Abstract
Introduction
The Comprehensive Care for Joint Replacement (CJR) model for total hip arthroplasty (THA) involves a target reimbursement set by the Center for Medicare and Medicaid Services (CMS). Many patients exceed these targets, but predicting risk for incurring these excess costs remains challenging, and we hypothesized that select patient characteristics would adequately predict CJR cost overruns.
Methods
Demographic factors and comorbidities were retrospectively reviewed in 863 primary unilateral CJR THAs performed between 2013 and 2017 at a single institution. A predictive model was built from 31 validated comorbidities and a base set of 5 patient factors (age, gender, BMI, ASA, marital status). A multivariable logistic regression model was refined to include only parameters predictive of exceeding the target reimbursement level. These were then assigned weights relative to the weakest parameter in the model.
Results
The overall cost of care for 225 patients (26.1%) exceeded the target price, and a comprehensive model containing all 36 parameters demonstrated adequate discrimination (AUC: 0.748). This model was narrowed to 12 parameters retained for their statistical value in predicting excess cost, without substantial loss of predictive ability (AUC: 0.735). A single score formed from the sum of each patient's weighted parameters also showed adequate discrimination (AUC: .732), with predicted risk for exceeding CJR targets ranging from 10% for a patient score of 10 to 80% for a score of 30. Average scores for patients exceeding the target price were significantly higher than those who did not (19.5 vs 15.0, p < 0.0001).
Conclusions
A model composed of weighted comorbidities and base demographics provides adequate discrimination in predicting whether THA costs will exceed CJR targets. This not only helps identify patients who may benefit from further pre-operative optimization, but also allows health systems to predict the likely minimum incurred costs for select patients scheduled for surgery.
