Types of bias:
(1) lead time
(2) selection
(3) length time
(4) overdiagnosis
Lead time bias:
(1) Basis: Diagnosing a disease sooner without altering actual life expectancy makes survival appear better than it actually is.
(2) Situations where encountered: A more sensitive diagnostic test may detect an unrelenting disease at an earlier stage without chance of cure.
(3) Example: Lung cancer imaging studies.
Selection bias:
(1) Basis: Failure to randomly place persons in the study and control groups may result in a cluster of patients with factors favoring a response unrelated to the intervention. This may be positive (favoring survival) or negative (favoring early death).
(2) Situations where encountered: Some people have a greater interest in health or resources for care, while other people are prone to more health problems.
(3) Example: Studies with athletes compared to obese inactives, or with wealthy, college educated populations compared to the unemployed.
Length time bias:
(1) Basis: Failure to distinguish aggressive from indolent forms of a disease, with a high percent of indolent forms, makes testing appear more effective than it actually is.
(2) Situations where encountered: An indolent form of disease is more likely to be detected while still asymptomatic.
(3) Example: Prostate cancer in men and screening with serum PSA.
Overdiagnosis:
(1) Basis: A group of persons are identified as having a disease but die from other causes, so that disease-related mortality is falsely low. This is regarded as an "extreme" form of length time bias.
(2) Situations where encountered: An indolent disease in patients with multiple concurrent conditions.
(3) Example: Prostate cancer in elderly males with heart disease or multiple comorbid conditions.
