EPID 301 Lecture Notes - Lecture 9: Confounding, Test Screening, Recall Bias
Document Summary
Not as expensive and lengthy as prospective cohort studies. Need a data source for exposure: case-cohort. *: if cases are a random sample of all cases, and the controls are sampled randomly from the source population. A hybrid between case-control and cohort studies. The sub-cohort is representative of the cohort from which cases arose. The sub-cohort is selected from the population at risk at the start of the cohort"s follow-up interval, and should consist of exposed and unexposed individuals. It is okay for controls to end up developing the disease during the study period = measurement was taken at the beginning of the study = it is okay for controls to end up in disease. Can test for hypotheses not thought of when original cohort was drawn. Ensures cases and non-cases sampled from same population. Not suitable if the exposure changes overtime (during long follow-up) study multiple outcomes: nested case-control. Estimates: odds ratio (can approximate an rr for rare disease)