Changes in Ultra-Processed Food Consumption and Lifestyle Behaviors Following COVID-19 Shelter-in-Place: A Retrospective Study

Abstract

Ultra-processed food (UPF) consumption poses a potential risk to public health and may be related to shelter-in-place orders. This study utilized the level of food processing as a lens by which to examine the relationships between diet, weight change, and lifestyle changes (including cooking, snacking, and sedentary activity) that occurred during regional shelter-in-place orders. This study used a cross-sectional, retrospective survey (n = 589) to assess baseline demographics, changes in lifestyle behaviors using a Likert scale, and changes in dietary behaviors using a modified food frequency questionnaire from mid-March to May 2020; data were collected in the California Bay Area from August to October 2020. Foods were categorized by level of processing (minimally processed, processed, and ultra-processed) using the NOVA scale. Stepwise multiple linear regression and univariate linear regression models were used to determine the associations between these factors. Increased snacking was positively associated with a change in the percent of the calories derived from UPF and weight gain (β = 1.0, p < 0.001; β = 0.8 kg, p < 0.001) and negatively associated with the share of MPF calories consumed (β = -0.9, p < 0.001). These relationships have public health implications as interventions designed around decreased snacking may positively impact diet and weight management and thereby mitigate negative health outcomes.

Keywords: COVID-19; diet; lifestyle behaviors; shelter-in-place; snacking; ultra-processed foods.

Figure 1

Univariate linear regression outputs presenting the relationships between behavioral, demographic, and anthropometric factors to change in percent calories from minimally processed (MPF), processed (PF), and ultra-processed foods (UPF). Results are presented as beta coefficients with 95% confidence intervals. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 2

Combined stepwise linear regression outputs presenting the relationships between behavioral, demographic, and anthropometric factors to change in percent calories from minimally processed (MPF), processed (PF), and ultra-processed foods (UPF). This analysis was conducted in six separate models for each type of calorie, separating factors into either demographic/anthropometric factors or behavioral factors. Changes in alcohol consumption, sedentary activity, cooking, takeout food consumption, baseline exercise, and ready-to-eat packaged food consumption were included in the behavioral model, while the remaining variables were included in the demographic/anthropometric model. Factors dropped from models are not shown. Results are presented as beta coefficients with 95% confidence intervals. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3

Univariate linear regression outputs presenting the relationships between behavioral, demographic, and anthropometric factors to change in weight in kilograms. Results are presented as beta coefficients measured in kg with 95% confidence intervals. * p < 0.05, *** p < 0.001.

Figure 4

Combined stepwise linear regression outputs presenting the relationships between behavioral, demographic, and anthropometric factors to change in weight in kilograms. This analysis was conducted in six separate models for each type of calorie, separating factors into either demographic/anthropometric factors or behavioral factors. Changes in alcohol consumption, sedentary activity, cooking, takeout food consumption, baseline exercise, and ready-to-eat packaged food consumption were included in the behavioral model, while the remaining variables were included in the demographic/anthropometric model. Factors dropped from models are not shown. Results are presented as beta coefficients in kg with 95% confidence intervals. * p < 0.05, ** p < 0.01, *** p < 0.001.