Sir.—We would like to comment on the article "The KIDS Chart: A Simple, Reliable Infant Development Screening Tool" by Holmes and Hassanein (Journal 1982;136:997-1001). While it would be advantageous to be able to assess development with three to five questions at each month of age, the validity of the Kansas Infant Development Screen (KIDS) chart has not been shown. There are three major methodological errors.
First, validity cannot be demonstrated by comparing mean chronologic and developmental ages. A screening instrument is valid if it detects all, or almost all, children with definite or suspected abnormalities (sensitivity) and, at the same time, calls a minimum number of normal children abnormal (specificity). Surely, Holmes and Hassanein have the data to provide this information.
Second, r values were calculated by comparing developmental and chronologic ages, or two sets of developmental ages. Statistically, when there are extreme values in a population, correlations are