Dietary fiber and total fat intake density and self-reported cancer history in U.S. adults: NHANES 2021-2022.

BACKGROUND

Diet-cancer associations are often small and sensitive to energy modeling. We tested whether harmonized energy-density exposures yield consistent inferences across design-based and Bayesian approaches in a nationally representative sample.

METHODS

We analyzed U.S. adults (≥20 years) in NHANES 2021-2022. Exposures were 2-day mean dietary fiber density and total-fat density (g/1000 kcal), standardized and mutually adjusted. Outcomes were self-reported history of any cancer and breast, prostate, and colorectal cancers. We fit survey-weighted logistic regression and a complementary hierarchical Bayesian logistic model.

RESULTS

For any cancer, associations were near null and concordant across methods (per 1-SD: fiber OR 1.03, 95% CI 0.91-1.17; total-fat OR 0.95, 0.82-1.09). Breast and prostate cancer estimates were similarly close to the null. In contrast, colorectal cancer showed a consistent inverse association with higher total-fat density (survey OR 0.46, 0.29-0.74; Bayesian OR 0.55, 95% credible interval [CrI] 0.35-0.84), whereas fiber density showed little evidence of association.

CONCLUSIONS

In NHANES 2021-2022, fiber and total-fat densities were not meaningfully associated with overall cancer history. The inverse colorectal cancer pattern for total-fat density was robust across analytic frameworks but should be interpreted cautiously given the cross-sectional, self-reported outcome and potential survivorship-related bias.