This Article  • References  • Full text PDF  • Send to a friend  • Save in My Folder  • Save to citation manager  • Permissions  Citing Articles  • Contact me when this article is cited  Related Content  • Similar articles in JAMA  Social Bookmarking   What's this?

Peter H. Gann, MD
Northwestern University Medical School Chicago, Ill

Charles H. Hennekens, MD; Meir J. Stampfer, MD
Brigham and Women's Hospital Boston, Mass

JAMA. 1995;274(8):608-609.

	Since this article does not have an abstract, we have provided the first 150 words of the full text PDF and any section headings.

In Reply.

—Drs Brunswick, Marshall, and Meyerhoff calculate predictive values from our data. We agree that predictive values are important, but chose not to present them in our article because their computation and interpretation, particularly given our study design, are not straightforward. Since predictive values are dependent on disease prevalence and factors affecting the occurrence of false-positive results, they can vary substantially in different study populations.¹ Moreover, in a nested case-control study, the artificial prevalence represented by the cases and controls alone (a prevalence determined by the case-control ratio) is obviously a problem. If the controls are an unbiased sample of the total nondiseased members of the cohort, as Brunswick, Marshall, and Meyerhoff assume, one can estimate predictive value. However, we used an "incidence density" approach for sampling controls (ie, the controls are randomly selected from similar-aged men at risk when each case is diagnosed), so our controls are . . . [Full Text PDF of this Article]

CiteULike    Connotea    Del.icio.us    Digg    Reddit    Technorati    Twitter     What's this?