Background Online grocery delivery services (OGDS) are becoming increasingly popular. The impact of OGDS on population diet and inequalities in food purchases is under-studied. This study examined how use of OGDS varies by sociodemographic characteristics and is associated with the amount and types of grocery purchased.

Methods Food and drink purchase data were available from households (n=1911) in the 2019 UK Kantar Fast Moving Consumer Goods Panel. Households were randomly sampled if residing in London and the North of England. Purchases were categorised as being bought online or in-store based on store type information. We used logistic regression to estimate the likelihood of above median frequency of OGDS use by sociodemographic characteristics. We then used Poisson regression to estimate (i) differences in overall energy purchased by households with above and below median OGDS use and (ii) the proportion of energy purchased from 35 food groups online versus in-store among households that used both shopping methods (n=665). All analyses were conducted in Stata SE 16.

Results Median use was five occasions in 2019 for households that used OGDS (n=668). Higher-income households were more likely to have above median use (OR 1.56, 95% CI 1.02 to 2.38 for ≥£50,000 versus <£20,000). Households with above, versus below, median OGDS use purchased a mean of 1,460.8 kcal (95% CI 1,447.7 to 1,473.8) more energy per person per week. Households that used both in-store and online shopping methods tended to have healthier purchases online. Online versus in-store purchases had a larger proportion of energy from vegetables (1.0%, 95% CI 0.2 to 1.8), healthy non-milk-based drinks (1.6%, 95% CI 0.7 to 2.4) and alcohol (1.2%, 95% CI 0.4 to 2.1), and a smaller proportion of energy from puddings and biscuits (-3.3%, 95% CI -4.1 to -2.5), and chocolate and confectionery (-1.5%, 95% CI -2.2 to -0.7).

Conclusion Although online purchases were healthier, households with above median OGDS use had greater total energy purchases. This may lead to increases in over-consumption or waste. Further, OGDS may intensify existing socioeconomic inequality in food purchasing because of differential OGDS use across socioeconomic groups. Thus, the online food environment may have positive and negative consequences for population diet. Study limitations include reliance on self-reported data, as households may have forgotten to report some purchases, and the lack of information on intra-household distribution of purchases.