This formula is from the textbook Modern Epidemiology (fourth edition) by Lash, VanderWeele, Haneuse, and Rothman. They provide the following citation: Kelsey JL, Thompson WD, Evans AS. Methods in Observational Epidemiology. New York, NY: Oxford University Press; 1986; chap 10.
This formula is from the textbook Modern Epidemiology (fourth edition) by Lash, VanderWeele, Haneuse, and Rothman. They provide the following citation: Kelsey JL, Thompson WD, Evans AS. Methods in Observational Epidemiology. New York, NY: Oxford University Press; 1986; chap 10.
Very helpful, i have downloaded the book (Modern Epi..), working on a case control study, do you think you could point me to the page number @@epidemiologystuff1178
If you mean an exploratory analysis where there is no “exposure of interest” I am unaware of a sample size calculation, although there may be one. One reason I think there might not be one out there is that these calculations rely on an anticipated strength of association, which implies a single association of interest. One option would be to pick the exposure that you expect to have the weakest relationship (e.g. the smallest odds ratio), and assume that the calculated sample size will be enough to detect all the associations your looking for
The most practical time to do a case control is when the disease is rare. That doesn't mean it CANT be carried out when the disease is common, the case control just loses all its practical value. A major reason is that the odds ratio no longer approximates a relative risk, and the other major reason is that when the disease is common, a cohort study becomes a far better option than a case control, because it is now reasonable to accumulate a sufficient number of cases over a reasonable follow-up period, and offers a ton of benefits over a case-control design such as ensured temporality between exposure and outcome, no recall bias, and resistance to selection bias.
