Y. Takahashi et al., A MULTICENTER STUDY ON THE PROGNOSIS OF FULMINANT VIRAL-HEPATITIS - EARLY PREDICTION FOR LIVER-TRANSPLANTATION, Hepatology, 19(5), 1994, pp. 1065-1071
To determine the risk of death at an early stage of fulminant viral he
patitis, we created severity indexes drawn from clinical data on the d
ay of development of encephalopathy in 128 patients with fulminant hep
atitis B and 103 with fulminant hepatitis non-A, non-B. In fulminant h
epatitis B, the risk score was 2.75 x BL + 2.75 x BR + 2.7 x AG c 2.3
x WB + 1.67 x CD + 1.56 x AL - 0.098 x PR - 0.88, where BL is 1 if tot
al bilirubin is higher than 20 mg/dl, BR is 1 if the ratio of total to
direct bilirubin exceeds 2.2, AG is 1 if age is above 40 yr, WB is 1
if white blood cell count is less than 4,000 cells/mm(3) or more than
18,000 cells/mm(3), CD is 1 if a hazardous disease coexists and AL is
1 if ALT is less than 100 times the upper limit of normal (otherwise a
ll are 0), and PR is prothrombin time (percentage of normal value). Us
ing a cutoff score of 0, we found the positive predictive value, negat
ive predictive value and predictive accuracy to be 0.90, 0.86 and 0.89
, respectively. Sensitivity and specificity were 0.94 and 0.77, respec
tively. In fulminant non-A, non-B hepatitis, the risk score was 2.66 x
BR + 2.25 x BL + 2.24 x DI + 2.05 x AL + 1.38 x AG + 0.00021 x WB - 6
.33. BR is 1 if the ratio of total to direct bilirubin is more than 1.
5, BL is 1 if total bilirubin is higher than 15 mg/dl,DI is 1 if the d
uration of illness before encephalopathy is less than 4 days or more t
han 12 days, AL is 1 if the ALT level exceeds 40 times the upper limit
of normal and AG is 1 if age exceeds 50; otherwise all variables are
0. WB is WBC count per cubic millimeter. Positive predictive value, ne
gative predictive value and predictive accuracy were 0.92, 0.80 and 0.
89, respectively. Sensitivity and specificity were 0.95 and 0.70, resp
ectively. These prediction equations were validated prospectively on a
population of testing samples comprising 52 patients with fulminant h
epatitis B and 47 with fulminant non-A, non-B hepatitis. Positive pred
ictive values were 0.92 and 0.92, negative predictive values were 0.71
and 0.78 and predictive accuracies were 0.87 and 0.89 for patients wi
th fulminant hepatitis B and fulminant non-A, non-B hepatitis, respect
ively. They may be used for selecting candidates for liver transplanta
tion and for evaluating efficacy of a new therapy.