The healthy participant effect
People who participate in epidemiological studies aren’t necessarily representative of the populations they come from. What’s more, in case-control studies different response rates in cases and controls can distort the findings. In a large study of breast cancer, with a participation rate of 43% for controls and 64% for cases, mortality among controls was only two thirds that of the reference population (Am J Epidemiol doi:10.1093/aje/kwae155). This may not have biased the findings in this particular investigation because, for causes of death other than breast cancer, mortality was similar in cases and controls. However, in many studies, adjusting for biases arising from …
