Machine-learning model successfully predicts patients at risk for prolonged postoperative opioid use following elective knee arthroscopy

Abstract

PurposeRecovery following elective knee arthroscopy can be compromised by prolonged postoperative opioid utilization, yet an effective and validated risk calculator for this outcome remains elusive. The purpose of this study is to develop and validate a machine‐learning algorithm that can reliably and effectively predict prolonged opioid consumption in patients following elective knee arthroscopy. MethodsA retrospective review of an institutional outcome database was performed at a tertiary academic medical centre to identify adult patients who underwent knee arthroscopy between 2016 and 2018. Extended postoperative opioid consumption was defined as opioid consumption at least 150 days following surgery. Five machine‐learning algorithms were assessed for the ability to predict this outcome. Performances of the algorithms were assessed through discrimination, calibration, and decision curve analysis. ResultsOverall, of the 381 patients included, 60 (20.3%) demonstrated sustained postoperative opioid consumption. The factors determined for prediction of prolonged postoperative opioid prescriptions were reduced preoperative scores on the following patient‐reported outcomes: the IKDC, KOOS ADL, VR12 MCS, KOOS pain, and KOOS Sport and Activities. The ensemble model achieved the best performance based on discrimination (AUC = 0.74), calibration, and decision curve analysis. This model was integrated into a web‐based open‐access application able to provide both predictions and explanations. ConclusionFollowing appropriate external validation, the algorithm developed presently could augment timely identification of patients who are at risk of extended opioid use. Reduced scores on preoperative patient‐reported outcomes, symptom duration and perioperative oral morphine equivalents were identified as novel predictors of prolonged postoperative opioid use. The predictive model can be easily deployed in the clinical setting to identify at risk patients thus allowing providers to optimize modifiable risk factors and appropriately counsel patients preoperatively. Level of evidenceIII.

Preoperative Predictors of Prolonged Opioid Use in the 6 Months Following Total Knee Arthroplasty

“…These data may point to a potential transition in factors driving prolonged use between the subacute and chronic periods and warrant further study in larger cohorts. Previously published predictive models for extended postoperative opioid use—all based on retrospective data—show an average AUC of 0.76 for preoperative opioid use 62–73 . Our prospective models show that at 6 weeks post-TKA, preoperative opioid use is a less accurate predictor (AUC=0.64) than prior retrospective models indicate, highlighting the importance of psychosocial and pain-related predictors at this time point.…”

“…Previously published predictive models for extended postoperative opioid use-all based on retrospective data-show an average AUC of 0.76 for preoperative opioid use. [62][63][64][65][66][67][68][69][70][71][72][73] Our prospective models show that at 6 weeks post-TKA, preoperative opioid use is a less accurate predictor (AUC = 0.64) than prior retrospective models indicate, highlighting the importance of psychosocial and pain-related predictors at this time point. In contrast, at 6-month follow-up, preoperative opioid use is an even better prospective predictor than prior work would suggest (AUC = 0.90), and the addition of the above phenotypic characteristics improves it even further.…”

Machine learning model successfully identifies important clinical features for predicting outpatients with rotator cuff tears

“…Although these features provided a solid basis for selecting factors and developing models, some important clinical features were inevitably missing. Several studies with clinical datasets suggested that the missing factors were collected with significant personnel resources or were not routinely collected into clinical datasets [7, 18]. In this study, individual comorbidities, such as hypertension or diabetes, were not documented for each patient as they did not have a medical examination.…”

Team Approach: Use of Opioids in Orthopaedic Practice

“…Other risk factors include age, tobacco use, and low educational level 33,36 . Given the many high-risk factors, multiple studies have also proposed machine learning models to predict postoperative opioid consumption [37][38][39][40][41][42] . The identification of this high-risk population should become an integral part of any examination provided in a clinical setting.…”