Validation of postinjury multiple organ failure scores

Most multiple organ failure (MOF) scores were developed over a decade ago, but little has been done in terms of validation and to understand the differences between populations identified by each of them. Given the lack of a gold standard, validation must rely on association with objective adverse outcomes. Thus, we propose to (a) validate two widely accepted MOF scores (Denver and Marshall), examining their association with adverse outcomes in a postinjury population; and (b) compare risk factors, characteristics, and outcomes of patients identified by each score. The Denver MOF score grades (from 0-3) four organ dysfunctions (lung, kidney, liver, and heart) and defines MOF as a total score more than 3. The Marshall score grades, in addition, central nervous system and hematologic dysfunction (total of six organs on a 0- to 4-point scale). Using a prospectively collected data set, MOF was scored daily by both scores for 1,389 consecutive trauma patients with Injury Severity Score of more than 15 admitted from 1992 to 2004. Risk factors, clinical outcomes (death, ventilator-free days), and resource utilization outcomes (mechanical ventilation time, length of stay in the intensive care unit) were evaluated. Both scores were associated with areas under the receiver operating characteristic curves of 80 or greater (ideal value = 100), with values for the Denver score being slightly greater (albeit not significantly) regarding prediction of most outcomes. Values of sensitivity and specificity were more than 70% for death and ventilator-free days (with the Denver score showing a consistent trend toward greater specificity), but either sensitivity or specificity was less than 70% for mechanical ventilation time and length of stay in the intensive care unit, suggesting that these scores are appropriately biased toward clinical outcomes as opposed to resource utilization. Both scores performed well, with the Denver MOF score showing greater specificity, which, coupled with its simplicity, makes it an attractive tool for both the research and clinical environments. Basic concepts of each score can probably be combined to produce an improved MOF score.