Vegan vs. Omnivorous Soul Food: What a 2-Year Trial Found

A culturally adapted vegan soul food diet did not beat a modified omnivorous soul food diet for weight loss or LDL cholesterol over two years in African American adults. That was the main result of the NEW Soul randomized trial, and it deserves to lead the story.

There was, however, a smaller and more complicated signal inside that null result: participants who adhered to the vegan diet lost about 5% of their body weight by 24 months. That finding came from a post-hoc analysis of roughly 22 adherent vegan participants, not from the trial's pre-specified primary comparison. It raises a useful question about what makes dietary change last. It does not prove that vegan soul food produces superior long-term weight loss.

A rare long-term diet trial in African American adults

The NEW Soul randomized clinical trial enrolled 159 African American adults with BMIs of 25.0 to 49.9: 77 were assigned to a culturally adapted vegan soul food diet and 82 to a modified omnivorous soul food diet. The intervention ran for two years at the University of South Carolina.

The researchers built both diets with community input, soul food chef demonstrations, and African American community facilitators. That design choice matters. Nutrition trials often ask people to adopt eating patterns that ignore their food traditions, budgets, kitchens, and social lives; then researchers act surprised when the plan becomes difficult to sustain.

Still, cultural fit did not create a between-group winner on the trial's co-primary outcomes. In the intention-to-treat analysis—the analysis that preserves the original random assignments—the groups had no statistically significant differences in body weight or LDL cholesterol at 12 or 24 months. Total cholesterol and insulin also did not differ significantly between groups.

The result answers a narrow but important question: offering the vegan protocol did not produce better average weight or lipid outcomes than offering the omnivorous protocol in this trial. It does not mean every participant had the same experience, nor does it make diet quality irrelevant.

The adherence finding is real, but it is not a verdict

A later post-hoc adherence analysis divided participants by dietary adherence. The researchers used a 0-to-5 food-group score and classified scores of at least 2.5 as adherent.

Fewer than 40% of participants in either group met that threshold at any measured time point. At six months, 27 of 77 vegan participants (35.1%) and 21 of 82 omnivorous participants (25.6%) met it; the between-group difference was not statistically significant (p=0.194). At 12 months, adherence was 36.4% versus 26.8% (p=0.196). At 24 months, it was 28.6% versus 24.4% (p=0.550).

Among the 24-month subgroups, adherent vegan participants lost approximately 5% of body weight. Their adjusted loss was significantly greater than the losses in less-adherent vegans, adherent omnivores, and less-adherent omnivores. No comparable adherence-stratified weight-loss pattern emerged in the omnivorous arm.

The primary trial result was null. The 5% figure is a hypothesis-generating subgroup result, not confirmation that vegan soul food wins when followed.

Why the restraint? Adherence was not randomly assigned. At the 12-month classification, adherent vegan participants were older than less-adherent vegans (53.1±7.4 versus 46.8±11.4 years; p=0.01) and began with higher total cholesterol (194.2±36.2 versus 175.0±30.0 mg/dL; p=0.02). The analysis adjusted for age, baseline weight, education, employment, food security, sex, medications, and other relevant factors. Adjustment helps, but it cannot fully account for unmeasured differences in motivation, health concerns, support, or life circumstances within a subgroup this small.

People with greater perceived cardiovascular risk may have had more reason to maintain the program. That possibility does not erase the result; it limits what the result can establish about cause.

Diet quality improved when the vegan plan was followed

The weight finding was uncertain, but a separate diet-quality analysis found that adherence to the vegan soul food diet was associated with higher Healthy Eating Index scores than the omnivorous diet. The Healthy Eating Index measures how closely an overall eating pattern aligns with U.S. dietary guidance. It is a diet-quality metric, not proof of fewer heart attacks or longer life, but it is a meaningful result.

This distinction helps. A trial can show better food-pattern quality without showing a statistically significant difference in weight or LDL cholesterol between assigned groups. Those outcomes answer different questions, and neither should be smuggled into the other.

Psychology does not explain the pattern

An analysis of dietary acceptability, restraint, disinhibition, and hunger found no significant differences between the vegan and omnivorous groups on those measures. In other words, the NEW Soul eating-behavior study did not find evidence that the vegan diet was more psychologically punishing, less acceptable, or more hunger-inducing.

That removes an easy explanation for the adherence result. If one diet was not measurably less acceptable or more hunger-provoking, the evidence does not tell us why adherence tracked with weight loss only in the vegan subgroup. A plausible story is not the same thing as a demonstrated mechanism.

COVID-19 and missing data complicate two-year results

The pandemic disrupted the second cohort around its 12-month assessment. Participant interviews in a qualitative COVID-19 study described stress, food-access problems, disrupted routines, and reduced capacity to cook. These conditions could affect adherence, weight, and follow-up in ways statistical adjustment cannot neatly repair.

Dietary-recall data were also increasingly incomplete: missingness rose from about 4% at three months to 37% at 24 months. The adherence analysis treated missing recalls as non-adherence. That conservative choice may underestimate adherence, but it can also bias results if missingness differed according to diet assignment or success on the diet.

Self-reported recalls add another limitation. Participants knew which diet they had been assigned, and social desirability can influence dietary reporting. The study also came from one site, so its results may not transfer cleanly to other regions, food environments, or socioeconomic circumstances.

What cultural adaptation can claim—and what it cannot

The trial does support testing dietary interventions within the culture of the people expected to use them. A related restaurant-based feasibility study found that the NEW Soul approach was feasible in a community setting and could be scaled. That is encouraging implementation evidence, not proof that restaurant delivery improves weight or cardiovascular outcomes.

There is no established consensus that plant-based diets outperform lower-fat omnivorous diets specifically among African American adults. A race-stratified plant-based diet trial reported differential responses by race, but it did not study an exclusively Black cohort and used a different protocol. Other nutrition interventions, including an African American cardiovascular-risk program based on ACC/AHA prevention guidance, help show the need for targeted research but do not answer the vegan-versus-omnivorous question.

The practical reading

If you enjoy and can sustain a vegan soul food pattern, this research offers a reasonable case that it can improve overall diet quality. The approximately 5% weight loss among adherent vegans is worth investigating in a larger trial that pre-specifies the adherence question and better protects against subgroup bias.

For now, the most defensible conclusion is less glamorous and more useful: the two assigned diets produced no significant average difference in weight or LDL cholesterol over 24 months, while sustained adherence to the vegan plan coincided with better diet quality and a small subgroup's meaningful weight loss. The next study needs to determine whether that loss came from the diet, the people able to maintain it, or both.