For six months, a bus converted into a single-aisle grocery store parked outside twelve subsidised housing sites in the Minneapolis–St. Paul area, once a week, on a schedule set with the buildings. It carried about 170 items, priced roughly level with Aldi and 17% to 39% below nearby supermarkets, with produce making up almost half of sales. Residents said they liked it. A lot of them did not shop there.

The trial that tested it, published 3 November in JAMA Network Open, is the first to measure a full-service mobile grocery market against overall diet quality rather than produce intake alone. The headline finding is a null: diet quality, fruit and vegetable intake, energy intake, added sugar, food security and household produce procurement were all statistically indistinguishable from a waiting-list control group. One outcome moved, and it moved in a direction the researchers had not centred in their hypothesis: sodium.

What the trial actually tested

This was a cluster randomised clinical trial run in three waves between January 2022 and October 2025. Twelve public and private subsidised housing sites were randomised to weekly market service or to a waiting list, with 242 adults completing baseline assessments — mean age 66.5, 64.5% women, and 56.6% receiving SNAP benefits. Retention was 90.0%, unusually strong for a community trial with three 24-hour dietary recall interviews per participant.

The intervention was not a produce stand. The market ran two buses, carried proteins, dairy and pantry staples alongside fruit and vegetables, limited shelf-stable items high in sodium and added sugar, matched SNAP produce purchases with coupons up to $10, and gave loyalty rewards every fourth trip. What it did not provide was nutrition education — a deliberate feature of the design, not an oversight.

Sites sat a mean of 0.9 miles from the nearest full-service grocery store (range 0.4 to 1.6 miles), and 7 of the 12 housed seniors or seniors and people with disabilities. In other words, this was a test of convenience, price and stocking inside neighbourhoods that were low-access by the study’s own definition, but not places without any supermarket within walking distance.

Six of seven secondary outcomes were flat

On the primary endpoint — the Healthy Eating Index-2020, a 0-to-100 score of how well a diet matches federal dietary guidance — the intervention group finished 1.25 points higher than controls, with a confidence interval running from −1.60 to 4.10 and a p value of 0.39. Both arms scored in the low 50s at follow-up. That is a small, imprecise difference, and the authors do not claim otherwise.

Daily fruit and vegetable intake was effectively identical at follow-up: 2.0 servings in the intervention group against 1.8 in controls, an adjusted difference of −0.07. Weekly produce brought into the home fell in both arms and the difference was not significant. Energy intake, added sugar as a share of calories, food security and food-security improvement all failed to separate.

Then there was sodium. Intervention participants averaged 2,526 mg a day at follow-up against 2,890 mg in controls — an adjusted difference of −397 mg (95% CI, −636 to −158; p = 0.001). In the complier-average causal effect model, shopping at the market once a month was associated with 761 mg less sodium per day (95% CI, −1,283 to −239).

Why salt may have been the one thing that shifted

The paper offers three plausible mechanisms, and is careful to label them speculation. First, the market’s stocking guidelines kept most items at or below 10% of the daily value for sodium and minimised anything at 21% or above — a form of choice architecture applied to the shelf rather than to the shopper. Second, buying fresh produce and proteins may have displaced canned vegetables and processed meats. Third, a full-service market in the building might have cut the quick corner-store trips that bring home eggs, milk and a great deal of packaged sodium along with them.

One detail complicates the story in a useful way. These participants started at roughly 2,800 mg a day, below the national average of more than 3,300 mg that CDC reports for Americans, and well below the federal recommendation of under 2,300 mg for teenagers and adults. A 397 mg reduction moves a person in that sample closer to the guideline without reaching it.

The number that explains the null result: 10.3%

Uptake was the constraint, not satisfaction: a mean of 110 of 1,071 households per site — 10.3% — used the market in a typical week, and sites averaged 18 weekly transactions. Among study participants specifically, 43.1% shopped at least once during the two months used to measure dose. Mechanical trouble took the buses off the road often enough that weekly service happened in 117 of 156 weeks during the first six months.

Yet satisfaction was not the problem. Of 94 intervention participants who reported shopping there, 91.5% were satisfied with prices, 88.3% with service reliability, and 95.7% said they would recommend it to a friend. The gap between “people like it” and “people use it weekly” is the most transferable finding in the paper. Residents screened for interest still shopped elsewhere, and the authors list the likely reasons without resolving them: more choice at a supermarket, established shopping habits, schedules, or simply adequate access already.

That uptake shortfall matters for interpretation. With fewer than half of participants shopping during the dose window, an intention-to-treat analysis dilutes whatever effect the market could have had — which is why the causal model’s 761 mg figure is larger than the headline 397 mg, and why the nulls on diet quality may partly reflect a trial of a market people did not visit rather than a market that did not work.

What the results sit alongside

Two of the three earlier cluster randomised trials of mobile produce markets found fruit and vegetable intake rose by roughly half a serving to a serving a day; a third found no change in intake or food security. The authors point out that this trial measured diet with three 24-hour recalls, as one of those earlier trials did, and still saw nothing. They also note that studies of new brick-and-mortar supermarkets have generally failed to shift diets either, which pushes the explanation away from the vehicle and toward what happens after the doors open.

Their leading hypothesis is competition on the plate. A full-service market puts produce next to a wider array of foods people may prefer or need more urgently, and purchases at the bus may simply displace purchases elsewhere rather than adding to the total. That reading is consistent with the flat procurement numbers.

The context is not small. USDA’s Economic Research Service puts food insecurity at 13.7% of US households in 2024 — 18.3 million households and 47.9 million people — with 18.4% of households with children affected, and very low food security at 5.4%. Those figures were statistically unchanged from 2023 and significantly above the 2016-to-2021 range. Access is one lever among several, and the trial suggests it is not the lever that moves diet quality on its own.

What we did not find

No trial in this literature has paired a mobile market with nutrition education and measured diet quality as the endpoint, though one earlier produce-market trial saw larger intake changes among participants who received education. This study did not test that combination, and its authors call for it. Nor does a six-month dietary trial say anything about heart attacks, strokes or longevity; sodium intake is the surrogate, and the sodium result, while the strongest in the paper, is one of seven prespecified secondary outcomes.

The honest summary is narrower than either the enthusiasm around mobile markets or the dismissal of “food deserts” as a concept would suggest. Bringing a cheap, well-stocked grocery to a building can change what people buy, and in this trial it changed salt intake by a measurable margin while leaving the rest of the plate roughly where it was. Whether that is enough depends on what a programme is for.

Sodium is worth the attention: CDC estimates most Americans exceed the federal limit, and salt is 40% sodium by weight, so a single teaspoon of table salt carries about 2,400 mg. Cutting back at the shelf — including through the potassium-enriched salt substitutes that have their own trial evidence, covered in our earlier look at the salt swap — is one of the few dietary changes with a plausible route to blood pressure. If produce is the goal instead, our guide to the signs you are not eating enough vegetables covers what a serving actually looks like.

This is a report on a population study, not personal guidance, and it does not diagnose or treat anything; the site’s medical disclaimer applies. Anyone weighing changes to diet or blood pressure management should talk to a healthcare professional.