Different versions of the German nutrient database affect energy and nutrient intake among adolescents

Abstract Updates of food composition data complicate comparisons of energy and nutrient intake between consecutive dietary surveys. The effect of an update of the German Nutrient Database (BLS) from version BLS2.3 to BLS3.02 on calculated energy and nutrient intake was evaluated among 1,272 adolescents aged 12 to 17 years from the first German children and adolescents nutrition survey EsKiMo. Food consumption during the past four weeks was assessed by computer assisted dietary history interviews. Energy and nutrient intake was calculated using BLS2.3 and recalculated using BLS3.02. Pair-wise differences in intakes derived from BLS2.3 and BLS3.02 were statistically significant for all nutrients except for cholecalciferol. Differences were less than 5% for energy and macronutrients but exceeded 10% for s-carotene (+12.3%), biotin (+12.9%), calcium (-18.7%), magnesium (-19.6%), iron (-17.6%), copper (-19.1%), and manganese (-11.8%). In total, 72.6 to 95.0% of participants were assigned to the same intake tertile based on both BLS-versions. Kappa-coefficients varied between 0.68 and 0.94. Median intakes were consistently above or below 100% of the reference values using both BLS versions, except for iron and calcium. In order to make more cautious interpretations of differences over time, recalculation of intakes according to the actual food composition database should be considered.