Two weeks of delivered meals, and a depression score that fell
- A pilot randomized study at the University of Michigan enrolled 31 adults with moderate to moderately severe depressive symptoms (PHQ-8 of 8 or more at the first visit). After one week of baseline observation, all were asked to eat a minimally processed diet for two weeks: 20 received three meals a day from a commercial delivery service plus nutritional guidance, and 11 received the guidance alone.
- Diet quality on the Diet ID scale rose in both groups, from a mean of 4.88 at the first visit to 8.05 at the last (d = 1.93). All 31 completed every assessment.
- PHQ-8 scores fell in the meal delivery group (d = 1.62) but not in the guidance group (d = 0.54). The authors say the group difference should be read cautiously: the study was a pilot with limited statistical power.
- Across all participants, greater improvement in diet quality went with a larger fall in PHQ-8 (semi-partial R2 = 0.17). PHQ-8 had also dropped between the first and second visits, before the intervention started. Published in Scientific Reports, online 25 April 2026.
A pilot randomized study at the University of Michigan gave 31 adults with moderate to moderately severe depressive symptoms two weeks of a minimally processed diet, after one week of baseline observation. Twenty received three delivered meals a day plus nutritional guidance; eleven received the guidance and prepared their own food.
Diet improved in both groups, the depression score in one
Diet quality on the Diet ID scale rose from a mean of 4.88 at the first visit to 8.05 at the last (d = 1.93), with no difference between the groups. PHQ-8 in the delivery group went from a mean of 13.45 at the second visit to 9.40 at the third (estimate 4.05; 95% CI 1.98 to 6.13; d = 1.62). In the guidance group it went from 11.55 to 10.18 (estimate 1.36; 95% CI -1.44 to 4.16; d = 0.54). The group-by-time interaction was reported at p = 0.046. The authors note that scores had already fallen between the first and second visits, and that regression to the mean and other study-related factors are hard to separate from the intervention.
Food, or being looked after
In the delivery group, nutrition arrived together with other things: meals at the door three times a day and no shopping or cooking to organize. There was no placebo, each participant knew whether meals were coming, and the window was two weeks. A d of 1.62 over that period does not show which of these carried the change.
This part is the author of this note speaking, not the study. When a client says a dietary change lifted their mood, ask what else changed that week: who became involved, how the day is structured, what they expected to happen. Treat the diet as one of several changes and ask about the rest before crediting any of them.
When a client says a change in eating has lifted their mood, ask what else changed that week: who became involved, how the day is structured and what they expected.
Thirty-one participants (20 and 11), two weeks of intervention, 80.6% women, no placebo arm. Diet quality was self-reported through an image-based tool, and eating the delivered meals was not verified. PHQ-8 fell before the intervention began. The study was not powered to compare the groups.