Types of selection bias

Neyman bias (prevalence-incidence bias)

Occurs depending on the selection of cases (prevalent/incident), s.t. there may be a time gap between exposure and selection of study participants

Unmasking bias

Occurs when an innocent exposure becomes suspect if it causes a symptom that causes a search for the disease, thereby raising the diagnosis rates overall

Differential referral

Occurs when cases tend to be referred to one hospital over another due to its expertise in dealing with exposure victim

Susceptibility; protopathic bias (Confounding by indication)

Occurs when one disease increases the likelihood for a second disease, and the treatment for the first disease erroneously appears to increase the likelihood of the second disease

Membership bias

Occurs when people who choose to be members of a group differ in important respects from those outside the group

Berkson bias (Admission rate bias)

Occurs when there are differential rates of hospital admission for cases and controls or exposed and unexposed

Self-selection bias

Occurs when there is either increased refusal/nonresponse or agreement to participate that is related to both the exposure and disease

How to minimize non-respondent bias?

Minimize this bias by achieving at least 80% response, so that the composition of responders can be compared to non-responders

Volunteer bias

More specific example of self-selection bias, in which those who agree to participate/volunteer have reasons related to the disease/exposure

Non-respondent bias

Specific case of self-selection bias, occurs when non-respondents are fundamentally different than respondents

Which type of study (in terms of the topic) is particularly vulnerable to Neyman bias?

Studies on clinical coronary heart disease

Which type of study (in terms of composition) is more likely to be susceptible to self-selection bias?

Studies with low participation

Referral bias

The chance of exposed cases being admitted to the study is different than that of exposed controls

Healthy worker effect

The rates of death/disease among the working are lower than that of the general population -- however, this is because the general population will include those who are too ill to work, raising the rates of death/disease in the general population, s.t. workers will appear to have overall lower rates of death and disease