Improved outcome prediction for patients with multiple organ failure undergoing continuous hemodiafiltration.

Abstract

A number of patients with multiple organ failure (MOF) regardless of accompanying acute renal failure have been treated with continuous hemodiafiltration (CHDF). However, despite its high cost, the costs/benefits of CHDF for MOF patients still need to be evaluated. Although many scoring systems were established to predict the outcome of MOF, their predictive powers were not estimated in MOF patients undergoing CHDF. Therefore, using 52 Japanese patients with MOF treated with CHDF for more than 1 week, we estimated the predictive powers of multiple organ dysfunction (MOD) scores and acute physiology and chronic health evaluation (APACHE) III scores, retrospectively. The patients were divided into 2 groups according to outcome at Day 28 after the initiation of CHDF. In both scoring systems, the median values at Day 0 were not significantly different between the survival (n = 19) and the nonsurvival (n = 33) groups. In contrast, at Day 3, the median values of MOD scores was 4 (0-14) in the survival group and 9 (1-12) in the nonsurvival group (p = 0.0035). The median value of APACHE III scores were 37 (19-97) and 87 (16-150) at Day 3, respectively (p < 0.0001). In the survival group, APACHE III scores significantly decreased from the median value of 64 (32-89) to 37 (p = 0.0269), and in the nonsurvival group, it increased significantly from the median value of 70 (29-103) to 87 (p = 0.0116). In contrast, no significant changes were observed in the MOD scores. In conclusion, the MOD score and the APACHE III score systems had less power to predict the outcome of MOF patients undergoing CHDF at Day 0. However, rescoring at Day 3 of each index was much more powerful to accurately predict the outcome of such patients.