1568. Evaluating the Elixhauser Comorbidity Index as a Risk Predictor for Post-Operative Complications in Posterior Cervical Discectomy and Fusion

Authors: Marcus Laroche; Sean Neifert; Daniel Snyder; Jonathan Gal, MD; Brian Deutsch; John Caridi, MD (New York, NY)


The Elixhauser Comorbidity Index (ECI) is used to risk stratify patients by their comorbidities. The present study examines the Elixhauser Comorbidity Index (ECI) as a predictor of post-operative complications for patients who underwent posterior cervical discectomy and fusion (PCDF).


All patients undergoing PCDF at a single institution from 2008-2016 and all hospitalizations in the National Inpatient Sample (NIS) from 2013-2014 were queried if they did not undergo concurrent anterior cervical spine surgery. Concordance statistics (c) representing the area under the receiver operating characteristic curve were calculated to understand the association between the ECI and various outcomes.


The ECI was a strong predictor (c>0.8) in the institutional database for PE (c = 0.906), dehiscence (c = 0.849), septic shock (c = 0.977), and death (c = 0.932). Good predictive relationships (c=0.7-0.79) were seen for DVT (c = 0.778) and sepsis (c = 0.73), and moderate relationships (c=0.6-0.69) were seen for renal failure (c = 0.667), MI (c = 0.645), cardiac arrest (c = 0.688), and nonhome discharge (c = 0.607).

Among hospitalizations in the NIS, the ECI was a strong predictor for sepsis (c = 0.805), septic shock (c = 0.805), and death (c = 0.817). The ECI had good predictive relationships with airway complications (c = 0.769), renal failure (c = 0.73), cardiac arrest (c = 0.707), DVT (c = 0.737), and PE (c = 0.785). It had moderate predictive relationships with bleeding (c = 0.651), MI (c = 0.662), cerebrovascular accident (c = 0.681), UTI (c = 0.675), nonhome discharge (c = 0.644), and prolonged LOS (c = 0.69).


The Elixhauser Comorbidity Index can be a useful tool for predicting risk of post-operative complications. However, further analysis needs to be done to prove its reliability throughout different data sources.