Dunham 1994.

Methods	RCT, single‐centre, 3‐arm, parallel design
Participants	Total number of randomized participants: 38 Inclusion criteria Blunt traumatic event, GCS ≥ 5, ISS ≥ 15 No spinal neuropathy above 8th thoracic spinal level No major fluid restriction requirement Aged 18 to 60 years Able to undergo upper gastrointestinal endoscopy Respiratory insufficiency that mandated need for mechanical ventilation for ≥ 48 hours Excluded criteria If randomization did not take place within 30 hours after admission or admission did not occur within 12 hours after injury Primary diagnoses Blunt trauma injuries Baseline characteristics EN group Age: not reported Gender: not reported APACHE II: not reported GCS, mean (SD): 11 (± 5) ISS, mean (SD): 34 (± 18) PN group Age: not reported Gender: not reported APACHE II: not reported GCS, mean (SD): 12 (± 3) ISS, mean (SD): 38 (± 12) EN + PN group Age: not reported Gender: not reported APACHE II: not reported GCS, mean (SD): 11 (± 4) ISS, mean (SD): 37 (± 15) Country: USA Setting: trauma centre
Interventions	EN group n = 12; 0 losses Details: transpyloric tube placement, feeding started within 24 hours of randomization and continued for 7 days. Investigators used Harris Benedict formula BEE x 1.3 to calculate caloric intake. Aimed to provide 50% projected calories by 24 hours after randomization; and 100% by 48 hours. Formula was Traumacal (Mead‐Johnson), NPC given in form of lipids (30%) and carbohydrates (70%). Ratio of NPC to nitrogen was 105:1. Protein load was 1.75 g/kg/day. Mean caloric intake: 1789 calories/day for 7 days PN group n = 16; 1 participant died on 4th study day, not included in study analysis but we have included in review outcome data. Details: feeding started within 24 hours of randomization and continued for 7 days. Investigators used Harris Benedict formula BEE x 1.3 to calculate caloric intake. Aim to provide 50% projected calories by 24 hours after randomization; and 100% by 48 hours. Formula consisted of dextrose‐lipid‐AA mixture; soybean solution, multi‐vitamin mixture. Mixture was 6.7% AA and 23.1% dextrose, soybean solution provided 30% of the NPC. Mean caloric intake: 1961 calories/day for 7 days EN + PN group n = 10; 0 losses Details: feeding started within 24 hours of randomization and continued for 7 days. Investigators used Harris Benedict formula BEE x 1.3 to calculate caloric intake. Aim to provide 50% projected calories by 24 hours after randomization; and 100% by 48 hours. EN formula provided 50% of calories and PN formula provided 50% of calories. EN consisted of Traumacal (Mead‐Johnson), NPC given in form of lipids (30%) and carbohydrates (70%). PN formula consisted of dextrose‐lipid‐AA mixture; soybean solution, multi‐vitamin mixture. Mean caloric intake: 2030 calories/day for 7 days
Outcomes	Number of ventilator days Number of ICU days Total hospital stay Presence of ARDS Respiratory infection Any infection Renal failure Icterus Death Hospital and professional charges
Notes	Funding/declarations of interest: not reported Study dates: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomized but no additional details
Allocation concealment (selection bias)	Unclear risk	No details
Blinding of participants and personnel (performance bias) All outcomes	High risk	No details and we assumed investigators made no attempts to blind personnel
Blinding of outcome assessment (detection bias) All outcomes (except mortality)	Low risk	Most outcome data were taken from the trauma registry. Quote: "All data from the trauma registry and the finance officers were blinded, since these sources had no knowledge of the patient's status relative to the research arm assigned."
Blinding of outcome assessment (detection bias) Mortality	Low risk	Outcome assessors were blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	1 participant was not included in data for PN group due to death during 7‐day intervention period; we have included this event in review analysis. No other loss of data
Selective reporting (reporting bias)	Unclear risk	Clinical trials registration or prospectively published protocol not reported; not feasible to judge risk of selective outcome reporting bias
Baseline characteristics	Low risk	Some usual baseline characteristics not reported (age, gender) but other characteristics all appeared comparable
Other bias	Unclear risk	Target rate of nutrition the same. Unclear if differences in formula were equivalent