Improvement in Mean CGM Glucose in Young People with Type 1 Diabetes During 1 Year of the COVID-19 Pandemic

Introduction

The coronavirus disease 2019 (COVID-19) pandemic has been associated with unprecedented changes in the lives of children and adolescents. On March 16, 2020, Massachusetts schools transitioned to remote learning ¹ and continued in a hybrid model through spring of 2021. Children with type 1 diabetes (T1D), who usually see their medical team in face-to-face visits at recommended 3-month intervals, ² switched very rapidly to a largely telemedicine-based model out of necessity. With remote care, use of diabetes technology, especially continuous glucose monitoring (CGM) devices, became fundamental to delivery of effective remote care, as cloud-based data acquisition supported meaningful interactions between families and health care team members. ³

CGM provides moment-to-moment assessment of glucose values and has been associated, in general, with better glycemic control. ⁴ Published studies in adults with diabetes during “lockdowns” in various countries in the spring of 2020 implied improved glycemic control ^{5 –7} ; similar studies in pediatric patients, with smaller samples, reported mixed results. Some pediatric studies reported improved glycemic control, ^{8 –10} and others showed no change. ¹¹ Given the sedentary nature of home schooling combined with fewer organized sports, one might expect higher glucose levels overall; in contrast, increased parent/guardian oversight, flexibility in schedules, and reductions in competing social activities could promote enhanced diabetes self-care. An analysis of a large clinic-based population of pediatric patients with T1D using CGM that compares metrics before and during the COVID-19 pandemic can offer insight into patterns of, and potential changes in, glycemic control during this unprecedented time as well as the potential value of remote telehealth care delivery.

We analyzed a large clinic-based population of pediatric patients with T1D using CGM to compare CGM metrics during the COVID-19 pandemic with those from the previous year to begin to gain insight into how changes in lifestyle and care delivery associated with the pandemic may have influenced glycemic outcomes. Significant among these changes is the rapid shift to telemedicine.

Methods

We conducted a retrospective review of electronic health record (EHR) data from pediatric patients aged 1 to <18 years, with T1D duration ≥6 months if <6 years old or T1D duration ≥1 year if ≥6 years old, both at the first visit in the time period. We included patients who were actively using CGM and had visits between April 1, 2019, and March 15, 2020 (prepandemic period) or between April 1, 2020, and March 15, 2021 (pandemic period). Data from the 2-week interval of March 16–31, 2020, were excluded. Since CGM data were generally reviewed from the past 2 weeks, we aimed to include only exclusively during or postpandemic data, without sets of CGM data that included a mix of both. The institution's committee on human studies approved the study (CHS #00000136) and granted waivers for informed consent and authorization for the use/disclosure of protected health information.

Mean and standard deviation (SD) glucose from CGM data, generally from the past 2 weeks, were extracted from the EHR; these values are manually entered into designated fields in the EHR during clinical encounters by pediatric health care providers. Sensor glucose, glucose SD, and coefficient of variation (CV) were collected from all visits for a patient during a certain time period and then averaged, such that each patient had one value for each of the time periods (prepandemic or pandemic) in which they were seen. Glucose management indicator (GMI) was calculated using the formula reported by the Jaeb Center for Health Research. ^12,13 Hemoglobin A1c (HbA1c) results were extracted from the EHR. Before the pandemic, most HbA1c levels were obtained in the clinical laboratory (Roche Cobas™, Indianapolis, IN) as part of routine in-person visits. During the pandemic, HbA1c levels were obtained either in the clinical laboratory or in a commercial Quest Diagnostics™ laboratory near the patient's home. Results from other outside laboratories were not included in the analyses as they could not be efficiently extracted from the EHR. At clinic intake, patients' families are invited to choose from the options of “American Indian or Alaskan Native,” “Asian,” “Black or African American,” “White,” or “Decline to Specify.” This selection is transcribed recorded in the EHR by clinic staff. Unpaired t-tests and chi-square were used to compare pre- and during-pandemic data. Owing to multiple comparisons, P < 0.01 was considered significant.

Results

There were 641 youth in the prepandemic sample and 648 youth in the pandemic sample, of which 555 youth were included in both observation periods. Prepandemic, only 0.1% of visits were virtual; during the pandemic, 93.5% of visits were virtual. Approximately three-quarters of both samples used pump therapy. Table 1 gives sample characteristics, revealing that mean age, diabetes duration, mean number of visits (in-person or virtual) per patient, and use of Dexcom^® G6 CGM were greater in the pandemic sample. There was no difference in gender, race, or pump use. In terms of outcomes, data reveal that mean glucose, SD glucose, glucose CV, and GMI were all significantly lower in the pandemic sample than those in the prepandemic sample, all P < 0.01.

As anticipated, laboratory-derived HbA1c data were universally available in the EHR during the prepandemic period, whereas there were fewer results available during the pandemic. The prepandemic sample had a mean ± SD HbA1c of 8.1% ± 1.1%. In the pandemic sample, only about one-third (n = 218) of the youth had HbA1c values available in the EHR data capture; mean was 7.7% ± 1.2%, significantly lower than in the prepandemic sample (P = 0.001). There was a mean of 3.5 HbA1c values per patient in the prepandemic sample and 1.2 HbA1c values per patient in the pandemic sample.

Given the imbalance in number of HbA1c values pre- and during the pandemic, the analyses focused on the CGM-derived GMI. However, there was a significant correlation between baseline prepandemic HbA1c and GMI, for which 98% were obtained on the same day (r = 0.75, P < 0.001). Only 1/10th of the prepandemic sample achieved a GMI <7%, whereas another 1/10th had values ≥9%. Figure 1 shows the favorable shift in GMI seen in the pandemic period. The percentage of patients with GMI <7% nearly doubled, whereas the proportion in all other GMI categories above the target of <7% decreased (chi square df = 3, P < 0.001).

OPEN IN VIEWER

FIG. 1.

GMI category pre- and during COVID-19 pandemic. COVID-19, coronavirus disease 2019; GMI, glucose management indicator.

Discussion

This analysis included EHR data extraction of CGM data in a fairly large sample of pediatric patients with T1D. We observed lower mean sensor glucose levels, lower GMI, and reduced glucose variability in the sample receiving primarily virtual care through telemedicine during the pandemic compared with the sample prepandemic receiving mainly in-person care. Furthermore, there was a clinically significant and favorable shift in the distribution of GMI values during the pandemic, with nearly 1 in 5 youth achieving the equivalent HbA1c goal of <7%, whereas only ∼1 in 10 achieved this glycemic outcome prepandemic.

This analysis cannot pinpoint the cause of the decreased GMI observed during the pandemic as there were likely many contributing factors. We hypothesize that greater parental oversight, more flexibility in school/activity schedule, and decreased competing external interests may have all contributed by allowing more attention to diabetes self-care activities. In addition, the ease of accessing telehealth did increase visit frequency, which could also be a contributing factor to the observed glycemic improvements, with a favorable change in the process of care leading to an improvement in the outcome of care. Another contributing factor to the observed improvements may relate to the approval of a commercially available hybrid closed-loop system (Tandem T-slim X:2™ with Control IQ™) 1 month before the shutdowns associated with the COVID-19 pandemic began. Use of this device has been associated with improvement in glucose time-in-range (TIR) ¹⁴ and other glycemic outcomes. ¹⁵ It should be noted that during the pandemic, the only way patients could avail themselves of this beneficial new device was through remote training and monitoring through telehealth. The T1D exchange group observed that the majority clinics, similar to our own, were able to do CGM and pump start visits remotely. ¹⁶

Recent pediatric reports from the EU and Middle East, comparing pre- and postlockdown outcomes, have been mixed, with some showing decreased mean sensor glucose ⁸ and no change observed in others. ¹¹ In addition, some publications have reported improvements in glucose TIR ^{7 –9} while others have not. ^11,17 In another publication, Alharthi et al. reported that those with telemedicine visits during lockdown had improved glucose outcomes, whereas those without telemedicine visits showed no change. ⁷ Of note, compared with these previous studies, our analysis assessed patients over a much longer time, 1 year versus only a few months.

When interpreting our findings, a number of limitations should be considered. First, by necessity, we only included CGM users to access metrics of glycemic control in the absence of laboratory measurements of A1c. The lack of such laboratory measurements motivated the use of GMI to describe glycemic control, method used in some other studies during lockdown when HbA1c could be difficult to obtain. ^5,8,18 Fortunately, CGM use has increased substantially in youth with T1D, allowing for a substantial sample for study. ^19,20 Nonetheless, exclusion of non-CGM users may yield a unique subsample given the recognized disparities in diabetes technology use. ^{21 –23} We recognize that our study sample may reflect a more advantaged group of individuals by virtue of their overall use of advanced diabetes technologies and their ability to access telehealth during the pandemic. Future research should include youth who do not use diabetes technologies and telehealth as often as observed on our sample. Another limitation relates to our inability to report glucose TIR, an important metric for diabetes care delivery. ^{24 –27} In our clinic, we review and discuss this metric with patients/families; however, it is currently not entered into the EHR in a unique data field like CGM glucose mean and SD; future studies could use downloadable raw CGM data. Another limitation stems from the reduced ability of our EHR data capture to report when patients may have changed their insulin programs or adopted a hybrid closed loop system. In contrast, we were able to identify telehealth visit frequency, which likely corresponds to management changes. Finally, racial and ethnic composition of our clinic sample was somewhat incomplete due to the elective nature of self-reporting these demographics.

This report includes one of the largest pediatric samples, observed for the longest period of time to date, describing glycemic outcomes in youth with T1D during the pandemic compared with identical calendar months in the year preceding the pandemic. There are multiple variables that may have affected the glycemic control of children and adolescents during the pandemic, including lifestyle changes, use of CGM, greater parental involvement, and the widespread introduction of telemedicine. By demonstrating improved mean sensor glucose levels during the COVID-19 pandemic, this analysis opens the door for future research, including more in-depth analysis of the benefits of telemedicine, even beyond the pandemic. In addition to such quantitative evaluations, qualitative work is needed to assess patient and family perceptions of telehealth and what factors may have contributed to changes in diabetes self-care and glycemic outcomes during the pandemic. Identifying which factors were associated with such improvements in glucose levels can help inform future interventions to improve glycemic control.

Conclusion

During this difficult and uncertain pandemic year, we observed a statistically significant decline in mean sensor glucose as well as a near doubling of the proportion of young people with T1D attaining the target GMI of <7% compared with that of the year before.