Information bias in epidemiological studies with a special focus on obstetrics and gynecology

Abstract

Information bias occurs when any information used in a study is either measured or recorded inaccurately. This paper describes some of the most common types of information bias, using examples from obstetrics and gynecology, and describes how information bias may affect results of observational studies. Non-differential misclassification occurs when the degree of misclassification of exposure status among those with and those without the disease is the same; in cohort studies, this type of bias is most likely and will bias estimates toward no association when exposure is dichotomized. Non-differential underreporting of an exposure with more than two categories may mask a true threshold effect as a dose-response relation and, if a true threshold effect exists, the threshold will be set at too low a level, if the exposure is underreported. Differential misclassification may cause bias in either direction and is particularly likely, when exposure status is reported after the outcome occurred. Misclassification of confounders is an issue that needs special attention by researchers, as failure to measure accurately one or more (strong) confounders may seriously bias the observed results. Misclassification of disease status may also cause bias of estimates of association in either direction. Information bias is probably best prevented during planning of data collection, as there are few and insufficient methods available for correcting inaccurate information.

Alcohol Intake in Early Pregnancy and Spontaneous Preterm Birth: A Cohort Study

“…Previous studies have indicated a U‐shaped association between average weekly alcohol intake and both spontaneous preterm birth (Kesmodel et al, ; McCarthy et al, ) and any preterm birth (Albertsen et al, ; Jaddoe et al, ; Kesmodel et al, ; Strandberg‐Larsen et al, ) with the lowest risks seen at an intake of 1 to 6 drinks per week compared to an intake of less than 1 drink per week. It has been suggested that the apparently beneficial effect of small doses of alcohol may be due to confounding by the “healthy drinker effect.” This may appear if drinking in small amounts is associated with a healthier lifestyle, or if women with chronic diseases are more likely to abstain from alcohol than healthy women (Howards, ; Howards, ; Kesmodel, ; Strandberg‐Larsen et al, ; Strandberg‐Larsen et al, ). In order to reduce the risk of unmeasured confounding by education, lifestyle, and medical/obstetric history, we chose to study a homogeneous population with a high educational level, and further restricted the analyses to nulliparous women or cohabiting women with ≥3 years of higher education.…”

“…As information about alcohol exposures was determined before the delivery of the child, we expect any misclassification to be nondifferential. Typically, nondifferential misclassification of a dichotomized exposure variable leads to bias toward the null hypothesis, whereas the mechanisms are harder to predict in an exposure variable with 3 or more categories (Kesmodel, ). Average weekly alcohol was dichotomized, and for binge drinking, we had findings close to null when dichotomizing the exposure.…”

Cross‐sectional studies – what are they good for?

“…This should not lead the reader to think in terms of cohort or case‐control studies. The analyses of association in cross‐sectional studies are evidently subject to the common types of bias in cohort and case‐control studies: selection bias , information bias and confounding .…”

“…inaccurate measurement or recording of a disease or characteristic, is also a key problem that needs to be addressed in cross‐sectional studies as in any other study design. Please refer to Nøhr & Liew and Kesmodel for details on selection and information bias, resepctively.…”

Methodological challenges in studies in obstetrics and gynecology

“…randomized clinical trials , cohort studies , and cross sectional studies . As for bias, we take you through the recent developments in selection bias, information bias and confounding , including a detailed introduction to directed acyclic graphs . Further, we provide a number of papers on specific topics in perinatal epidemiology.…”