Abstract
Prognostic factors and the efficacy of therapy were studied on 518 patients with fulminant viral hepatitis collected as a joint study from the active members of the Japanese Gastroenterological Society during the period from 1983 to 1988. Using five independent prognostic variables (patients' age, occurrence of infection, gastrointestinal bleeding, renal failure and coexistence of accompanying diseases), a prognosis discriminating logistic model was constructed. A risk score was calculated from the model and patients were classified into two groups with high and low probabilities of survival according to the score.
In patients with the duration of illness more than ten days before development of encephalopathy, survival rate of patients given Fischer's amino acid solution was significantly low compared with those not given in the group of low survival probability.
A similar deleterious effect of the amino acid solution was proven with another logistic model comprising three more covariates (prothrombin percent, total birirubin level on the day of development of hepatic encephalopathy and the duration of illness before encephalopathy) on 391 patients without missing data on these items. No significant life saving effect was observed on plasma exchange, charcoal hemoperfusion, glucagon-insulin therapy, H2 receptor antagonist and steroid.
By Cox's proportional hazard model, plasma exchange was found to double the survival period of patients after development of encephalopathy (p<0.001).
