August 1988

Surface Cultures in the Newborn and Deeper Implications

Author Affiliations

PO Box 1269 Los Lunas, NM 87031

Am J Dis Child. 1988;142(8):812. doi:10.1001/archpedi.1988.02150080017005

Sir.—The editorial by Fulginiti and Ray1 interested me because I had previously published an article on the Bayes' formula.2 This comment on the report by Evans et al3 was admirable, but use of the modified formula suggested by Katz4 is in error.

The notation that you used, "PA," is conceptually the same as that customarily called the positive predictive value, or PV+. The "simplified Bayesian formula" is only accurate when the sensitivity is very near 1.00, ie, when nearly every person with the condition has a positive test for the condition. Sensitivity in the Evans et al report was 0.56 or below, depending on the day of the surface culture. When the sensitivity is this low, the Katz modification should not be used.

For those intimidated by the formidable appearance of the usual Bayes' formula for PV+ (which Katz shows in his report), there is

First Page Preview View Large
First page PDF preview
First page PDF preview