Accurate assessments of skeletal maturity is of critical importance to guide type and timing of orthopaedic surgical interventions. Several quantitative markers of the proximal tibia were recently developed using historical knee radiographs. The purpose of the present study was to determine which marker would be most effective in assessment of full-length radiographs in a modern pediatric patient population.
All full-length radiographs at our institutions between 2013 and 2018 were reviewed. Inclusion criteria for our study required that the child reached final height as defined by 2 consecutive unchanged heights, at least 6 months apart, after age 16 for boys and 14 for girls. Patients with metabolic bone disease, prior surgery such as epiphysiodesis, or previous infections around the knee were excluded. Summary statistics for each of the 3 proximal tibial ratios were calculated and multiple linear regression was performed with percent of growth remaining as a dependent variable. A recommended regression model is presented and evaluated.
A total of 692 full-length radiographs met inclusion criteria. Proximal tibial ratios were calculated and averaged values for each percent of growth remaining was presented. Multiple linear regression demonstrated that using all 3 variables led to overfitting of the model so tibial metaphyseal width/lateral tibial epiphyseal height was selected as the optimal ratio for use by clinicians. The optimal model for determining growth was found to have R2=0.723 in the developmental set and R2=0.762 in an excluded validation set.
This study demonstrates that the proximal tibial metaphyseal width/lateral tibial epiphyseal height is the ideal measurement for clinicians seeking to determine growth remaining in children. It presents average values between 0% and 25% of growth remaining. This study also develops and validates a multivariable regression model for determining percentage of growth remaining in children that will allow for quantitative determination of growth using full-length radiographs.
Level of Evidence: