Predicting adult height from a child’s current height

BMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d6032 (Published 27 September 2011) Cite this as: BMJ 2011;343:d6032
  1. Jan M Wit, emeritus professor of paediatrics ,
  2. Wilma Oostdijk, senior lecturer of paediatric endocrinology
  1. 1Department of Paediatrics, Leiden University Medical Centre, 2300RC Leiden, Netherlands
  1. j.m.wit{at}lumc.nl

New tool is useful for parents, but has little value in clinical practice

The physical examination of infants, children, and adolescents includes the measurement of weight and longitudinal growth (supine length or body stature), and plotting these measurements on growth charts is regarded as a useful tool for detecting health disorders.1 Growth charts are used for almost every child seen in primary or secondary care. Abnormal growth is usually arbitrarily defined as a weight, length (height), or head circumference below the −2 or above the +2 standard deviation score (SDS) lines (equivalent to less than the 2.3rd centile or above the 97.7th centile, so that 2×2.3%=4.6% can be considered “abnormal”), a large distance to midparental height, or crossing the SDS lines. Poor growth can be caused by many disorders, of which persistent short stature after a low birth size, growth hormone deficiency, and Turner’s syndrome are most common. A recent study by Cole and Wright reports a height prediction chart that could provide a simple and unbiased estimate of adult height for children of all ages.2 How valid is this approach, and how does it compare with alternative methods of predicting adult height?

The method the authors use is essentially a refinement of that used in primary care, which is that the adult height centile or SDS position (height SDS is defined as the distance between the observed height and the mean height for age and sex, divided by the SDS at that age) will usually be close to that of the child’s current height SDS.2 Although this unadjusted method has a wide range (in the example given the range was 184-196 cm, with a mean of 190 cm), adjustment for the correlation between a child’s current height and their adult height (which changes with age) shrank this range to 182-191 cm (with a mean of 187 cm); the observed adult height was 188.6 cm. In addition, this adjustment partially compensated for the exaggerated height prediction that occurred in the example because of a relatively early puberty. The example shows that although this adjusted method is better than the unadjusted one, the range is still wide. From a clinical point of view, there is little reason to predict adult height in normal children, and the shrinking of the prediction range may have little effect on the perception of parents and children of the uncertainty of the prediction.

In clinical practice adult height prediction is based on a combination of attained height and the most common measure of maturation, which is bone maturation. Bone maturation is assessed on a radiograph of the left hand and wrist, and it is expressed as bone age (skeletal age) by comparing the radiograph with an atlas.3 4 On the basis of height and bone age, predicted adult height is calculated using various mathematical formulas.4 5 Two recent reviews summarised the use of bone age and adult height prediction in clinical practice.6 7 Adult height predictions using these measures are more accurate than those based on height alone, but there is still a substantial prediction error up to early adolescence.8

The new method reported by Cole and Wright was primarily compared with midparental target height, which the authors described as another method to predict adult height.2 However, midparental target height only indicates the mean height (corrected for sex) of all children of these parents, not the height of an individual child. As far as we know, target height was never meant to be used as a prediction tool, but rather as an indicator of the adult height adjusted for genetic background, provided that there are no intrauterine or postnatal factors to disturb growth potential. Target height expressed as SDS is a valuable tool for clinicians. If a child’s height is more than 2.0 standard deviations below the population mean and more than 1.6-2.0 standard deviations below the target height SDS, the likelihood of poor growth having a pathological cause is great enough for the child to be referred for expert examination.9 The distance to the target height SDS which offers the best sensitivity and specificity is dependent on the formula used.9 10

In summary, the tool developed by Cole and Wright is helpful to parents who are curious about the adult height of their child but has limited clinical value. In a clinical setting bone age is needed to predict adult height, although even this is not very accurate in the age range where it is used most. Midparental target height is not a method for predicting adult height but is necessary to assess whether the child’s height is lower than expected for the genetic background of the parents. The distance between the child’s height SDS and midparental target height SDS is the best screening tool for children with short stature.


Cite this as: BMJ 2011;343:d6032


  • Competing interests: Both authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; not externally peer reviewed.


View Abstract

Log in

Log in through your institution


* For online subscription