OP0071 PREDICTING PERSISTENT HIGH-DOSE OPIOID USE AFTER TOTAL KNEE REPLACEMENT

Abstract

Background:Patients undergoing total knee replacement (TKR) are at increased risk of persistent opioid use and dependenceObjectives:To identify patients with persistent high-dose opioid use after TKR using group-based trajectory models (GBTM) and determine predictors of persistent high-dose opioid users using pre-TKR patient characteristicsMethods:Using US Medicare claims (2010-2014), we identified patients aged ≥65 years who underwent a TKR and had no history of cancer or high-dose opioid use (>25 mean morphine equivalents (MME)/day) in the year prior. All patients were continuously enrolled in Medicare for ≥360 days prior to and ≥30 days after the TKR. To determine opioid filling patterns after the surgery, patients were followed up to 360 days from the day of TKR. We modeled 12 monthly indicators of opioid prescription fills as a continuous (MME/day) variable using a censored normal GBTM and categorized patients into 4 groups. The primary outcome was persistent high-dose opioid use defined as patients in trajectory Group 3 (38.8 MME/day) or Group 4 (22.4 MME/day). We split the data into training (2010-2013 data) and test (2014 data) sets and used logistic regression to predict high-dose opioid use vs low-dose opioid use (Groups 1 and 2) as a binary outcome utilizing pre-TKR patient characteristics as candidate predictors using the least absolute shrinkage and selection operator (LASSO) regression for variable selection. A reduced model with only 10 pre-specified variables readily available for clinical use was also consideredResults:The final study cohort included 142,089 patients. The GBTM identified 4 distinct trajectories (Group 1- Short-term, low-dose, Group 2- long-term, low-dose, Group 3- medium-term, high-dose, Group 4-long-term, high-dose) of opioid use in the year after TKR(Figure). Using logistic regression and LASSO, we predicted the probability of persistent high-dose opioid use (N=17,171) (vs. low-dose opioid use) in the training set (N=101,810) for an AUC=0.80. The AUC in the test set (N=40,279) predicting high opioid use (N=5,893) was 0.77. The final model selected 33 variables and identified baseline history of opioid use as the strongest positive predictor of high-dose persistent opioid use. The reduced model with only ten predictors also performed equally well (AUC=0.77)(Table).Conclusion:In this cohort of older patients with no history of cancer or high-dose opioid use at baseline, 16.2% became high dose (28.1 MME/day) opioid users during the year after TKR. Our prediction model with 10 readily available clinical factors may help identify patients at high risk of future adverse outcomes from persistent opioid use and dependence after TKRFigure. Trajectories of opioid use patterns after TKRTable.Predictors of persistent high-dose opioid use in the reduced modeVariableMultivariable Odds Ratio (95% CI)Predicting High dose vs.Low dose opioid useP-valueAge (in years)0.94 (0.93-0.94)<0.001Females (Ref=Males)0.99 (0.93-1.06)0.78White race (Ref=Other)1.25 (1.04-1.50)0.02Baseline opioid use (MME/day)1.22 (1.22-1.23)<0.001Substance use (Yes/No)1.10 (1.02-1.20)0.02Benzodiazepine use (Yes/No)1.22 (1.12-1.32)<0.001Anxiolytic use (Yes/No)1.30 (1.19-1.43)<0.001Anticonvulsant use (Yes/No)0.94 (0.87-1.03)0.19Antidepressant use (Yes/No)1.03 (0.96-1.11)0.36NSAID use (Yes/No)1.07 (1.00-1.14)0.04Disclosure of Interests:Chandrasekar Gopalakrishnan: None declared, Jessica Franklin: None declared, Yinzhu Jin: None declared, Daniel Solomon Grant/research support from: Funding from Abbvie and Amgen unrelated to this work, Jeffrey Katz Grant/research support from: Dr Katz reported receiving grants from Samumed and Flexion Therapeutics outside the submitted work., Yvonne Lee Shareholder of: Cigna-Express Scripts, Grant/research support from: Pfizer, Consultant of: Highland Instruments, Inc., Patricia Franklin: None declared, Joyce Lii: None declared, Rishi J Desai Grant/research support from: Dr. Desai reported receiving grants from Bayer, Novartis, and Vertex Pharmaceuticals outside the submitted work., Seoyoung Kim Grant/research support from: Seoyoung C Kim has received research grants from AbbVie, Roche, Bristol-Myers Squibb and Pfizer.