Naveau 1986

Methods	Randomised trial comparing parenteral nutrition to no parenteral nutrition in hospitalized patients with alcoholic cirrhosis. Geographical location Bicetre, France. Study published 1986.
Participants	Inclusion criteria: Alcoholic cirrhosis on biopsy or, if not possible, at least 2 our of 5 clinical characteristics (firm liver, ascites, hepatic encephalopathy, splenomegaly, varices at endoscopy) AND bilirubin > 5mg%. Exclusion criteria: Hepatocellular carcinoma, creatinine > 2mg%, sodium < 130 meq/l, septicemia, spontaneous bacterial peritonitis, gastrointestinal bleeding within 3 days, hepatic coma. 40 patients (25 men/15 women, mean age 52).
Interventions	Intervention group received intravenous formulation (20 kcal/kg glucose, 20 kcal/kg lipid, 0.2 gm nitrogen/kg, minerals, vitamins) + oral diet; Controls received oral diet (40 kcal/kg, 0.2 gm nitrogen/kg). Duration therapy 28 days. Patients in both groups received neomycin for encephalopathy.
Outcomes	Mortality, ascites resolution, development of encephalopathy, infections (sepsis), serum bilirubin, triceps skinfold thickness, midarm muscle circumference. The only adverse effects noted were four episodes of sepsis in patients receiving parenteral nutrition; no apparent attempt to look for such events in all of the patients in the trial.
Category of study	Parenteral nutrition/Medical.
Sample size calculation	Sample size based on previously reported trial (Nasrallah 1980).
Full paper or abstract only	Full paper.
Notes	Attempt made to follow patients for 2 years; decision made to confine analysis to in‐hospital period. Request for further information sent to Dr Naveau via e‐mail on September 11, 2011 (Address = Sylvie.naveau@abc.ap‐hop‐paris.fr). No response has been received as of March 20, 2012.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated.
Allocation concealment (selection bias)	Low risk	Serially numbered, opaque, sealed envelopes.
Blinding (performance bias and detection bias) All outcomes	High risk	Not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Three dropouts in treatment group and 2 in control group accounted for.
Selective reporting (reporting bias)	Low risk	Mortality and morbidity outcomes reported.
Other bias	Unclear risk	Unclear funder.
Intent to treat analysis	High risk	All information on dropouts not available, although in‐hospital mortality reported.
Baseline imbalance?	Low risk	No differences identified.
Early stopping?	Low risk	Achieved planned number.