Fig. 3
Potential downstream influence of the enrollment factors, illustrated using the healthy worker bias as an example. The healthy worker bias (HWB) is a long-recognized study design flaw first described in research attempting to understand the influence of exposure to occupational hazards [33] . HWB occurs when selection bias results in a particular characteristic being over-represented in the study population. This is especially problematic if the characteristic is associated with the variable of interest and/or the incident event. In the case of the healthy worker bias, group A represents individuals healthy enough to be part of the work force; group B represents those too unwell to work. In our analyses, group A could be individuals recruited from the community, that is, referred to an Alzheimer’s Disease Center (ADC) by a friend. Group B would be individuals seen in a dementia…
