In a case-only study, which of the following is true regarding interaction effects?

• A. Interactions can be estimated; main effects cannot
• B. Main effects can be estimated; interactions cannot
• C. Neither interactions nor main effects can be estimated
• D. Both interactions and main effects can be estimated

Answer: (A)

In a case-only study, the design focuses on cases to assess whether there is a gene–environment interaction influencing disease risk. The key idea is that if genotype (G) and exposure (E) are independent in the source population, then any association between G and E observed among the cases arises from an interaction between G and E on disease risk, not from main effects. This lets you estimate the interaction term efficiently. However, because there is no control group, you cannot separate the individual main effects of G or E on disease risk from their joint effect, so main effects cannot be estimated reliably in this design. If the independence assumption holds, you gain information about interaction; if it doesn’t, the interaction estimate can be biased.