Dietary transition proxies and ischemic heart disease burden: a country-year panel analysis with a systematic...

Dietary trans fatty acids (TFA) are established risk factors for ischemic heart disease (IHD) at the individual level, but the robustness of country-level ecological associations across dietary transition proxies has rarely been examined systematically. We constructed a country-year ecological panel (8,536 country-years; 194 countries; 1980-2023) linking GBD 2021 total trans-fat exposure, GBD 2023 IHD/stroke outcomes, FAOSTAT dietary supply proxies, WHO policy indicators, and World Bank covariates. Two-way fixed-effects regressions with country-clustered standard errors were estimated at 0-, 3-, 5-, and 10-year lags and subjected to a systematic robustness program: alternative exposure constructions, population weighting, leave-one-country-out refitting, sample-composition exclusions, formal tests of income- and country-size interaction, and multiple-comparison adjustment. At the 5-year lag, higher estimated total trans-fat exposure was associated with higher IHD incidence (β = 5.77 per 0.1% point energy/day; 95% CI: 1.29-10.26; p = 0.012). This association was the most robust examined: it held in all 186 leave-one-country-out refits and across simple, population-weighted, and age-standardized exposure constructions (r ≥ 0.9997) and across OLS and Gamma GLM. However, it could not be separated from the development transition. The unadjusted two-way fixed-effects association remained significant after excluding island-states, whereas the fully covariate-adjusted estimate in the island-excluded sample was attenuated to non-significance (p = 0.111, β = 3.51). Sequential adjustment localized this attenuation to entanglement with urbanization and adiposity rather than to island states. Weighting each observation by population produced null but underpowered estimates (about 39% power to detect the unweighted effect; effective sample 12.9 of 186 countries), with confidence intervals containing the unweighted estimate. Formal interaction tests showed no significant modification by income group or, for IHD incidence, by country size. Sugar/sweeteners supply was positively associated with IHD incidence (β = 3.27; 95% CI: 0.25-6.29; p = 0.034) but did not survive multiple-comparison adjustment. This panel documents a positive trans-fat/IHD association that is consistent across estimators and exposure constructions and at lag windows up to 5 years. However, it is entangled with the development transition, cannot be isolated from it, and could be explained by modest unmeasured confounding (E-values 1.15-1.20); the association is therefore hypothesis-generating and does not, on its own, support causal or policy conclusions.