Response to Letter to the Editor on “Does dietary supplementation with polyunsaturated fatty acids really help prevent erectile dysfunction? Findings from a mixed-exposure analysis”

We thank the correspondents for engaging with our mixed-exposure analysis of dietary polyunsaturated fatty acids (PUFAs) and erectile dysfunction (ED) in National Health and Nutrition Examination Survey (NHANES) 2001-2004. Their comments appropriately highlight key design and interpretation issues in nutritional epidemiology, particularly when correlated nutrients are modeled as a mixture. First, regarding propensity score matching in a complex survey: we agree that matching trades sample size—and potentially population representativeness—for covariate balance. Our primary aim was not to re-estimate a nationally weighted prevalence effect, but to reduce model dependence and make the ED and non-ED groups more comparable before applying mixture models that are sensitive to correlation structure. In that setting, the matched sample is best viewed as an internally balanced analytic cohort. We also agree with the correspondents’ point that the observed effect estimates (eg, an odds ratio around 0.86 with wide confidence intervals in Weighted Quantile Sum (WQS)) are compatible with anything from a modest benefit to no effect. Accordingly, our intended inference is narrower: the NHANES 2001-2004 data do not provide strong or consistent evidence for a clinically meaningful protective association once the PUFA mixture is modeled jointly, rather than proving a complete absence of any association.