Quantitative Methods 2017

Epidemiological study design

 Cohort study (observational, longitudinal, forward, incidence)

 clean design (e.g. eligibility, standards)  time of event and directionality  multiple endpoints ! confounders ! Lost to follow-up ! not always feasible (rare disease) ! expensive

Time

Disease ? ?

Exposure

 Case-control (observational, retrospective, case-referent)

 quick  cheap  sample size  feasible for rare disease or diseases with long latent time ! prone to bias (e.g. recall) ! not observational (reliant on records)

Historical data

Disease

Exposure

? ?

 Cross-sectional (single time point, prevalence)

Time 1 population cross-section

Time N population cross-section

 quick  cheap  easy (e.g. ethics) ! association, not causality ! prone to bias (e.g selection)

Exposure

Exposure

! unbalanced ! confounders

FMB

Disease

Disease