1 Smith Kline & French Laboratories, 709 Sedeland Rd, Swedeland, PA 19479
To the Editor.
My interest in the article by Rachelefsky et al (Pediatrics 1986;78:1133-1138) stems in part from the heightened media coverage it received. I must admit as a professional biostatistician that I have some difficulties with regard to its conclusions.
The authors report a P value equal to .004 for the intergroup comparison of change scores for the teacher's behavioral assessment. However, the statistically significant change (P = .04), a reduction for placebo and an increase for theophylline, in the posttreatment score when compared to the pretreatment score is completely ignored.