Table 2.

Potential biases in gene-gene and gene-environment investigations of coronary artery disease (CAD)

Bias	General description	Application to CAD
Selection bias	Skew in the selection of study participants	Patients with strong family history may self-select for study participation; patients with strong family history may be more likely to be referred to tertiary care and research centers
Survivor bias (prevalence-incidence bias)	Selection of study participants may miss mild disease or severe fatal cases	Patients whose first myocardial infarction is fatal are less likely to be studied
Recall bias	Patients are more likely to recall an environmental exposure if it was linked to a negative outcome	Patients with CAD may be more likely to remember an environmental exposure because of its negative consequences
Respondent bias	Patients answer in the way they believe they should answer, not the true answer	Patients with CAD and knowledge of potential CAD risk factors will be more motivated to report those exposures
Family information bias	Individuals become more aware of exposure if it is prevalent in their family	Many CAD risk factors and environmental exposures cluster in families
Exposure suspicion bias	Disease status can affect the amount of environmental exposure history collected	If data collection is not standardized, investigators may more thoroughly query patients with CAD
Publication bias	Statistically significant findings are more likely to be published	Gene-gene and gene-environment interaction findings in CAD are more likely to be published if significant
Measurement bias	Systematic errors of measurement	Platform- or laboratory-dependent genotyping errors; errors of laboratory values; errors of environmental exposure measurement
Population stratification	Differences in allele frequencies between groups resulting from ancestry not outcome status	CAD prevalence varies between ethnicities; but this can be tested and corrected for using methodological and statistical techniques