Global Burden of Disease worldwide cohort analysis of dietary and other risk factors for cardiovascular diseases: lipid hypothesis versus fat-soluble vitamin hypothesis

Cundiff, D. K.; Wu, C.

Background Regarding diets contribution to cardiovascular disease, Ancel Keys, MD proposed his lipid hypothesis in the 1950s. Despite USDA Dietary Guidelines endorsing the lipid hypothesis, debate about whether dietary saturated fat and cholesterol cause cardiovascular disease has continued. Methods Using Global Burden of Disease (GBD 2017) data on cardiovascular disease deaths/100k/year, ages 15-69 years old in male and female cohorts (CVD) and dietary and other risk factors, we formatted and population weighted data from 195 countries. Each of the resulting 7846 rows of data (cohorts) represented about 1 million people, projected to total about 7.8 billion people in 2020. We correlated CVD with dietary and other risk factors worldwide and in appropriate subsets. Findings Low density lipoprotein cholesterol (LDL-c) correlated negatively with worldwide CVD (r= -0.282, 95% CI -0.302 to -0.262, p<0.0001). All foods were expressed in kilocalories/day (KC/d). We summed the KC/d of processed meat, red meat, fish, milk, poultry, eggs, and added (saturated fatty acid, polyunsaturated fatty acid, and trans fatty acid) to create a fat-soluble vitamins variable (FSVV) high in vitamins A, D, E, and K2 (menaquinones). The FSVV correlated negatively with CVD (r= -0.329, 95% CI -0.349 to -0.309, p<0.0001, n=7846 cohorts). We averaged CVD data from pairs of male and female cohorts to determine the 500 pairs of male and female cohorts with the lowest CVD (n=1000 cohorts arbitrarily, approximately 1 billion people). In this subset, mean CVD=227.8 and mean FSVV=531.4 KD/d versus worldwide mean CVD=542.4 and mean FSVV=285.4 KC/d. Meat and poultry negatively correlated with CVD worldwide (e.g., red meat mean=50.27 KC/d, r= -0.232, 95% CI -0.253 to -0.211, p<0.0001) and positively with CVD when FSVV >=531.4 KC/d (e.g., red meat mean=121.9 KC/d, r=0.633, 95% CI 0.597 to 0.668, p<0.0001). Fish and eggs negatively correlated with CVD both worldwide (e.g., mean eggs =19.4 KC/d, r= -0.390, 95% CI -0.408 to -0.371, p<0.0001) and when FSVV>=531.4 KC/d (e.g., mean eggs=39.8 KC/d, r= -0.261, 95% CI -0.315 to -0.206, p<0.0001). Milk product intake was high in 20/34 countries with the lowest CVD (range milk: 68.8 - 124.1 KC/d, including high levels of fermented cheeses with high vitamin K2, versus worldwide mean milk=25.0 KC/d). Japan, South Korea, and Taiwan had low CVD (CVD<=237.4) and low milk intake (mean milk<=29.0 KC/d but high intakes of fermented legumes containing vitamin K2. Interpretation: Since LDLc correlated negatively with CVD worldwide, the lipid hypothesis was not supported. These data support the fat-soluble vitamins hypothesis because FSVV correlated negatively with CVD worldwide, meaning the more fat-soluble vitamin containing animal foods and fat for gut absorption the less the CVD. Vitamin K2 containing milk products, particularly fermented cheeses, fermented legumes, particularly soy, and eggs, appear to be most protective against CVD. In high FSVV countries, reducing meat and poultry intake by at least half would likely reduce CVD significantly. This GBD analysis methodology could be used in developing food policy and education strategies for reducing CVD and improving public health. Funding: None