Isolating Factors for the Prediction of Rebound After Guided Growth with Tension Band Plating for the Valgus Deformity of the Knee

Purpose: The conditions leading to growth rebound after hemiepiphysiodesis are still poorly understood. This article analyzes the radiographical outcomes after guided growth with tension band plating, using plates in idiopathic genu valgum patients and attempts to generate a predictive model of growth rebound.

Methods: Patients with idiopathic genu valgum deformity who received tension band plating were selected for evaluation. We only analyzed coronal plane deformities. Only patients with a long-standing X-ray before tension band plating surgery, a long-standing X-ray at tension band plating removal, and a long-standing X-ray at the latest follow-up after tension band plating removal were considered for this study. The change of mechanical axis deviation between the tension band plating removal and the last follow-up was evaluated for rebound, and ordinal logistic regression was performed to determine the relevant variables for predictive modeling rebound growth.

Results: Overall, 100 patients (189 legs) were analyzed. The mean mechanical axis deviation at tension band plating removal was 8.4 mm in varus direction, and the mean mechanical axis deviation at the last follow-up was -3.4 mm (p ≤ 0.001). However, 111 legs (59%) showed rebound growth, 57 (30%) stayed stable, and 21 (11%) showed a continuous correction. Six significant factors significantly influencing rebound were isolated which are clinically relevant: sex, age, baseline mechanical axis deviation, mechanical lateral distal femoral angle, and mechanical medial proximal tibial angle, and mechanical axis deviation correction rate. Mechanical axis deviation correction rate had the highest odds ratios. The machine learning classification model for predicting rebound growth built from the study data showed a misclassification rate of 39%.

Conclusion: There was a high rate of rebound growth in this cohort, especially for patients at a young age at implantation. The highest risk factors for rebound growth were male sex, and high correction rates, such as found during peak growth spurt. The proposed classification model needs more data to improve its predictive power before it can be used in clinics.

Level Of Evidence: Level III.