This article is only available in the PDF format. Download the PDF to view the article, as well as its associated figures and tables.
To the Editor:
Breneman's stimulating communication, "Postoperative Thromboembolic Disease" (193:-576, 1965) raises a number of statistical and practical issues in the application of computer techniques to medical data and practice. Breneman first isolated seven preoperative variables (age, overweight, preanesthetic immobilization, etc) in which there were significant differences between patients who developed postoperative thromboembolic complications and those who did not. This in itself is a worthwhile contribution.This was followed by a multivariate analysis yielding an equation by which each patient's measures on these seven variables could be weighted and summed, producing an individual overall score. This score correlates maximally, for this sample, with the occurrence of complications, and is recommended to identify high-risk patients and initiate prophylactic measures.This application of relatively novel statistical techniques should require a cautionary discussion of the mathematical assumptions underlying the technique and their fulfillment in the case at hand. In this case the
Klein DF. Application of Computer Techniques to Medical Data. JAMA. 1965;194(4):472-473. doi:10.1001/jama.1965.03090170150039