Growth Mixture Modeling of Patient-reported Outcomes After Total Knee Arthroplasty: No Recovery Trajectory Shows
Kareem Omran1, Colleen Wixted, Daniel Waren
1From the Department of Public Health and Primary Care, University of Cambridge, Cambridge, United Kingdom (Mr. Omran), and the Department of Orthopedic Surgery, NYU Langone Health, New York, NY (Dr. Wixted, Waren, Dr. Rozell, and Dr. Schwarzkopf).
Background:
Recovery after total knee arthroplasty (TKA) shows considerable variability in both pain relief and functional improvement. The Knee Injury and Osteoarthritis Outcome Score (KOOS-JR) is a widely used measure for evaluating these outcomes. This study aimed to identify distinct latent recovery trajectories, which represent underlying, unobserved patterns of postoperative recovery inferred from KOOS-JR scores, and to explore patient characteristics associated with these trajectories.
Methods:
This retrospective cohort study analyzed patients who underwent primary TKA for osteoarthritis at a tertiary academic center from January 2020 to March 2023. Inclusion criteria required patients to have completed a preoperative KOOS-JR questionnaire and at least two postoperative follow-ups at 1, 3, 6, or 12 months. Exclusion criteria included bilateral or revision procedures. Collected characteristics included age, sex, Body Mass Index, American Society of Anesthesiologists physical status classification, race, smoking status, procedure type, anesthesia type, length of hospital stay, and discharge disposition. Growth mixture modeling was used to model recovery trajectories, with associations evaluated using the "three-step approach." Model fit was assessed using the Akaike and Bayesian Information Criteria, Vuong-Lo-Mendell-Rubin likelihood ratio, posterior probabilities, and entropy values.
Results:
Of 700 eligible patients, growth mixture modeling identified two recovery trajectories: 95.4% of patients (trajectory 1 [T1]) demonstrated steady improvement, while 4.6% (trajectory 2 [T2]) began with lower KOOS-JR scores (mean 9.7 vs. 47.9 for T1) but recovered to near T1 levels by 1 month. Trajectory 2 patients were markedly younger (mean 64 vs. 67 years), had higher Body Mass Index (36 vs. 31), included more Black or African American individuals (38% vs. 20%), and were more frequently discharged to rehabilitation facilities (16% vs. 3.3%; all P < 0.05). Each additional year of age reduced the likelihood of following T2 by 4% (odds ratio = 0.96, 95% confidence interval, 0.92 to 0.99; P = 0.016), while discharge to rehabilitation increased the likelihood 6-fold (odds ratio = 6.22, 95% confidence interval, 1.89 to 17.8; P = 0.001).
Conclusion:
This study identified two distinct recovery trajectories after TKA, with notably no trajectory emerging showing decline or stagnation from preoperative levels. Despite lower baseline scores, patients in T2 achieved substantial recovery, suggesting TKA provides meaningful improvement even for those with substantially compromised function. The findings also highlight the need to explore whether rehabilitation discharge directly influences the observed postoperative gains.
