Abstract
OBJECTIVE
Automated insulin delivery (AID) systems improve glycemic outcomes in youth with type 1 diabetes and are now the recommended mode of insulin delivery. Previous studies highlighted racial disparities in the use of continuous glucose monitoring and insulin pump therapy. The purpose of this study was to evaluate the use of AID systems and A1C outcomes in youth with type 1 diabetes by race.
RESEARCH DESIGN AND METHODS
This was a single-center cross-sectional study. We included youth and young adults with type 1 diabetes aged 2–21 years who had at least two clinic visits between December 2022 and December 2023. Demographics, diabetes device use, and A1C data were gathered from chart review, based on the latest office visit records available in the electronic medical record system for 2023.
RESULTS
Out of 668 youth aged 3.1–19.9 years with type 1 diabetes, 435 (65%) were AID users. The prevalence of AID use was 70% (341 of 483) in White youth compared with 47% (60 of 129) among Black youth and 62% (34 of 55) among youth of other racial groups (P <0.001). Black youth using AID achieved significantly lower A1C levels (median 8%, interquartile range [IQR] 7.5–8.8%) compared with Black youth who did not use an AID system (median 9.6%, IQR 8.1–11.6%) (P <0.001).
CONCLUSION
These findings support the persistence of racial disparities in diabetes technology utilization.
Graphical Abstract
Automated insulin delivery (AID) systems are now considered the standard of care for people with type 1 diabetes (1). Several clinical trials showed that consistent use of AID significantly reduced A1C levels by 0.4–0.7% (2–4), with larger improvement in A1C (0.91–1.18%) in individuals with a baseline A1C >8% (3). There are few data available on real-world use of AID systems and clinical outcomes in youth from diverse backgrounds.
KEY POINTS
This study assessed the prevalence of automated insulin delivery (AID) system use among youth with type 1 diabetes by race and analyzed associated A1C outcomes.
Black youth were less likely to use AID systems than non-Black youth. After adjusting for socioeconomic status and potentially confounding biological factors, the lower use of AID persisted in Black youth. Even when using an AID system, Black youth had higher A1C levels compared with their counterparts of other racial groups.
To promote equitable care, health care systems and clinical researchers should continue to explore interactions between racial social constructs and health care services and seek opportunities to close race-related care gaps and health disparities.
Several studies highlight disparities in the use of previous generations of insulin pumps and continuous glucose monitoring (CGM) systems (5–9) and identify barriers to uptake among youth from diverse backgrounds and low socioeconomic groups, such as limited access to health care and insurance coverage, negative perceptions of diabetes technology, and provider biases (10–12).
AID use has increased dramatically in the past 6 years, but, as of 2022, disparities have not decreased (13), and A1C remains higher in Black and Hispanic youth compared with White youth using AID (14). A recent study found that, even when Black youth used AID, they had lower glycemic time in range compared with their White counterparts using AID (15).
Recognizing that race serves as a proxy for structural and social experiences rather than biological differences, the social ecological model (SEM) of health (16) is a useful framework that teams can use to disentangle contextual contributors to disparities in health outcomes related to AID use, including race and socioeconomic status (SES). For example, barriers to AID initiation and continuation related to insurance coverage and uninterrupted device access (operating at the community/society level of the SEM) motivate different strategies and target audiences than do barriers to sustained AID use resulting from family cultural beliefs (operating at the interpersonal/community level of the SEM).
Given the interplay of social and environmental factors at interpersonal, institutional, community, and societal levels (17) that contribute to disparities in AID-related diabetes outcomes, the purpose of this study was to evaluate the prevalence of AID use and associated glycemic outcome in youth with type 1 diabetes by race and determine the extent to which SES influences those relationships.
Research Design and Methods
This study was conducted in a tertiary pediatric diabetes program in the Midwest region of the U.S. and administered at five distinct geographical locations within a city megapolis. The study was approved by the Institutional Review Board of the University Hospitals/Case Western Reserve University in Cleveland, OH (study ID: 20231414). We performed a cross-sectional chart review from the electronic medical record (EMR) system; some of the data had been transferred from a prior EMR system. We included youth and young adults with type 1 diabetes aged 2–21 years who had at least two clinic visits between December 2022 and December 2023, the year when three major commercially available AID systems became available. We gathered data on demographics, including age, sex, self-identified race, ZIP code of residence, and type of insurance. Clinical information was obtained from providers’ notes, including date of type 1 diabetes diagnosis, type of glucose monitoring and insulin delivery used, and, if applicable, duration and type of CGM system, insulin pump, or AID system used. Ascertainment of technology use and glycemic control was based on the latest office visit records available in the EMR for 2023. The primary outcome measure was the use of AID, analyzed by race. We evaluated the impact of SES, which was extrapolated from estimated income based on the ZIP code of residence. We also performed a sensitivity analysis using insurance as a proxy for SES. Secondary outcomes included the association between AID use and A1C.
Statistical Analysis
We used descriptive statistics to present the unadjusted effects of characteristics, with the frequency and proportion for categorical variables and median and interquartile range (IQR) for continuous variables. Statistical significance between the groups was tested using a Kruskal-Wallis rank sum test or a χ2 test, depending on the type of variable. We performed bivariate tests of association between the use of AID and other covariates, including age, sex, insurance type, ZIP code, median income, A1C, AID system type and start date, number of visits per year, and diabetes duration. Of note, data on AID use, insurance type, and ZIP code were collected for 100% of the youth included in the study.
To control for SES confounders when estimating the effect of race on AID usage, we applied propensity score weighting (Figure 1). This methodology is designed to reduce confounding in observational studies to emulate a clinical trial. The propensity score combines multivariate covariate information into a single metric, reduces confounding bias, and facilitates visual assessment of covariate balance (18). Propensity scores were estimated using an ensemble model from the R statistical software package SuperLearner, incorporating a diverse set of regression-based methods and machine learning algorithms. Covariates with <30% missing data—median income, sex, insurance type, age, number of visits, visits in the EMR system, A1C, and diabetes duration—were included. Covariate balance after weighting was assessed with standardized mean differences (SMDs). A marginal structural model was fitted for AID usage using the same ensemble model setting of propensity scores, incorporating race and the covariates from the propensity score model. Augmented inverse probability weighting was used to estimate average differences in AID usage across racial groups. Analyses were conducted in R, v. 4.4.4, statistical software, using its gtsummary, WeightIt, npcausal, SuperLearner, and survey packages.
Figure 1.
Standardized mean differences in covariates.
Results
We included a total of 668 youth (55% male) in the analysis (Table 1). Of those included, 72% were White, 19% were Black, and the remaining 8% were of other races. The median age for all groups was 14.8 years (IQR 11.8–17.5 years), and the age range was 3.1–19.9 years; there was no difference in median age among racial groups. The median diabetes duration for all cohorts was 6.3 years (IQR 3.8–9.6 years). Additionally, 59% of the individuals had private insurance.
Table 1.
Characteristics of the Study Population
| Characteristic | Overall (n = 668) | White (n = 484) | Black (n = 129) | Other Race* (n = 55) | P |
|---|---|---|---|---|---|
| Sex Female Male |
302 (45) 365 (55) |
211 (44) 272 (56) |
68 (53) 61 (47) |
23 (42) 32 (58) |
0.2 |
| Age, years | 14.8 (11.8–17.5) | 15.4 (12.3–17.8) | 13.5 (11.5–16.7) | 13.1 (9.8–15.2) | <0.001 |
| Income, $† | 62,700 (46,648–77,433) | 65,004 (53,775–80,068) | 42,711 (36,161–60,432) | 60,022 (44,331–68,071) | <0.001 |
| Insurance type Private Public |
396 (59) 272 (41) |
334 (69) 150 (31) |
34 (26) 95 (74) |
28 (51) 27 (49) |
<0.001 |
| Diabetes duration, years | 5.6 (3.1–8.9) | 5.8 (3.2–9.1) | 5.5 (3.3–8.2) | 3.5 (2.3–6.4) | 0.003 |
| AID use | 435 (65) | 341 (70) | 60 (47) | 34 (62) | <0.001 |
| A1C, %‡ | 7.40 (6.70–8.60) | 7.20 (6.60–8.10) | 8.60 (7.60–10.30) | 7.40 (6.70–8.95) | <0.001 |
| A1C, mmol/mol‡ | 57 (50–70) | 55 (49–65) | 70 (60–89) | 57 (50–74) | <0.001 |
Data are n (%) or median (IQR). Statistical significance was determined using a Pearson χ2 test or Kruskal-Wallis rank sum test.
*Other races included multiracial (n = 16 [29%]), Asian (n = 3 [5%]), and unknown race (n = 36 [66%]).
†Median income estimated based on ZIP code.
‡One patient had missing A1C data and was not included.
Cross-sectional analysis showed that AID was used by 435 (65%) of the youth with type 1 diabetes at their last appointment in 2023. The most common type of AID insulin pump was tubeless (63%), followed by a tubed AID system (35%). Youth using AID had significantly lower A1C levels compared with non-users of AID, with median A1C of 7.1% (IQR 6.6–7.9%) and 8.4% (IQR 7.2–10.0%), respectively (P <0.001).
Racial Disparities in AID Use
We found significant race-based disparities in AID use in our program; less than half (47%) of the Black youth used AID versus 70% of the White youth and 61% of youth of other races (P <0.001) (Table 1).
To address the effect of SES on the racial disparity in AID use, we used propensity score weighting. Figure 1 displays the unbalanced/balanced distribution of covariates between races based on SMDs. After propensity score weighting, we observed balanced distribution of age, sex, household income, insurance type, and diabetes duration among the racial groups. Nevertheless, the adjusted absolute risk difference between groups remained statistically significant. The estimated difference in AID use was 6.9% lower in Black youth compared with White youth (95% CI 11.6 to –2.2%, P = 0.004) and 4.6% lower compared with youth of other races (95% CI –8.6 to –1.0%, P = 0.026). There was no significant difference in AID use between White youth and youth of other races.
We also performed a sensitivity analysis without median income and using only insurance as an individual level proxy for SES, potentially influencing AID use, which did not significantly change the outcome; the difference in AID use between Black and White youth was −11.6% (95% CI −22.4 to −1.0%, P = 0.034), and the difference was −11.7% (95% CI −26.2 to 2.8%, P = 0.115) between Black youth and youth of other races. There was no significant difference in AID use between White youth and youth of other races.
Glycemic Outcomes and Pump Preferences by Racial Group
Based on unadjusted data, Black youth had a higher median A1C than White youth in the AID group (8% [IQR 7.5–8.8%] vs. 7% [IQR 6.5–7.6%], P <0.001) and in the non-AID group (9.6% [IQR 8.1–11.6%] vs. 8% [IQR 6.9–9.1%], P <0.001) (Table 2). Black youth using AID had a significantly lower median A1C compared with Black youth who were non-AID users (8% [IQR 7.5–8.8%] vs. 9.6% [IQR 8.1–11.6%], P <0.001). The mean difference in A1C was 1.05% (P <0.001), even after adjustment for SES factors. Seventy-three percent of the Black youth (44 of 60) and 60% of the White youth (204 of 327) used a tubeless AID system.
Table 2.
A1C Levels in Youth With Type 1 Diabetes of Different Racial Groups by AID Status
| Race | A1C of AID Users, % | A1C of AID Nonusers, % | P |
|---|---|---|---|
| Black | 8.0 (7.5–8.8) | 9.6 (8.1–11.6) | <0.001 |
| White | 7.0 (6.5–7.6) | 8.0 (6.9–9.1) | <0.001 |
| Other* | 7.0 (6.4–7.8) | 8.4 (7.6–9.3) | <0.001 |
Data are median (IQR).
*Other races included multiracial (n = 16 [29%]), Asian (n = 3 [5%]), and unknown race (n = 36 [66%]).
Discussion
Our study evaluated the prevalence of AID use among different racial groups from a single institution in 2023, the first full year with commercial availability of three different AID systems in the United States. Our results revealed significantly lower use of AID and higher A1C levels among Black youth with type 1 diabetes, even after accounting for differences in SES, as proxied by income estimates based on ZIP code or health insurance. These findings with AID use are consistent, with previous evidence showing disparities in CGM and insulin pump use between Black and non-Black youth (5–9), persisting even after adjusting for SES (9,13).
We used extensive statistical modeling techniques to perform propensity weighting, which allowed us to show that the racial differences in AID use persisted after the effects of biological and SES factors were statistically eliminated. Persistent statistically significant differences in AID use between Black and White youth suggest the influence of structural inequities operating at multiple levels of the SEM (17). Applying the SEM, this could include “gatekeeping” at the institutional level, where arbitrary rules are set regarding who is a “good candidate” to use diabetes technologies, despite data showing that those with highest A1C have the largest improvement with technology use (9,15). The health care team-patient relationship is within the interpersonal level and can be affected by medical mistrust, lack of discussion about AID, and provider biases (11). At the community level, there may not be as much familiarity with diabetes devices, because these devices are less likely to be discussed, prescribed, or used by Black individuals (8,10), leading youth with type 1 diabetes to feel “othered” even within their community and therefore not use AID. Lower familiarity with technology in Black communities highlights the interplay of social and structural experiences at other levels mentioned above (8,10,19). We also must consider that other unmeasured factors also contribute (20), such as food and housing insecurity (21), digital inclusion (22), and school nurse support (23). There are ongoing efforts focused on health system navigation support (24) and community-based interventions (25) to meet the unique needs of diverse groups.
To our knowledge, this is one of the first pediatric reports about racial disparities in AID use in the United States. Prior studies showed racial disparities in the adoption of earlier AID options through 2022 (26), including a large population-level dataset from the T1D Exchange Quality Improvement Collaborative (13,14). Marks et al. (15) reported lower initiation of tubeless AID systems among Black youth and publicly insured individuals. They also found that AID use was associated with the greatest glycemic outcome changes in individuals with higher baseline A1C levels and lower initial glycemic time in range. Our study incorporates more recent data and a wider range of commercially available AID systems, including both tubed and tubeless options.
Strengths and Limitations
Our study’s strength is its large and diverse cohort representing a diverse mix of social and racial groups. The retrospective design and reliance on EMR documentation is a limitation of the study. However, we collected data for AID use, insurance, and ZIP codes for 100% of included youth.
Conclusion
Our study highlights racial disparities in the use of state-of-the-art diabetes technologies, which are shown to improve glycemic control while also decreasing the burden of diabetes management. This finding calls for action by all parties involved in diabetes care delivery and policy development. Researchers play a key role in exploring the underlying causes of these disparities and identifying solutions to address modifiable factors. Health care systems must continue to collect and analyze patient data, collaborate with community resources, implement targeted programs, and ensure fair reimbursement for services to promote equity. Clinicians should receive training on disparities and inclusive, person-centered care to reduce implicit bias. Together, these groups share responsibility for driving meaningful changes for equitable care.
Funding Statement
The work of E.G.Y. is supported by the Fellowship Research Award Program of Rainbow Babies and Children’s Foundation, Cleveland, OH.
Duality of Interest
S.A.M. has received speaking fees from Insulet and nonfinancial research support from Dexcom and Insulet. No other potential conflicts of interest relevant to this article were reported.
Author Contributions
E.G.Y. participated in writing the manuscript, creating the graphical abstract, and managing the project. K.T. participated in conceptualizing the study; developing the methodology; writing, reviewing, and editing the manuscript; creating the graphical abstract; and supervising the project. S.R. participated in developing the methodology, reviewing and editing the manuscript, and supervising the project. Y.Z. participated in developing the methodology, conducting the formal analysis of data, and reviewing and editing the manuscript. S.A.M. participated in conceptualizing the study; developing the methodology; writing, reviewing, and editing the manuscript; and supervising the project. E.G.Y. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
References
- 1. American Diabetes Association Professional Practice Committee . 14. Children and adolescents: Standards of Care in Diabetes—2024. Diabetes Care 2024;47(Suppl. 1):S258–S281 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Breton MD, Kanapka LG, Beck RW, et al. ; iDCL Trial Research Group . A randomized trial of closed-loop control in children with type 1 diabetes. N Engl J Med 2020;383:836–845 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Brown SA, Forlenza GP, Bode BW, et al. ; Omnipod 5 Research Group . Multicenter trial of a tubeless, on-body automated insulin delivery system with customizable glycemic targets in pediatric and adult participants with type 1 diabetes. Diabetes Care 2021;44:1630–1640 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Carlson AL, Sherr JL, Shulman DI, et al. Safety and glycemic outcomes during the MiniMedTM advanced hybrid closed-loop system pivotal trial in adolescents and adults with type 1 diabetes. Diabetes Technol Ther 2022;24:178–189 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Everett EM, Wright D, Williams A, et al. A longitudinal view of disparities in insulin pump use among youth with type 1 diabetes: the SEARCH for Diabetes in Youth study. Diabetes Technol Ther 2023;25:131–139 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Tremblay ES, Bernique A, Garvey K, Astley CM.. A retrospective cohort study of racial/ethnic and socioeconomic disparities in initiation and meaningful use of continuous glucose monitoring among youth with type 1 diabetes. J Diabetes Sci Technol 2024;18:1433–1444 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Baboun D, Solano N, Del Toro V, Alvarez-Salvat R, Granados A, Carrillo-Iregui A.. Technology use and clinical outcomes in a racial-ethnic minority cohort of children and adolescents with type 1 diabetes. J Pediatr Endocrinol Metab 2023;36: 1128–1132 [DOI] [PubMed] [Google Scholar]
- 8. Agarwal S, Schechter C, Gonzalez J, Long JA.. Racial–ethnic disparities in diabetes technology use among young adults with type 1 diabetes. Diabetes Technol Ther 2021;23:306–313 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lipman TH, Smith JA, Patil O, Willi SM, Hawkes CP.. Racial disparities in treatment and outcomes of children with type 1 diabetes. Pediatr Diabetes 2021;22:241–248 [DOI] [PubMed] [Google Scholar]
- 10. Agarwal S, Crespo-Ramos G, Long JA, Miller VA.. “I didn’t really have a choice”: qualitative analysis of racial-ethnic disparities in diabetes technology use among young adults with type 1 diabetes. Diabetes Technol Ther 2021;23:616–622 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tanenbaum ML, Adams RN, Hanes SJ, et al. Optimal use of diabetes devices: clinician perspectives on barriers and adherence to device use. J Diabetes Sci Technol 2017;11: 484–492 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Mencher SR, Weinzimer SA, Nally LM, Van Name M, Nunez-Smith M, Sadler LS.. Technology utilization in Black adolescents with type 1 diabetes: exploring the decision-making process. Diabetes Technol Ther 2022;24:249–257 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ebekozien O, Mungmode A, Sanchez J, et al. Longitudinal trends in glycemic outcomes and technology use for over 48,000 people with type 1 diabetes (2016–2022) from the T1D Exchange Quality Improvement Collaborative. Diabetes Technol Ther 2023;25:765–773 [DOI] [PubMed] [Google Scholar]
- 14. Noor N, Kamboj MK, Triolo T, et al. Hybrid closed-loop systems and glycemic outcomes in children and adults with type 1 diabetes: real-world evidence from a U.S.-based multicenter collaborative. Diabetes Care 2022;45:e118–e119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Marks BE, Meighan S, Zehra A, et al. Real-world glycemic outcomes with early Omnipod 5 use in youth with type 1 diabetes. Diabetes Technol Ther 2023;25:782–789 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. McLeroy KR, Bibeau D, Steckler A, Glanz K.. An ecological perspective on health promotion programs. Health Educ Q 1988;15:351–377 [DOI] [PubMed] [Google Scholar]
- 17. Addala A. Making a good thing even better: expanding access and applicability of automated insulin delivery systems to benefit all youth with type 1 diabetes. Diabetes Care 2023;46:2126–2128 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Chen JW, Maldonado DR, Kowalski BL, et al. Best practice guidelines for propensity score methods in medical research: consideration on theory, implementation, and reporting: a review. Arthroscopy 2022;38:632–642 [DOI] [PubMed] [Google Scholar]
- 19. Kanbour S, Jones M, Abusamaan MS, et al. Racial disparities in access and use of diabetes technology among adult patients with type 1 diabetes in a U.S. academic medical center. Diabetes Care 2023;46:56–64 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Willi SM, Miller KM, DiMeglio LA, et al. ; T1D Exchange Clinic Network . Racial-ethnic disparities in management and outcomes among children with type 1 diabetes. Pediatrics 2015;135:424–434 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Vitale M, Dorado L, Pais V, Sidani S, Gucciardi E.. Food insecurity screening among families of children with diabetes. Diabetes Spectr 2019;32:338–348 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Hölgyesi Á, Luczay A, Tóth-Heyn P, et al. The impact of parental electronic health literacy on disease management and outcomes in pediatric type 1 diabetes mellitus: cross-sectional clinical study. JMIR Pediatr Parent 2024;7:e54807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Tonyushkina K, Cobb V, Lawson G, et al. Challenges and opportunities of engaging school nurses in a diabetes care team in a culturally diverse low-income community: a mixed-methods feasibility study. Health Behav Policy Rev 2021;8:465–476 [Google Scholar]
- 24. Pemberton J, Collins L, Drummond L, et al. Enhancing equity in access to automated insulin delivery systems in an ethnically and socioeconomically diverse group of children with type 1 diabetes. BMJ Open Diabetes Res Care 2024;12:e004045 [Google Scholar]
- 25. Lai CW, Craven M, Hershey JA, Lipman TH, Hawkes CP.. Adverse social determinants of health in children with newly diagnosed type 1 diabetes: a potential role for community health workers. Pediatr Diabetes 2024;2024:8810609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Conway RB, Gerard Gonzalez A, Shah VN, et al. Racial disparities in diabetes technology adoption and their association with HbA1c and diabetic ketoacidosis. Diabetes Metab Syndr Obes 2023;16:2295–2310 [DOI] [PMC free article] [PubMed] [Google Scholar]