Probability of surviving postoperative acute renal failure. Development of a prognostic index.

Sixty-five patients who developed postoperative acute renal failure requiring hemodialysis were retrospectively analyzed to identify variables that could be used to predict outcome. Our aim was to identify patients who would have an unfavorable outcome despite hemodialysis and to identify those factors that might be altered to improve outcome. A linear discriminant function capable of segregating survivors from nonsurvivors in the retrospective analysis was subsequently validated in a prospective fashion using a second patient population. Variables used were age, sex, number of transfusions, interval from onset of acute renal failure to dialysis, type of surgery, preoperative hypotension, and the presence of cardiac failure. Scores were formulated for each patient and then segregated into three groups: patients with no precedence for survival, patients with an intermediate risk of dying, and patients with low risk of dying. Based on the univariant analysis, the interval from onset of acute renal failure to first dialysis and the maximum serum creatinine prior to first dialysis were the only factors that might be altered to change mortality. The prognostic index we have developed enables one to select patients without a chance of survival.