Exercise-based rehabilitation interventions for kinesiophobia after total knee arthroplasty: a systematic review and meta-analysis of randomized controlled trials

Abstract

Affecting 21–35% of patients after total knee arthroplasty (TKA), kinesiophobia critically impedes postoperative rehabilitation. Whether exercise-based interventions reduce kinesiophobia more effectively than routine postoperative rehabilitation remains unclear. We systematically searched 11 databases from inception to July 11, 2026, for RCTs comparing exercise-based rehabilitation with routine postoperative rehabilitation in adults undergoing primary TKA. The primary outcome was kinesiophobia. Secondary outcomes included knee function, pain, balance, mobilization time, and adherence. Random-effects meta-analyses were performed using RevMan 5.4. Certainty of evidence was assessed using GRADE. Eleven RCTs (N = 1,096) were included. Exercise-based rehabilitation was associated with a reduction in kinesiophobia (SMD = − 1.22, 95% CI [− 1.72, − 0.72]; I2 = 93%; low certainty) and improvements in knee function (SMD = 1.19, 95% CI [0.67, 1.71]; I2 = 93%; low certainty). For pain, the ≤ 4-week pooled estimate (SMD = − 1.09, 95% CI [− 2.03, − 0.14]; I2 = 94%) was not interpretable due to extreme heterogeneity and was not robust: after exclusion of the single acute in-hospital trial, the confidence interval crossed zero (SMD = − 1.09, 95% CI [− 2.28, 0.09]). Leave-one-out sensitivity analyses otherwise supported the directional conclusions for kinesiophobia and knee function, although effect sizes were unstable and certainty remained low. Low-certainty evidence suggests that exercise-based rehabilitation may be associated with reductions in kinesiophobia at ≤ 4 weeks and an interpretable 12-week improvement in knee function after TKA; the ≤ 4-week and ≥ 24-week functional subgroup estimates were not interpretable because of extreme heterogeneity, and the early pain signal was possible but non-robust. Effect sizes should be interpreted with caution due to substantial heterogeneity, risk of bias, and sensitivity to outlying studies. Well-designed, multicenter RCTs with rigorous blinding and standardized protocols are urgently needed.