ORIGINAL ARTICLE The Multiplier Method for Prediction of Adult Height Jonathan Paley, Jonathan Talor, Anna Levin, Anil Bhave, PT, Dror Paley, MD, FRCSC, and John E. Herzenberg, MD, FRCSC Abstract: The authors propose the ‘‘height multiplier method,’’ a new, simple, universal method to predict adult height. The authors calculated age- and gender-specific height multiplier values and validated their use for height prediction. Average height multiplier values (M) for boys and girls were calculated by dividing height at skeletal maturity (Ht m ) by present height (Ht) for each age, gender, and height percentile (M = Ht m /Ht) using data published by the Cen- ters for Disease Control and Prevention (CDC). The accuracy of the multiplier method was compared with the CDC growth charts using longitudinal growth data from 52 normal children. No significant dif- ferences were noted between predictions made using the CDC growth charts versus the multiplier method. The authors also compared the CDC-derived multiplier values with other growth databases from around the world and from different eras. Height multipliers were independent of height percentile, race, and generation, indicating that the multiplier method is universal. Key Words: height, prediction, growth, development (J Pediatr Orthop 2004;24:732–737) I nterest in human growth has produced various methods of predicting the adult height of a growing child. Height prediction systems are based on either longitudinal or cross- sectional height databases. Because average adult height varies among nationalities and races, numerous height databases have been developed. The simplest method for height predic- tion uses growth curves; the clinician plots the child’s height at the appropriate age and determines the relative percentile height for that child. 12,13,22 The clinician then follows the percentile line to skeletal maturity to obtain an estimate of adult height. This works well provided the child is of the same race/ethnicity as the database used to construct the growth chart. In an effort to improve this method based on chronologic age, Bayley 3 published tables to predict adult height based on skeletal age and current height. This method was later adapted by Bayley and Pinneau 4 to incorporate the newer Greulich- Pyle atlas 11 of skeletal age standards. Other studies 16,17,27,28 have tried to improve the accuracy of predictions by using complex regression formulas for adult height prediction that incorporate factors such as recumbent length, weight, skeletal age, and mid-parent height. Tanner et al 37 presented a method to estimate skeletal age based on the Greulich-Pyle atlas. More recently, Tanner et al 36,38 published the TW2 and TW3 methods using growth data for British children and the Radius-Ulna-Small Bone (RUS) score to assess skeletal age. They consider the RUS score to be more accurate from population to population than conventional skeletal age determinations 36 ; nevertheless, they caution against the application of their system to non-British populations. All these methods are subject to a wide range of error, at least partly because of the use of bone age estimations. 11 For these reasons, clinicians typically rely on growth charts, which differ among races. With the progressive blending of the Amer- ican population, 20 it has become impossible for pediatricians to keep up with the multitude of growth charts needed for their racially diverse patients. Furthermore, which growth chart is most appropriate for the multi-ethnic/multi-racial child? A more universal method of height prediction is needed. A universal method called the multiplier method has been developed for predicting lower limb length. 24 It has been shown that lower limb growth follows a biologic constant based on age and gender. For example, the leg length of boys at 4 years of age can be doubled to predict leg length at maturity. The leg length multiplier values are age- and gender-specific coefficients calculated from the normal growth charts of the femur and tibia. 1 The length of the femur or tibia at skeletal maturity, divided by the length of the bone at each age and gender, is the age- and gender-specific multiplier. Interestingly, little variability exists among the leg length multipliers calcu- lated for different height percentiles. In other words, a child in the 5th percentile has a lower limb multiplier similar to that of a child of the same age and sex in the 50th or 90th percentile. The study also demonstrated that the lower limb multipliers were similar among different nationalities and races. It seems reasonable to expect that a height multiplier would have similar properties to the lower limb multiplier, considering that the lower limbs contribute more than half of adult height. We postulated that average height curves would also follow universal biologic constants based on age and gender. The purposes of this study were to calculate age- specific height multipliers for boys and girls; to determine whether they are independent of height percentile, population, and generation; to compare the accuracy of this prediction method with that of other methods; and to validate its use for height prediction. Study conducted at the Rubin Institute for Advanced Orthopedics, Sinai Hospital of Baltimore, Baltimore, Maryland. From the Rubin Institute for Advanced Orthopedics, Sinai Hospital of Baltimore, Baltimore, Maryland. None of the authors received financial support for this study. Reprints: John E. Herzenberg, MD, FRCSC, Rubin Institute for Advanced Orthopedics, Sinai Hospital of Baltimore, 2401 West Belvedere Avenue, Baltimore, MD 21215-5721 (e-mail: jherzenberg@lifebridgehealth.org). Copyright Ó 2004 by Lippincott Williams & Wilkins 732 J Pediatr Orthop  Volume 24, Number 6, November/December 2004 JOBNAME: jpo 24#6 2004 PAGE: 1 OUTPUT: Mon October 4 13:48:59 2004 lww/jpo/87469/6038