The relative validity of nutrition assessment methods for estimating infant carotenoid intake differs by assessment tool, nutrient database, and milk carotenoid adjustment method

Abstract

Validated carotenoid assessment methods are needed to study infant carotenoid nutrition. This is a secondary analysis of repeated diet assessments of healthy participants collected at 4- (n = 21), 6- (n = 12), and 8- (n = 9) months of age in Houston, TX between April 2019 and June 2020. Intake was assessed with 3 assessment tools, analyzed with 3 nutrient databases, and underwent 3 adjustments to account for milk composition variability. We hypothesized that manual adjustment of milk carotenoid intake based on laboratory measurements would improve the validity of all assessment approaches and that using a database with greater coverage of infant food carotenoid compositions would improve accuracy. Generalized linear mixed models assessed associations between tool, nutrient database, age, and milk carotenoid adjustment variables with carotenoid, energy, fruit, and vegetable intakes. The effect of the number of food diary days on intake estimate precision was evaluated by testing the correlation between intake estimates derived from 1, 3, or 5, vs. 7 days. Visit age influenced energy intake estimates (p = .029), along with assessment tool (p = .020). Estimates of vegetable intake were influenced by tool (p = .009). Combined fruit and vegetable intake differed by nutrient database (p = .007). Carotenoid intake differed by age (p =<.0001), tool (p = .002), and nutrient database (p = .004). A minimum of 3 food diary days strongly correlated (rho = 0.79-1) with reference estimates across ages. Milk carotenoid adjustment was most influential in estimating 4-month olds’ carotenoid intake, while nutrient database and tool were important for 6- and 8-month-olds’, highlighting the dynamic nature of infant diet assessment validity across feeding stages.