We define racial and ethnic groups in this Series paper using the Office of Management and Budget Standards (OMB) definition.¹³ The racial groups are American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, and White. The ethnic groups are Hispanic and non-Hispanic. From here on, we will refer to non-Hispanic Black as Black, and non-Hispanic White as White. American Indian or Alaska Native might also be referred to as Native American on the basis of study descriptions.

Parallel to the growing epidemic of childhood obesity, prevalence and incidence of type 2 diabetes in children and adolescents has increased since the turn of the 21st century, with continued and persistent earlier onset of diagnosis, which has been exacerbated by the COVID-19 pandemic.¹⁵ The burden of type 1 and type 2 diabetes varies by race and ethnicity, as reported in the SEARCH for Diabetes in Youth Study (SEARCH; figure 1). Although this Series paper is focused on type 2 diabetes, type 1 diabetes continues to play an important role in diabetes in young people. We note here that data in young people are limited by racial and ethnic group differences and not within racial and ethnic groups; therefore, this topic will not be discussed in the young people section (figure 2).

SEARCH has conducted diabetes surveillance in young people in the USA for nearly two decades. In 2001, 6379 individuals aged 0–19 years were identified with diabetes within a population of approximately 3·5 million, with Native American young people having the largest proportion (76%) of type 2 diabetes.¹⁷ Population projections from this study estimated roughly 154 000 young people aged 0–19 years were living with diabetes in the USA in 2001. Since 2001, diabetes prevalence in young people has increased among all racial and ethnic groups, with the largest increases among Asian and Pacific Islander (these groups were not separated in SEARCH), Black, and Hispanic children. Between 2009 and 2017, the estimated prevalence of diabetes in Black children increased by 7·1% (95% CI 4·8 to 9·4), compared with 7·3% (2·7 to 12·1) in Asian and Pacific Islander children, 4·8% (0·7 to 9·0) in American Indian children, 3·2% (1·4 to 5·0) in Hispanic children, and 1·4% (–1·2 to 4·1) in White children.^12,16 A separate study showed a pronounced increase in diabetes among American Indian and Alaska Native young people (aged 15–19 years), with a 68% increase in cases of diagnosed type 2 diabetes from 1994 to 2004.^12,18 The COVID-19 pandemic has been associated with a rise of incident paediatric diabetes cases, with one study finding incident type 2 diabetes cases decreasing moderately by 7·4% in the 2 years before the pandemic and increasing 182% during the pandemic (n=141; 1·45 cases per month; p<0·001). Type 2 diabetes cases increased by more in Black children (76·6%) than in White children (56·7%; p=0·001) and more in boys (58·9%) than in girls (40·4%; p=0·005).¹⁹ Among American Indian children and adolescents, metabolic abnormalities led to greater relative risk of diabetes in children (aged 5–11 years) compared with adolescents (aged 12–19 years).^20,21 With one in five adolescents in the USA having prediabetes, the problem of diabetes among young people will probably continue to grow.²²

Overall, adolescents have more rapid and complete onset of β-cell failure compared with adults, increasing urgency for the few available treatment options.²³ Furthermore, the onset of diabetes-related complications in specific racial and ethnic groups is accelerated in young people. The Pediatric Diabetes Consortium Registry reported significant race and ethnic differences in HbA_1c and C-peptide levels, with Black young people (aged <18 years) having higher HbA_1c and lower C-peptide than White young people. Black children are significantly more likely than White children (odds ratio [OR] 4·85, 95% CI 1·49–15·77) to have poor glycaemic control (HbA_1c >9·5%) at 1 year from time of diagnosis.²⁴ Non-alcoholic fatty liver disease was diagnosed in 9% of Hispanic children and 11% of White children, compared with 2% of Black children.²⁵ Similarly, Hispanic children have significantly elevated liver aminotransferases compared with Black children.²⁶

Here, we describe within-group differences in diabetes prevalence, burden, and management among Hispanic, Black, and Asian individuals. Data on differences within White, American Indian, and Pacific Islander populations are scarce. White people are often viewed as one large, homogeneous group; however, many people who identify as White in the USA might represent different and culturally distinct subgroups, such as people born within the USA and those born overseas (eg, Europe or the Middle East), or religious subgroups. Although studies exploring these subgroups are scarce, differences in diabetes prevalence by rural versus urban residency have been found.^27,28 Additionally, demographic-adjusted models have shown increased diabetes risk among White populations based on residency in the so-called stroke belt of the southeastern USA (OR 1·22, 95% CI 1·09–1·36) and buckle of eastern North Carolina, South Carolina, and Georgia (1·26, 1·10–1·44), compared with residency in other parts of the USA.²⁹

Within the Hispanic population, the national diabetes statistic report of 2020 highlights differences in age-adjusted prevalence of diagnosed diabetes among Hispanic Americans, with Cuban Americans having the lowest prevalence at 6·5%, compared with 12·4% in Puerto Rican American and 13·4% in Mexican American adults.² These differences have been shown using different datasets as well, including the nationally representative NESARC-III study.²² The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) corroborates these findings with a population-based cohort. A longitudinal analysis from HCHS/SOL showed persistent differences in diabetes incidence within the Hispanic population with higher incidence among those with Puerto Rican and Mexican backgrounds.^30,31

US-born Black individuals and non-US born Black individuals of sub-Saharan African or Caribbean descent are often analysed as one racial group (Black) in the literature, although research disaggregating these groups has shown important differences in diabetes prevalence.³² Several studies have indicated the higher prevalence of diabetes among US-born versus African-born or Caribbean-born Black adults.^33,34 Ford and colleagues¹⁰ used National Health Interview Survey data to show that Black individuals born in Africa or the Caribbean had significantly lower odds of having diabetes (OR 0·75, 95% CI 0·62–0·89) compared with US-born Black individuals. Within the population of Black individuals born in Africa or the Caribbean, studies have shown little to no difference between those born in the Caribbean and those born in sub-Saharan Africa. Nevertheless, increasing time in the USA and higher levels of acculturation were associated with increased odds of diabetes diagnosis among foreign-born Black individuals.³³ This increase has highlighted the importance of furthering our understanding of the differences between Black people born within and outside of the USA, as a unique opportunity to examine genetic differences, sociocultural contexts, migration experience, and population characteristics.³⁵

Asian Americans are a large and diverse population, although they are often grouped into one large racial and ethnic group.³⁶ South Asian Americans have higher age-adjusted prevalence of diabetes and gestational diabetes than other Asian American subgroups.³⁷ Filipino American and Korean American adults have the next highest prevalence of diabetes among Asian Americans;^38,39 prevalence of diabetes among Korean Americans was 10% in New York City and 11·9% in California,^40,41 with Korean Americans having very low rates of physical activity compared with other Asian subgroups.^42,43 Of all Asian American subgroups, Chinese Americans continue to have the lowest rates of diabetes.⁴⁴

Factoring in within-group differences will allow us to better understand trends in diabetes prevalence epidemiologically. We can better target resources and interventions if we look at subgroups within racial and ethnic groups whose rising rates might be otherwise masked. These described differences within racial and ethnic groups are crucial considerations when designing interventions for diabetes prevention or management.

During the past decade, there have been substantial advances in our understanding of the genetic basis for type 2 diabetes.⁷² Identified individual genetic variants can only provide a modest effect on risk, but when combined into a polygenic risk score, they provide information on disease predisposition and could be useful for guiding clinical management. These polygenic risk scores are determined using ever-growing genome-wide association studies. Polygenic risk scores have been used to identify subtypes of diabetes, predict risk for type 2 diabetes, and predict effectiveness of interventions. In a diabetes prevention programme post-hoc analysis, polygenic risk scores identified subgroups at high risk for whom successful lifestyle modification was associated with an improved absolute reduction in the risk of incident diabetes.⁷³ As polygenic risk scores become more widely available, it is important to remember that genetics represent only one contributor to individual disease risk and profile.73 Additionally, there is a need to ensure that the benefits of these risk scores are equitably available to all. Equitable access to polygenic risk scores is especially important as much of the genome-wide association studies and sequence data are derived from individuals of European descent. As such, there is an urgency to build a repository for other ethnic groups as well.

The complex interplay of factors across all five domains and all levels of influence contribute to disparities between racial and ethnic groups. Structural racism play a prominent role at all levels. Past policies, laws, and economies contribute to where people live, their access to sufficient and healthy food, and their access to health-care services.^{106,110,111,113,115} For example, redlining and predatory lending affected where and what type of housing was available to Black individuals, forcing them to live in less healthy and worse resourced neighbourhoods compared with other racial and ethnic groups. Furthermore, these same areas received fewer government and private sector investments, leading to fewer facilities, such as health facilities and food access points.^161,162 These factors, stemming largely from structural racism and discrimination, play an important role in driving racial and ethnic inequities in diabetes and diabetes-related complications.¹⁶³ At the community level, neighbourhoods with primarily Black and Hispanic individuals tend to have little space for physical activity, more food deserts (ie, neighbourhoods of 5000–15 000 people served by two or fewer big supermarkets), and restricted access to health care compared with other neighbourhoods. These factors have been linked to physical inactivity, poor diet quality, and reduced use of health-care services, which in turn have been found to contribute to the development of diabetes and poor diabetes management.¹¹² Additionally, neighbourhoods with Black and Hispanic individuals have higher levels of toxic environmental exposures than other neighbourhoods, and a growing body of evidence indicates that these exposures are associated with the development of diabetes.¹¹² The interpersonal level reflects relationships between individuals in a family, within a social network, or with health-care professionals. The many forms of interpersonal racism and discrimination can play an important role in health disparities including in those with diabetes. For example, perceived racism has been associated with increased incidence of diabetes in Black women.¹⁵² This association is hypothesised to be through biological mechanisms including increase in chronic stress. These interpersonal relationships in turn affect the gut microbiome, lifestyle behaviours, mental health, and access to care.¹⁵⁶ The determinants are all inter-related and overlapping. The factors within the structural racism domain overlap with the economic development domain and the access to high-quality care domain as outlined.

Native Americans have the highest age-adjusted prevalence of diabetes compared with all other racial and ethnic groups in the USA. Native Americans have 2·0 times higher diabetes prevalence and 1·9 times higher diabetes-related mortality than White Americans.¹⁶⁴ The disproportionate prevalence and disease burden of diabetes has been associated with factors in the economic development domain with social disadvantages faced by American Indian and other Native American communities, including high rates of neighbourhood poverty, inadequate access to health services, and long-standing economic adversity. These factors in turn are related to the health behaviours and social norms domain, where they facilitate unhealthy day-to-day living practices that ultimately contribute to disease, and subsequently hinder optimal management of chronic illnesses. The structural racism domain is also relevant in this case as discrimination has been associated with increased prevalence of diabetes among American Indians and other Native Americans. Some findings have shown that microaggressions experienced by American Indians and other Native Americans with diabetes in health-care settings were associated with greater reports of worse physical and mental health, including increased hospitalisations and depressive symptoms within the previous year.¹⁵⁶

Although pathophysiology of diabetes is important to consider, and there might be a role of genome-wide association studies, it is important to recognise that biological factors are not a cause of disparities. There is a growing body of evidence on biological and epigenetic differences that contribute to the risk of developing diabetes.¹⁶⁵ However, race is a socially constructed variable, not a biologically constructed one.¹⁶⁶ Existing literature suggests that acknowledging the admixture of multiple ethnic groups is crucial in efforts to further define any genetic predisposition to diabetes.¹⁶⁷ Admixture is also an important consideration when looking at the performance of diagnostic tests such as HbA_1c in different racial and ethnic groups.^168,169

Using the modified framework proposed by Agarwal and colleagues,¹⁴ we can see that factors at the individual, interpersonal, community, and societal levels together contribute to differences within racial and ethnic groups (table). The health behaviours and social norms domain plays an important role as a contributor to diabetes disparities. Diversity within racial and ethnic groups in cultural practices, knowledge, and beliefs can contribute to differences in dietary and lifestyle habits that in turn can drive within racial and ethnic group differences.¹⁷⁰ Cultural practices, knowledge, and beliefs can also contribute to different relationships with the health-care profession, which might lead to different levels of trust and adherence to medical advice.^132–135 Economic development is also important. Previous studies have consistently found that people with lower socioeconomic status are more likely to develop diabetes and experience more diabetes-related complications compared with those with higher socioeconomic status.⁹⁸ There is also a rural and urban divide that can help explain disparities within groups given rural counties have the highest diabetes-related mortality.¹⁷¹ Examining trends in diabetes-related mortality in urban and rural areas in the USA between 1999 and 2018, Kobo and colleagues found that diabetes-related age-adjusted mortality rates of both Black and White adults decreased in urban but not rural areas.¹⁷² The socioeconomic and built environment can also contribute to disparities within racial and ethnic groups. In 2014, Gaskin and colleagues described the “nexus of race, poverty, and place” and showed that both individual poverty and living in a poor neighbourhood increased the odds of having diabetes for Black and White adults.¹⁷³ In 2019, a systematic review and meta-analysis of longitudinal studies found strong evidence for relationships between neighbourhood walkability and diabetes outcomes.¹⁷⁴ Results from an analysis of nationally representative county-level data showed that counties with higher rates of unemployment and poverty had higher incidence rates of diabetes compared with counties with lower levels of unemployment and poverty, whereas counties with greater access to healthy food and more opportunities for exercise had fewer cases of diabetes.⁴

The association between social determinants of health and diabetes differs by ethnicity within the Black population. For example, the association of health insurance, income, and education with diabetes was different among African Americans, African immigrants, and African Caribbeans.³⁴ The differences described within the Hispanic population are most likely multifactorial but are still not fully understood.³⁰ Finally, within the Asian American population, we described how south Asians have the highest prevalence of diabetes. The causes of this are multifactorial and are likely due to not only patterns of immigration and sociodemographic characteristics, but also might be related to distinct cardiometabolic phenotypes that increase the risk of diabetes among south Asians.¹⁷⁵

Risk factors for type 2 diabetes in children and adolescents (aged <18 years) are similar to those of adults. Studies have found that intergenerational factors, malnutrition, little physical exercise, obesity, family history, and low or very high birthweight elevate diabetes risk in children.¹⁷⁶ Poor sleep has also been linked to diabetes risk among both adults and children, with one multiethnic study in the UK finding that children aged 9–10 years had reduced risk of type 2 diabetes for every additional hour of sleep duration.¹⁷⁷

Obesity and insulin resistance in children portends an increased risk of developing diabetes. There is now good evidence to show that stress, obesity, and hyperglycaemia in lactating mothers is associated with increased rates of obesity and insulin resistance in their offspring.¹⁷⁸ Additionally, overnutrition as a consequence of an infant being born small for gestational age, as well as early weaning from breastmilk are associated with obesity and insulin resistance.¹⁷⁸ Lastly, on the other end of the spectrum, a combination of early-life undernutrition and later development of obesity is associated with increased risk of type 2 diabetes in adulthood.¹⁷⁹

Economic development also has a role for racial and ethnic disparities in young people. For example, lower socioeconomic status and living in poor areas are associated with higher diabetes risk, a 34% increased likelihood of developing obesity, and a 60% increased likelihood of developing diabetes, when compared with living in high income areas.^180,181 Additionally, compared with White women, Asian and Hispanic women have a higher prevalence of gestational diabetes, which has been associated with increased risk in offspring of developing type 2 diabetes later in life.^182,183

Diabetes disparities are now evident across the lifespan with the emerging concept of intergenerational risk of developing diabetes.¹⁸⁴ Young people who identify as Native American, American Indian, or Alaska Native; Black; and Hispanic have an increased burden of diabetes and obesity. This burden is in part due to social determinants of health within the structural racism domain that are “systemic, population-based, cyclical, and intergenerational.”¹⁸⁵ These same social determinants of health are rooted in racism that is internalised, or at the interpersonal, institutional, and societal levels. These factors are experienced by the family (including children) and not only the adult. Additionally, intergenerational poverty makes it harder for children to find a way out of the cycle of intergenerational risk. Although socioeconomic status is complex and multidimensional, all indicators are strongly patterned by racial and ethnic group: “for every dollar of wealth that Whites have, Asian households have 83 cents, Black households have 6 cents, and Hispanic households have 7 cents”.¹⁸⁶

This Series paper has thus far highlighted the complexity and multifaceted nature of diabetes disparities. Solutions must, therefore, meet this complexity by ensuring a multidisciplinary approach that carefully engages affected communities in an equitable manner. Increased attention to more detailed diabetes phenotypic structures (see previous sections) is important to guide tailored prevention and management strategies. Understanding diabetes phenotypes and the resulting tailored management strategies can provide the health-care community with the needed information to move towards effective and equitable precision medicine in diabetes care. As we give the pathophysiology of diabetes more attention, researchers should be sure not to confuse social constructs such as race with biological factors such as genetics.¹⁸⁹ As we promote ideas around precision medicine for diabetes care, it is crucial to curb any misconceptions of this idea and ensure emergent technologies and advancements overcome current inequities in access. Community-engaged and community-based participatory research is essential for designing acceptable and feasible strategies of promoting uptake of diabetes treatment interventions among disadvantaged groups. These strategies will need to be tailored to groups across the lifespan (strategies related to young people [aged <18 years] will differ from those targeting older people [aged >65 years]). These strategies will also need to be tailored across and within racial and ethnic groups given much of the social and behavioural norms that differ between these groups. To do this well, research needs to continue to disaggregate the large groupings of Black, Asian, and Hispanic individuals.

As we see the growth of the fields of environmental justice, climate, and health, we must pay attention to their effects on diabetes prevalence, management, and thereby disparities. We need more community-engaged research to document the attribution of environmental factors on diabetes risk and management. Moreover, we need community-engaged research to co-design strategies with communities to mitigate the effects of environmental hazardous exposures and climate change on health. Researchers need to apply a health equity lens to ensure that novel therapeutics and diagnostics are implemented without worsening disparities.

The field of implementation science increasingly offers approaches to integrating health equity into intervention design and implementation and should be embraced along with equitable partner and community engagement as the field moves to develop, implement, and scale evidence-based interventions to reduce diabetes disparities.^190,191

Funders of community or clinic-based diabetes research should require a health equity plan for all phases of diabetes research—ie, a statement on how the research will promote health equity or at the very least not increase disparities. To achieve this goal, principles of implementation science should be considered even at the effectiveness and clinical trial stages of research to ensure an understanding of how to promote equity as the research progresses. Practically speaking this means evaluating key implementation outcomes such as reach, acceptability, appropriateness, feasibility, and adoption early on. Studying implementation outcomes enables investigators to start to identify early on potential barriers for equitable implementation in the real world. Defining implementation outcomes also provides an opportunity to tailor implementation strategies to age groups and differences within racial and ethnic groups.

Upon review of the multilevel causes of diabetes disparities across the lifespan, it becomes apparent that multilevel and multidisciplinary approaches are crucial to solving the problem of diabetes disparities. Funders should recognise this importance and provide opportunities for multidisciplinary teams to address this need in collaboration with communities. To promote community-based participatory research that addresses these multilevel factors, funders should invest in strengthening capacity among early stage investigators in the intersection of diabetes and health equity with a focus on trainees from under-represented groups. Moreover, given the value of implementation science in improving equity in health service delivery, training, and funding for early-stage investigators to conduct health-equity focused implementation science research could accelerate the implementation and dissemination of evidence-based interventions in routine practice.¹⁸⁹ Furthermore, there is a need to provide training opportunities for more experienced scientists to embrace and incorporate principles of community engagement and health equity into existing research programmes.

Front-line providers play an important role in addressing disparities through their one-on-one interaction with patients, through institutional change, and through community engagement. At an individual level, the role of health-care providers’ implicit bias has long been studied and the effect of training on reducing disparities examined.¹⁹² Continued attention to implicit bias with focus on understanding gaps in translational effectiveness is important.¹⁹³ At an institutional level, front-line practitioners have a role in addressing structural barriers to access to care for their patients. It is crucial to develop evidence-based approaches to address structural barriers to health-care access such as inadequacies of language services, variable dependence on technology, transportation challenges, and financial constraints. Lastly, engaging with community members to actively understand the challenges faced by patients and develop equitable solutions is essential.^194–196 We need participatory action research engaging community partners in prioritising and developing approaches for overcoming these health system barriers. This participatory action research provides an opportunity for health providers to understand differences across the lifespan as well as tailoring approaches to within group differences.

In their approach to counselling and management of patients, health providers must learn to recognise differences in phenotypic presentation of diabetes to tailor therapy. Recognising the structural and environmental factors that contribute to diabetes and poor management of diabetes in some communities is imperative as we design therapies, set policy, and provide advice. Training providers on the role of social and environmental determinants of health is crucial to ensuring future practitioners overcome their own and their institution’s contribution to health disparities.

In this Series paper, we summarised the various ways in which policy decisions now (eg, scarcity of policies to drive investment in disadvantaged neighbourhoods) and in the past (eg, redlining), differentially affect the health of minority race and ethnic communities and, therefore, drive health disparities. Addressing social determinants of health holds great potential for progress towards reducing diabetes disparities.^98,197 As health-care quality improvement efforts to address diabetes care advance, there is a need for rigorous studies to design and test delivery models that integrate social care into medical care; for example, through screening of individuals’ social risks and assets and strategies to address or mitigate the adverse effects of social adversity on diabetes management.¹⁹⁸ In addition to addressing social determinants of diabetes within the health-care system, there are opportunities to fill gaps in evidence on the effectiveness of large-scale policies and programmes such as food and agricultural policies and programmes in promoting healthy eating and physical activity and thereby contributing to the prevention of diabetes and diabetes-related complications, including their effect on diabetes disparities.¹⁹⁹

Recognising the role of institutional racism on existing policies is crucial to revising and developing evidence-based solutions to policy change that can overcome these barriers to care. Further research is needed to inform policies that can address some of the structural and social determinants of health that are the root causes of health disparities, both between and within racial and ethnic groups.²⁰⁰ To achieve this goal, policy and programme monitoring should integrate health equity indicators so that the differential effect of policies can be documented and thus the impact of policy change reported on health disparities. This type of attention would help promote accountability and generate data needed to inform change.