Between about age 4 and the start of puberty, most children grow at a fairly steady 5 to 6 cm per year and stay close to one percentile. Puberty changes both. Height velocity roughly doubles during the growth spurt, and the age at which the spurt occurs varies by two years or more between healthy children.
The child height predictor has its widest error in this period for that reason.
The pubertal growth spurt
Tanner and Whitehouse (1976) published longitudinal standards that remain the reference description. In girls, peak height velocity occurs at about age 12 on average and reaches about 8 to 9 cm per year. In boys, it occurs at about age 14 and reaches about 9 to 10 cm per year. Girls typically gain 20 to 25 cm during the spurt and boys 25 to 30 cm. The later and larger spurt in boys accounts for most of the adult difference in average height between the sexes.
Growth then slows as the growth plates close, and adult height is reached about two years after peak velocity in girls and about three years in boys. Abbassi (1998) summarized the normal range. Puberty begins between about 8 and 13 years in girls and 9 and 14 years in boys. The spurt follows onset by one to two years.
Why timing varies
The age at which puberty begins is partly inherited and partly influenced by nutrition, body composition, and health. Two healthy children of the same sex and age can be at different stages: one may have completed the spurt while the other has not started. Their heights at that moment differ by a large amount, and neither height is informative about adult height without knowing where each child is in the sequence.
Early-maturing children are tall for their age during puberty and stop growing sooner. Late-maturing children are short for their age and continue growing longer. Both patterns can end at the same adult height.
What this does to a prediction
A prediction made from current height and age treats height as a function of age. During puberty, height depends more on maturational stage than on age. A method that does not know the stage has to average over early, on-time, and late maturers.
Khamis and Roche (1994) reported that their equation's error was smallest in mid-childhood and rose during the pubertal years. Cole and Wright (2011) found the same for prediction from current height percentile, with the standard error increasing in early adolescence. The predictor applies the published age-specific error bounds, and users comparing a result at age 9 with one at age 13 will see the range widen.
The equation's weight input partly compensates. Weight tends to rise ahead of the growth spurt, and including it captures some information about maturation that height alone does not. It does not capture enough to remove the effect.
Bone age
The information the equation lacks is skeletal maturity. A hand and wrist X-ray read against a bone-age atlas gives a direct estimate of how much growth remains. Methods that use bone age predict adult height in adolescence more accurately than methods that do not. The predictor does not accept bone age, because it requires a clinical radiograph and expert reading. It does not ask for pubertal stage either, because self-assessment of stage is unreliable. A family that has a clinician's bone-age estimate should use that estimate rather than the predictor.
Reading the result during adolescence
A prediction made during puberty is best treated as provisional. The predictor's range is wider by design. The center of the range is also more likely to move after a measurement a year later than a prediction made at age 7 would be. For an early-maturing child, the estimate tends to run high. For a late-maturing child, it tends to run low. Once the spurt has ended, the remaining growth is small and the estimate becomes stable.
A question about whether puberty is early, late, or progressing normally is separate from adult-height prediction, and the predictor is not built to answer it. Abbassi's ranges above are the general guide, and the child's pediatrician is the appropriate source for an assessment.