The modified Maddrey's Discriminant Function (DF) is used to identify poor prognosis acute alcoholic hepatitis (AAH).
This calculation requires conversion of serum bilirubin from mg/dl to µmol/l, as well as a corrected calculation of prothrombin time (PT) prolongation. The DF is therefore awkward to use in a clinical setting.
In this study, researchers from Glasgow sought to identify variables associated with mortality.
|Variables for predicting short-term mortality include blood urea, serum bilirubin, serum albumin, age, and encephalopathy.|
The research team collected clinical and laboratory data was from 256 patients with clinical AAH.
They used stepwise logistic regression identified variables associated with mortality.
The researchers identified several day 1 variables associated with short-term (28 day) mortality. These were blood urea, serum bilirubin, serum albumin, age, and encephalopathy.
Day 7 variables associated with medium-term (84 day) mortality were serum bilirubin, PT, age, and leukocyte count.
The team found that by using their set of day 1 variables they were able to predict short-term mortality more reliably than by using the day 1 DF.
The team determined that day 1 variables correctly predicted 49% of patients as dead at 28 days. The variables also predicted 51% dead who remained alive at 28 days, and 5% alive who died.
In comparison, day 1 DF correctly predicted 16% of patients dead at 28 days, however predicted 84% dead who remained alive and 2% alive who were dead at 28 days.
The team also found that using day 7 variables to predict day 84 mortality was also more reliable than using day 7 DF.
Day 7 variables correctly predicted day 84 mortality in 53% of patients, but predicted 47% dead who remained alive and 7% alive who were dead at day 84.
Day 7 DF correctly predicted day 84 mortality in 32%, but predicted 68% dead who remained alive and 8% alive who died.
Dr Forrest's team concluded, "The DF failed to identify the majority of patients who died".
"The variables we identified significantly improved the identification of patients with poor short or medium term prognosis."