The Khamis-Roche Percentage of Predicted Adult Height as a Measure of Somatic Maturation: A Scrutiny of Fundamental Simple Ratio Scaling Assumptions.

Abstract

To scrutinize the fundamental assumptions that underpin the formulation of the Khamis-Roche percentage of predicted adult height as the simple ratio of standing height (numerator) divided by the Khamis-Roche predicted adult height (denominator) for assessing maturity status in youth athletes. A secondary data analysis of publicly available, de-identified, cross-sectional data relevant to 139 male youth soccer players (age range, 12.2 to 15.2 yr). Assumptions for accurate simple ratio scaling are satisfied if the bivariate regression line between the numerator and denominator passes through the origin, the ratio of the numerator and denominator coefficient of variations (CV x /CV y ) equals the correlation coefficient between these variables, and the relationship between the ratio and its denominator yields a zero correlation. The y-intercept value of -62 cm (95% confidence interval [CI], -99 to -25 cm) from the linear regression between standing height and Khamis-Roche predicted adult height indicated violation of the zero y-intercept assumption. The correlation coefficient between standing height and Khamis-Roche predicted adult height of 0.72 (95%CI, 0.63 to 0.79) was not equal to the ratio of coefficient of variations between these variables (CV x /CV y = 0.52), indicating the failure to meet the proportionality assumption. The positive, non-zero correlation between the ratio and the denominator of 0.28 (95%CI, 0.12 to 0.42) suggested that the Khamis-Roche percentage of predicted adult height was biased low for players with generally shorter predicted adult height, and vice versa . The percentage of predicted adult height, whether determined using the Khamis-Roche method or any alternative protocol, is an inconsistent simple ratio statistic for assessing maturity status, with the potential to bias the understanding of completed growth in adolescents.