) Treating physicians	High risk	Review authors' judgement: There is no mention of blinding of the treating physicians. Figure 1 (p.17) The barbiturate group received the additional intervention at regular intervals, so the treating physicians were aware of the treatment. Correspondence with the author in 2012. Quote: "It was a blinded study."
Blinding of outcome assessment (detection bias) All outcomes	High risk	Review authors' judgement: High risk of bias as no details provided as to who was blinded in the original report or through correspondence. Correspondence with the author in 2012. Quote: "It was a blinded study."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Complete outcome data were reported.
Selective reporting (reporting bias)	Unclear risk	Review authors' judgement: It was a multi‐centre study and the authors wrote it was "coordinated and monitored by investigators from the University of Texas School of Public Health, Houston, and from the NINCDS Division of Stroke and Trauma" (p.16) so there must have been a protocol. Correspondence with the author in 2012 clarified there was a study protocol but the university's data storage policy is to destroy records after seven years and so the protocol is no longer available.
Other bias	Unclear risk	None known.