Premium
An overview of issues related to the correction of non‐differential exposure measurement error in epidemiologic studies
Author(s) -
Willett Walter
Publication year - 1989
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.4780080903
Subject(s) - covariate , confidence interval , observational error , statistics , computer science , nominal level , econometrics , mathematics
Abstract Procedures to correct estimates of association in epidemiologic studies for the effects of exposure measurement error have rarely been employed in practice. The application of correction procedures would be enhanced by methods that allow the inclusion of covariates, provide corrected confidence intervals, are compatible with commonly employed analytic methods, and that are clearly communicated to potential users. Before using such a procedure, it is important to clearly specify the conceptual ‘true’ exposure, determine the nature of the measurement error, and decide whether a reproducibility study or validity study is required to quantify the error. The careful use of correction procedures promises to improve our knowledge of the quantitative relationships between many exposures and disease since it is likely that we have substantially underestimated the effects of many exposures and overstated our confidence in null results.