Prediction of fatality in fulminant hepatic failure.

Thirty-three consecutive patients admitted to the intensive care liver unit of Rigshospitalet with acute hepatic encephalopathy induced by viral hepatitis, drugs, or pregnancy were studied. All were treated with a standard anticoma regime. The 20 patients (61%) who died had a higher bilirubin level and lower total cholic acid conjugation and glycine cholic acid conjugation (p less than 0.05) than the surviving patients. Antipyrin clearance and galactose elimination capacity tended to be lower in the non-survival group than in the survival group (p = 0.09 and 0.11, respectively). Of single variables a bilirubin level of greater than 384 mumol/l gave the best prediction of non-survival (sensitivity, 0.80; specificity, 0.69; PVpos, 0.80; PVneg, 0.69; kappa, 0.49). However, a discriminant score based on combination of variables distinguished completely between non-survivors and survivors when validated by an unbiased method in which each patient is classified on the basis of the other patients' data. It is suggested that the discriminant score is used to select patients with very low probability of survival for liver transplantation or liver assistance procedures of unknown value.