False Positive Exposure Errors and Low Exposure Prevalence in Community-Based Case-Control Studies

Abstract

The community-based case-control study (CBCC) offers a potentially efficient strategy for identifying putative causes of disease and has special relevance to the surveillance of occupationally related diseases such as cancer. The primary limitation in using the CBCC design for surveillance purposes is that defining exposure is highly prone to error. This paper quantifies the bias to the odds ratio caused by nondifferential exposure misclassification under a range of conditions common to occupational studies.

When the exposure prevalence is low (< 10%), a condition common to community-based occupational and other case-control studies, and assuming nondifferential misclassification, false positive exposure errors (i.e., nonexposed are classified as exposed) are the principal source of bias to the odds ratio. In contrast, even a relatively large number of false negative exposure errors results in negligible bias.

Knowledge of the major source of bias to the odds ratio offers the means to develop specific and efficient strategies to correct for exposure misclassification errors. Three strategies are recommended to minimize the effect of false positive errors: making use of exposure groups rather than occupational title groups in the analysis; repeat interviews of subjects designated as exposed; and methodological research regarding the manner in which occupational histories are obtained and exposure decisions are made. Stewart, W.F.; Correa-Villaseñor, A.: False Positive Exposure Errors and Low Exposure Prevalence in Community-Based Case-Control Studies. Appl. Occup. Environ. Hyg. 6:534–540; 1991.