Data_Sheet_1.docx

Abstract

Heterogeneity in meat food groups hinders interpretation of research regarding meat intake and chronic disease risk. Our objective was to investigate how heterogeneity in red meat (RM) and poultry food groups influences US population intake estimates. Based on a prior systematic review, we created an ontology of methods used to estimate RM [1= unprocessed RM; 2 (reference)= unprocessed RM + processed RM; 3= unprocessed RM + processed RM + processed poultry; and 4=unprocessed RM + processed RM + processed poultry + chicken patties/nuggets/tenders (PNT)] and three for poultry [A=unprocessed poultry; B= unprocessed poultry + PNT; C (reference)= unprocessed poultry + processed poultry + PNT). We applied methods to 2015-18 National Health and Nutrition Examination Survey data to estimate RM and poultry intake prevalence and amount. We estimated and compared mean intake and usual intake distributions within RM methods and within poultry methods via the NCI Method for individuals ≥2 years old (n=15,038), adjusted for age, sex, and race/Hispanic origin. We compared the percent of the population that exceeded age- and sex-specific RM and poultry allotments from the Dietary Guidelines for Americans recommended eating patterns. The percent of those ≥2 years old consuming RM ranged from 47±1.2% to 75±0.8% across RM methods and mean amount ranged from 10.5±0.28 to 18.2±0.35 lean oz-equivalents/week; 38±1.2% to 71±0.7% and 9.8±0.35 to 13.3±0.35 lean oz-equivalents/week across poultry methods. Estimates for higher, but not lower, percentiles of the intake distribution differed across RM methods. Compared to the reference, Method 1 was ≥3.0 oz-equivalents/week lower from 20th-70th percentiles, ≥6.0 oz-equivalents/week lower from 75th-90th percentiles, and ≥9.0 oz-equivalents/week lower for the 95th percentile. Method 4, but not Method 3, differed from the reference by ≥3.0 oz-equivalents/week higher from 50th-95th percentiles. The population percentage that exceeded allotments compared to the reference was 27±1.8% lower for Method 1, 9±0.8% higher for Method 3, and 14±0.9% higher for Method 4. Differences were less pronounced for poultry. Our analysis quantifies the magnitude of bias introduced by heterogeneous meat food group methodology. Explicit descriptions of meat food groups are important for development of dietary recommendations to ensure that research studies are compared appropriately.