Predicting Patient-Reported Outcome Measures, Satisfaction, Healthcare Utilization, Mortality, and Return to Work After Total Knee Arthroplasty Using Machine Learning: A 14,900-Patient Study.
Jul 2026· Journal of Arthroplasty· 0 citations· 33 references
Medicine
TL;DR
This study developed and validated machine learning models to predict postoperative patient-reported outcomes, satisfaction, healthcare utilization, mortality, and return to work after primary TKA.
Abstract
Background
Patient-reported outcome measures and healthcare utilization metrics are increasingly used to evaluate the success of total knee arthroplasty (TKA). Predictive models may improve preoperative planning, patient counseling, and resource allocation. This study developed and validated machine learning models to predict postoperative patient-reported outcomes, satisfaction, healthcare utilization, mortality, and return to work after primary TKA.
Methods
A prospective cohort of 14,900 patients who underwent primary unilateral TKA at a large tertiary academic center from 2016 to 2022 was analyzed. Patients undergoing bilateral TKA or missing baseline patient-reported outcome measures were excluded. Random Forest and XGBoost models were developed using baseline demographic, clinical, socioeconomic, and surgical variables. Outcomes included postoperative Knee injury and Osteoarthritis Outcome Score (KOOS) Pain, Physical Function, Joint Replacement, and Quality of Life subscales; Patient Acceptable Symptom State; length of stay; discharge disposition; 90-day readmission; 1-year mortality; and return to work. Model performance was evaluated using root mean square error for continuous outcomes and accuracy for categorical outcomes.
Results
Predictive performance was moderate to strong across outcomes. For KOOS outcomes, root mean square error ranged from 15.13 to 23.49. Accuracy for healthcare utilization outcomes ranged from 55 to 71%. Accuracy reached 73% for 1-year mortality and 78% for return to work. Important predictors across models included baseline KOOS Joint Replacement score, patient-reported outcome phenotype, age, body mass index, Area Deprivation Index, race, and surgery start time. The Charlson Comorbidity Index also contributed to the prediction of select clinical outcomes.
Conclusion
Machine learning models demonstrated moderate to strong performance for predicting patient-reported outcomes, satisfaction, healthcare utilization, mortality, and return to work after TKA. Integration of these tools may allow population-level variable parsing - identifying distinct socioeconomic, functional, and operational signals within an optimized cohort. External validation is needed before widespread clinical implementation.
Background: Knee osteoarthritis (OA) is a leading cause of chronic pain and functional disability, disproportionately affecting elderly individuals in low- and middle-income countries. Total knee arthroplasty (TKA) is the definitive treatment for end-stage disease, yet outcome data from tertiary care hospitals in Uttar Pradesh using validated scoring tools remain limited. Objectives were to assess the preoperative and postoperative functional outcomes of patients undergoing primary TKA using the Knee Society Score (KSS), and to examine associations between functional outcome and patient-level demographic variables.
Methods: A hospital-based prospective observational study was conducted in the Department of Orthopaedic Surgery, Balrampur Hospital, Lucknow, from September 2024 to August 2025. Sixty-five consecutive patients with primary knee OA undergoing primary TKA were enrolled. Patients with inflammatory or post-traumatic arthritis, revision surgery, active infection, or neuromuscular disorders were excluded. Functional outcome was assessed using the two-component KSS (Knee score and functional score) preoperatively and at 12-month follow-up. Preoperative-postoperative comparison was performed using paired t-test; associations between functional outcome category and demographic variables were assessed using the chi-square test (p<0.05 considered significant).
Results: Mean patient age was in the sixth decade, with female predominance (63.1%). The mean Knee Score improved from 41.6±8.9 preoperatively to 86.9±6.7 at 12 months (mean difference 45.3; p<0.001), and the mean functional score from 38.2±9.4 to 82.4±7.9 (mean difference 44.2; p<0.001). Excellent outcomes were achieved in 73.8% of patients and good outcomes in 18.5%, yielding a combined excellent-to-good rate of 92.3%. Patients aged ≥60 years demonstrated significantly better functional outcomes than younger patients (p=0.024). Gender and body mass index did not independently influence postoperative functional scores. No cases of deep infection or revision surgery were recorded during follow-up.
Conclusions: Primary TKA produces significant, sustained improvement in knee function and mobility at one year, achieving excellent-to-good outcomes in over 90% of patients. The KSS is a reliable and responsive instrument for routine outcome audit. Appropriate patient selection, structured rehabilitation, and targeted expectation management-particularly in younger patients-are key determinants of optimised postoperative results.
A. Yadav, S. Yadav, A. K. Yadav· International Journal of Res...· 0 citations
Patient-reported outcome measures (PROMs) are increasingly used for benchmarking hip and knee arthroplasty, yet evidence on their feasibility for public reporting of short-term rehabilitation outcomes remains limited. International initiatives such as ICHOM recommend combining disease-specific and generic PROMs, but practical frameworks for risk-adjusted benchmarking in routine rehabilitation are scarce. In this prospective pilot study, 23 German rehabilitation facilities contributed data on patients following total hip (THA) or knee arthroplasty (TKA). Patients completed the HOOS-PS or KOOS-PS and the VR-12 PCS/MCS at admission and discharge after a standardized three-week multimodal rehabilitation program. For primary analyses, 961 THA and 961 TKA cases with complete pre- and post-data on at least one PROM were available. Facility-level unadjusted change was quantified using Cohen’s dz. Risk adjustment used pooled multiple linear regression including baseline PROMs, age, sex, and additional covariates retained via backward elimination (p < 0.10), yielding standardized residual effect sizes (dres) per facility. A composite Patient-Reported Outcome Quality Index (ProQI) weighted disease-specific standardized residuals at 50% and VR-12 PCS/MCS at 25% each, rescaled to a 0–100 metric with a normative mean of 90 and SD 30. Pooled pre-post effect sizes indicated clinically relevant improvements (THA: dz = 0.76 HOOS-PS, 0.76 PCS, 0.45 MCS; TKA: 0.83 KOOS-PS, 0.71 PCS, 0.39 MCS), with between-facility variability exceeding discrimination thresholds. Risk-adjusted dres identified facilities performing above or below expectations. ProQI values showed substantial dispersion (THA range 75.4–100; TKA 76.9–100), enabling intuitive differentiation of providers. Overall, ProQI completeness was 69% (THA) and 72% (TKA). Non-completer analyses showed small baseline differences but no systematic ProQI bias. PROM-based, risk-adjusted benchmarking of short-term rehabilitation outcomes after hip and knee arthroplasty is feasible and meaningfully discriminative. The ProQI composite offers an interpretable, biopsychosocially grounded index for public reporting and quality management. Future work should enhance data completeness, refine risk adjustment, and validate the approach across broader cohorts and settings.
Objective The aim of this study was to examine temporal trends in metabolic syndrome–associated osteoarthritis (MetS-OA) among patients undergoing revision total knee arthroplasty (RTKA) and to identify patient- and clinical-level factors associated with this condition and related outcomes. Methods A retrospective cohort analysis was conducted using data from the Nationwide Inpatient Sample from 2010 through 2019. Patient demographics, hospital characteristics, length of stay, total hospitalization charges, in-hospital mortality, comorbid conditions, and perioperative complications were assessed. All analyses incorporated NIS discharge weights, and multivariable logistic regression models were used to assess associations between MetS-OA and clinical outcomes among patients undergoing RTKA. Results Among 1,361,454 RTKA hospitalizations identified, 1,330,099 RTKA hospitalizations were included in the analysis. The overall prevalence of MetS-OA was 16.1%, demonstrating a progressive increase from 2011 through 2019. Factors independently associated with MetS-OA included advanced age, male sex, non-White racial background, and the presence of comorbid conditions such as chronic pulmonary disease, depression, and hypothyroidism. Patients with MetS-OA experienced slightly longer hospital stays and incurred higher median total hospitalization charges, exceeding those without MetS-OA by $1,445.50. In-hospital mortality did not differ significantly between groups. MetS-OA was also associated with a higher likelihood of postoperative complications, including acute myocardial infarction, severe malnutrition, acute cerebrovascular disease, postoperative delirium, acute respiratory distress syndrome, prolonged mechanical ventilation, urinary tract infections, acute renal failure, and lower limb nerve injury. Conclusion The prevalence of MetS-OA among patients undergoing RTKA has increased over time and is associated with a higher burden of postoperative complications and healthcare utilization. Patient- and hospital-level factors play a substantial role in shaping these outcomes. Targeted preoperative optimization and standardized perioperative management strategies for patients with MetS-OA may mitigate adverse events, enhance postoperative recovery, and reduce overall hospitalization charges.
Miaolan Yuan, Hao Xie, Yanjie He et al.· Frontiers in Endocrinology· 0 citations
BACKGROUND
Many patients who have bilateral hip osteoarthritis eventually undergo contralateral total hip arthroplasty (THA), yet limited data exist on how recovery differs between staged procedures performed within one year. This study evaluated differences in patient-reported outcomes and healthcare utilization following staged bilateral THA and identified predictors of suboptimal outcomes after the second surgery.
METHODS
We retrospectively reviewed a prospective institutional registry to identify patients who underwent staged bilateral primary THA within one year (n = 680) and completed one-year postoperative patient-reported outcome measures (PROMs) for both procedures. The primary outcomes included discharge disposition, prolonged lengths of stay (LOS ≥ three days), 90-day readmission, and one-year reoperation. The secondary outcomes included Hip Disability and Osteoarthritis Outcome Score (HOOS) Pain, HOOS Physical Function Shortform (PS), and Joint Replacement subscale (JR) scores. Minimal clinically important difference (MCID) and patient acceptable symptom state (PASS) thresholds were applied. Logistic regression identified predictors of suboptimal second-side outcomes.
RESULTS
Readmission (4.9 versus 2.8%, P = 0.045) and reoperation (3.1 versus 0.6%, P = 0.001) were more frequent after the second THA. Patients who had a prolonged LOS after the first THA had 20-fold greater odds of prolonged LOS after the second. The one-year PROM scores were similar across surgeries, but symptom improvement was consistently lower after the second THA (HOOS Pain: 58 versus 50, P < 0.001; PS: 43 versus 34, P < 0.001; JR: 46 versus 38, P < 0.001). Failure to achieve MCID was more common after the second THA, while PASS rates remained stable. Failure to reach PASS thresholds after the first THA strongly predicted failure after the second (ORs [odds ratios] 7.7 to 14.8).
CONCLUSION
Patients experienced diminished improvement and greater healthcare utilization rates after their second surgery. Poor recovery following the first THA was highly predictive of second-side failure. These findings support individualized surgical timing and counseling strategies based on early recovery trajectories.
Shujaa T Khan, Ahmed Emara, Chao Zhang et al.· Journal of Arthroplasty· 0 citations
BACKGROUND
Patient-reported outcome measures (PROMs) are increasingly used to evaluate disease-specific and overall health improvements following total hip and knee arthroplasty (THA/TKA). While disease-specific changes are well documented, the extent to which these procedures influence perceived overall health remains unclear. The aim of this study was to analyze changes in both disease-specific and overall health PROMs from the preoperative visit to one-year postoperatively in patients who underwent THA and TKA.
METHODS
A retrospective cohort study was conducted using data from patients who underwent primary THA or TKA between 2016 and 2022 at a tertiary-care academic center and completed pre- and postoperative PROMs. Propensity score matching was applied to balance the two groups on demographic and clinical variables. Correlations were assessed between changes in Hip dysfunction and Knee injury Osteoarthritis Outcome Score for Joint Replacement (HOOS JR/KOOS JR) and Patient-Reported Outcomes Measurement Information System Global Health (PROMIS-10) subscales (physical health, mental health, and quality of life [QOL]) using Pearson correlation (r) and Spearman's rank coefficients (⍴).
RESULTS
Matched cohorts included 609 patients who underwent a THA procedure and 609 patients who underwent a TKA procedure. Changes in HOOS JR/KOOS JR were moderately to strongly correlated with physical health (THA: r = 0.63, 95% confidence interval [CI] = [0.58 to 0.68]; TKA: r = 0.56 [CI, 0.51 to 0.62]), but also weakly correlated with mental health (THA: r = 0.33 [CI, 0.26 to 0.40]; TKA: r = 0.24 [CI, 0.16 to 0.31]) and QOL (THA: ⍴ = 0.31 [CI, 0.24 to 0.38]; TKA: ⍴ = 0.22 [CI, 0.15 to 0.30]; all P-values for the correlations < 0.001). There were no statistically significant differences between the THA and TKA groups for the strength of correlations between joint-specific scores and physical health (P = 0.088), mental health (P = 0.103), and QOL (P = 0.101).
CONCLUSION
Findings from this study support the use of PROMs in evaluating THA and TKA surgical outcomes and emphasize the extent to which hip and knee health impacts patient-perceived overall health.
Jaeyoung Park, Emilie N. Miley, Xiang Zhong et al.· Journal of Arthroplasty· 0 citations
INTRODUCTION
Total knee arthroplasty (TKA) improves pain and function in osteoarthritis patients, but recovery varies with age. TKA patients average at 68.7 years old, but with wide variation, the impact of age on outcomes remains unclear. This study investigated the impact of age on recovery after TKA to provide guidelines for TKA in different age groups.
METHODS
This retrospective study included patients who underwent TKA between 2018 to mid-2023 and had 1 year post-operative patient-reported outcome measures (PROMs) data. Patients completed the Oxford Knee Score (OKS), Forgotten Joint Score (FJS), VR-12 Mental Component Score (MCS) and Physical Component Score (PCS) pre-operative and 1 year post-operative. Patient Medical Care Satisfaction, Knee Normality Score, and Visual Analogue Scale (VAS) (0-100) were recorded 3 months and 1 year post-operative. Data were analyzed and adjusted for sex, body mass index (BMI), range of motion (ROM), implant type, surgeon volume, and joint alignment.
RESULTS
Of the 1,723 participants (mean age 69.4 years, 50.7% males, average BMI 30 kg/m2), pre-operative PROMs were higher with age [OKS (p<0.001), FJS (p<0.001), MCS (p<0.001)]. At 3 months post-operative, older patients had lower Patient Medical Care Satisfaction (p=0.005) and Knee Normality Score (p=0.008) and higher VAS (p=0.001). At 1 year, older age was linked to lower OKS (p=0.002), PCS (p<0.001), and Patient Medical Care Satisfaction (p=0.02). After adjusting for covariates, OKS (p<0.0001), PCS (p<0.0001), and FJS (p=0.006) were statistically significant. ANCOVA modelling predicted cumulative difference in adjusted score between the youngest and oldest patients to exceed published minimal clinically important difference (MCID) thresholds for FJS and VR-12 PCS, but oldest age groups still experienced meaningful improvement in PROMs between pre- and post-operative, meeting all MCIDs.
CONCLUSION
Older age is linked to better pre-operative PROMs, but worse outcomes after surgery. These results highlight the relationship between age and recovery following TKA, indicating worse improvements in knee function for older patients. However, all post-operative PROMs in older age groups still met MCIDs, thus showing the substantial impact of TKAs in patients of all ages.
LEVEL OF EVIDENCE
Level IV retrospective study using historical controls or having more than one negative criterion.
Tess L. Y. Tsoi, H. Ahedi, David A. Parker· Journal of ISAKOS Joint Diso...· 0 citations