Transient growth deceleration in normal short children. A potential source of bias in growth studies.

We have examined the records of 21 short children evaluated for growth hormone deficiency and found not to be deficient. Their growth velocity was evaluated for at least 6 months, both before and after testing. Without any specific therapeutic intervention, growth velocity was significantly higher after testing, as compared with before. We attribute this apparent "therapeutic" effect of testing to a selection bias, due to the fact that, in normal clinical practice, children are selected for testing immediately following a period of slow growth, and that decelerations of growth are very often transient. Studies of growth-stimulating treatments using children as their own controls should, for this reason, be interpreted with caution.