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?

David C. Parish, MD, MPH; Francis C. Dane, PhD
Mercer University Macon, Ga

JAMA. 1991;265(11):1393.

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

To the Editor.—

The statistical analysis used by Zwerling et al¹ is flawed. The major defect is that "the six covariates and four drug categories were included simultaneously in the [Cox proportional hazards] models" (p 2642, emphasis added). The purpose of introducing covariates is, of course, to remove their influence prior to testing the influence of the independent variable (in this case, drug category). However, removing the influence of covariates requires that they be entered into the model in hierarchical, not simultaneous, fashion.^2-4 Thus, the risk estimates for drug categories reported in Table 4 are not adjusted fully for the covariates.

The authors use a classification of jobs in their statistical model that renders their analysis meaningless. Six job classes are collapsed into two or three categories for each end point; the job classifications used in each analysis are almost certainly different.

The authors describe the study conducted . . . [Full Text PDF of this Article]

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