Bias in relative odds estimation owing to imprecise measurement of correlated exposures

Abstract

A series of graphs is presented that show the estimated degree of bias in logistic coefficient estimates for two correlated continuous exposures measured with imprecision. These graphs indicate that even when the correlation coefficient between the exposure of interest and a correlated exposure is as low as 0.2, imprecision in the measurement of the latter exposure can result in at least as serious bias in the logistic coefficient estimate for the exposure of interest as measurement imprecision in the exposure of interest itself. The implications for the design and interpretation of epidemiological studies are discussed.