using an odds ratio to predict prevalence/incidence in different populations

[Rachel Sklar]

Say you have an odds ratio which represents the ratio of outcome given a particular exposure to the outcome given no particular exposure. How can you use this ratio to make an estimate about the incidence in a population that has that known particular exposure?

For instance, say there is an OR that classifies the relationship between exposure factor E and disease D in population P. Now you have population y that has an exposure to E and you want to predict the incidence or prevalence of the disease in population Q. How can the OR for P be used to make a prediction about disease D in population Q?

 

[hlsmith]

You are mixing terms, not sure if it is intentional or not. Incidence (also prevalence) are calculated differently than odds. However, there is the rare disease assumption that states the Odds Ratio can be a proxy for Relative Risk if outcome is rare (<10%).


Are you talking about just applying the numbers (risk) to the population? If so, the biggest questions you have to ask is what was the study design and sample scheme used to get data and results. Were the observations randomly sampled from the population? If not, direct application may be problematic. Also, could there be any systematic errors in the collection of sample or data (e.g., could people miss-report exposures and outcomes, etc.).