Sir.—Brown is to be congratulated for a lucid and timely article delineating the nature of types I and II statistical errors in medical practice.1 However, in the same issue of the AJDC, an abstract by Jaffe and colleagues2 commits exactly the type II error to which Brown refers. Although this work was published only as an abstract, it already has been quoted widely after its presentation at the 1983 annual meeting of the Ambulatory Pediatric Association and, undoubtedly, now will be referenced in the literature as having appeared in a pediatric journal.
Jaffe et al2 studied the effects of oral antibiotic therapy in almost 500 febrile children without localizing signs, 28 of whom had bacteremia. In analyzing their data regarding these 28 patients, these investigators found no significant difference with regard to outcome and stated "there were no significant differences between these groups in frequency of
LORIN M, SMITH EO. Statistical Error. Am J Dis Child. 1984;138(2):206-207. doi:10.1001/archpedi.1984.02140400084024