Foodsmart™ is a digital nutrition platform that encourages lasting behavior change through personalization of nutrition and meal/recipe recommendations and through altering the food purchasing environment to provide healthy eating options. The two main components are FoodSmart and FoodsMart, which both use behavior change theory to facilitate access and engagement with affordable, tasty, and healthy food (Fig. 1).
The FoodSmart component contains the in-app Nutriquiz, a food frequency questionnaire (based on the National Cancer Institute’s Diet History Questionnaire) which users take to report their dietary habits, and that provides immediate feedback on areas they can improve upon as well as personalized meal and recipe planning based on the Nutriquiz results. The Nutriquiz also ascertains demographic information, weight, and clinical conditions. The user can retake the Nutriquiz at any time, allowing them to monitor their progress on diet and weight. The other component is FoodsMart, which helps alter the user’s food purchasing environment through personalized meal planning. Users can add to their grocery list within the app and then use integrated online ordering and delivery of groceries, meal kits, and prepared foods. Customized grocery discounts on healthier options help the user save money and further nudges the user to make healthier choices.
Foodsmart is available through certain health plans and employers, who provide this product as an option or benefit for their members/employees to enroll in. It is available on web, iOS, and Android.
We conducted a longitudinal, retrospective analysis of 1,740 adults with obesity living in the U.S. who used Foodsmart between January 2013 and April 2020. As of April 2020, there were 888,999 users who had signed up for Foodsmart. Among all of the users, we excluded individuals who did not report weight (n = 562,276), individuals who reported extreme values for height (< 54 or > 78 inches) or weight (< 60 or > 400 pounds) (n = 25,946), and individuals who did not have obesity [body mass index (BMI) < 30 kg/m2] at first weight entry (n = 200,308). We further excluded individuals who did not report weight at least three times and participants with less than 1 month between first and second or second and third weight report (n = 98,729). The final analytic sample included 1,740 users.
Dietary data were self-reported through Foodsmart. Upon registration, users were prompted to fill out a dietary questionnaire called “Nutriquiz”, a 53-item food frequency questionnaire adapted from the National Cancer Institute Diet History Questionnaire, which has been previously validated . Information on sex, age, weight, and usual frequency of dietary intake (fruits, vegetables, whole grains, proteins, carbohydrates, fats, fiber, sodium, and water) are collected through the Nutriquiz. A healthy diet score created by the Foodsmart research team called Nutriscore was calculated, which is derived from the Alternative Healthy Eating Index-2010, a previously validated score among several U.S. cohorts, and the Commonwealth Scientific and Industrial Research Organization (CSIRO) Healthy Diet Score [22, 23]. Participants were assigned a score from 0 to 10 (with 10 being optimal) for each of seven components: fruits, vegetables (excluding potatoes), protein ratio (white meat/vegetarian protein to red/processed meat), carbohydrate ratio (total fiber to total carbohydrate), fat ratio (polyunsaturated to saturated/trans fat), sodium, and hydration (percent of daily fluid goal). A total Nutriscore (possible scores ranging from 0 to 70) was calculated by summing the scores of the seven components. Change in Nutriscore was calculated as the difference between a participant’s first and last Nutriscores. We categorized participants by whether their Nutriscore decreased or was stable (no improvement in diet quality) versus increased (improvement in diet quality).
Measurement of Weight
Users were given the option to add weight and height data when they first created their Foodsmart account and could update their weight at any time during usage of the platform. Baseline BMI was calculated as first weight entry in kilograms divided by height in meters squared (kg/m2). We categorized participants by baseline obesity class. Class 1 obesity was defined as a BMI between 30 and 34.9 kg/m2, class 2 was defined as a BMI of 35 to 39.9 kg/m2, and class 3 was defined as a BMI of 40 kg/m2 or higher.
Our primary outcome was sustained weight loss, which we defined as losing 5 % of initial weight between first and second reported weights and additional weight loss or no change between the second and third reported weights.
Duration of enrollment (in months) in Foodsmart was calculated as the number of months between the first activity date and last activity date.
We used descriptive analyses to examine baseline characteristics of the total study population and by whether participants sustained weight loss or not. We reported categorical variables as frequencies (%) and continuous variables as mean ± standard deviation (SD).
To investigate long-term efficacy of Foodsmart on sustained weight loss, we examined the percent of participants who sustained weight loss by the duration of their enrollment time (by 12, 24, and 36 months). Further, we examined the percent of participants by each category of age, baseline obesity class, and change in Nutriscore. We used chi-square tests to determine whether differences within each category were statistically significantly different.
Multivariate logistic regression models were used to estimate odds ratios (OR) and 95 % confidence intervals (CI) of sustained weight loss adjusted for gender, age category, baseline obesity category, baseline Nutriscore (per 2-point increase), and change in Nutriscore (per 2-point increase).
We considered a P-value smaller than 0.05 to be significant for all tests. Stata version 16 was used for all analyses (StataCorp, College Station, Texas).
The study was declared exempt from Institutional Review Board oversight by the Pearl Institutional Review Board given the retrospective design of the study and less than minimal risk to participants.