ON THE DISTORTION OF RISK ESTIMATES IN MULTIPLE EXPOSURE LEVEL CASE-CONTROL STUDIES
- 1 April 1981
- journal article
- research article
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 113 (4), 464-473
- https://doi.org/10.1093/oxfordjournals.aje.a113114
Abstract
Marshall, J. R. (Dept of Sociology, State U. of New York at Buffalo, Buffalo, NY 14261), R. Priore, S. Graham, and J. Brasure. On the distortion of risk estimates in multiple exposure level case-control studies. Am J Epidemiol 1981;113:464–73. The Impact of error in assessing exposure to hypothesized pathogens on distortion of the appearance of relative risk is considered in the case-control study design. We assume there is a true dose-response relationship which links exposure to disease status. We also assume that multiple categories are used to classify levels of exposure, and that the errors which occur are those between adjacent exposure categories. The findings suggest that, if the pattern of errors is random, even though they are numerous, the resulting bias in risk will not be severe enough to mask a substantial risk increment A large and consistent understatement of exposure by cases, and/or consistent overstatement of exposure by controls, can effectively mask even a strong pattern of risk enhancement. To produce the appearance of risk where there is none would require a substantial error in the opposite direction among the cases and controls.Keywords
This publication has 4 references indexed in Scilit:
- SPOUSE-SUBJECT INTERVIEWS AND THE RELIABILITY OF DIET STUDIESAmerican Journal of Epidemiology, 1980
- ASSURING THE QUALITY OF QUESTIONNAIRE DATA IN EPIDEMIOLOGIC RESEARCHAmerican Journal of Epidemiology, 1979
- ADEQUACY OF SURVEY DATA COLLECTED FROM SUBSTITUTE RESPONDENTSAmerican Journal of Epidemiology, 1977
- Effects of Errors in Classification and Diagnosis in Various Types of Epidemiological StudiesAmerican Journal of Public Health and the Nations Health, 1962