 |
The 130th Annual Meeting of APHA |
|
The 130th Annual Meeting of APHA |
Jayawant Mandrekar, MS, Division of Epidemiology and Biometrics, School of Public Health, The Ohio State University, B-107, Starling Loving Hall, 320 W 10th Avenue, Columbus, OH 43210, 614-228-2454, mandrekar.1@osu.edu and Melvin Moeschberger, PhD, School of Public Health, The Ohio State University, 320 W. 10th Ave., B-107 Starling-Loving Hall, Columbus, OH 43210.
In the analysis involving data from clinical or epidemiological studies, significant attention is given to continuous variables such as patient's age, blood pressure, cholesterol etc., but the predictive importance of such variables cannot be established easily. Transforming a continuous variable into a categorical variable, usually binary, makes the model more interpretable.
The choice of a cutpoint to convert the continuous covariate to binary covariate needs special attention and is often based on biological knowledge about the particular risk factor or on the results already published in other studies. However in the case when the cutpoint is not readily available, statistical methods that determine the cutpoint need to be used. Several data oriented techniques such as median and upper quartile, and outcome oriented techniques based on score, Wald and likelihood ratio tests are commonly used in the literature. Contal and O'Quigley (1999) presented a technique that uses log rank test statistic in order to estimate the cutpoint and its significance. Their method is computationally intensive and hence is overlooked due to the unavailability of built in options in standard statistical software. We critically compare the results from this method with the existing methods and provide an S-plus code that is easy to implement and does all the necessary computations in a relatively short amount of time. Some of the issues on the use and misuse of categorizing a continuous covariate and dichotomization will be discussed.
Learning Objectives:
Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.