Predicting ICU length of stay using APACHE-IV in persons with severe sepsis – a pilot study

Abstract

Introduction: Accurate length of stay (LOS) prediction of severe sepsis patients in intensive care unit (ICU) is critical for resource management. Acute Physiology and Chronic Health Evaluation-IV (APACHE-IV) model is commonly used forpredicting LOS. This study assesses the ICU-LOS predictability of APACHE-IV system for severe sepsis patients.Methods: Following ethical clearance, we used ICU data (06/2006 – 08/2008: from a hospital in India) to compare APACHE-IV score and predicted LOS of severe sepsis patients with actual observed ICU-LOS. We employed t-test, correlations, ANOVA andlinear regression of suitably transformed variables as needed.Results: Out of 3,949 ICU admissions, 198 were severe sepsis admissions where 134 patients (80%) had usable data. Of these 75 had verifiable APACHE-IV scores (final sample) with 55% men; median age: 67 years (IQR: 21) 53% did not have dialysis; 87% were on mechanical ventilation (MV). Mean ICU-LOS (10.1 days + 6.4) was significantly greater than predicted ICU-LOS (5.6days + 1.8 ; p<.001). ICU-LOS was very strongly correlated with days on MV (r=0.9). Mean ICU-LOS was significantly greaterfor those receiving blood transfusion (p<.001); on MV (p<.001); having surgery (p<.001) and having high frequency of dialysis (p<.001) – differences not predicted by APACHE-IV. Overall, the predicted ICU-LOS underestimation was by 4.5 days.Conclusions: The results provide a preliminary indication that APACHE-IV model may be a poor predictor of ICU-LOS insevere sepsis cases.