1563. Elixhauser Comorbidity Index Predicts Complications Following Anterior Cervical Discectomy and Fusion

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

Introduction:

Despite a significant body of literature regarding individual predictors of poor outcomes for patients undergoing anterior cervical discectomy and fusion (ACDF), there is a paucity of information regarding composite comorbidity indicators, such as the Elixhauser Comorbidity Index (ECI), and their ability to predict postoperative complications in spine surgery. The present study investigates whether the ECI can predict postoperative complications in patients undergoing ACDF.

 

Methods:

All patients undergoing ACDF at a single institution from 2008-2016 and all patients in the National Inpatient Sample (NIS) from 2013-2014 who did not receive a concurrent posterior cervical surgery were included. The ECI was calculated and concordance statistics (c-statistics) representing the area under the receiver operating characteristic curve were calculated for postoperative complications to demonstrate whether the ECI is an accurate risk stratification tool.

 

Results:

From 2013-2014, 278,070 hospitalizations from the NIS were identified, with an average age of 55.15, 51.17% female, and 2.45% with an APR-DRG Risk of Mortality score >2. Strong predictive relationships (c≥0.8) were seen for pulmonary embolism (c=0.82), sepsis (c=0.83), septic shock (c=0.83), and death (c=0.83). Good predictive relationships (c=0.7-0.79) were seen for airway complications (c=0.78), and renal failure (c=0.75). Moderate predictive relationships (c=0.6-0.69) were seen for bleeding complications (c=0.65), cardiac arrest (c=0.65), urinary tract infection (c=0.67), and nonhome discharge (c=0.67).

 

From 2008-2016, 2,408 patients from the single institution were included with an average age of 51.92, 51.16% female, and 1.17% with an APR-DRG Risk of Mortality score >2. A strong predictive relationship was seen for sepsis (c=0.87). Good predictive relationships were seen for septic shock (c=0.75) and death (c=0.75). Moderate predictive relationships were seen for renal failure (c=0.66) and urinary tract infection (c=0.61).

 

Conclusions:

The ECI has a variable ability to predict complications following ACDF.