Effectiveness of Interactive Remote Rehabilitation After Total Knee Arthroplasty: Systematic Review and Meta-Analysis of Randomized Controlled Trials.

Background: Remote rehabilitation has become increasingly relevant after total knee arthroplasty (TKA). However, existing reviews have often pooled interventions with markedly different levels of clinician involvement, making it difficult to determine whether interactive, feedback-enabled models provide distinct clinical value. Objective: This systematic review and meta-analysis evaluated the effectiveness of interactive remote rehabilitation (IRR), defined as technology-enabled rehabilitation involving bidirectional communication between patients and health care providers, compared with conventional rehabilitation after TKA. Methods: PubMed, Cochrane CENTRAL, Embase, Web of Science, CINAHL, Scopus, and CNKI were searched from inception to May 5, 2026. Randomized controlled trials enrolling adults after TKA and comparing IRR with conventional rehabilitation were eligible. Outcomes included pain, patient-reported function, range of motion, quadriceps muscle strength, mobility, general health status, and health-related quality of life. Risk of bias was assessed using the Cochrane Risk of Bias 2 tool. Random-effects meta-analyses used Hartung-Knapp-Sidik-Jonkman CIs and Nagashima-corrected 95% prediction intervals (PIs); analyses were stratified as short-, mid-, and long-term follow-up. The protocol was prospectively registered in PROSPERO (CRD420251049015). Results: In total, 23 randomized controlled trials involving 2607 participants were included. A total of 7 studies were judged to be at low risk of bias, 7 raised some concerns, and 9 were at high risk of bias. Across most primary and secondary outcomes, IRR did not show statistically significant advantages over conventional rehabilitation, including pain, Western Ontario and McMaster Universities Osteoarthritis Index, Knee Injury and Osteoarthritis Outcome Score, Timed Up and Go test, and EQ-5D outcomes, across follow-up periods. A very small benefit was observed for short-term active extension range of motion (mean difference 0.26, 95% CI 0.04-0.48; 95% PI 0.00-0.52), which remained statistically significant after excluding high-risk studies but was of limited clinical magnitude. Short-term quadriceps muscle strength favored IRR in the primary analysis (standardized mean difference 0.60, 95% CI 0.01-1.19; 95% PI -0.63 to 1.87), but the effect was not robust after sensitivity analysis, and the PI crossed the null. Evidence for 36-Item Short Form Survey and 6-minute walk test outcomes was insufficient for quantitative synthesis. Heterogeneity was substantial for several outcomes, PIs were frequently wide, and certainty of evidence was generally low to very low. Conclusions: This review is innovative in focusing specifically on bidirectional, IRR rather than treating all remote or technology-assisted rehabilitation as a single category, and it differs from prior reviews by combining this conceptual distinction with time-stratified analyses and more conservative random-effects inference. This synthesis brings to the field a clearer and more clinically interpretable assessment of what IRR currently adds after TKA: available evidence does not establish superiority over conventional rehabilitation, although IRR may represent a feasible care model where access to in-person rehabilitation is limited. These findings can inform service planning and trial design while underscoring the need for higher-quality studies to identify which interactive features, patient groups, and contexts are most likely to benefit.