Disparities in Hemoglobin A_1c Testing During the Transition to Adulthood and Association With Diabetes Outcomes in Youth-Onset Type 1 and Type 2 Diabetes: The SEARCH for Diabetes in Youth Study

The troubling increase in youth-onset type 1 and type 2 diabetes (1) is compounded by disparities in glucose control (2) and a high prevalence of complications and comorbidities in the first decade following diagnosis (3). In addition to self-monitoring of blood glucose levels, provider monitoring of glucose control over time is believed to facilitate individualized care and treatment adjustments to reach management goals. Accordingly, the American Diabetes Association recommends hemoglobin A_1c (HbA_1c) testing at least every 6 months for individuals meeting glycemic goals (generally <7.0%) and every 3 months for individuals not meeting goals, with recent medication changes, and <18 years with type 2 diabetes (4,5).

However, only one-half of youth and young adults with youth-onset type 1 or type 2 diabetes report ≥3 HbA_1c tests/year (6,7), despite notably poor glycemic control (3,8). Testing frequency appears to decrease with age in the first 5 years postdiagnosis, especially among lower-income individuals (7). This is concerning given evidence that more frequent HbA_1c testing is associated with lower HbA_1c levels in youth-onset type 1 diabetes (6,9) and adult-onset type 2 diabetes (10). Maintaining glycemic control throughout the course of youth-onset diabetes is critical, as lifetime risk of complications is greater than adult-onset because of a longer duration of disease (11).

We examined factors potentially related to infrequent HbA_1c testing in youth-onset type 1 and type 2 diabetes at an average of 8 and 13 years postdiagnosis, a time when many individuals are transitioning to adulthood with an increased responsibility for their own health and health care. We considered socioeconomic and clinical factors with the potential to inform strategies for increasing testing frequency and, ultimately, successful disease management. To confirm the relevance of HbA_1c testing for diabetes-related outcomes during this transitional period, we also examined testing frequency in relation to HbA_1c levels and microvascular complications.

The SEARCH for Diabetes in Youth (SEARCH) study population and protocol have been described previously (12). Briefly, SEARCH identifies individuals newly diagnosed with any type of diabetes before age 20 years through a population-based registry network covering 5.5 million youth from five sites in the U.S.: South Carolina; Cincinnati, Ohio, and surrounding counties; Colorado with southwestern U.S. American Indian sites; Seattle, Washington, and surrounding counties; and Kaiser Permanente Southern California membership in seven counties. Individuals diagnosed in 2002–2006 or 2008 were invited to join an observational study of the natural course of youth-onset diabetes. Baseline visits were conducted shortly after diagnosis (2002–2010: n=4,095, mean age 11.0 years, mean diabetes duration 9.9 months), with a first follow-up visit conducted ≥5 years after diagnosis (2011–2015: n=2,777, mean age 17.9 years, mean diabetes duration 8.0 years) and a second follow-up visit conducted ≥8 years after diagnosis (2015–2019: n=2,668, mean age 21.5 years, mean diabetes duration 11.2 years). The study was approved by the institutional review boards with jurisdiction in each study location. Written informed consent and/or assent were obtained from all participants and parents, as appropriate.

We abstracted date of birth and date of diagnosis from medical records and calculated age of diagnosis and diabetes duration for each visit. At the baseline visit, parents reported child sex, race, ethnicity, highest parental education, and annual household income. At follow-up visits, participants (or parents if the participant’s age was <14 years) self-reported their diabetes treatment regimen, glucose self-monitoring practices, and number of HbA_1c tests in the prior year using categories of 0, 1, 2, or ≥3. Further, participants (or parents if participant’s age was <18 years) reported type of diabetes provider, type of insurance, insurance status in the prior year, problems with cost of diabetes care in the prior year, occurrence of hypoglycemia events requiring hospitalization in the prior year, and meetings with a diabetes nurse or diabetes educator in prior year. Finally, participants were asked about the provider-patient relationship as follows: “A personal doctor or nurse is the health care provider who knows you best. This can be a general doctor, a specialist doctor, a nurse practitioner, or a physician assistant. Do you have one person you think of as your personal diabetes doctor or nurse?” At the second follow-up, the question was modified to ask about a personal diabetes doctor. Blood and urine samples were collected after an overnight, ≥8-h fast with no medications taken the morning of the visit (including short-acting insulin). Participants were asked to bring a first morning urine void; when not provided (∼8% of participants), a spot urine specimen was collected. Blood samples were analyzed for HbA_1c and urine for albumin, creatinine, and cystatin C at the Northwest Lipid Metabolism and Diabetes Research Laboratory (Seattle, WA).

Complications were ascertained at both follow-up visits. Diabetic kidney disease was defined as moderate albuminuria (≥30 μg/mg of creatinine) or low glomerular filtration rate (≤60 mL/min/1.73 m² as estimated by the Chronic Kidney Disease Epidemiology equation with serum creatinine and cystatin C) (13). Diabetic retinopathy was assessed with 45° color digital fundus images taken by trained research staff using a nonmydriatic camera (Visucam Pro N; Carl Zeiss Meditech), centered on the disc and macula of both eyes. Photos masked to all clinical characteristics were graded by the Wisconsin Ocular Epidemiology Reading Center. Diabetic retinopathy was defined as mild, moderate, or proliferative retinopathy (codes of 30–80) in at least one eye (14). Our threshold for diagnosing retinopathy was set higher than typically used clinically to distinguish diabetic retinopathy from milder forms that may have an etiology other than diabetes retinopathy. Peripheral neuropathy was defined as a score >2 on the Michigan Neuropathy Screening Instrument (15), as described previously (16).

Analyses were conducted separately by diabetes type. We included SEARCH participants aged ≥10 years at the first follow-up visit with a provider diagnosis of type 1 (excluding 19 reporting no insulin use) or type 2 diabetes and data for HbA_1c testing frequency, HbA_1c levels and/or microvascular complications, and all socioeconomic and clinical variables of interest (Supplementary Fig. 1). We conducted a cross-sectional analysis in participants with complete data from the first follow-up (n=1,885 type 1 and n=230 type 2) and a longitudinal analysis in the subset of participants from the cross-sectional analysis also having sufficient data from the second follow-up (n=649 type 1 and n=84 type 2). We dichotomized HbA_1c testing frequency as reporting three or more versus fewer than three times in the prior year. For the cross-sectional analysis, we classified participants as reporting ≥3 versus <3 tests/year in the year prior to the first follow-up. For the longitudinal analysis, we classified participants as reporting ≥3 tests in the year prior to both visits versus <3 tests in the year prior to either visit.

We examined socioeconomic and clinical factors associated with HbA_1c testing using logistic regression, modeling the odds of reporting ≥3 tests/year at the first follow-up and the odds of reporting ≥3 tests/year at both follow-ups. Models included core variables of age at diagnosis, diabetes duration at the visit, sex, race/ethnicity, SEARCH clinical site, and (for the longitudinal analysis only) time interval between the follow-up visits. Additional socioeconomic and treatment-related variables were considered via stepwise selection, whereby variables were retained in the model only when statistically significant (P < 0.05). Because of the small sample size in the type 2 diabetes longitudinal analysis, all core and candidate variables were subject to stepwise selection. Candidate variables included parental education at baseline, household income at baseline, diabetes treatment, diabetes care provider, having a personal doctor, having problems with the cost of care in the prior 12 months, type of health insurance, and having continuous health insurance in the prior year. For the longitudinal models, levels of the candidate variables reflected status over time (e.g., having a consistent regimen vs. a change in regimen in diabetes treatment between visits).

We evaluated HbA_1c testing frequency in relation to diabetes outcomes using linear regression for HbA_1c levels and logistic regression for microvascular complications. For complications, we used a composite variable reflecting the manifestation of one or more microvascular complications. The cross-sectional analysis modeled the HbA_1c level or odds of having one or more complications at the first follow-up among those reporting ≥3 tests/year at the first follow-up relative to those reporting <3 tests/year. The longitudinal analysis modeled the HbA_1c level or odds of having one or more complications at the second follow-up among those reporting ≥3 tests/year at either follow-up relative to those reporting <3 tests/year at both follow-up visits. For both analyses, we adjusted for covariates sequentially as follows: Model 1 was unadjusted, model 2 was adjusted for core variables, and model 3 was additionally adjusted for candidate variables that passed stepwise selection for the analysis of HbA_1c tests per year as well as for the variables of frequency of glucose self-monitoring and receipt of diabetes education. Sociodemographic and clinical characteristics are presented as mean (SD) or count (%). Linear and logistic regression model estimates are presented as β and odds ratio (OR), respectively, with 95% CI. All analyses were conducted with a two-sided α of 0.05 for statistical significance.

Participant characteristics of the analytic cross-sectional sample (n=1,885 type 1 and n=230 type 2) and longitudinal subset (n=649 type 1 and n=84 type 2) are presented in Tables 1 and 2. On average, the first follow-up occurred at 8.0 years diabetes duration (range 3.3–13.0 years) and the second follow-up at 12.6 years’ duration (range 7.8–17.4 years). These descriptive data suggest a decline in quality of diabetes care and management over time, which was affirmed in the longitudinal subset (Supplementary Table 1).

Socioeconomic and clinical factors significantly related to HbA_1c testing frequency cross sectionally at 8 years diabetes duration are presented in Fig. 1. Participants with type 1 diabetes were less likely to report ≥3 HbA_1c tests/year if they had been older at diagnosis (OR 0.88 [95% CI 0.86–0.91] per 1-year increase), had a greater duration of diabetes (OR 0.85 [0.80–0.90] per 1-year increase), did not have a personal doctor (OR 0.46 [0.36–0.59]), did not have continuous health insurance in the prior year (OR 0.64 [0.43–0.96]), were not seeing an endocrinologist for their diabetes care (OR 0.71 [0.55–0.92]), had problems with the cost of care in the prior year (OR 0.78 [0.64–0.97]), or were not using an insulin pump (OR 0.67 [0.54–0.84]). Odds of reporting ≥3 tests/year also differed by household income (P=0.02), although only the comparison between higher income households and those not reporting their income (i.e., declined to report or did not know) was statistically significant (OR 1.68 [1.08–2.62] for those without income data). No differences in HbA_1c tests per year were observed by sex or race/ethnicity for participants with type 1 diabetes (data not shown). Participants with type 2 diabetes were less likely to report ≥3 tests/year when they were not seeing an endocrinologist for their diabetes care (OR 0.18 [0.08–0.41]), were using noninsulin medications (OR 0.33 [0.15–0.73] vs. insulin), or were not using medication for their diabetes treatment (OR 0.06 [0.01–0.32] vs. insulin). No differences in HbA_1c testing frequency were observed by age at diagnosis, diabetes duration, sex, or race/ethnicity for type 2 diabetes (data not shown).

Sociodemographic and clinical factors significantly associated with HbA_1c testing frequency longitudinally from 8 to 13 years diabetes duration are presented in Fig. 2. Participants with type 1 diabetes were less likely to report ≥3 tests/year across this period if they were older at diagnosis (OR 0.91 [95% CI 0.88–0.95] per 1-year increase), had a greater duration of diabetes (OR 0.90 [0.82–0.99] per 1-year increase), did not have a personal doctor at one or more visits (OR 0.44 [0.30–0.65]), or did not have continuous health insurance in the year prior to one or more visits (OR 0.51 [0.27–0.96]). Odds of reporting ≥3 tests/year over time also differed by household income (P=0.01), although only the comparison between higher income households and those not reporting their income (i.e., declined to report or did not know) was statistically significant (OR 2.06 [1.03–4.11] for those without income data). No differences were observed in HbA_1c tests per year over time by sex or race/ethnicity for type 1 diabetes (data not shown). For type 2 diabetes, where all core and candidate variables were subject to stepwise selection because of the small sample size, only type of diabetes provider was retained in the model. Participants were less likely to report ≥3 tests/year over time when they were not seeing an endocrinologist for their diabetes care at one or more visits (OR 0.19 [0.06–0.63]). Sensitivity analyses using ordinal HbA_1c testing categories (0, 1, 2, or ≥3 tests/year) yielded similar results (Supplementary Table 2).

Associations of frequency of HbA_1c testing with HbA_1c levels and microvascular complications cross sectionally at 8 years diabetes duration are presented in Supplementary Fig. 2. For type 1 diabetes, participants reporting ≥3 HbA_1c tests/year had significantly lower HbA_1c levels than those reporting <3 tests/year (unadjusted model 1: 9.03% vs. 9.36%, P < 0.001; adjusted model 3: 9.67% vs. 9.83%, P=0.07)). In model 3, covariates that were statistically significant included age at diagnosis (β=−0.05 [95% CI −0.07 to −0.03] per 1-year increase), sex (β=0.27 [0.12–0.43] for females vs. males), race/ethnicity (overall P < 0.001; only non-Hispanic Black vs. non-Hispanic White, β=1.24 [0.97–1.52], was statistically significant), not using an insulin pump (β=0.59 [0.42–0.76]), household income <$50,000/year (β=0.32 [0.14–0.49] vs. ≥$50,000/year), not having a personal doctor (β=0.29 [0.09–0.49]), and glucose self-monitoring <4 times/day (β=0.61 [0.45–0.78]). Participants who reported ≥3 tests/year also had a lower odds of any microvascular complication (unadjusted model 1: OR 0.53 [95% CI 0.41–0.69]), but this was attenuated to nonsignificance after covariate adjustment (adjusted model 2: OR 0.82 [0.62–1.09]; additionally adjusted model 3: OR 0.97 [0.72–1.31]). In model 2, covariates that were statistically significant included age at diagnosis (OR 1.13 [1.09–1.17] per 1-year increase) and duration of diabetes (OR 1.34 [1.25–1.45] per 1-year increase). In model 3, covariates that were statistically significant included age at diagnosis (OR 1.11 [1.07–1.15]), duration of diabetes (OR 1.33 [1.23–1.43]), using an insulin pump (OR 0.67 [0.49–0.90]), and not having a personal doctor (OR 1.59 [1.16–2.18]). For type 2 diabetes, no significant association was observed between HbA_1c testing and HbA_1c levels or microvascular complications at 8 years’ duration in any model. Again, sensitivity analyses using ordinal HbA_1c testing categories (0, 1, 2, or ≥3 tests/year) yielded similar results (Supplementary Table 3).

Associations of HbA_1c testing longitudinally with HbA_1c levels and microvascular complications at 13 years diabetes duration are presented in Supplementary Fig. 2. For type 1 diabetes, participants reporting ≥3 tests/year at both visits had lower HbA_1c levels than those reporting <3 tests/year at either or both visits, although the comparison was statistically significant only in model 2 (8.78% vs. 9.13%, P=0.02). Participants reporting ≥3 tests/year also had a lower odds of any microvascular complication (unadjusted model 1: OR 0.51 [95% CI 0.35–0.75]; adjusted model 2: OR 0.64 [0.43–0.97]), but this was attenuated to nonsignificance after additional adjustment (model 3: OR 0.70 [0.46–1.07]). In model 3, covariates that were statistically significant included age at diagnosis (OR 1.06 [1.01–1.12]), duration of diabetes (OR 1.29 [1.16–1.44]), race/ethnicity (overall P=0.05; only non-Hispanic Black vs. non-Hispanic White, OR 2.23 [1.25–3.99] was statistically significant), noncontinuous health insurance (OR 2.15 [1.22–3.79]), and glucose self-monitoring <4 times/day (OR 1.78 [1.19–2.66]). For type 2 diabetes, HbA_1c testing over time was not observed to be related to HbA_1c levels or microvascular complications at 13 years’ duration in any model. For microvascular complications, models 2 and 3 did not converge (no data shown).

Our data demonstrate disparities in HbA_1c testing frequency among individuals with youth-onset type 1 and type 2 diabetes, predominately according to health care–related factors. We previously reported that participant-reported receipt of three or more HbA_1c tests per year and screening for complications at 6 years’ duration of youth-onset diabetes occurred less frequently among young adults, those from lower-income households, and those receiving diabetes care from general practitioners (7). Our present analysis extends this work to 8 and 13 years postdiagnosis, when individuals with youth-onset diabetes are increasingly responsible for their own health care (17). Again, HbA_1c testing decreased with age and diabetes duration in type 1 diabetes, aligning with national reports of notably poor adherence to diabetes care goals in young adults (18). Considered alongside the other correlates of HbA_1c testing we identified, this is likely due to a reduction in clinic visits driven by multiple health care–related factors.

For both diabetes types, not receiving care from an endocrinologist reduced odds of reporting ≥3 HbA_1c tests/year. Endocrinologists are most familiar with diabetes management recommendations and more likely to be in clinics where quarterly visits with HbA_1c testing (point of care or venipuncture) is standard care (19). Yet, fewer participants reported seeing an endocrinologist after 13 years of diabetes duration: only two in three individuals with type 1 diabetes and one in four individuals with type 2 diabetes were consistently receiving care from an endocrinologist. Others have similarly reported that ∼70% of young adults with type 1 diabetes visit an endocrinologist in a 1-year period, which further decreases with age (20). Barriers to specialty care for young adults with type 1 diabetes include inconvenient scheduling, perceived lack of need or benefit of specialty care, and prior negative experiences (21). Adults with type 2 diabetes are typically treated by general practitioners (22,23), with the more severe cases (i.e., higher HbA_1c levels) more frequently referred to specialists (24). Yet, the more aggressive nature of youth-onset type 2 diabetes relative to adult onset (25) may warrant specialty care for most individuals. Further research into the key factors limiting specialty care for both types of youth-onset diabetes would be beneficial to improve access. Given projected increases in incidence, it is equally important to equip general practitioners to successfully manage these individuals (26).

Diabetes treatment was also a strong correlate of HbA_1c testing for both types of diabetes. Individuals with type 1 diabetes not using an insulin pump were 31% less likely to have ≥3 tests/year than their peers who use a pump, with pump use likely indicating access to higher quality of care and resources to cover costs. This disparity is even more pronounced in type 2 diabetes, where individuals not using insulin were 64–93% less likely to report ≥3 tests/year than their peers who use insulin. This may be explained by the increased number of clinic visits and, therefore, more opportunities for HbA_1c testing among individuals who are using insulin compared with those successfully managing type 2 diabetes with noninsulin medications or diet/exercise alone. However, this likely applies to few of our participants with type 2 diabetes given that ≥60% warranted quarterly testing based on HbA_1c levels and/or age. While we interpret the absolute ORs with caution given the wide CIs, the pattern of results suggests that individuals who receive more intensive diabetes treatment also receive increased glycemic monitoring, with both factors likely facilitating better outcomes.

For type 1 diabetes, not having a personal doctor had the greatest negative impact on HbA_1c testing, suggesting that individuals who do not believe that any single provider knows them best may be less motivated to pursue clinic visits. Alternatively, this finding may simply reflect that individuals who are not receiving regular care (and thereby HbA_1c testing) do not have the opportunity to identify a provider who knows them best. This may be particularly relevant during the transition from pediatric to adult care, wherein a reduction in clinic visits often occurs, but structured transition programs may increase visits and improve HbA_1c (27). Participants reporting lapses in health insurance coverage or problems with the cost of care were also less likely to report ≥3 tests/year, affirming reports that individuals with diabetes receive less care during insurance gaps or financial hardship (24,28–30). Interestingly, neither parental education nor household income had clear relationships with HbA_1c testing, indicating that affordable health care and/or health insurance have greater implications for regular diabetes care than more general indicators of socioeconomic status. Given evidence that affordable health care and insurance can become more accessible to low-income individuals (31), our study identifies potential intermediate targets that may facilitate access to regular diabetes care. While these “cost-of-care” factors are likely interrelated, they each exhibited independent effects on HbA_1c testing frequency in our analysis, highlighting the substantial barriers faced by low-resourced individuals with type 1 diabetes. We note that race/ethnicity itself was not related to HbA_1c testing, but there are known racial/ethnic disparities in the health care–related factors that were related to testing frequency disparities (32–34). Taken together, these findings suggest that eliminating disparities in such health care factors may be a potential strategy for addressing racial/ethnic disparities in diabetes outcomes.

HbA_1c testing ≥3 times/year, in turn, was associated with a 0.16–0.36% decrease in HbA_1c levels and a 36–49% reduced odds of microvascular complications. A similarly sized shift in HbA_1c levels was observed with more frequent HbA_1c testing over 1–2 years in German and Austrian young adults with type 1 diabetes (35). However, in our study, these associations were attenuated after adjustment for the socioeconomic and clinical factors found to be related to HbA_1c testing frequency. We interpret these sequential results as indicating that HbA_1c testing ≥3 times/year is a proxy for receipt of frequent high-quality care, which, in turn, depends on access. When access to HbA_1c testing is threatened, such as by a lapse in health insurance or problems with the cost of care, diabetes management is impaired, with potentially long-term consequences.

In contrast, limited HbA_1c testing was not observed to be associated with HbA_1c levels or microvascular complications in type 2 diabetes. As a prior study of nearly 200,000 older adults (mean age >60 years) reported improved HbA_1c levels following quarterly testing (36), we note that our analysis was likely underpowered with <250 participants. Alternatively, it may have been confounded by more severe (uncontrolled) cases of diabetes having poorer outcomes, even with more intensive clinical monitoring, while less severe cases could achieve better outcomes with more intensive clinical monitoring. Nonetheless, it remains concerning that nearly 60% of participants reported <3 HbA_1c tests/year over time despite average HbA_1c levels >9%. We observed increases in mean HbA_1c over time, as has been reported previously (37), which may be the result of treatment failure in highly aggressive youth-onset type 2 diabetes (38). Diabetes provider type and treatment type, but not sex, age, race/ethnicity, and socioeconomic factors, were related to HbA_1c testing frequency, suggesting that interventions targeting providers rather than patients may be needed to increase testing and, ultimately, quality of care. Specifically, general practitioners who more frequently treat adult-onset type 2 diabetes may need education on the aggressive nature of youth-onset type 2 diabetes and support for adjusting care processes appropriately for younger individuals.

Our study has limitations and strengths. Our analysis was limited by the small sample size for type 2 diabetes, as noted above. We were also unable to examine microvascular complications individually because of relatively low prevalence. We estimated the number of participants warranting quarterly testing based on research study visit HbA_1c ≥7% and/or age <18 years (just for type 2 diabetes) but focused our analysis on HbA_1c testing ≥3 times/year because we did not have data on clinical HbA_1c or medication changes in the full year prior to the SEARCH visit (which also informs testing frequency [4,5]). Less frequent testing may have been appropriate for participants meeting treatment goals. We also did not have information on number of clinic visits in the prior year and, thus, cannot draw conclusions on whether HbA_1c testing was low because clinic visits overall were low or whether testing was not done at clinic visits. Misclassification of testing frequency may have been a result of reliance on participant report, although patient recall of HbA_1c testing correlates well with medical record data (39,40). We included self-monitoring of blood glucose as a covariate, with testing ≥4 times/day (including continuously) serving as a proxy for a higher level of engagement with diabetes self-management. Further study of how continuous glucose monitoring specifically (vs. more or fewer tests per day) is related to HbA_1c levels and complications would be warranted in a future study. Key strengths of this analysis include the large sample size for type 1 diabetes, the racial/ethnic and socioeconomic diversity for both types, clinical assessments of microvascular complications, consideration of socioeconomic and clinical factors, and the longitudinal design.

In conclusion, our study affirms the recommendations for frequent HbA_1c testing in type 1 diabetes to facilitate reaching glycemic goals and potentially to reduce microvascular complications as the disease progresses. No consistent association was observed between HbA_1c testing and glycemic control or microvascular complications in type 2 diabetes as of 13 years diabetes duration. Nonendocrinology providers appear to monitor HbA_1c less frequently, particularly in type 2 diabetes, highlighting a target population for interventions to increase testing and, in turn, potentially improving glycemic control. Our results also suggest that youth and young adults may forgo HbA_1c testing during health insurance lapses or financial instability, while those having a personal doctor report more frequent testing. Thus, continued access to quality health care appears critical to successful management of diabetes, especially during the transition from pediatric to adult care.

Acknowledgments. SEARCH is indebted to the many youth and their families, and their health care providers, whose participation made this study possible.

Funding. The authors acknowledge the involvement of the Kaiser Permanente Southern California Clinical Research Center (funded by Kaiser Foundation Health Plan and supported in part by the Southern California Permanente Medical Group), the South Carolina Clinical and Translational Research Institute at the Medical University of South Carolina (National Institutes of Health [NIH], National Center for Advancing Translational Sciences [NCATS], grants UL1-TR-000062 and UL1-TR-001450), Seattle Children’s Hospital and the University of Washington (NIH/NCATS grant UL1-TR-00423), University of Colorado Pediatric Clinical and Translational Research Center, (NIH/NCATS grant UL1-TR-000154), the Barbara Davis Center at the University of Colorado at Denver (Diabetes Endocrinology Research Center NIH grant P30-DK-57516), the University of Cincinnati (NIH/NCATS grants UL1-TR-000077 and UL1 TR001425), and the Children with Medical Handicaps program managed by the Ohio Department of Health. SEARCH 1, 2, and 3, are funded by the Centers for Disease Control and Prevention (CDC) (PA nos. 00097, DP-05-069, and DP-10-001) and supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The SEARCH 4 cohort study (1-UC4-DK-108173) is funded by NIH/NIDDK and supported by the CDC. The population-based registry of Diabetes in Youth Study (1-U18-DP-006131, U18-DP-006133, U18-DP-006134, U18-DP-006136, U18-DP-006138, and U18-DP-006139) is funded by the CDC and supported by NIH/NIDDK. Funding for the SEARCH 1–4 sites is as follows: Kaiser Permanente Southern California (U18-DP-006133, U48/CCU-919219, U01-DP-000246, and U18-DP-002714), University of Colorado Denver (U18-DP-006139, U48/CCU-819241-3, U01-DP-000247, and U18-DP-000247-06A1), Cincinnati Children’s Hospital Medical Center (U18-DP-006134, U48/CCU-519239, U01-DP-000248, and 1-U18-DP-002709), University of North Carolina at Chapel Hill (U18-DP-006138, U48/CCU-419249, U01-DP-000254, and U18-DP-002708), Seattle Children’s Hospital (U18-DP-006136, U58/CCU-019235-4, U01-DP000244, and U18-DP-002710-01), and Wake Forest University School of Medicine (U18-DP-006131, U48/CCU-919219, U01-DP-000250, and 200-2010-35171).

This study includes data provided by the Ohio Department of Health, which should not be considered an endorsement of this study or its conclusions. S.S. (CDC) was a participating member of the study steering committee and writing group for this article because of the cooperative funding agreement. She was involved in the design of the study, but not the conduct of the study; she was not involved in the collection, management, and analysis of the data, but was involved in interpretation of the data; and she was involved in the preparation, review, and approval of the manuscript and the decision to submit the manuscript for publication. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC and NIDDK.

Duality of Interest. R.B.D. reports personal fees from Amgen and KCI outside the submitted work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. K.A.S. wrote the first draft of the manuscript. K.A.S., J.M.S., R.B.D., and D.D. designed the present research question. J.M.S. and R.B.D. analyzed the data. J.M.L., A.K.M., C.P., and D.D. supervised data collection. S.S. contributed to the study design and data interpretation. All authors interpreted the results and reviewed, critically revised, and approved the manuscript. K.A.S. 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.

Prior Presentation. Parts of this study were presented in poster form at the virtual 81st Scientific Sessions of the American Diabetes Association, 25–29 June 2021.