Introduction Several papers have discussed which effect measures are appropriate to capture the contrast between exposure groups in cross-sectional studies, and which related multivariate models are suitable. Although some have favoured the Prevalence Ratio over the Prevalence OR—thus suggesting the use of log-binomial or robust Poisson instead of the logistic regression models—this debate is still far from settled and requires close scrutiny.
Method In order to evaluate how accurately true causal parameters such as Incidence Density Ratio (IDR) or the Cumulative Incidence Ratio (CIR) are effectively estimated, we present a series of scenarios in which a researcher happens to find a preset ratio of prevalences (eg, 2.0) in a given cross-sectional study.
Results Provided essential and non-waivable structuring conditions for causal inference are all met, results show that the CIR is most often inestimable whether through the prevalence ratio or the prevalence cross-product ratio, and that the latter is the measure that consistently yields an appropriate measure of the Incidence Density Ratio. Debating the role of multivariate regression models in cross-sectional studies, we contend that such models should be avoided when the structuring assumptions for causal inference do not hold. Nevertheless, if these assumptions are reasonably met, it is the logistic regression model that is best suited for this task as it provides a suitable estimate of the Incidence Density Ratio.
Discussion In closing, we discuss the (un)favourable arguments raised in the literature in the light of our findings.