Psychological and Knee Function Measures at 1 Month Predict High Kinesiophobia at 6 Months After Anterior Cruciate Ligament Reconstruction: Development and Internal Validation of a Prognostic Model

Abstract Background To develop and internally validate a prognostic model based on psychological and knee-function measures obtained 1 month after surgery to identify patients with high kinesiophobia 6 months after anterior cruciate ligament reconstruction (ACLR). Methods Data from a single-center prospective cohort were used, with postoperative month 1 defined as the prediction landmark. High kinesiophobia was defined as a 17-item Tampa Scale for Kinesiophobia (TSK-17) score ≥ 37 at postoperative month 6. Based on prior evidence, clinical relevance, construct representation, and sample-size constraints, seven predictor parameters were prespecified: age, sex, TSK-17, ACL-RSI, K-SES, HADS-A, and IKDC. A demographic model, a model adding TSK-17, and a multidimensional full model were fitted sequentially. All three models underwent 2,000 paired bootstrap resamples for internal validation, and the optimism-corrected incremental performance of the full model over the TSK model was estimated. Flexible calibration curves, inverse-probability-weighted sensitivity analysis for attrition, post hoc sample-size precision assessment under the Riley framework, and checks for nonlinear effects of continuous predictors were performed. Results Of 209 patients who met the age criterion, 177 entered the primary analysis; 84 (47.5%) had high kinesiophobia. Apparent AUCs for the three models were 0.600, 0.821, and 0.874, respectively, and optimism-corrected AUCs were 0.578, 0.812, and 0.852. For the full model versus the TSK model, the corrected ΔAUC was 0.040 (bootstrap 95% interval, -0.015 to 0.074) and the corrected ΔBrier was − 0.017 (-0.036 to 0.013), suggesting possible incremental value but limited precision. The optimism-corrected Brier score of the full model was 0.162, and the calibration slope was 0.853. Coefficients and performance changed little after inverse probability weighting. Conclusion TSK-17 at postoperative month 1 provided the core predictive information for high kinesiophobia at month 6. Multidimensional measures such as ACL-RSI may provide additional information, but the interval estimates for incremental performance included no improvement. The current sample is insufficient for precise individual risk estimation; the full model should be regarded as a preliminary development model and externally validated in a larger independent cohort.