Literatures showed that when an outcome is common (high prevalence), the odds ratio exaggerates the strength of the association compared to the relative risk (RR). RR aligns more closely with the absolute risk difference, making it a better choice for understanding the impacts. Using OR can mislead by suggesting a stronger association than what exists.
In cross-sectional studies, researchers typically assess the prevalence of an outcome at a single point in time rather than incidence. If you are estimating the association between an exposure and a prevalent outcome, prevalence ratio (PR) (which is similar to RR) is more intuitive than OR. However, logistic regression, which naturally produces OR, is often used because of its mathematical convenience. If OR is reported, it should be interpreted with caution, especially if the outcome is common. In such cases, converting OR to RR using Zhang & Yu’s correction or using log-binomial regression can provide more meaningful estimates.
These articles may be helpful for further information:
10.1001/jama.280.19.1690
Article Estimating the Relative Risk in Cohort Studies and Clinical ...
Article The Relative Merits of Risk Ratios and Odds Ratios
Article When to use the odds ratio or the relative risk?
Article Alternatives for logistic regression in cross-sectional stud...
That said, OR is still useful when the focus is on case-control studies or when analyzing rare outcomes (
- Badges
- Science topic