Comparative Risk Stratification for Prediction of Early Postoperative Morbidity and Mortality after Open Fixation of Periarticular Lower Extremity Fractures

Abstract

Category: Trauma; Ankle; Hindfoot Introduction/Purpose: The standardized identification of patients who are at higher risk of early postoperative adverse events has implications for quality improvement, preoperative medical optimization, and cost reduction through bundled payments. The purpose of the present study was to develop points-based risk stratification systems for predicting 30-day adverse events (AEs) and mortality after open fixation of periarticular hip, knee, and ankle fractures. Methods: Query of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database yielded 65,529 patients who underwent periarticular hip, knee, and ankle fracture repair from 2010-2019. We collected patient demographics and preoperative risk factors. To generate our risk stratification systems, 60% of patients were randomly designated as the development cohort and analyzed with multivariable regression plus bootstrap modeling to identify independent risk factors for early AE and mortality. A nomogram analysis was then conducted to assign scores for each risk factor and generate two points-based risk stratification systems, for AE and mortality. To validate our models, the systems were applied to the remaining 40% of patients (the validation cohort) and tested for predictive ability. Results: In total, 13,212 patients (20.2%) experienced any AE and 3,613 patients (5.5%) mortality within 30 days of fracture fixation. Patients were assigned points for each of the following in both risk stratification systems: fracture type (+4 hip, +2 knee, +0 ankle), male gender (+1), age (>=80 years +5, 60-79 years +3, 40-59 years +2), functionally dependent (+2), anemia (+2 for AE, +1 for mortality), pulmonary disease (+3 for AE, +1 for mortality), congestive heart failure (+3 for AE, +2 for mortality), and end- stage renal disease (+3 for AE, +1 for mortality). Corticosteroid use (+1), hypertension (+1), and insulin-dependent diabetes (+2) were additional predictors for only AEs. The AE and mortality models had maximum scores of 27 and 17 points, and Harrell C statistics of 0.66 and 0.75, respectively. The estimated risk of developing early AE ranged from 3.4-79.5% and mortality from 0.08- 54.4%. Conclusion: Fracture type, male gender, age >=40 years, corticosteroid use, functional dependence, anemia, hypertension, insulin-dependent diabetes, pulmonary disease, congestive heart failure, and end-stage renal disease can be used in the prediction of early AE or mortality following open fixation of periarticular lower extremity fractures, with a marked disparity in estimated risks depending on the number of risk factors possessed by a patient.