Limitations of the pediatric risk of mortality score in assessing children with acute renal failure

The pediatric risk of mortality score (PRISM) incorporates 14 physiological and laboratory variables to calculate a patient's score, which is then adjusted for operative status and age to determine the probability of death. Because of the ethical issues surrounding the initiation of dialysis in critically ill children, a scoring system which could differentiate survivors from nonsurvivors prior to the initiation of dialysis would be useful to the clinician. Similarly, a score which could accurately estimate the probability of mortality in children with acute renal failure would be useful to third party payors attempting to evaluate the performance of individual care providers. We calculated PRISM scores on the day dialysis was initiated, retrospectively, in 31 children seen from 1984–1988 with the diagnosis of acute renal failure and requiring dialysis, in order to determine if the PRISM score was accurate in prediction of mortality. In addition, we calculated scores on the day of admission to the intensive care unit (DICU) in order to see if DICU scores accurately reflected mortality risk. The mean PRISM scores of nonsurvivors were significantly higher than the mean scores of survivors on the day dialysis therapy was initiated. However, overlap in the scores of survivors and nonsurvivors would limit the applicability of PRISM scores for clinical decision making. Children that developed acute renal failure requiring dialysis due to extrarenal diseases had a higher mortality rate than those that had primary renal disease (57% versus 12.5%,P<0.05). DICU scores underestimated the mortality of these patients. The decision to institute dialysis for children with acute renal failure cannot be based on PRISM scores calculated during the hospital course. The use of PRISM scores in quality assurance activities for children with acute renal failure is inappropriate.