Derivation of a prediction rule for posttraumatic organ failure using plasma DNA and other variables

Abstract

The early identification of patients at high risk of developing posttraumatic organ failure would allow preventive therapies to be studied. In this study, highly sensitive and specific guidelines for the early prediction of posttraumatic organ failure (OF) and multiple organ dysfunction syndrome (MODS) using cell-free (plasma) DNA and other predictors of posttraumatic complications were derived. As plasma DNA increases after injury and may be used to predict acute lung injury (ALI), we hypothesized that in combination with other predictors it would predict the later development of OF and MODS. Eighty-three patients (69 males; median age, 36 years) were studied as a consequence of major trauma within 3.5 hours of injury (median time to sampling and assessment, 60 min). Plasma DNA was measured using a real-time, quantitative, polymerase chain reaction assay for the β-globin gene. OF and MODS occurred in 20/83 (24%) and 9/79 (11%) cases, respectively. At selected cutoff points, the sensitivity of plasma DNA for predicting OF and MODS ranged from 50% to 100%, specificity ranged from 74% to 95%, and the likelihood ratio ranged from 3.89 to 10.50. Other variables studied included serum albumin, creatine kinase, aspartate transaminase, lactate dehydrogenase, leukocyte count, hematocrit, injury severity score, maximal abbreviated injury score, and shock index. Using a classification and regression tree, plasma DNA and aspartate transaminase at optimal cutoffs predicted OF and MODS with an overall correct classification of 93% and 87%, respectively.