Preoperative Risk Index for Patients Undergoing Head and Neck Cancer Surgery

Presentation: AHNS-QS-101
Topic: Other
Type: Quickshot
Date: Thursday, May 2, 2019
Session: 5:30 PM - 6:00 PM
Authors: Marco A Mascarella, MD, MSc, Keith Richardson, MD, MSc, Nader Sadeghi, MD, MSc, Nancy Mayo, PhD
Institution(s): McGill University

Background: Seniors are the largest demographic users of operative resources and most vulnerable to postoperative adverse events. Within head and neck cancer surgery, frailty indices are increasingly being utilized for risk stratification; however, most models lack a multifactorial basis and cannot be directly applied to clinical practice.  

Methods: A cohort analysis of inpatient head and neck cancer surgeries recorded in the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) participant user file from 2006-2016 was performed. The primary outcome was a composite variable of major adverse events including death within 30 days of surgery. The secondary outcome was discharge location to any facility other than home. Sociodemographic, frailty-related and surgical factors in the derivation cohort were evaluated using simple and multiple logistic regression. Predictor variables were subsequently integrated into a preoperative head and neck surgery risk index (HNSRI) and compared to existing models using the validation cohort.

Results: Of the 31 399 operations reviewed, 4556 (14.5%) patients had a major postoperative AE with 209 (0.7%) deaths. Older age, male sex, smoking, anticoagulation, recent weight loss, functional dependence, free tissue transfer, tracheotomy, length of surgery, wound classification, anemia, leukocytosis and hypoalbuminemia were independently associated with major AEs or death on multiple regression analysis (c statistic 0.83, Table 1). The area under the curve (AUC) of the HNSRI to predict major adverse events including death using the validation cohort was 0.84 (95% CI 0.83 – 0.85) with sensitivity 80.1% (95% CI 79.4 – 80.8) and specificity 72.3% (95% CI 70.3 – 74.2, Table 2). The AUC for the HNSRI to predict any morbidity was 0.82 (95% CI 0.82 – 0.83) with sensitivity of 74.1 % (95% CI 72.3 – 75.9) and specificity of 77.4% (95% CI 76.6 – 78.1, Table 3). The HNSRI outperformed existing risk models for prediction of major adverse events including death (P < 0.0001, Figure 1). The predictive ability of the HNSRI for discharge destination was 84% (95% CI 83 – 86) with a sensitivity of 79.7% (95% CI 76.2 – 82.8) and specificity of 76.6% (95% CI 75.9 – 77.3). 

Conclusion: The head and neck surgery risk index accurately predicts major postoperative adverse events including death in the studied population. This risk index can be used to counsel patients awaiting head and neck cancer surgery.

Table 1Table 2Table 3Figure 1