Original Contribution A Dynamic Panel Model of the Associations of Sweetened Beverage Purchases With Dietary Quality and Food-Purchasing Patterns

Investigating the association between consumption of sweetened beverages and dietary quality is challenging because issues such as reverse causality and unmeasured confounding might result in biased and inconsistent estimates. Using a dynamic panel model with instrumental variables to address those issues, we examined the in-dependent associations of beverages sweetened with caloric and low-calorie sweeteners with dietary quality and food-purchasing patterns. We analyzed purchase data from the Homescan survey, an ongoing, longitudinal, na-tionally representative US survey, from 2000 to 2010 (n = 34,294). Our model included lagged measures of dietary quality and beverage purchases (servings/day in the previous year) as exposures to predict the outcomes (macro-nutrient (kilocalories per capita per day; %), total energy, and food purchases) in the next year after adjustment for other sociodemographic covariates. Despite secular declines in purchases (kilocalories per capita per day) from all sources, each 1-serving/day increase in consumption of either beverage type resulted in higher purchases of total daily kilocalories and kilocalories from food, carbohydrates, total sugar, and total fat. Each 1-serving/day in-crease in consumption of either beverage was associated with more purchases of caloric-sweetened desserts or sweeteners, which accounted for a substantial proportion of the increase in total kilocalories. We concluded that consumers of both beverages sweetened with low-calorie sweeteners and beverages sweetened with caloric sweeteners had poorer dietary quality, exhibited higher energy from all purchases, sugar, and fat, and purchase