Sir.—Pollitt and Eichler reported a higher prevalence of atypical behavior in failure-to-thrive children (age 12 to 60 months) than in a contrast group of normal-growth children (Am J Dis Child 130:24, 1976). They summarized their observations: "The number of index children exhibiting atypical behaviors... is ten; the number of contrast children is one (χ2 = 8.19; df = 1; P <.01)." The authors found atypical behavior in a group of failure-to-thrive children with a prevalence of ten in 19 (52.6%), while only one in 19 (5.3%) of the normal-growth group showed such behavior.
In the same issue (130:56, 1976), I pointed out the possibility of spurious conclusions from surveys in which uneven selection of patients might have occurred. Could such "selection factors" alone have caused the difference between the "index group" and the "contrast group"? Could such a marked difference have been observed if atypical behavior had the same prevalence