Usually conduct an analytic study to quantify relationships between exposures and disease. Key is comparison group.

Cohort study – fine when population is known, finite. Calculate and compare attack rates (cumulative incidence) among those exposed and those not exposed.

Case-control study – advantageous when true denominator is not known. Choice of control group is critical – friends? Other patients? Population-based such as random-digit dialing? Compare prevalence of exposure among cases and controls.