Use of Acute Physiology and Chronic Health Evaluation (APACHE)-II and Red Cell Distribution Width (RDW) for Assessment of Mortality of Patients with Sepsis in ICU.

Abstract Critically ill patients of Intensive Care Unit (ICU) need highest level of monitoring, intense nursing care and integrated management which are very expensive and consume significant part of hospital resources. Prediction of outcome from disease has become an essential component of health science. So, various scoring systems have been developed to predict outcome of critically ill patients in ICU. There is no perfect model of severity score to predict ICU mortality. Search for new system is still remaining as continuous efforts to find the best model to get accurate information about the prognosis and outcome of critically ill patients. This observational prospective cohort study was carried out in ICU of Bangabandhu Sheikh Mujib Medical University (BSMMU), Dhaka, Bangladesh from March 2015 to September 2015 to evaluate the ability of mortality prediction of Acute Physiology and Chronic Health Evaluation (APACHE)-II after adding RDW. Total 62 patients, clinically diagnosed as sepsis with positive culture were included in this study after analyzing selection criteria. APACHE II score model was compared with APACHE II plus RDW score model in relation to mortality outcome assessment. Sensitivity, Specificity, Positive predictive value (PPV), Negative predictive value (NPV) and Receiver Operating Characteristic (ROC) curve were used as parameter to compare the predictive ability of the two models. The derived model APACHE II- RDW was found with higher predictive power (Pearson's correlation coefficient - 0.915) than APACHE II (Pearson's correlation coefficient - 0.885) in relation to mortality (p<0.01). Accuracy was compared by using Receiver Operating Characteristic (ROC) curve between the two models and AUROC was found higher (AUC-0.87) in case of new model compared with conventional model (AUC-0.85). So combination of RDW with APACHE-II increases the predictive ability of the scoring model in relation to mortality.