Abstract P345: A Diet-Wide Association Study on Blood Pressure- the INTERMAP Study

Background: Epidemiological studies have examined the relationship between dietary habits and blood pressure. However, studies often test one or only a few factors at a time or examine food patterns (e.g. the Mediterranean diet) not well assessed or defined. Here we use the previously suggested Environmental-wide Association Study (EWAS) paradigm, which evaluates multiple factors concurrently for associations, adjusting for multiplicity of comparisons. Methods: We used data from the INTERMAP cross-sectional epidemiological Study with 4680 persons ages 40 to 59 years from 17 random population samples in China, Japan, the United Kingdom, and the United States. BP was measured 8 times at 4 visits; dietary data (83 nutrients) were obtained from 4 standardized, multipass, 24-hour dietary recalls and 2 timed 24-hour urine collections. We examined association between each nutrient and systolic and diastolic blood pressure in separate linear regression models using a random 50% test sample, stratified across centers, of the INTERMAP population. The most significant results were subsequently validated in the remaining 50%. The false discovery rate (FDR), the expected proportion of false positive results among statistically significant hypotheses, was calculated. We considered an examined nutrient as validated when the FDR was 5% in the training set and P-value for association was <0.05 in the validation set. Results: In models adjusted for multiple confounders, dietary alcohol and urinary sodium-to-potassium ratio had positive associations and non-heme iron intake, dietary phosphorus, vegetable protein, magnesium, total carbohydrate, pantothenic acid, folacin, thiamin, riboflavin and urinary calcium inverse associations with systolic blood pressure (FDR<5% in training and P<0.05 in validation). With the same set of confounders, dietary alcohol had positive and non-heme iron intake, dietary calcium, phosphorus, protein, thiamin and glutamic acid and urinary potassium and magnesium, inverse associations with diastolic blood pressure (FDR<5% in training and P<0.05 in validation). Conclusions: Using the methodology originally proposed for genome-wide association studies studies (GWAS), we replicated previously reported associations between dietary habits and blood pressure, also found some novel associations for B vitamins. As epidemiological studies increasingly collect large amount of data, it is essential to report transparently all analyses, account for multiple tests, and replicate findings.