Prevalence ratios offer a more interpretable alternative to odds ratios in cross-sectional studies
In short: Researchers who compared different statistical models for cross-sectional studies found that logistic regression may significantly overestimate the prevalence ratio. Instead, they found that log-binomial regression and Cox or Poisson regression with robust variance estimators provided more accurate and interpretable results for binary outcomes.
In cross-sectional studies, the prevalence ratio is often preferred over the odds ratio because it is easier for non-specialists to understand. This study evaluated several modeling techniques using data from a Brazilian study involving three different outcome prevalence levels: 4%, 31%, and 52%.
The researchers compared Cox, Poisson, and log-binomial regressions against Mantel-Haenszel estimators. While log-binomial and robust variance models performed well when correctly specified, the study noted that specific precautions are necessary to avoid estimation problems in certain situations.
Why it matters: Using prevalence ratios instead of odds ratios can make epidemiological findings easier to communicate and more accurate.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Research: Barros and Hirakata (2003). Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Medical Research Methodology.
To cite this card, link to https://intellectplus.co.uk/app/science-research-prevalence-ratios-offer-a-more-interpretable-alternative-to.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Epidemiology · More in Health · Latest · A–Z · All topics