Disparities in Continuous Glucose Monitoring Use Among Women of Reproductive Age with Type 1 Diabetes in the T1D Exchange (2024)

Associated Data

Supplementary Materials

Abstract

We identified characteristics associated with continuous glucose monitoring (CGM) use in women of reproductive age with type 1 diabetes (T1D) in the T1D Exchange clinic registry from 2015 to 2018. Among 6643 assessed women, the frequency of CGM increased from 2015 to 2018 (20.6% vs. 30.0%; adjusted odds ratios [aOR]: 1.72; confidence interval [95% CI]: 1.51–1.95) and was more likely with recent pregnancy (45.3% vs. 25.8%; aOR: 1.63; 95% CI: 1.23–2.16). Non-Hispanic Black and Hispanic race and ethnicity, younger age, lower educational attainment, lower income, and Medicaid insurance were associated with lower odds of CGM. The use of CGM was associated with lower odds of diabetic ketoacidosis and lower hemoglobin A1c without any difference in the odds of symptomatic severe hypoglycemia. In conclusion, although CGM use was associated with better glycemic control, the majority of reproductive-age women still did not use it. Those who did not use CGM were more likely to be those at greatest risk of adverse pregnancy outcomes.

Keywords: Type 1 diabetes, Periconception, Pregnancy, Women, Disparities, Continuous glucose monitoring

Introduction

Optimal glycemic control with type 1 diabetes (T1D) in the periconception period is critical to prevent adverse pregnancy outcomes.^1,2 Guidelines suggest a preconception hemoglobin A1c (A1c) target of <6.5%.³ Hyperglycemia even before the first trimester is a causative factor for miscarriage and congenital malformations, and later in pregnancy increases the risk of adverse pregnancy outcomes.^4–6 Nearly one in two pregnancies with T1D are not planned,⁷ and this frequency is over twofold higher among non-Hispanic Black and Hispanic women.⁸ The problem is many women with T1D are not aware that they are pregnant during the early first trimester,⁹ prenatal care with intensive glycemic control frequently does not begin until the late first trimester,^1,10,11 and the risks of congenital malformation and miscarriage due to inadequate glycemic control are determined before prenatal care starts.^4,6

Continuous glucose monitoring (CGM) improves glycemic control and reduces adverse pregnancy outcomes in women with T1D.^12,13 CGM use for T1D in the United States has increased significantly from 3% in 2006 to 40% in 2018.¹⁴ Recent data on CGM use and related patient characteristics from women of reproductive age with T1D are lacking.¹⁵ CGM use has continued to lag for many individuals who experience a higher burden of adverse social determinants of health, including inadequate insurance, lower educational attainment, systemic racism, and provider biases.^14,16–18

The objective of the current analysis was to identify sociodemographic, clinical, and glycemic-related characteristics associated with CGM use versus no CGM use in women of reproductive age with T1D.

Methods

We conducted a secondary analysis from the T1D Exchange clinic registry from >80 clinics across the United States.¹⁹ This prospective cohort collects clinical and laboratory data on adults and children with T1D. In this analysis, we included reproductive-age girls and women with T1D aged 15–45 years with data between 2015 and 2018. Each site received approval from an Institutional Review Board (IRB), and informed consent was obtained. For the current analysis, we utilized the publicly available dataset at https://public.jaeb.org/datasets/diabetes, which was reviewed and deemed exempt by the Biomedical Institutional Review Board at The Ohio State University (No. 2022E0847; date: September 12, 2022).

Data was collected for the clinic registry central database from participants' medical records. Demographic information was collected through self-reported questionnaires. Data on age and diabetes duration were collected from medical chart review. Diabetes management habits, including CGM use, and pregnancy status were reported by participants and confirmed by medical records. A1c measurements closest to the assessment (3 months before 1 month after) were obtained from clinic medical records, and were assessed as a continuous measure and categorically as an A1c <6.5%, consistent with recommendations in the periconception period to prevent adverse pregnancy outcomes.^3,20 Consistent with prior analyses, a diagnosis of symptomatic severe hypoglycemia required loss of consciousness or seizure and a diagnosis of diabetic ketoacidosis required an overnight hospitalization. This analysis does not use data from more detailed questionnaires, including pregnancy surveys, from the T1D Exchange.

The outcome was CGM use versus no CGM use at the time of survey assessment reported by participants and confirmed by medical records. We assessed sociodemographic (age, education, income, self-reported race, and ethnicity), clinical (body mass index [BMI], insulin delivery, T1D duration, T1D complications), and glycemic-related factors (diabetic ketoacidosis, symptomatic severe hypoglycemia, A1c) that were associated with CGM use compared to no CGM use. Unadjusted and adjusted odds ratios (OR, aOR) with confidence intervals (95% CIs) were calculated. Models were adjusted for age, insurance status, educational attainment, and T1D duration. Imputation for missing data was performed (n = 30 imputations) and estimates were combined using Rubin's rule. All statistical analyses were performed using STATA (version 16.1; STATACORP, College Station, TX).

Results

Among 24,576 individuals with T1D, we excluded those who identified as male sex (n = 17,888), were aged <15 or >45 years (n = 5712), without sex recorded (n = 32), identified as transgender (n = 13), and did not have CGM use recorded (n = 165), resulting in a final analytic sample of 6478 women (Supplementary Fig. S1). Overall, 5552 of these women were assessed in 2015–2016, 4761 in 2016–2017, and 5877 in 2017–2018.

The median age was 20.0 years (interquartile range [IQR]: 17.0, 28.0), and 5.9% and 9.3% self-identified as non-Hispanic Black and Hispanic, respectively, and 19.6% were enrolled in Medicaid. The median duration of living with diabetes 11.5 years (IQR: 7.5, 16.8). Over two thirds (67.3%) reported current insulin pump use. A total of 5.9% reported diabetic ketoacidosis and 1.6% symptomatic severe hypoglycemia within the past 12 months. The mean A1c was 8.3% (IQR: 7.4, 9.6), which was lower with CGM use (7.6% [IQR: 7.0, 8.5] vs. 8.6% [IQR: 7.6, 10.0]; P < 0.05)

CGM use increased over time (2017–2018 vs. 2015–2016 [reference]: 30.0% vs. 20.6%; aOR: 1.72; 95% CI: 1.51–1.95) (Fig. 1). Women who had recently been pregnant in the past 12 months were more likely to use CGM (45.3% vs. 25.8%; aOR: 1.63; 95% CI: 1.23–2.16). But, CGM use in those who had been pregnant remained stable across time (44.7% 2015–2016, 43.5% 2016–2017, and 46.5% 2017–2018.

Sociodemographic characteristics generally indicative of adverse social determinants of health were associated with lower adjusted odds of CGM use, including non-Hispanic Black (11.3% vs. 29.1%; aOR: 0.40; 95% CI: 0.28–0.56) and Hispanic compared to non-Hispanic White race and ethnicity (16.2% vs. 29.1%; aOR: 0.73; 95% CI: 0.57–0.92), lower educational attainment among those >18 years of age (for example, high school or less vs. graduate school or higher: 17.6% vs. 44.3%; aOR: 0.45; 95% CI: 0.34–0.60), lower income (for example, <$35,000/year vs. $75,000/year: 24.1 vs. 34.6%; aOR: 0.48; 95% CI: 0.39–0.59), and Medicaid insurance (9.6% vs. 31.1%; aOR: 0.30; 95% CI: 0.24–0.36) (Fig. 1). Odds of CGM use increased with age (<20 years vs. >35 years: 18.6% vs. 43.1%; aOR: 2.66; 95% CI: 2.14–3.32).

CGM use was associated with higher adjusted odds of insulin pump use (34.0% vs. 10.2%; aOR: 3.90; 95% CI: 3.32–4.57), and a lower adjusted odds of current smoking (11.8% vs. 27.2%; aOR: 0.36; 95% CI: 0.23–0.58) (Fig. 1). CGM use was not associated with the duration of living with T1D and BMI. CGM use was not associated with clinical conditions associated with worse diabetes, including renal disease (glomerular filtration rate <60 mL/min), retinopathy, neuropathy, and gastroparesis (data not shown).

CGM use was associated with improved glycemic control, including higher adjusted odds of A1c <6.5% versus ≥6.5% (49.4% vs. 24.6%; aOR: 1.95; 95% CI: 1.57–2.44), as well as incrementally with each 1% improvement in A1c from ≥9.5% to ≤6.5% (Fig. 1). CGM use was associated with lower adjusted odds of diabetic ketoacidosis (12.9% vs. 27.2%; aOR: 0.57; 95% CI: 0.41–0.78). CGM use did not vary by symptomatic severe hypoglycemia, of which few cases occurred with an A1c <6.5% (5.7%).

Discussion

In the T1D Exchange registry, reproductive-age women with T1D increased their use of CGM from 2015 to 2018. Moreover, those who used CGM demonstrated better glycemic control. Yet, most were still not using CGM, including those who were recently pregnant. Disconcertingly, women who were the least likely to use CGM were also those who were exposed to more adverse social determinants of health.

These findings among women of reproductive age are consistent with prior studies conducted among adults and adolescents with T1D, including from the T1D Exchange Registry and T1DX-QI Collaborative, that found that those of lower socioeconomic status, younger age, and who identified as non-Hispanic Black and Hispanic were less likely to use CGM.^18,21,22 Low CGM use by women who experience more adverse social determinants of health could contribute to persistent disparities in glycemic control, as well as adverse pregnancy outcomes. Possible reasons for disparities in CGM use may include systemic racism, diminished access to health care, psychosocial stressors, provider bias, inadequate patient-provider communication, and strict eligibility criteria that add additional hurdles to use for those who experience more unmet social needs.^23–26

In the current study, CGM use was associated with a higher likelihood of achieving an A1c in the optimal range for preconception. In addition, women who were recently pregnant were more likely to use CGM. A prior analysis using baseline data from the TID Exchange Registry between 2010 and 2013 found that CGM use was higher and median A1c was slightly lower among recently and ever pregnant individuals compared to those who were never pregnant.²⁷ Nevertheless, most women were not at an optimal A1c level for preconception. The current analysis demonstrated that even if CGM use was more likely periconceptionally, CGM use still only occurred in a minority of these women.

In the pregnancy arm of the T1D CONCEPTT trial, those randomized to CGM versus self-monitoring blood glucose had a significant reduction in A1c, more time in target range, and less hyperglycemic events.¹⁹ Episodes of severe hypoglycemia and diabetic ketoacidosis were similar between groups. In the smaller CONCEPTT trial arm of nonpregnant individuals planning pregnancy who were randomized to CGM, the point estimates were consistent with the pregnancy arm but with wider CIs that included the null value.

These findings of the current study emphasize that further research is needed to understand how to increase access to and uptake of CGM in general, and in particular among women who experience a higher burden of adverse social determinants of health.¹⁸ Encouragingly, CGM use may be increasing as among a subset of 47 pregnant women who completed a questionnaire as part of the TID Exchange, CGM use was 70% in 2016–2018.²⁸ Nevertheless, preconception care could be an opportunity to increase CGM use by optimizing glycemic control before pregnancy. The problem remains that only 1 in 10 women of reproductive age with T1D in the United States receive any preconception care, and those who do are more likely to have higher educational attainment, be employed, and have private insurance.¹¹

CGM continues to lack U.S. Food and Drug Administration approval for use in pregnancy, which may also limit access in the periconception period. In addition, patient-provider relationships that emphasize person-centered decision-making, open communication, and enhance patient engagement could increase CGM use.²⁹ Finally, standard and equitable care delivery that provides equal opportunity and access to diabetes technology could also increase CGM use.

There are several limitations of this study. We conducted a serial cross-sectional analysis, and hence cannot comment on the temporality as well as causality between observed associations between patient characteristics and CGM use. We cannot determine whether patients had used but discontinued CGM or whether they had been offered CGM but then declined this option. Because the T1D Exchange Registry does not collect detailed reproductive health outcomes on follow-up surveys, those outcomes could not be assessed. We assessed individual factors associated with CGM use, and further studies are needed to understand interactions among the multiple factors associated with CGM use. The current analysis did not address additional technology features associated with CGM use, including duration of use, alarms, use of CGM to decide/adjust insulin dosing, device downloading, and mobile medical applications.

The T1D Exchange Registry is a clinic-based and not population-based cohort, which may affect the generalizability of the findings. The data source primarily included individuals treated at endocrinology centers that focused on TID care, and hence CGM use was likely higher than in the broader population of individuals with T1D. It is also possible that the proportion meeting glycemic targets may be greater in this study population. These data are now at least 4 years old, and CGM use has likely increased with greater insurance coverage. However, obtaining CGM access for those individuals who experience more adverse social determinants of health, including adequate insurance coverage, remains a persistent problem and may further increase disparities in equitable diabetes outcomes.

In conclusion, CGM use has recently increased in reproductive-age women with T1D, and results in better glycemic control. Yet, most of these women were not using CGM, including in the periconception period, and the likelihood of use was related to multiple adverse social determinants of health. As innovative technologies raise the standards of care in T1D, further efforts that drive toward equitable care and outcomes will be needed not only to increase CGM use among reproductive-age women, but to then translate that use to improved maternal and neonatal outcomes.

Supplementary Material

Authors' Contributions

K.K.V., N.F., and W.A.G.: review and editing, K.K.V., C.E.P., E.B., and N.F.: Conceptualization; K.K.V., N.F., S.G., and W.A.G.: writing—original draft; K.K.V., J.W., and W.A.G.: formal analysis; M.B.L., S.G., and K.G.: review and editing; K.K.V., J.W., N.F., and W.A.G.: methodology.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

K.K.V. was supported by the Care Innovation and Community Improvement Program at The Ohio State University. K.K.V. and N.F. were supported by AHRQ grant no. R01HS028822.

Supplementary Material

Supplementary Figure S1

References