Efficacy of supermarket and web-based interventions for improving dietary quality: a randomized, controlled trial

Supermarkets with expansive footprints and evolving healthcare operations may offer opportunities to deliver clinical care services beyond traditional medical settings, including addressing unhealthy diets. Over the past decade, supermarket-based retail clinics have incorporated registered dietitians, and new retail technologies (online shopping, delivery, nutrition-support applications) can help overcome barriers to healthy eating. Electronically collected purchasing data can provide insight into dietary behaviors for consumers and healthcare providers. The SuperWIN trial was designed to test two in-person, dietitian-led education interventions focused on the DASH dietary pattern, individualized using each participant’s food-purchasing data. The first intervention emphasized the in-store shopping environment (Strategy 1), while the second added online shopping, home delivery and other technologies (Strategy 2) to improve grocery purchases and diet quality. Rigorous academic–retail collaborations to improve dietary quality have been limited; this study aimed to evaluate the efficacy of these supermarket- and web-based interventions on improving DASH adherence.

The discussion situates the study within evidence that suboptimal diet is the leading global risk factor for mortality. Increased adherence to DASH is consistently associated with improved clinical outcomes; a meta-analysis found each five-point DASH score increase associated with 5% lower all-cause mortality and 3–4% lower cardiovascular, stroke and cancer mortality. Trials such as PREDIMED and observational studies indicate that even modest improvements in diet quality yield substantial cardiovascular risk reductions. While DASH was validated in feeding studies, the PREMIER trial demonstrated benefits in free-living populations through intensive, center-based programs. The 2021 American Heart Association guidance endorses heart-healthy dietary patterns, including DASH. Despite recommendations, DASH adoption remains low in the United States, particularly among individuals with hypertension. An AHA Science Advisory called for innovative approaches and partnerships with retailers, and research into benefits of online shopping and smart technologies. The present study leverages such a supermarket-based partnership with dietitian-delivered education and online tools to extend prior dietary trial findings.

Design: Randomized, parallel-assignment, active-control efficacy trial (SuperWIN), registered at ClinicalTrials.gov (NCT03895580), approved by the University of Cincinnati IRB, with written informed consent. Setting and timeframe: Conducted across 13 Kroger supermarket locations in Ohio and Kentucky (each with in-store clinic). Recruitment March 2019–February 2021, with a temporary COVID-19 pause beginning March 13, 2020, and resumption June 2020. Participants: UC Health primary care patients aged 21–75 who were primary household food planners, current shoppers at a study Kroger store, able to shop and prepare food independently, and had a home computer. Inclusion required at least one cardiovascular risk factor: SBP >130 mmHg, DBP >80 mmHg and/or antihypertensive treatment; obesity (BMI ≥30 kg/m²); or non-HDL-C ≥130 mg/dl and/or lipid-lowering therapy. Key exclusions: concurrent dietary/weight-loss intervention, use of Kroger online shopping within 12 months, prior Kroger dietary counseling, baseline SBP ≥190 mmHg, DBP ≥110 mmHg, or non-HDL-C ≥190 mg/dl. Recruitment via database queries, mailings, phone/text/email, and flyers. Participants could receive up to two $25 incentives. Run-in and randomization: After run-in (baseline dietary phone recalls and surveys), participants attended an in-store visit with a supermarket registered dietitian. After confirming eligibility and interest, participants were randomized 2:2:1 to Strategy 1, Strategy 2, or control using REDCap with stratification by age group, gender, and household size; block size 5. Principal investigators and key staff were unblinded. Interventions: All participants first received a 30-minute enhanced medical nutrition therapy session (standard of care) before randomization, covering DASH education, personal goals, and action plan, enhanced by visualizations of baseline dietary intake. Strategy 1 (in-store education): six additional in-store educational sessions at 2-week intervals over 3 months, using the supermarket environment and individualized purchasing data (from Kroger loyalty card) to guide education. Strategy 2 (in-store education plus online tools): same six in-store sessions plus stepwise training in the store’s online shopping platform, free home delivery, and two additional health applications; tools to facilitate healthier purchases, meal planning, and recipes. For Strategies 1 and 2, updated individualized purchasing data (food groups/items, counts, spend) were reviewed at each session. All participants had in-store assessments at baseline, 3, and 6 months; dietary recalls were collected by phone. Endpoints: Primary endpoint was change in DASH score (0–90 scale) from baseline to 3 months; also assessed at 6 months. DASH score derived from 11 component scores aligning intake with DASH serving recommendations, calculated from three 24-hour dietary recalls (two weekdays, one weekend) at each time point after study completion. Prespecified secondary endpoints: SBP, DBP, BMI, non-HDL-C, total cholesterol, and triglycerides measured at baseline, 3, and 6 months. COVID-19 impact analyses were prespecified; the prepandemic subgroup comprised those completing 3-month follow-up before March 13, 2020. Statistical analysis: Intention-to-treat. Regression-based multiple imputation for missing follow-up outcomes. Mixed model for repeated measures with DASH score as dependent variable; independent variables included treatment group, time, and group×time interaction; models controlled for age group, sex, household size, race, income, and baseline BMI. Two hierarchical primary hypotheses: (1) combined Strategies 1 and 2 vs enhanced control on 3-month change in DASH score; (2) Strategy 2 vs Strategy 1 on 3-month change (tested only if hypothesis 1 P < 0.05). Two-sided alpha 0.05; no further multiple-comparison adjustments. Power: n=100 per intervention group and n=50 control provided 93% and 97% power to detect a 5-point between-group difference (assumed SD=9). Analyses conducted in SAS 9.4.

Participants: 267 randomized; due to COVID-19, 20 participants without 3-month data were withdrawn per prespecified plan. Analysis cohort n=247 (139 women, 108 men); prepandemic subgroup n=109. Engagement and data completeness: Overall, 91% provided 3-month dietary data. In Strategies 1 and 2, 96% attended ≥1 intervention visit; 80% completed all six visits. Prepandemic visit completion: 97.8% (Strategy 1) and 98.4% (Strategy 2); during pandemic: 84.5% and 81.1%, respectively. Prepandemic 3-month dietary and biometric data collection: 100%; during pandemic: 84.1% and 79.0%, respectively. Baseline characteristics: Median age 58 years; 69% female; 21% Black/African American; 64.8% married/partnered; 53.4% full-time employed; 63.6% with ≥ bachelor’s degree. Mean SBP 129.3 (SD 16.7); 72.9% on antihypertensive meds; 63.6% with obesity. Median non-HDL-C 111 mg/dl; 41.7% on lipid-lowering meds. Groups were balanced. Primary endpoint (DASH score): Baseline mean DASH 45.4 (SD 10.4). From baseline to 3 months, mean DASH score increased by 5.8 (95% CI 2.5, 9.2) in control, 8.6 (6.4, 10.8) in Strategy 1, and 12.4 (10.3, 14.6) in Strategy 2. Combined Strategies 1 and 2 increased by 10.5 (9.0, 12.1). Between-group differences: Combined Strategies 1 and 2 vs control, +4.7 points (0.9, 8.5), P=0.02 (primary hypothesis met). Strategy 2 vs Strategy 1, +3.8 points (0.8, 6.9), P=0.01 (hierarchical secondary hypothesis met). Sustainability (6 months): Within-group increases persisted: control +4.4 (0.6, 8.1), Strategy 1 +6.6 (4.0, 9.2), Strategy 2 +8.4 (5.9, 11.0). Combined Strategies 1 and 2 remained elevated by +7.5 (5.7, 9.3), P<0.0001 vs baseline; however, between-group difference vs control at 6 months was not significant: +3.1 (−1.0, 7.3), P=0.14. Strategy 2 vs 1 at 6 months: +1.8 (−1.9, 5.5), P=0.34. DASH components (3 months): Strategy 2 vs 1 exhibited greater increases in fruit servings per 1,000 kcal: +0.5 (0.0, 0.9), P=0.03; combined Strategies vs control also showed higher increases in fruits (+0.8, P<0.01) and whole grains (+0.9, P<0.01). Other components showed improvements but between-group differences were generally nonsignificant; sodium decreased similarly across groups (−550 to −615 mg/day). Secondary endpoints: At 3 months, SBP decreased within Strategy 1 (−6.6 mmHg, 95% CI −9.8, −3.4) and Strategy 2 (−5.7, −8.7, −2.8); DBP decreased by −2.4 and −2.0 mmHg, respectively; BMI decreased by −0.4 and −0.8 kg/m², respectively. Control showed no decreases at 3 months. Between-group comparisons for SBP, DBP, BMI found no significant differences. At 6 months, SBP decreased in control (−5.4), Strategy 1 (−5.2), Strategy 2 (−4.2); DBP decreased in control (−3.7) and Strategy 1 (−4.4) but not in Strategy 2; BMI decreased in Strategy 1 (−0.6) and Strategy 2 (−0.7). Lipids (non-HDL-C, total cholesterol, triglycerides) showed no reductions in combined Strategies vs control at 3 or 6 months. COVID impact analyses: In the prepandemic subgroup, combined Strategies 1 and 2 vs control increased 3-month DASH by +8.3 (3.4, 13.3), P=0.001; the 6-month difference was not significant (+5.1, −0.8 to 11.1, P=0.09). Strategy 2 vs 1 at 3 months showed a nonsignificant difference (+3.1, −1.3 to 7.6, P=0.17). No significant between-group reductions were observed for SBP, DBP, BMI, or lipids in prepandemic analyses. Subgroups: Greater improvements in DASH adherence (Strategies 1 and 2 vs control) were associated with older age (51–75 vs 21–50), white race (white vs nonwhite), and baseline hypertension (with vs without); no differences by gender (interaction P ≤ 0.01 for age, race, hypertension).

The trial addressed whether supermarket-based, data-guided dietitian interventions, with and without online tools, improve adherence to the DASH diet in free-living adults with cardiovascular risk factors. All groups, including an enhanced standard-of-care control, improved DASH adherence at 3 months, with improvements persisting at 6 months. Compared with control, combined in-store interventions produced a clinically meaningful additional 4.7-point increase at 3 months, and adding online shopping and supportive technologies yielded a further 3.8-point gain over in-store education alone. These findings highlight that delivering individualized nutrition education within participants’ home supermarkets, guided by purchasing data, and augmented by online tools can meaningfully enhance diet quality. While between-group differences were not sustained at 6 months, within-group improvements remained, suggesting potential value in extending contact (additional sessions, telenutrition, ongoing feedback of purchasing data) to maintain gains. Secondary cardiometabolic outcomes improved within groups, but no between-group differences were detected, indicating that larger or longer studies or combined interventions may be needed to translate dietary improvements into measurable changes in blood pressure, adiposity, or lipids. The study supports retailer–academic partnerships as a scalable model to improve dietary quality, leveraging the convenience of supermarkets and the growing adoption of online shopping platforms.

Individualized, in-person, dietitian-led nutrition education delivered in supermarkets and guided by electronic purchasing data significantly improved adherence to the DASH diet at 3 months compared with an enhanced control. Incorporating online shopping technologies and training produced additional improvements beyond in-store education alone. Although between-group differences diminished by 6 months, diet quality remained improved from baseline across all groups. Further research should assess strategies to sustain and amplify dietary improvements over longer periods, evaluate effects on clinical outcomes (blood pressure, lipids, weight) with adequate power, and test integration with complementary components such as exercise interventions, detailed medication assessments and titration, adherence counseling via retail pharmacies, and continued remote support (telenutrition, ongoing purchasing feedback). These results demonstrate the promise of supermarket-based interventions and online tools for enhancing dietary quality and underscore the value of collaborations with the retail sector.

Generalizability may be limited as participants were receiving routine primary care and were predominantly middle-aged, female, married or partnered, and had relatively higher household incomes. The COVID-19 pandemic disrupted intervention delivery and data collection, potentially affecting outcomes and shopping behaviors. Medication data were limited: doses were not collected, use was assessed via survey (not rigorously verified), and medication use at 3 months was not collected; these factors limit interpretation of blood pressure and lipid changes. Blood pressure measurements were limited in number at each time point, potentially reducing precision. Baseline levels and cohort characteristics may have reduced power to detect changes in secondary endpoints. The study did not assess long-term maintenance beyond 6 months.