A Systematic Review of Literature on Representation of Racial and Ethnic Minority Groups in Clinical Nutrition Interventions

There is a disproportionate increase in the incidence of diet-related cardiometabolic disorders in racial and ethnic minority groups. This systematic review examines the extent to which diet-induced changes in health outcomes have been discussed by race or ethnicity in randomized controlled trials recruiting both minority and non-Hispanic White groups. Databases i.e. PubMed, Cochrane library and Web of Science were searched up to November 2019. Studies Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 29 January 2021 doi:10.20944/preprints202101.0629.v1


INTRODUCTION
Approximately 40% of the US population comprises of non-White racial/ethnic groups 1 .
The prevalence of major chronic diseases such as coronary heart disease, hypertension, obesity, and diabetes in racial and ethnic minorities is much higher than in Non-Hispanic White groups 2 .
For example, Black people are 1.5 times more likely to have hypertension, American Indians are 1.5 times more likely to have coronary heart disease, and most racial and ethnic minority groups (except for Asians Americans) are 1.2-1.8 times more likely to have obesity, and 1.6-2.9 times more likely to have diabetes than non-Hispanic White individuals. In addition, despite lower body weights, Asian Americans have a higher prevalence of diabetes.
Racial and ethnic groups such as Black, Hispanic, Asian, and American Indian people or Pacific Islanders experience diet-related disparities. High saturated fat and salt intake, and low fruit, vegetable, and whole grain intake, resulting in suboptimal nutrient profiles, are especially evident in these racial and ethnic minority groups compared to Whites 3 . Diet-related disparities can arise at the biological, behavioral, environment (physical/built, sociocultural) and systems levels and contribute to chronic disease risk. Although, genetic predisposing factors may contribute to this increased prevalence of cardiometabolic diseases among these groups 4 , these differences could result more often from interactions of genetic variants with environmental 5 and dietary factors, for example, carbohydrate and fiber 6 and dietary fat and monounsaturated fatty acids 7 . Poor diet quality can exacerbate the expression of the genes involved in metabolic dysfunctions such as insulin resistance 8 . Nutritional status is in part determined by food choices, which in turn are influenced by the environment. It is documented that ethnic and racial minority groups are systematically exposed to physical, socioeconomic, and political environments that hinder their ability to sustain healthful choices, including consistent consumption of nutritious food 9-15 .
Findings on White groups cannot be assumed to be applicable to other groups and failing to recognize specific contextual factors that distinctively affect the nutritional status of ethnic and racial minorities can perpetuate disparities in their representation in research. Given the disproportionate increase in the incidence of diet-related cardiometabolic disorders in racial and ethnic minority groups, clinical nutrition research on these populations is critical. In fact, one of the stated research priorities of the American Society of Nutrition is to determine the variability in responses to diet and food components by population subgroups, including ethnic and racial minority groups 16 .
The purpose of this review is to report on clinical nutrition research on ethnic and/or racial minority groups in countries with predominantly non-Hispanic White populations. More specifically, the systematic review examines the extent to which diet-induced changes in health outcomes have been studied in randomized controlled trials with ethnic or racial groups vs. non-Hispanic Whites. In the text of this review, we use the same term to describe a specific ethnic or racial group as in the study being referenced.

Search strategy
A comprehensive search strategy was developed in accordance with the Cochrane Handbook of systematic reviews 17 . This strategy employed a mixture of controlled vocabulary and natural language to reflect the focus of the analysisto identify dietary studies that

Study selection
The search was conducted across 3 databases (PubMed, Web of Science, and Cochrane Library) and the results were compiled in Zotero (version 5.0.87). Reference lists of related systematic reviews and meta-analyses were hand-searched to identify additional relevant articles.
Three authors (JD, AJ, SO) reviewed the articles in a systematic manner for inclusion in the review. Any discrepancies were resolved via a vote for inclusion.
In the first pass, the titles and the abstracts of articles were independently screened to identify potentially relevant articles based on the criteria. Articles were excluded if the studies 1) were duplicates, 2) were not clinical dietary interventions or the dietary intervention was not defined; 3) included children or pregnant women; 4) did not assess health outcomes; 5) did not have full texts; or 6) were published as conference abstracts. In the second pass, full-text of articles were screened and articles were excluded if the studies 1) recruited exclusively non-Hispanic White groups, 2) did not mention race or ethnicity, or 3) were conducted in countries where the predominant population is not non-Hispanic White. In the third and final pass, only studies that discussed health outcomes by racial or ethnic groups were included in the systematic review. Clinical dietary interventions included in the final pass are defined as studies that manipulate the dietary composition of participants' diets via a specific dietary prescription, foods, but not supplements or drugs. The study selection process is documented in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram (Figure 1).

Data extraction
Data collection tables were developed by 1 author (JD) and variables were finalized in discussion with co-authors. Data extraction was completed by 2 authors (AJ, SO) and was reviewed for accuracy by JD. The quality of each study article was assessed using the Oxford Centre for Evidence-Based Medicine Levels of Evidence 18 . The risk of bias was evaluated with use of the Academy of Nutrition and Dietetics Quality Criteria Checklist (QCC) for Primary Research 19 .

RESULTS
In total, 9943 studies were screened for inclusion, 9871 of which were excluded based on the established criteria ( Figure 1). Twenty-nine randomized controlled interventions that discussed health outcomes by racial/ethnic group were included in the systematic review. Some studies had multiple publications discussing different outcomes and are grouped together in Table 1. Sixteen pertinent studies that accounted for race and/or ethnicity in their analyses were acute trials (n=6) i.e. not assessing changes in health outcomes over a period of time and non-RCTs (n=10), and hence did not fully align with established criteria for inclusion in the systematic review but are discussed separately. Twenty-seven studies that focused on a single racial/ethnic group are also discussed separately. The detailed inclusion and exclusion criteria are discussed in Figure 1.

Study design and duration
Twenty-seven studies were conducted in the US, 1 in the United Kingdom (UK) 20 and 1 study was a multi-center trial conducted in Australia, Canada and the UK 21 .
Included studies varied in design, duration, and dietary intervention tested (Table 1) Eleven studies 34,54-67 followed a crossover design in which participants were exposed to 2-5 dietary interventions, with interventions lasting between 1 week and 6 months.

Effects of dietary intervention on health outcomes by race or ethnicity
Most interventions were designed to modify the macronutrient composition of the diets and health outcomes largely included anthropometric markers (6 studies), blood pressure (BP, 8 studies), markers of glucose metabolism (6 studies), and lipids (10 studies) (Table 1). Study stratification by health outcomes is depicted in Figure 2. markers of glucose metabolism were outcomes in 6 studies, including blood glucose, glucose tolerance, insulin, insulin sensitivity, insulin resistance, glycated albumin, fructosamine, beta-cell responsiveness; anthropometric markers were outcomes in 6 studies, including, body weight, waist-to-hip ratio; inflammation markers were the outcomes in 3 studies; markers of kidney function were outcomes in 3 studies; leptin, adiponectin, or ghrelin were outcomes in 2 studies; markers of cardiovascular disease were outcomes in 3 studies; markers of breast cancer were main outcomes in 2 studies; markers of calcium metabolism, or metabolomics profiles were outcomes in 2 studies. A few studies examined responses to dietary interventions in multiple ethnic or racial groups.
For example: a high fiber, low fat and low sodium diet resulted in similar decreases in diastolic BP and urinary sodium excretion in West Indian and White individuals, but no changes were observed for these outcomes in Asians 20 . Another study which grouped all non-Caucasian people together observed similar reductions in LDL with oat cereal with beta-glucan consumption in both Caucasian and non-Caucasian people 21 .

Risk of bias and quality of assessment of studies included in systematic review
Characteristics of the included studies are summarized in Table 2. According to the Oxford Centre for Evidence-Based level of evidence, 39 study articles were classified as 1b i.e. highquality individual RCTs and 6 were classified as 2b i.e. RCTs with less than 80% completion or follow-up. According to the Academy of Nutrition and Dietetics Quality Criteria Checklist for Primary Research risk of bias assessment, 31 articles were rated positive, indicating that they had adequately addressed issues of bias, generalizability, and data collection and analysis. The remaining 14 were found to be neutral, meaning that they were neither exceptionally weak nor exceptionally strong. It should be noted that several of these studies did not report or include diet by race interaction effects in the statistical model and instead present subgroup analyses for race or diet groups. Two studies adjusted for race or ethnicity in the statistical model but failed to adequately explain how the dietary intervention effects were stratified by race. Only 19 articles report conducting diet by race interaction analyses (Table 2).  3 The statistical analyses rating is independent of and not accounted for in the QCC rating.

Acute studies
Six studies that accounted for race or ethnicity in their analyses were acute trials. Four studies assessed the glycemic responses to specific foods (including drinks) [70][71][72][73] , one examined the effects of a high-fat vs. low-fat meal on cardiovascular outcomes 74 and one examined the effects of a high-glycemic vs. low-glycemic load meal on appetitive hormones 75,76 . Although, collectively, these studies do not demonstrate an appreciable difference in the acute responses to foods or meals by ethnicity or race, Woelver et al. 71 report potential differences in the glycemic index of starchy foods by Caucasian vs. non-Caucasian race but no mechanisms were discussed.

Studies with only one diet group or non-RCTs
Ten studies that accounted for race and/or ethnicity in their analyses either examined the pre-post effects of a dietary intervention such as low-fat, high-fiber, fruits, and vegetables diet 77 , verylow calorie diet 78,79 , high-fat high-calorie diet 80 , or alternate-day fasting 81 ; or were a post-diet analysis of an individualized diet 82 , crossover studies of sodium dietary restriction (one dietary intervention) with sodium or placebo supplementation 83,84 , or studies examining one dietary intervention with different physical activity conditioning 85,86 . Notably, one of those studies demonstrated greater skeletal muscle mass loss during diet-induced weight loss in European American in comparison to African American women who preserved muscle mass 85 . Another study found different metabolic adaptations for South Asian men compared to European men during energy restriction, i.e., improved glucose disposal rate and decreased shift from glucose to lipid oxidation in South Asians 79 .

Studies with only one ethnic or racial minority group
Twenty-seven studies focused on only one ethnic/racial group. Eleven of such studies included only Black people, 5 studies included Hispanics or people of Mexican origin, 10 studies were on people of Asian origin, and 1 was on Native American people. Dietary studies for East Asian people comprised of cereal based diets [107][108][109] , acute effects of kiwifruit preloads 107 , American heart association step 1 and 2 diets 110,111 , and intermittent energy restriction combined with Mediterranean diet (IER+MED) 112 . Notable results include improvement in metabolic risk factors when substituting brown rice for white rice for 3 months in pre-diabetic Chinese American population 108 . Moreover, consuming an IER+MED diet for 12 weeks improved indices of liver function more than the DASH diet among East Asians in Hawaii.
Lastly, a flaxseed intervention in hypercholesterolemic Native American women resulted in reduced total and LDL cholesterol 113 .  114,115 . Possible explanations include higher baseline weight 28 and fat-free mass 116 and preservation of skeletal muscle mass during weight loss 85 .

Summary of evidence
Despite smaller weight-loss, greater improvements in waist-hip ratio 28 were observed.
Interestingly, a decrease in adiposity with dietary interventions is also seen independent of weight loss in Black individuals 97 . Another possible explanation is limited consideration for group-specific eating patterns, cultural preferences, and lifestyle factors in the design of the diet treatment to achieve comparable adherence [117][118][119][120] . Many of the studies reviewed did not report whether the diet treatment was tailored for acceptability by all the groups studied.
The high prevalence of high blood pressure among Black individuals has led to the development of several dietary interventions, the most popular being the DASH diet and its reduced sodium version). The DASH approach is a diet rich in fruits, vegetables, and low-fat dairy products with reduced saturated and total fat. Overall evidence supports greater reductions in sodium intake and BP with the DASH diet in Black vs. White people. Several explanations have been proposed for these differences, such as Black people having increased salt sensitivity, greater body mass, and lower baseline dietary potassium intake. Nonetheless, more research is needed to further characterize the factors contributing to these differences and their potential relevance to other diet-related outcomes and interventions 121 .
Because human nutrition is complex and dynamic, factors beyond intake need to be considered when designing research interventions. Thus, future studies need to explore the effectiveness of culturally sensitive interventions, controlling for social, economic, and environmental factors to investigate racial and ethnic disparities in nutritional outcomes.
Accounting for the socioecological context that inform intake allows for robust interpretation of findings and can provide key information to ascertain adoption and sustainability in real life circumstances.

Strengths and limitations
This review is unique in that it profiles the representation of racial and ethnic minority Most studies examined the differential effects of dietary interventions by race or ethnicity in subgroup analyses that ignored diet by race interaction effects. There is a need for adequately powered studies for accurate interpretation of race by diet effects. Differences in outcomes by race and ethnicity cannot be suitably studied with insufficient sample sizes for population subgroups. Factors that help explain variability in responses and those that can offer mechanistic insight, such as omics techniques, and socioecological context, should be included and adequately powered 115 . Lastly, an added strength of this overall review is that studies of lower evidence quality-for example uncontrolled interventions and acute studies and studies focusing on one ethnic/racial group-were not dismissed but reported separately to provide a more comprehensive review of the available evidence.

AUTHOR CONTRIBUTIONS
The authors' responsibilities were as follows-JD designed the study; JD, AJ, SO extracted data; and JD, AJ, KDR wrote the paper. All authors read and approved the final manuscript and take responsibility for the final content.