A 12-Week Randomized Intervention Comparing the Healthy US, Mediterranean, and Vegetarian Dietary Patterns of the US Dietary Guidelines for Changes in Body Weight, Hemoglobin A1c, Blood Pressure, and Dietary Quality among African American Adults

Abstract

Background

The US Dietary Guidelines (USDG) form the basis of nutrition guidelines, but the research informing the 3 USDG dietary patterns (Healthy US-Style [H-US], Mediterranean [Med], and vegetarian [Veg]) has been drawn largely from observational studies among White populations.

Objectives

The Dietary Guidelines 3 Diets study was a 3-arm, 12-wk randomly assigned intervention among African American (AA) adults at risk of type 2 diabetes mellitus that tested the 3 USDG dietary patterns.

Methods

The AAs (ages 18–65 y, BMI 25–49.9 kg/m², and BMI was measured in kg/m²) with ≥3 type 2 diabetes mellitus risk factors were recruited. Weight, HbA1c, blood pressure, and dietary quality (healthy eating index [HEI]) were collected at baseline and 12 wk. In addition, participants attended weekly online classes that were designed using material from the USDG/MyPlate. Repeated measures, mixed models with maximum likelihood estimation, and robust computation of standard errors were tested.

Results

Of the 227 participants screened, 63 were eligible (83% female; age 48.0 ± 10.6 y, BMI 35.9 ± 0.8 kg/m²) and randomly assigned to the Healthy US-Style Eating Pattern (H-US) (n = 21, 81% completion), healthy Mediterranean-style eating pattern (Med) (n = 22, 86% completion), or healthy vegetarian eating pattern (Veg) (n = 20, 70% completion) groups. Within-group, but not between groups, weight loss was significant (−2.4 ± 0.7 kg H-US, −2.6 ± 0.7 kg Med, −2.4 ± 0.8 kg Veg; P = 0.97 between group). There was also no significant difference between groups for changes in HbA1c (0.03 ± 0.05% H-US, −0.10 ± 0.05% Med, 0.07 ± 0.06% Veg; P = 0.10), systolic BP (−5.5 ± 2.7 mmHg H-US, −3.2 ± 2.5 mmHg Med, −2.4 ± 2.9 mmHg Veg; P = 0.70), diastolic blood pressure (−5.2 ± 1.8 mmHg H-US, −2.0 ± 1.7 mmHg Med, −3.4 ± 1.9 mmHg Veg; P = 0.41), or HEI (7.1 ± 3.2 H-US, 15.2 ± 3.1 Med, 4.6 ± 3.4 Veg; P = 0.06). Post hoc analyses showed that the Med group had significantly greater improvements in HEI compared to the Veg group (difference = −10.6 ± 4.6; 95% CI: −19.7, −1.4; P = 0.02).

Conclusions

The present study demonstrates that all 3 USDG dietary patterns lead to significant weight loss among AA adults. However, none of the outcomes were significantly different between groups.

This trial was registered at clinicaltrials.gov as NCT04981847.

Introduction

The US Dietary Guidelines (USDGs) form the basis of federal nutrition programs and policy and provide valuable guidance to health initiatives, industries, and community-based nutrition programs [1]. The USDG state that healthy eating goals can be met through a variety of dietary patterns but presents 3 main dietary patterns in line with recommendations: 1) a Healthy US-Style Eating Pattern (H-US), 2) a Healthy Mediterranean-style eating pattern (Med), and 3) a Healthy Vegetarian Eating Pattern (Veg) [2]. Currently, US adults fail to meet the nutrition recommendations set forth by the USDG as measured by the healthy eating index (HEI). On a scale of 0–100, the mean HEI score for Americans is 58 [3]. In addition, African Americans (AAs) have lower HEI scores (mean score 54) than Hispanic adults (mean score 60) [4]and, among adults with type 2 diabetes, AAs have lower HEI scores than White adults (mean score 51 compared with 55, respectively) [5].

In the US, AAs have the highest rates of obesity compared to Whites and Hispanics [6]. Overweight and obesity are associated with a number of chronic diseases, including T2DM [7, 8]. AA men and women also have higher rates of T2DM (12.1% prevalence) than non-Hispanic Whites (7.4% prevalence) [9]. Although there are numerous causes of obesity and chronic disease, poor dietary quality has been implicated in causing both obesity and T2DM [[10], [11], [12]].

Adherence to and adoption of the 3 diet patterns presented in the 2020 USDG varies among AA and White populations. Previous studies have shown that AA adults have lower adherence to the 3 USDG healthy diet patterns than White adults, such as a lower HEI score (reflecting adherence to the Healthy US-Style) [4]and lower adherence to a Mediterranean diet pattern [13]. On the other hand, more AAs report following a vegan or vegetarian diet (8%) than Whites (3%) or Hispanics (1%) [14]. In addition, a recent poll found that 19% of White adults reported consuming less meat over the previous 12 mo, whereas 31% of AA adults reported consuming less meat [15]. Although there might be an increased interest in following meat-reduced or meat-free diets among AAs, these percentages are still low. All 3 diets may be viewed as different from usual or culturally familiar dietary intake for most adults in the US, making adoption of these dietary patterns difficult.

Prior studies have indicated that plant-based eating styles are associated with significant health benefits for AAs. The Adventist Health Study-2 (AHS-2), a large prospective observational study that has a considerable population of both vegetarians/vegans and AAs, has examined diet and health outcomes among both Whites and AAs. Research from the AHS-2 examining different plant-based eating styles among AAs found that compared with AAs who followed omnivorous diets, AAs that followed vegetarian/vegan diets had a significantly lower risk of hypertension, diabetes, and high total and LDL cholesterol [16]. In addition, another AHS-2 study found that vegan diets were protective against overall cancer incidence (HR: 0.84; 95% CI: 0.72, 0.99) and female-specific cancers (HR: 0.66; 95% CI: 0.47, 0.92) as compared with 4 other plant-based eating styles [17]. Another AHS-2 cohort study examined diabetes risk, diet pattern, and race and found that AA vegans (OR: 0.381; 95% CI: 0.236, 0.617), vegetarians (OR: 0.618; 95% CI: 0.503, 0.760), and semi-vegetarians (OR: 0.486; 95% CI: 0.312, 0.755) had a lower risk of diabetes than omnivores [18]. There is a possible biological basis for reducing chronic disease risk among AAs. Research has indicated that individuals of African descent (or from other countries that have primarily subsisted on plant-based diets for thousands of years, i.e., South Asians) may be genetically predisposed to respond much better to the adoption of plant-based diets in the prevention of cardiovascular disease than those of European descent [19, 20].

Although the USDGs are the basis of nutrition guidelines, the research informing these dietary pattern recommendations has been drawn largely from observational studies among mostly White populations [2]. In addition, there has been very limited research examining these dietary patterns among AAs living in the southeastern US, who experience a disproportionate burden of chronic disease, especially T2DM [21, 22]. Therefore, little is known about how adopting these 3 dietary patterns impact the dietary quality and T2DM risk factors among AAs living in the south.

The objective of this study was to test the 3 dietary patterns, as presented by the USDG, among AA adults at risk of T2DM. In this 12-wk intervention, we hypothesized that participants assigned to the Mediterranean and Vegetarian patterns (both diets have an emphasis on plant-based foods) would have greater improvements in body weight, HbA1c, blood pressure (BP), and dietary quality (HEI score) than participants assigned to the healthy US-style dietary pattern. Although there has been little work examining health outcomes among AAs following a Mediterranean-style diet, we hypothesized that both the Mediterranean and Vegetarian patterns would produce similar outcomes in the present study. This is because a Mediterranean diet overlaps with some of the same foods consumed or avoided on other plant-based diets that have been examined among AAs (such as vegan or semivegetarian diets), and those plant-based diets have shown to be associated with a lower risk of diabetes than omnivores [18]. The findings of this 12-wk study will be used to refine and culturally tailor a year-long intervention for an AA study population. Postintervention, participants took part in focus groups for their assigned diet group and provided feedback on the diets and dietary recommendations in general, the topics discussed in class, and the recipes that were chosen for each group. We will utilize the Designing Culturally Relevant Intervention Development Framework, developed by Joseph et al. [23], to refine the resulting 1-y intervention. This framework provides guidance for cultural tailoring on the basis of 1) developmental considerations (e.g., age), 2) cultural considerations (e.g., visual appeal/packaging of the program, staff demographics, religious ties, and experiential knowledge), and 3) intervention delivery channels (e.g., preferred means of communication). The results of these focus groups will be published separately.

Methods

The Dietary Guidelines: 3 Diets (DG3Ds) study was a 12-wk randomly assigned intervention that tested the 3 examples of healthy dietary patterns that are presented in the USDG: 1) H-US, 2) Med, and 3) Veg [2]. Participants were recruited via local community events, workplace listservs, radio advertisements, and local television interviews with media outlets. AA adults [18–65 y old) with overweight or obesity (BMI: 25.1–49.9 kg/m²), who were willing to be randomly assigned to any of the 3 diet patterns, had access to the internet (to attend remotely delivered videoconference classes), and agreed to complete all assessments were recruited for the study. The intervention focused on improving dietary quality and reducing risk factors for T2DM. Therefore, in addition to having overweight and obese, participants were required to have ≥2 risk factors for T2DM (family history of T2DM, high BP, not physically active, etc.) as identified by the National Institute of Diabetes and Digestive and Kidney Diseases [24]. Eligibility criteria also included a stable medical status (e.g., no uncontrolled thyroid conditions or diabetes), not currently following a vegan, vegetarian, or Mediterranean diet, not breastfeeding or pregnant or planning to become pregnant in the next 3 mo, not in a weight loss program or had recently lost ≥10 lbs, and available Monday and Tuesday evenings for meetings.

Participants were recruited in the summer of 2021, and the intervention began in September 2021. Orientation sessions and assessment visits were conducted in person. Due to an increase in COVID-19 cases, the intervention classes were delivered online via Zoom video conference. Participants who were interested in the study completed an online screening questionnaire, and eligible participants were called to complete additional screening questions. Participants who qualified then attended a Zoom-delivered information session, where they learned study intervention details and signed a consent form. Participants were then invited to a small-group in-person baseline data collection session that was held in a computer laboratory. At this session, participants received training on how to complete their dietary recalls, completed baseline questionnaires and 1 dietary recall, and signed up for a time to have their baseline laboratory assessment completed. Participants were randomly assigned in blocks of 9 and stratified by sex to ensure the diet groups had an equal distribution of men and women. A study statistician created the blocks using a computerized random-number generator. Once all baseline data were collected, a study coordinator, who was blinded to the participant, assigned participants to the next number in the block. Participants did not learn of their randomly assigned assignment until they attended their first online class meeting, and all baseline data had been collected. The study was approved by a university institutional review board and overseen by a 3-member Data Safety and Monitoring Board. All participants provided informed consent and received financial incentives for completing all 3-mo assessments ($30).

Measures

The study’s primary outcomes were changes in body weight, HbA1c, and HEI score, with BP as a secondary outcome. A wall-mounted stadiometer model S100 (Ayrton Corp.) was used to assess height at baseline only. Weight was assessed in light street clothes using a calibrated digital scale model 500 KL (Health o Meter). Changes in blood glucose control were assessed via HbA1c. The DCA Vantage® Analyzer was used for this analysis and is considered suitable for laboratory or point-of-care measurement of HbA1c [25], because it is one of the few instruments that meet the accepted analytical performance criteria of having a total CV <3% and error <0.85% in the clinically relevant range [26, 27]. Whole blood was obtained by finger stick and placed directly into a glass capillary for analysis using the reagent cartridges. The assay measured both the concentration of HbA1c and total hemoglobin, with the ratio reported as percent HbA1c. All measurements and calculations are performed automatically by the DCA Vantage Analyzer. Quality control and calibration procedures were performed each day of testing and included running normal and abnormal (i.e., diabetic) control samples provided by the manufacturer. BP was also assessed at each visit after a 5-minute rest using an Omron Hem 705 CP Auto Inflate BP Monitor. At least 2 readings were taken with the mean of those readings used [28]. If there was a >5 mmHg difference between the first and second readings, then subsequent readings (up to 4) were taken. Participants completed 3 unannounced 24-hour dietary recalls at baseline and 3 at 12 wk using NCI’s Automated Self-Administered 24-hour Dietary Recall (ASA24) [29], including 1 weekend day (Friday–Sunday) and 2 weekdays (Monday–Thursday). The 3 recalls at each time point were averaged, and the National Cancer Institute’s SAS program was used to calculate the HEI score at each time point. All assessment staff was blinded to participant group assignments.

Dietary interventions

To test the 3 healthy patterns presented by the USDG, materials were gathered from the USDG [2], MyPlate.gov, and any other resources that were informed by the USDG. In order to include evidence-based behavior change strategies for adopting healthy dietary patterns, topics from the Diabetes Prevention Program (DPP) were also included in the classes [30]. All recipes that were demonstrated in the classes were retrieved from the recipes section of MyPlate.gov. The recipes were selected for each of the 3 dietary patterns through the cuisine filter function. For the H-US group, the “American” filter was used (116 available recipes); for the Med, both the “Latin American/Hispanic” (125 recipes) and the “Mediterranean” (19 recipes) filters were used because of the low number of recipes categorized as strictly Mediterranean. For the Veg group, the “vegetarian” filter was used (629 recipes). The MyPlate website allows for the creation of a cookbook of selected recipes. Recipes for each group were selected ahead of time and compiled into a cookbook specifically for each diet. Serving sizes for each of the food groups recommended by the USDG were derived from the 2020–2025 Dietary Guidelines for Americans [2]. Tables of serving sizes by different calorie levels are provided for each of the 3 dietary patterns (table A3-2 H-US, table A3-4 Veg, and table A3-5Med). To promote weight loss, energy intakes were personalized for each participant based on their baseline body weight, height, age, activity level, and sex using the MyPlate plan calculator [31]. For example, a 48-y-old sedentary female who is 167.6 cm and 100.2 kg (the average age, height, and weight in the study) would be given an 1800 kcal/d meal plan to achieve a healthier weight.

Participants had access to an online portal where they could access their handouts and recorded classes which could be viewed to make up missed sessions. Participants attended all intervention classes online via Zoom. The group assignment was revealed to participants during their first class, which occurred on either a Monday or Tuesday evening. Classes were led by an AA nutrition interventionist and an AA-registered dietitian. Among all 3 groups, each class generally followed a similar structure: 1) welcome, 2) sharing of successes and challenges from the previous week (mostly conducted in smaller groups in Zoom breakout rooms), 3) nutrition topic from the USDG, 4) behavior change topic from the DPP, 5) cooking demonstration, and 6) setting a Specific, Measurable, Achievable, Relevant, and Time-Bound (SMART) goal [32] for the week related to the topic covered in class. To encourage attendance, a prize was raffled off at the end of each class. In addition, participants were invited to pick up a dish sample prepared in class on the alternate day of their respective class: Monday or Tuesday. This allowed participants to sample the foods demonstrated during the Zoom-delivered classes while practicing social distancing during COVID-19. In addition to using the materials from the USDG to design class topics, participants were instructed to utilize the “Start Simple with MyPlate” application that complements the MyPlate website. Participants took the MyPlate quiz that helped create personalized goals for them within the application. Goals were related to the MyPlate food groups. Participants could then earn badges for completing those goals. Prizes were awarded to participants for earning badges through the MyPlate application.

Supplemental Table 1 provides the outline of the topics for the 12 weekly classes, the USDG materials used for the classes, the DPP topics utilized, the MyPlate recipes that were demonstrated and those that were provided, and the handouts or recipes that were given to participants in their online website portal. All handouts provided to participants were from the USDG website or the MyPlate website (unless otherwise indicated in Supplemental Table 1).

Statistical analysis

The statistician for the study was blinded to group assignments when completing all analyses. Descriptive statistics were used to present baseline characteristics, and a 1-factor ANOVA was conducted to compare differences in class attendance between groups, with Tukey honestly significant difference (HSD) used for post hoc comparisons. Analyses followed the intention-to-treat principle. Primary study aims were addressed using repeated measures, mixed models with maximum likelihood estimation, and robust computation of standard errors as provided by PROC MIXED in the SAS software version 9.4 (SAS Institute). The models included time, group, and a group × time interaction. The full information from the available data was used in each model to provide unbiased estimates of the intervention effect in the presence of attrition between both time points. In addition, contrasts were constructed comparing body weight (or other outcomes) at 3 mo between groups.

Although the primary goal of this study was to refine the intervention for delivery in a longer study, this study was also powered based on the change in HEI score. Based on both the HEI scores observed in a previous study conducted by our team of 5 dietary patterns. The New Dietary Interventions to Enhance the Treatments for weight loss (New DIETs) study [33] and a study that examined HEI scores among 5 different dietary patterns (vegan, veg, pesco-veg, semi-veg, and omnivorous) [34], we estimate a pooled SD of 9 points for HEI change. Therefore, the study aimed to recruit 21 per diet group, which after attrition corresponds to 19 participants per group. A clinically significant difference of 10 (Cohen’s d = 1.11) points between diet groups would achieve 95% power, and a smaller difference of 7.5 points (Cohen’s d = 0.83) yields 80% power.

Results

Figure 1 provides the CONSORT diagram for the study. Of the 63 participants who began the study, the outcome of body weight was obtained from 50 (79% completion) at 3 mo. Table 1 provides the baseline demographics of participants in the DG3D study, as well as relevant baseline measures, such as body weight and diet outcomes.

Figure 1.

CONSORT diagram of the 12-wk dietary intervention among African American adults comparing 3 diets (healthy US, Mediterranean, or vegetarian).

Table 1.

Baseline characteristics of African American adult study participants in the Dietary Guidelines: 3 Diets 12-week intervention comparing a Healthy US, Mediterranean, and Vegetarian diet

Baseline variables	Healthy US group (n = 21)	Mediterranean group (n = 22)	Vegetarian group (n = 20)
n (%)
Sex
Female	18 (86%)	18 (82%)	16 (80%)
Male	3 (14%)	4 (18%)	4 (20%)
Race
Black	21 (100%)	22 (100%)	20 (100%)
Black and other race	0	0	0
Education
Some college	3 (14%)	4 (18%)	4 (20%)
College graduate	10 (48%)	7 (32%)	9 (45%)
Advanced degree	8 (38%)	11 (50%)	7 (35%)
Occupation
Employed for wages	14 (68%)	14 (64%)	9 (69%)
No current employment	0 (0%)	0 (0%)	2 (10%)
Retired	6 (27%)	4 (18%)	2 (10%)
Other	1 (5%)	4 (18%)	7 (11%)
Marital status
Single	7 (33%)	4 (18%)	6 (30%)
Married	8 (38%)	12 (55%)	11 (55%)
Partnered/living with someone	0 (0%)	0 (0%)	1 (5%)
Divorced or separated	6 (29%)	6 (27%)	2 (10%)
Means ± SD
Mean age (y)	46.8 ± 11.2	49.4 ± 11.3	47.8 ± 9.3
BMI (kg/m2)	35.7 ± 4.9	36.5 ± 8.1	35.9 ± 6.1
Body weight (kg)	98.2 ± 17.3	101.6 ± 24.1	101.4 ± 16.9
HbA1c (%)	5.9 ± 0.3	5.8 ± 0.4	5.5 ± 0.4
Systolic BP (mmHg)	124.1 ± 11.4	130.9 ± 16.7	128.9 ± 12.2
Diastolic BP (mmHg)	81.6 ± 8.7	80.9 ± 6.1	80.4 ± 9.0
HEI score (out of 100)	55.0 ± 13.0	54.6 ± 12.0	57.8 ± 12.7
Energy intake, kcal/d	1869 ± 323	1836 ± 626	1493 ± 387
Carbohydrate intake, g/d	203 ± 70.6	186 ± 67.3	150 ± 41.1
Fat intake, g/d	83.8 ± 15.0	86.1 ± 32.5	68.3 ± 25.6
Protein intake, g/d	78.1 ± 22.7	77.7 ± 33.3	67.4 ± 22.5

BMI, body mass index; BP, blood pressure; HbA1c, hemoglobin A1c; HEI, healthy eating index; SD, standard deviation; US, United States.

A total of 63 participants enrolled in the study (83% female; mean ± SD: age 48.0 ± 10.6 y, BMI 35.9 ± 0.8 kg/m²) and were randomly assigned to the H-US (n = 21, 81% completion), Med (n = 22, 86% completion), or Veg (n = 20, 70% completion) groups. There were no significant differences in attrition between groups (χ² = 1.76, P = 0.42). Participants attended a mean of 9.5 ± 3.4 of the classes (79%). However, there was a significant difference in the mean total number of classes attended by the group [F (2, 60) = 3.19; P = 0.048]. Participants in the Med group attended significantly more classes (10.6 ± 2.4 classes out of 12) than the Veg group (8.1 ± 4.1; P = 0.04). H-US participants attended a mean of 9.6 ± 3.2 classes.

Table 2 provides the results of the study’s main outcomes (mean ± SE). After 12 wk, weight loss did not differ between groups; however, within-group weight loss was significant within each diet group. Participants across the H-US, Med, and Veg groups lost a mean of -2.5 ± 2.9 kg (equal to −2.6 ± 3.1%) of their body weight. There was also no significant difference between groups for changes in HbA1c, systolic BP, diastolic BP, or HEI score. Only the H-US diet resulted in a significant decrease in systolic and diastolic BP, and only the Med diet resulted in a significant improvement in HEI score. Post hoc analyses showed that the Med group had significantly greater improvements in HEI than the Veg group. Examining energy intake (kcals) and macronutrients, only change in carbohydrate grams differed among the groups, with H-US participants decreasing their carbohydrate intake more than the Veg group. There were significant within-group decreases in kcals and carbohydrates for both the H-US and Med groups, and all groups had significant within-group decreases in fat intake. No within-group changes were observed for protein.

Table 2.

Changes in body weight, hemoglobin A1c, blood pressure, and healthy eating index score among African American adults randomly assigned to the Healthy US, Mediterranean, or Vegetarian diets for 12 weeks¹

Outcomes	Healthy US group (n = 21)	Mediterranean group (n =22)	Vegetarian group (n = 20)	P-interaction (diet × time)2
Delta body weight, kg	−2.4 ± 0.7 (−3.8, −1.0)	−2.6 ± 0.7 (−3.9, −1.3)	−2.4 ± 0.8 (−3.9, −0.8)	0.97
Delta HbA1c, %	0.03 ± 0.05 (−0.08, 0.14)	−0.10 ± 0.05 (−0.21, 0.01)	0.07 ± 0.06 (−0.06, 0.20)	0.10
Delta systolic BP, mmHg	−5.5 ± 2.73 (−10.8, −0.2)	−3.2 ± 2.5 (−8.3, 1.9)	−2.4 ± 2.9 (−8.2, 3.4)	0.70
Delta diastolic BP, mmHg	−5.2 ± 1.83 (−8.8, −1.8)	−2.0 ± 1.7 (−5.4, 1.3)	−3.5 ± 1.9 (−7.3, 0.3)	0.41
Delta HEI score	7.1 ± 3.2 (0.7, 13.5)	15.2 ± 3.13,4 (9.0, 21.4)	4.6 ± 3.4 (−2.1, 11.4)	0.06
Delta energy intake, kcal/d	−487.3 ± 111.83 (−711.1, −263.5)	−449.9 ± 108.93 (−668.0, −231.8)	−219.6 ± 117.6 (−454.9, 15.7)	0.22
Delta carbohydrates, g/d	−56.5 ± 14.03,5 (−84.5, −28.4)	−30.6 ± 13.63 (−58.0, -3.2)	0.2 ± 14.7 (−29.3, 29.8)	0.03
Delta total fat, g/d	−23.9 ± 6.53 (−40.0, −10.8)	−29.1 ± 6.43 (−41.9, −16.3)	−14.1 ± 6.93 (−27.9, −0.4)	0.28
Delta protein, g/d	−9.1 ± 7.4 (−23.8, 5.6)	−13.5 ± 7.2 (−27.8, 0.9)	−15.3 ± 7.7 (−30.7, 0.1)	0.83

BP, blood pressure; HbA1c, hemoglobin A1c; HEI, healthy eating index; US, United States.

¹Values are mean ± SE (95% CI).

²P values are group x time (df = 2)

³Significantly different within-group changes (P < 0.05)

⁴Significantly different from the vegetarian diet: Difference = −10.6 ± 4.6; 95% CI: −19.7, −1.4; P = 0.02.

⁵Significantly different from the vegetarian diet: Difference = 56.7 ± 20.3; 95% CI: 15.9, 97.4; P = 0.01.

Discussion

The goal of the DG3D study was to test a 12-wk intervention of the 3 healthy dietary patterns presented by the USDG. All materials and recipes were obtained from MyPlate. The 12-wk study will help refine a year-long intervention that is tailored to an AA audience. We hypothesized that the Med and Veg groups would lose more weight and have greater improvements in HbA1c and HEI scores than the H-US group but did not find any differences among the 3 diet groups in this study. The only difference found in post hoc analyses was that the Med group had a greater HEI score improvement than the Veg group. Despite being randomly assigned, the Veg group had the highest baseline HEI score and the lowest baseline energy intake among the 3 groups (although not significantly), so it is possible they had less room for improvement. Focus groups conducted after the study will provide more insight as to why the Veg group may not have had significant increases in HEI similar to the Med group. All groups saw significant within-group weight loss and the H-US group saw significant within-group decreases for both systolic and diastolic BP.

Weight loss of >5% is generally required to promote metabolic improvements [35], and across groups, participants lost half that amount. Therefore, participants were on track to achieve clinically significant weight loss if the study had continued for a longer duration. These findings demonstrate that all the healthy diet patterns presented in the USDG have the potential to promote clinically meaningful weight loss. Given the short-term length of the study and remote delivery, the weight loss observed was slightly higher than in previous research comparing these diet patterns using in-person delivery methods. The Cardiovascular Prevention With Vegetarian Diet (CARDIVEG) study used a 3-mo cross-over design to compare Med and Veg diets, finding similar weight loss between the groups (−1.88 kg Veg and −1.77 kg Med) [36]. A meta-analysis of vegetarian diets for weight loss found a mean of −2.21 kg body weight loss among participants assigned to an energy-reduced vegetarian diet [37]. In another meta-analysis, the Mediterranean diet led to a −1.75 kg greater weight loss than control diets [38]. In a systematic review examining Mediterranean diet interventions, researchers found that although the Mediterranean diet generally produced greater weight loss than a low-fat diet, the impact on BP was similar across comparison diets [39].

The present study found a mean change in HbA1c of 0.03% among the H-US group, −0.10% among the Med group, and 0.07% among the Veg group. These changes were not significant within or between groups, indicating either a longer intervention would be necessary to promote significant changes or the use of different diet patterns. In a meta-analysis of diet pattern interventions that lasted longer than 6 mo and targeted reductions in HbA1c, randomization to a Mediterranean diet resulted in a mean decrease of −0.47% in HbA1c [40]. This decrease was greater than the other examined diets (−0.28% high protein, −0.12% low carbohydrate, and −0.14% low glycemic index) [40]. In another meta-analysis of the impact of the Mediterranean diet on glycemic control, researchers found that the Mediterranean diet produced a range of −0.3% to −0.47% decrease in HbA1c as compared with a control diet [41], which was greater than what was observed in the present study. A meta-analysis of vegetarian diet interventions lasting longer than 4 wk found a mean reduction in HbA1c of −0.39% among those randomly assigned to a vegetarian diet [42]. The present study found no significant change in HbA1c in the Veg group, as well as the H-US or Med group.

At baseline, participants had a mean systolic BP of 128.0 ± 13.7 mmHg and a mean diastolic BP of 81.0 ± 7.9 mmHg, which is considered elevated. Although the change in BP did not differ significantly between groups, only the H-US diet saw significant within-group decreases in BP. According to the 2020–2025 Dietary Guidelines for Americans, the Dietary Approaches to Stop Hypertension (DASH) diet “is an example of a healthy dietary pattern and has many of the same characteristics as the healthy US-style dietary pattern” [2]. A systematic review found that the DASH diet led to reductions in systolic (−3.2 mmHg) and diastolic (−1.5 mmHg) BP [43], which is slightly less than what was observed in the present study. In another study among White and AA adults examining the DASH diet for BP reduction, researchers found that adherence to the DASH diet predicted reductions in BP but that AAs in the study were less likely to be adherent to the diet than Whites [44]. Another study found that DASH scores were low among AAs and that DASH score was not associated with BP among those without chronic kidney disease [45]. Among AAs with chronic kidney disease, each point increase in the DASH score was associated with a 1.6 mmHg decrease in systolic BP and a 0.9 mmHg decrease in diastolic BP [45]. In a Cochrane systematic review of the use of a Mediterranean diet for cardiovascular disease risk reduction, a mean decrease of −2.99 mmHg was found for systolic BP and −2.0 mmHg for diastolic BP [46]. This is similar to the −3.2 mmHg systolic BP, and −2.0 mmHg diastolic BP found in the present study for participants randomly assigned to the Med diet. A meta-analysis on the use of a Mediterranean diet for BP reduction found that there were only differences in change in BP when compared to a usual diet (compared with an active comparator diet), with mean decreases of −1.5 mmHg for systolic and −0.9 mmHg for diastolic BP [47]. Very few studies have examined a Mediterranean diet among AAs [48], although future research is being conducted in this area [49]. The decrease in BP among Veg participants in the present study was not as great as what has been observed in previous studies. In a meta-analysis of 7 randomly assigned trials (duration range of 6–52 wk), randomization to a vegetarian diet as compared with an omnivorous diet led to a mean reduction in systolic BP of −4.8 mmHg (more than observed in the present study) and a mean decrease in diastolic BP of −2.2 mmHg (less than observed in the present study) [50]. In another meta-analysis (duration range of 4–74 wk), participants randomly assigned to follow a vegetarian diet had a mean difference of −2.66 mmHg systolic and −1.69 diastolic BP than those randomly assigned to an omnivorous diet [51].

Among US adults, the mean HEI score is 59 [3] and is lower for AAs (mean score of 54) [4]. The present study found a similar HEI score among participants at baseline (mean of 55.8 across groups). Although all groups saw increases in their HEI score, only the Med group saw significant within-group changes, with an improvement in the HEI score that was greater than the Veg group. A meta-analysis of dietary interventions examining changes in HEI found an average of a 5-to-6-point increase [52]. That is in line with the Veg group in the present study but lower than what was observed among the H-US and Med groups. The same study found that improvements in HEI scores were greater among studies that focused on multiple food behaviors, which is similar to the focus of the DG3D study [52]. In a study examining the HEI score and the Mediterranean Diet Score (MDS) of participants following a vegan, vegetarian, pesco-veg, or omnivorous diet, the vegan diet group had the highest HEI and MDS as compared with the omnivorous group, with the vegetarian group’s score falling in between [34]. The pesco-veg group ranked second highest for the MDS [34]. Except for carbohydrates, there were no differences in the dietary variables examined. Future work from this study conducting a more in-depth dietary and adherence analysis will be published elsewhere.

Most of the previous studies that have examined changes in body weight, HbA1c, BP, or HEI score among participants following different dietary patterns have been conducted among mostly White participants. In addition, few previous studies have examined the impact of adopting various healthy dietary patterns on weight loss or other health outcomes among AA adults. The Heart Healthy Lenoir Project is 1 previous study that examined the use of a Mediterranean diet tailored to the south (Med-South diet) among a predominantly (65%) AA group [53]. During the weight loss phase of the study, participants in the in-person group-based intervention lost a mean of −3.1 kg over 16 wk [53]. To our knowledge, the present study is 1 of the first to compare an H-US, Med, and Veg diet solely among AAs, who lost a mean total of 2.47 kgs over 12 wk.

The present study has many strengths. The study was conducted solely among AA participants, and the intervention was delivered by AA program staff. Objective measures were used to collect outcomes related to diabetes, BP, and body weight, and 3 d of dietary recall were obtained at both baseline and 3 mo. The intervention used all materials from the USDG, such as MyPlate resources, to deliver the nutrition and recipe content of the intervention. This allowed for rigorous testing of how the USDGs are presented to the public. The study also had limitations. The population was mostly female and educated; therefore, findings may not be generalizable to AAs more broadly. In addition, there were no true control groups because all participants received a dietary intervention. The study was short-term and may not have been enough time to allow participants to adopt each diet fully. In addition, although HbA1c typically measures blood glucose control over the previous 3 mo [54], the study may not have been long enough to assess the full changes that would have occurred in HbA1c if the study had lasted longer. Although not significantly different, there was higher attrition from the Veg group than the other groups. Attrition was higher in all groups than anticipated and may have impacted the study’s power to detect differences. In addition, the study was originally designed to be delivered in person. Due to rising COVID-19 cases, the study was moved to an all-online delivery. This may have impacted engagement in the study, especially because of a lack of ability for participants to learn hands-on cooking skills. However, online classes may have addressed barriers to attendance, such as transportation or childcare needs. Although overall attendance was high (close to 80% of the classes attended across groups), attendance among Veg participants was lower than in the Med group. Previous research has demonstrated that class attendance during a behavioral weight loss program is associated with weight loss outcomes [55]. Future studies should examine ways to keep participants in plant-based diet interventions more engaged. Participants were randomly assigned to each diet without consideration for dietary preferences, values, or personalized treatment goals. Prior weight loss interventions using various dietary patterns have not found an impact on baseline insulin secretion [56], genotype pattern [56], or preference for the diet assigned [57, 58] as predictors of weight loss. Future studies examining the dietary patterns highlighted by the USDG should consider the role preference or personalized goals and values may have on weight loss and other health-related outcomes. Lastly, even though the groups were randomly assigned, the Veg group had a nonsignificantly higher baseline HEI score and lower baseline energy intake than the other 2 groups. So it is possible the Veg group was closer to meeting recommendations at baseline.

In conclusion, the present study demonstrated that all 3 dietary patterns highlighted in the USDG led to significant weight loss among AA adults. In addition, all changes in outcomes were in the expected direction (i.e., improvements in weight, BP, and HEI), with the exception of HbA1c. Only the Med group saw decreases in HbA1c, although changes in HbA1c across groups were minimal. Future studies should examine a longer-term intervention that includes in-person cooking classes to help with engagement and cooking skills for dietary patterns that may be unfamiliar to participants.

Author contribution

The authors’ responsibilities were as follows: GMTM, ADL, SW, DBF, and MJW: designed the research. GMTM, JC, MD, HA, NO, and MJW: conducted the research. BH and SB: analyzed the data. GMTM: wrote the paper and had primary responsibility for the final content. All authors have read and approved the final manuscript.

Funding

This study was supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under award number R01DK128057. However, the content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author disclosures

GTM, MJW, JC, NO, HA, SB, MD, BH, SW, DBF, MAS, and ADL

Conflicts Of Interest

No conflicts of interest.

Data availability

The datasets generated during and/or analysed during the current study are not publicly available yet due to authors using the data for additional manuscripts. The data will be made available once all papers from the study have been published.

Appendix A. Supplementary data

The following is the Supplementary data to this article: