Exploring Racial and Ethnic Differences in Arterial Stiffness Among Youth and Young Adults With Type 1 Diabetes

Nonstandard Abbreviations and Acronyms

NHB	non‐Hispanic Black
NHW	non‐Hispanic White
PWV	pulse wave velocity
T1D	type 1 diabetes

Cardiovascular disease (CVD) occurs ≈10 times more frequently among individuals with type 1 diabetes (T1D) than among peers without diabetes,¹ with higher CVD rates emerging as early as 28 to 38 years of age.² Abnormalities such as arterial stiffness, left ventricular hypertrophy, and diastolic dysfunction that precede overt CVD (coronary heart disease and stroke) are also more prevalent in T1D, even in childhood.^{3, 4, 5, 6} Moreover, individuals who identify as non‐Hispanic Black (NHB) experience 8 times the CVD event rates than peers who identify as non‐Hispanic White (NHW).² Given continued annual increases in T1D incidence in the United States (+1.9% annually), which are notably larger in Hispanic and NHB populations (+4.0% and +2.7%, respectively),⁷ a better understanding of the modifiable factors contributing to differences in CVD is needed to improve care for this growing population.

A joint scientific statement issued by the American Heart Association and American Diabetes Association in 2014 highlighted the limited evidence base for CVD prevention and treatment specific to T1D.⁸ Most of this evidence comes from studies of NHW participants with ≥25 years T1D duration who preceded diabetes management technologies (continuous glucose monitors and insulin pumps),^{2, 9, 10} and thus do not represent the current increasingly diverse T1D population receiving care in the modern clinical landscape. These Associations called for research on the relative contributions of risk factors for CVD in T1D, data on risk factors specific to children with T1D, and racial and ethnic differences in risk factors and CVD events. Subsequent studies have confirmed that traditional CVD risk factors (eg, obesity, hypertension, dyslipidemia, and hyperglycemia) increase with T1D duration and differ by race and ethnicity even in childhood.¹¹ However, the relative contribution of modifiable factors from T1D diagnosis onward in relation to the differences that emerge early in the course of CVD is unclear. Such information is critical to developing evidence‐based guidelines to promote cardiovascular health beginning soon after T1D diagnosis for this increasingly diverse population.

Thus, we examined how modifiable risk factors in the first 2 decades of T1D are related to arterial stiffness in early adulthood in a contemporary sample of individuals with T1D. We focused on arterial stiffness, given its prognostic significance for clinical CVD.¹² We examined clinical and social risk factors, recognizing that the social conditions in which people live can influence health and disease.¹³ We aimed to identify the modifiable risk factors that explain, at least in part, racial and ethnic differences in arterial stiffness, and thus can inform interventions and clinical management strategies to reduce differences in CVD.

METHODS

Data Availability

Anonymized data and materials for the SEARCH for Diabetes in Youth study have been made publicly available at the NIDDK Central Repository and can be accessed at https://repository.niddk.nih.gov.

Participants

The SEARCH for Diabetes in Youth Study population and protocol have been described previously.¹⁴ Briefly, SEARCH identified individuals newly diagnosed in 2002 to 2018 with any type of diabetes at age <20 years via a population‐based registry network covering 5.5 million youth from 5 sites: the state of South Carolina; Cincinnati, Ohio and surrounding counties; the state of Colorado with southwestern United States American Indian sites; Seattle, Washington and surrounding counties; and Kaiser Permanente Southern California membership in 7 counties. Individuals diagnosed in 2002 to 2006 or 2008 were invited to join an observational cohort study of the natural course of youth‐onset diabetes. Baseline visits were conducted shortly after diagnosis (2002–2010), with follow‐up visits conducted ≥5 years after diagnosis (2011–2015) and ≥8 years after diagnosis (2015–2019). The study was approved by the institutional review boards with jurisdiction in each location; participants and/or parents provided written informed consent and/or assent.

Data Collection

We abstracted date of birth and date of diagnosis from medical records, and calculated age of diagnosis, current age, and diabetes duration for each visit. At baseline, parents reported child sex, race, ethnicity, highest parental education, household income, type of health insurance, and whether diabetic ketoacidosis occurred at diagnosis. At the follow‐up visits, participants (or parents of participants younger than 18 years) self‐reported current household income, type and continuity of insurance, problems with the cost of care, missing medications because of cost, insulin regimen, and glucose self‐monitoring practices. They self‐reported smoking status, and completed questionnaires regarding diet,¹⁵ physical activity,¹⁶ depression,¹⁷ and food security.¹⁸ They also self‐reported whether they had met with a dietitian in the past year, and how frequently their provider checked their hemoglobin A1c, blood pressure, cholesterol; evaluated for nephropathy, neuropathy, and retinopathy; and recommended (when applicable) smoking cessation, hypertension therapy, or dyslipidemia therapy. Last, participants rated their perceptions of their T1D clinical care as excellent, good, fair, poor, or not applicable for 5 areas: getting answers to T1D questions, courtesy and communication style of the provider, explanation of laboratory results, access during emergencies, and overall T1D care.

Anthropometric, metabolic, microvascular, and macrovascular profiles were assessed by trained research staff as described previously.¹⁴ Briefly, height and weight were measured in light indoor clothing without shoes and used to calculate body mass index (kg/m²). Blood pressure was measured after 5 minutes of seated rest, with 3 readings taken 1 minute apart averaged for analysis. Participants were instructed to fast overnight for ≥8 hours and abstain from medications (including short‐acting insulin) on the morning of the visit. Blood samples were obtained and analyzed for hemoglobin A1c, lipids, creatinine, and cystatin C at the Northwest Lipid Metabolism and Diabetes Research Laboratory, Seattle, WA. At the follow‐up visits, participants were asked to bring a first morning urine void; if not done, a spot urine specimen was collected at the time of the visit (≈8% of participants). Urine samples for albumin and creatinine were also analyzed at the Central Laboratory. All analyses were performed using standardized methods as previously reported.¹⁴ Diabetic kidney disease was defined as the presence of albuminuria (>30 μg/mg of creatinine) or low glomerular filtration rate (<60 mL/min per 1.73 m² as estimated by the CKD‐EPI equations with serum creatinine and cystatin C).¹⁹ Also at the follow‐up visits, diabetic retinopathy was assessed with 45° color digital fundus images taken with a nonmydriatic camera (Visucam Pro N, Carl Zeiss Meditech) and centered on the disc and macula of both eyes. Photographs masked to all clinical characteristics were graded by the Wisconsin Ocular Epidemiology Reading Center. Diabetic retinopathy was defined as mild, moderate, or proliferative retinopathy in at least 1 eye.²⁰ Peripheral neuropathy was defined as a score >2 on the Michigan Neuropathy Screening Instrument.²¹ Cardiovascular autonomic neuropathy was assessed by heart rate variability using the SphygmoCor‐Vx device (AtCor Medical). Electrocardiographic R‐R intervals measured with the patient in a supine position for 10 minutes were used to measure heart rate and estimate 5 heart rate variability indices: the SD of the intervals, root mean square differences of successive intervals, normalized high‐frequency power, normalized low‐frequency power, and the low‐to‐high‐frequency ratio.

Our primary outcome of arterial stiffness is highly relevant to CVD: every SD increase is associated with 30% increased risk of CVD events, independent of conventional risk factors.¹² We measured carotid–femoral pulse wave velocity (PWV) with the SphgymoCor‐Vx device (AtCor Medical, Lisle, IL) as previously described.²² We used the average of >10 beats to cover a complete respiratory cycle, and obtained 3 recordings per participant, which were averaged for analysis. PWV coefficients of variability were 7%. Higher PWV values indicate greater arterial stiffness.

Statistical Analysis

Our analytic sample included participants with a provider diagnosis of T1D who had PWV measured at the second follow‐up visit and self‐identified as Hispanic, NHB, or NHW. We did not include participants who identified as American Indian, Alaskan Native, Asian, other races, or ethnicities, or did not report race or ethnicity, because the sample sizes were too small to support meaningful analyses. However, we do present descriptive data of these populations historically under‐represented in research in Table S1. Unadjusted differences in prevalence of ideal risk factors were evaluated using χ² test of differences, first for the overall sample with all 3 racial and ethnic groups, and then with pairwise comparisons when the overall test for differences reached statistical significance. We used general linear models to analyze differences in PWV according to self‐identified race and ethnicity. The base model was adjusted for sex; age, heart rate, and mean arterial pressure at PWV measurement; and SEARCH study site. We also explored sex differences by including an interaction term for sex by race and ethnicity in all models. However, this term did not reach the statistical significance threshold in any model; thus, it was removed and only main effects are reported and interpreted. Because of our a priori plans to examine differences between racial and ethnic group, we abstracted adjusted means of PWV for each racial and ethnic group from the model and the corresponding P values. Then, we explored risk factors that potentially explain differences in PWV between these racial and ethnic groups by adding variables to the base model as grouped within risk factor domains (socioeconomic, T1D characteristics, CVD risk factors, behavioral health, quality of clinical care, and perception of care). We constructed separate models for each risk factor domain that included all variables in the base model and all variables within each domain. Last, we constructed a final model that included all variables from all domains, including the base model. We examined the R² value and change in mean difference to understand variation in apparent differences across models. We examined how differences in PWV between racial and ethnic groups changed as additional modifiable variables were added to the analytic models. This was calculated by comparing the difference in PWV between each pair in the base model with the difference in PWV between each pair in the domain models. For example, if PWV was 0.40 m/s higher in NHB participants than NHW participants in the base model, but only 0.33 m/s higher in the socioeconomic model, then we conclude adjustment for socioeconomic factors reduced the NHB versus NHW disparity in PWV by 17% (calculated as: [0.40–0.33]/0.40=0.17×100=17%).To account for the longitudinal nature of our data, we classified participants as having the ideal levels of each risk factor at all visits versus all other combinations (fully detailed in Data S1). Jackknifed studentized residuals from all models were inspected for normality to confirm that the assumptions for general linear modeling were met. Analyses were conducted in SAS v9.4 (Cary, NC) with 2‐sided alpha set at 0.05 for statistical significance.

RESULTS

PWV measurements were conducted on 1224 participants with T1D at the second follow‐up visit at a mean T1D duration of 11.2 years. Of these, 1162 identified as Hispanic, NHB, or NHW and thus were included in this analysis. Their characteristics at each visit are presented in Table S2. The proportion with ideal characteristics at all 3 visits is presented in the Table, overall and stratified by race and ethnicity. Unadjusted mean PWV (m/s) at the second follow‐up visit was 6.01 (SD 1.16) overall, 5.97 (1.08) in Hispanic participants, 6.48 (1.46) in NHB participants, and 5.89 (1.05) in NHW participants.

In the base model, PWV was significantly higher in Hispanic versus NHW participants (Figure). This initial model, which included age, heart rate, mean arterial pressure, sex, race, ethnicity, and site, accounted for 40% of the variability in PWV (R²=0.40). Statistically significant differences in PWV between Hispanic and NHW participants persisted in all domain models except those for cardiovascular risks and socioeconomic factors. In the cardiovascular risks model, the mean difference in PWV decreased 15% (indicated by a reduction in the beta estimate for the pairwise difference from 0.17 in the base model to 0.14 in the cardiovascular model) and was attenuated to statistical nonsignificance (P=0.056). In the socioeconomic model. The mean differences in PWV decreased 27% (beta estimate reduced from 0.17 in the base model to 0.12 in the socioeconomic model) and was statistically nonsignificant (P=0.12; Table S3). In the full model, there were no significant differences in PWV between Hispanic and NHW participants (P=0.07), and the mean difference decreased 15% from the base model (beta estimate reduced from 0.17 in the base model to 0.14 in the socioeconomic model). Variables in the full model account for 43% of the variability in PWV (R²=0.43).

• Download figure
• Download PowerPoint

Also in the base model, PWV was significantly higher in NHB versus both Hispanic and NHW participants (Figure). Statistically significant differences in PWV between Hispanic and NHB participants persisted in all models except the full model, wherein the mean difference for NHB versus Hispanic participants decreased by 25% (beta estimate reduced from 0.23 in the base model to 0.17 in the full model) and was no longer significant (P=0.08). Statistically significant differences in PWV between NHB and NHW participants persisted in all models. In the full model, the mean difference for NHB versus NHW participants decreased by 21% (beta estimate reduced from 0.40 in the base model to 0.32 in the full model) but remained statistically significant (P<0.0001).

Beta estimates associated with each variable included in the full model are presented in Table S4. Variables associated with significantly lower PWV included never missing medications in the prior year because of cost, body mass index <25 kg/m², hypertension therapy prescribed when applicable (or normotensive), and excellent or good rating of provider's communication. Variables associated with significantly higher PWV included NHB race, older age at PWV measurement, and excellent or good rating of overall T1D care.

DISCUSSION

We report that PWV at ≈20 years of age among participants with T1D varies by self‐identified race and ethnicity. PWV varied by 0.17 to 0.40 m/s in the base model, which is similar to the difference previously reported for youth and young adults with versus without T1D,²³ and corresponds to as much as a 9% increased risk of CVD morbidity and mortality.²⁴ However, these differences were attenuated by up to 27% after accounting for cardiovascular and socioeconomic‐related factors. Our results suggest that a large portion of racial and ethnic differences in subclinical CVD may be attributable to modifiable factors present in the first 11 years of T1D. Of note, PWV was significantly elevated in NHB compared with NHW participants in all analytic models, indicating that other factors must be considered to address these persistent differences.

Prior studies have reported differences in heart disease and stroke among NHW and NHB middle‐aged adults (with or without diabetes) in a general population, which were largely attenuated after adjustment for clinical, socioeconomic, lifestyle, and/or neighborhood factors.^{25, 26} Our work demonstrates a similar phenomenon in a population of young adults with T1D who have high risk for CVD.¹ While our participants with T1D are not yet experiencing major adverse cardiovascular events, these data demonstrate that an important precursor to CVD (arterial stiffness)^{12, 24} manifests with apparent racial and ethnic differences as early as the third decade of life. Because race and ethnicity are social (not biological) constructs,²⁷ our analysis clarified important modifiable contributors that explain, at least in part, these apparent racial and ethnic differences in this population. Unlike studies in the general population that noted clinical factors as the largest contributor to differences in CVD,^{25, 26} socioeconomic factors explained the largest percent of the PWV disparity in our study of young adults with T1D. This could be because our participants are younger, and have not accumulated as much of the age‐related CVD risk factors as older adults. In contrast, socioeconomic factors influence health from early life, including when they limit one's ability to access care or obtain medications.¹³ We do note that in the socioeconomic domain model and the full adjusted model, the only socioeconomic variable that was significantly related to PWV was missed diabetes medication in the prior year because of cost. Thus, our work underscores the importance of improving access to affordable care across the increasingly diverse population⁷ of individuals with youth‐onset T1D.

Moreover, our analysis could be an example of how the racial and ethnic differences in socioeconomic position in the United States may contribute to racial and ethnic differences in health.²⁸ Yet our results do suggest that socioeconomic factors have differential impact on health‐related outcomes between racial and ethnic groups. We observed the greatest reduction in PWV differences (27%) for Hispanic versus NHW participants after adjustment for socioeconomic factors. In contrast, socioeconomic adjustment reduced PWV differences by only 17% for NHB versus NHW participants, and only 10% for Hispanic versus NHB participants. Thus, socioeconomic factors appear to be less relevant to PWV differences evident among NHB participants, suggesting other factors need further attention. Our study did not collect data on broader experiences and sources of racial and ethnic discrimination and inequities in the health care system, community, or society at large, which should be explored in future studies.

While not explaining the largest amount of variability, cardiovascular risks did explain 5% to 15% of the racial and ethnic differences in PWV. Variables in this domain included traditional risks (weight, lipids, blood pressure, glycemic control, heart rate variability) and more specific, diabetes‐related, microvascular complications (retinopathy, neuropathy, and nephropathy). These traditional risk factors are more common in populations with type 2 diabetes,²⁹ but are increasingly evident in T1D, starting in childhood.¹¹ Yet evidence‐based clinical guidelines for managing these risk factors in youth with T1D are limited, with some for hypertension and dyslipidemia treatment supported only by “C” or “E” grades (C=supportive evidence from poorly controlled or uncontrolled studies; E=expert consensus or clinical experience), versus clear or supporting evidence from rigorous trials (“A”).³⁰ Evidence‐based guidelines for glycemic control in T1D are more robust, and largely informed by the Diabetes Control & Complications Trial showing that intensive insulin therapy to achieve A1c to <7% among youth and young adults with T1D reduced CVD events by 40%.¹⁰ However, this target is achieved by only 1 of 5 youth and young adults in SEARCH (similar to the present subset where ≈21% achieved A1c <7.5% by age 20 years),¹⁴ with notable differences by race, ethnicity, and other social determinants of health.³¹ Moreover, glycemic control in the United States has declined since 2000 in age‐matched samples,³¹ despite the fact that more youth and young adults are using diabetes management technologies including insulin pumps and continuous glucose monitors, indicating that these modern advances in diabetes technologies are insufficient for addressing CVD risks in this population. Given that our results showing cardiovascular risks explain at least some of the racial and ethnic differences in arterial stiffness, further research to identify risk factor targets and treatment strategies that can improve cardiovascular profiles in youth‐onset T1D while limiting differences is warranted.

We did not find health behaviors, quality of clinical care, or perception of clinical care to explain any notable portion of the racial and ethnic differences in arterial stiffness. This is in contrast to a prior report that lifestyle factors did explain, in part, differences in CVD, particularly among men.²⁵ However, diet and activity behaviors are notably poor in youth and young adults with T1D,^{15, 32} despite increased clinical care relative to peers without diabetes. It is possible that in the late teens and early 20s years, poorer lifestyle behaviors have not yet translated into poorer cardiovascular risk profiles within a population with T1D; this may change as our participants age. While our study does not rule out the possibility that these factors could contribute to racial and ethnic differences in T1D, it suggests that focusing on socioeconomic factors and cardiovascular risks to reduce differences may be more appropriate.

We acknowledge that the data analyzed here explained only 40% of the variability in arterial stiffness, leaving 60% yet to be explained, including more than half of the apparent racial and ethnic differences. Other factors at play could include the impact of systemic racism and discrimination,³³ biological effects of adversity or other sources of excessive stress in childhood,^{34, 35} environmental exposures,³⁶ neighborhood factors,³⁷ or developmental programming induced by early life exposures,³⁸ or interactive effects that may operate differently across subgroups of participants. Future studies should aim to quantify the degree to which these factors exacerbate cardiovascular abnormalities in T1D, especially in disparate patterns, so that targeted interventions and clinical management strategies can be developed and implemented. Multifaceted programs that provide support for clinical and nonclinical challenges, including communication barriers, family conflict, and food insecurity, have shown some promise for improved outcomes in youth with T1D, and are likely needed on a broader scale.³⁹

Our work has some limitations. Analysis of hard outcomes of CVD rather than precursors is needed to confirm that the differences we observed at young ages have clinical significance as these participants age. However, understanding how such precursors develop and the modifiable factors that contribute to their manifestation is important for improving cardiovascular risk management in youth with T1D, for which evidence is limited. Our definitions of “ideal” modifiable factors were informed by clinical guidelines or research standards wherever possible, though somewhat arbitrary for some variables. Exploring different definitions of the modifiable factors, as well as expanded domains, can increase the precision of such estimates in future studies. However, our large sample size and notable diversity in participants with regard to identity, socioeconomic status, cardiovascular profiles, and other domains provided a rich data set to explore potential contributors to arterial stiffness and racial and ethnic differences therein in the context of youth‐onset T1D. We did not consider genetics in our analysis given our interpretation of race and ethnicity as social, not biological constructs;²⁷ future studies could examine how genetic ancestry is related to CVD in T1D⁴⁰ or how genetics influences PWV.⁴¹

In conclusion, we report that socioeconomic‐ and cardiovascular‐related factors explain a notable portion of racial and ethnic differences in arterial stiffness in young adults with T1D, especially between Hispanic and NHW individuals. However, NHB individuals experienced the most arterial stiffness, which was not explained by the factors examined here. A better understanding of the pervasive inequities in health care, personal resources, and social standing that could be driving these racial and ethnic differences is needed.

Sources of Funding

The SEARCH for Diabetes in Youth Cohort Study (1R01DK127208‐01, 1UC4DK108173) is funded by the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, and supported by the Centers for Disease Control and Prevention. The Population Based Registry of Diabetes in Youth Study (1U18DP006131, U18DP006133, U18DP006134, U18DP006136, U18DP006138, and U18DP006139) is funded by the Centers for Disease Control and Prevention (DP‐15‐002) and supported by the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention and the National Institute of Diabetes and Digestive and Kidney Diseases.

Disclosures

None.

Footnotes