Improved CGM Glucometrics and More Visits for Pediatric Type 1 Diabetes Using Telemedicine During 1 Year of COVID-19

Abstract

Purpose

The COVID-19 pandemic led to rapid adoption of telemedicine for the care of youth with type 1 diabetes (T1D). We assessed the utility of a primarily virtual care model by comparing glucometrics from a pediatric sample with T1D using continuous glucose monitoring (CGM) both before and during the pandemic.

Methods

Pediatric patients aged 1 to 17 years with T1D duration ≥ 1 year if ≥ 6 years old or ≥ 6 months if < 6 years old, with ≥ 1 visit with recorded CGM data both prepandemic (April 1, 2019-March 15, 2020) and during the pandemic (April 1, 2020-March 15, 2021) were included. Data were extracted from the electronic health record.

Results

Our sample comprised 555 young people (46% male, 87% White, 79% pump-treated), mean age 12.3 ± 3.4 years, T1D duration 5.9 ± 3.5 years, baseline glycated hemoglobin A1c 8.0 ± 1.0% (64 ± 10.9 mmol/mol). Diabetes visit frequency increased from 3.8 ± 1.7 visits/prepandemic period to 4.3 ± 2.2 visits/pandemic period (P < 0.001); during pandemic period, 92% of visits were virtual. Glucose management indicator (GMI) improved slightly from 7.9% (63 mmol/mol) prepandemic to 7.8% (62 mmol/mol) during the pandemic (P < 0.001). Those with equal or greater visit frequency (n = 437 [79% of sample]) had significant improvement in GMI (8.0% to 7.8% [64 to 62 mmol/mol], P < 0.001), whereas those with lower visit frequency did not (7.8 [62 mmol/mol], P = 0.86).

Conclusions

Children and adolescents with T1D using CGM before and during the pandemic showed an overall increase in visit frequency using primarily telemedicine-based care and improved CGM glucometrics. Further research is needed to understand factors associated with successful use of telemedicine for pediatric T1D.

The COVID-19 pandemic upended routine care for chronic diseases such as type 1 diabetes (T1D). This prompted many around the world to investigate changes in glycemic control associated with pandemic lockdowns. Many studies noted improvements in glycemia for adults and children with T1D during the initial months of lockdowns compared with prepandemic data (1-10). In the United States, telemedicine-based care was approved by emergency order on March 17, 2020 (11). Although the most strict closures concluded within a few months of this date, the concern to reduce the spread of the novel coronavirus persisted over the next year and beyond (12, 13), including the need for hospitals and clinics to minimize onsite care when possible. Thus, telemedicine continued as a significant portion of ambulatory pediatric diabetes care in our clinic over the next year, similar to other centers (14).

Prepandemic, telemedicine had been explored in research settings as a potentially valuable vehicle for diabetes care delivery (15-18), and early reports during the pandemic observed benefits of virtual diabetes care (14, 19, 20). As the pandemic persisted through 2021, telemedicine remained an important part of ongoing diabetes care delivery for more than 1 year in many areas (14), including provision of quarterly visits as recommended by the American Diabetes Association (21). Further, the use of technologies such as continuous glucose monitors (CGM), which incorporate cloud-based data sharing capabilities, helped facilitate the delivery of virtual care during the pandemic, as clinicians were able to review details of glycemic excursions virtually and provide insulin dose adjustments and other guidance (5, 22). Telemedicine was also used to educate new users of diabetes technology such as CGM and hybrid closed-loop systems remotely during the pandemic, allowing patients to have continued access to these technologies despite the need to distance (14, 23).

In addition to providing clinicians necessary data for clinical care, CGM became a valuable tool for research to understand glycemic control during the pandemic (2, 3, 5-8, 24-26) especially in light of the decreased access to and utilization of laboratory-based glycated hemoglobin measurements (HbA1c) during this time (14). CGM allows an estimation of average HbA1c (glucose management indicator [GMI]) (27), which overcomes many of the recognized challenges of HbA1c as a reflection of mean glucose, including varying rates of glycation and shorter red blood cell lifespan (28-30). Previous consensus guidelines have advocated for the value of CGM glucometrics as a means to assess glycemia and to target therapeutic changes (31). Although many with T1D continue to use fingerstick glucose monitoring for day-to-day diabetes care, CGM use has risen consistently in pediatrics (32), specifically during the COVID-19 pandemic (33).

Given the unexpected duration of the need for pandemic precautions, we aimed to understand changes in visit frequency and glycemic control observed after the addition of virtual care options. To assess the utility of telemedicine care, we performed a paired comparison using a clinic-based sample of pediatric patients with T1D who were using CGM both before and during the COVID-19 pandemic, comparing visit frequency and glycemic control during the first year of the pandemic with the preceding year.

Materials and Methods

We conducted a retrospective review of electronic health record (EHR) data from pediatric patients at the Joslin Diabetes Center in Boston, Massachusetts, who had established T1D and were using CGM both before and during the pandemic. We included those aged 1 year to less than 18 years, with T1D duration ≥ 6 months if < 6 years old or T1D duration ≥ 1 year if ≥ 6 years old. Patients had to have at least 1 visit with recorded CGM data both between April 1, 2019, and March 15, 2020 (prepandemic period), and between April 1, 2020, and March 15, 2021 (pandemic period). Virtual visits were conducted using a combination of telephone and various videoconferencing platforms depending on availability of software, patient factors, and national, state, and local guidance on need for Health Insurance Portability and Accountability Act-compliant software. Our center switched to an overwhelmingly telemedicine-based care model on March 16, 2020, and continued to offer virtual care over the next year and beyond. For this analysis, we focused on medical diabetes visits (physician, nurse practitioner, and registered nurse visits) because these most consistently recorded CGM data. Because CGM data were typically reviewed from the proximal 2 weeks before the medical visit, we excluded the 2-week interval of March 16 through 31, 2020, to avoid mixing pre- and during pandemic glucose data. Age and diabetes duration were assessed based on the first visit in the prepandemic period. The institution’s Committee on Human Studies approved the study (CHS #00000136) and granted waivers for the informed consent process and for the requirement of authorization for the use/disclosure of protected health information.

Glucose mean/SD and CGM wear time as days per week were extracted from the EHR; these CGM-derived values were manually entered into designated fields in the EHR during clinical encounters by pediatric health care providers. If patients had multiple visits during the pre- or pandemic periods, the mean of all glucometric measurements was calculated for the pre- and pandemic periods separately. GMI was calculated using the formula reported by Bergenstal et al (27). HbA1c results were also extracted from the EHR. Before the pandemic, the majority of 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 a commercial Quest Diagnostics laboratory. Occasionally, results were obtained from other outside laboratories; these were not included in these analyses. Sample characteristics by baseline HbA1c group were compared using nonparametric tests (Kruskal-Wallis ANOVA, paired sign tests). To understand visit frequency and CGM glucometrics during the pandemic, we divided patients in 2 ways. First, we divided into groups of glycemic risk based on mean HbA1c in the prepandemic period: < 7%, 7% to 9%, or > 9% (< 53, 53-75, > 75 mmol/mol). Next, we divided into groups of visit frequency: 1 to 2, 3, or 4 or more visits per year in each period. Paired t tests and χ ² were used to compare prepandemic with pandemic data. Nonparametric tests were used to compare HbA1c groups. Because of multiple comparisons, P < 0.01 was considered significant.

Results

There were 555 youth in our sample. Mean (±SD) age was 12.3 ± 3.4 years, T1D duration was 5.9 ± 3.5 years, and 46% were male. Of those who elected to report race at clinic intake, 87% identified as White (9.7% declined to specify race). Our sample’s mean baseline HbA1c was 8.0 ± 1.0% (64 ± 10.9 mmol/mol), and 79% were pump-treated. These data are summarized in Table 1. In this sample of CGM users, 95% used a Dexcom CGM.

Table 1.

Characteristics and CGM glucometrics, for total sample and by baseline laboratory HbA1c value

	Total sample (N = 555)	Baseline HbA1c < 7% (n = 90)	Baseline HbA1c 7-9% (n = 392)	Baseline HbA1c > 9% (n = 73)
Age, ya	12.3 ± 3.4	11.9 ± 3.6	12.2 ± 3.4	13.2 ± 3.2
Sex, male	46%	48%	46%	45%
Race, Whiteb	87%	90%	88%	78%
Diabetes duration, y	5.9 ± 3.5	5.2 ± 3.8	5.9 ± 3.4	6.7 ± 3.7
HbA1c, % (mmol/mol)	8.0 ± 1.0 (64 ± 10.9)	6.5 ± 0.4 (48 ± 4.4)	7.9 ± 0.5 (63 ± 5.5)	9.7 ± 0.6 (83 ± 4.4)
Pump use	79%	79%	80%	73%
Mean glucose, mg/dL (mmol/L)
- Prepandemic	193 (10.7)	155 (8.6)	193 (10.7)	234 (13.0)
- Pandemic	187 (10.4)	150 (8.3)	188 (10.4)	223 (12.4)
- P	<0.001	0.01	<0.001	0.002
SD glucose, mg/dL (mmol/L)
- Prepandemic	72 (4.00)	59 (3.28)	72 (4.00)	87 (4.83)
- Pandemic	69 (3.83)	56 (3.11)	69 (3.83)	82 (4.56)
- P	<0.001	0.004	<0.001	0.001
Glucose CV, %
- Prepandemic	37.7	37.9	37.6	37.8
- Pandemic	37.0	36.8	36.9	37.6
- P	<0.001	0.01	0.001	0.61
GMI, % (mmol/mol)
- Prepandemic	7.9 (63)	7.0 (53)	7.9 (63)	9.0 (75)
- Pandemic	7.8 (62)	6.9 (62)	7.8 (62)	8.6 (70)
- P	<0.001	0.13	<0.001	0.002

Prepandemic period is defined as April 1, 2019-March 15, 2020, and during pandemic is defined as April 1, 2020-March 15, 2021. Data are presented as mean ± SD or percent. P < 0.01 was considered significant; these values are in bold. Characteristics were not significantly different between groups. HbA1c groups: <7%, <53 mmol/mol; 7%-9%, 53-75 mmol/mol; >9%, >75 mmol/mol.

Abbreviations: CGM, continuous glucose monitor; CV, coefficient of variation; GMI, glucose management indicator; HbA1c, glycated hemoglobin.

^a P value for difference between mean age was 0.02.

^b 9.7% declined to specify race at clinic registration.

Telemedicine comprised the majority of care delivery during the pandemic, with 92.2% of diabetes visits (physicians, nurse practitioners, registered nurses) conducted virtually. Mean glucose, SD glucose, glucose coefficient of variation, and GMI were all significantly lower during the pandemic when compared with prepandemic data. Mean glucose declined from 193 to 187 mg/dL (10.7 to 10.4 mmol/L), SD glucose from 72 to 69 mg/dL (4.00 to 3.83 mmol/L), coefficient of variation from 37.7% to 37.0%, and GMI from 7.9% to 7.8% (63 to 62 mmol/mol) (all P < 0.001, Table 1). HbA1c data were limited during the pandemic; less than one-third (182 of the 555 patients) had HbA1c values in the pandemic period. HbA1c levels between the groups with pandemic HbA1c data (n = 182, mean 8.0%) and without pandemic HbA1c data (n = 373, mean 8.0%) (P = 0.63). Of those with values in both periods, HbA1c was reduced significantly from 8.0% to 7.7% (P < 0.001). There was no significant difference in prepandemic information. Given this absence of complete HbA1c data during the pandemic, GMI values provided an estimate of HbA1c values pre- and during the pandemic. Baseline HbA1c (prepandemic) correlated significantly with baseline GMI (r = 0.76, P < 0.0001), supporting the reliability of GMI during the pandemic. Of our sample of 555 patients, 540 had CGM wear time values in both the pre- and during pandemic periods. The mean wear time varied from 6.7 days per week prepandemic to 6.8 days per week during the pandemic (P = 0.003), a statistically but nonclinically significant difference.

Diabetes visits increased significantly during the pandemic, from 3.8 visits per patient to 4.3 (P < 0.001, Fig. 1). When patients were divided based on the aforementioned visit categories of 1 to 2, 3, or 4 or more visits in the year, we found that in the prepandemic period, 77 had 1 to 2 visits (13.9% of the sample), 136 had 3 visits (24.5%), and 342 had 4 or more visits (61.6%). In the pandemic period, 91 patients had 1 to 2 visits (16.4%), 111 had 3 visits (20.0%), and 353 had 4 or more visits (63.6%). The majority of patients, 79%, had similar or increased visit frequency (Table 2). These patients had statistically significant improvements in glucose and GMI (194 to 187 mg/dL [10.8 to 10.4 mmol/L] and 8.0 to 7.8% [64 to 62 mmol/mol], respectively; P < 0.001 for both). For those with decreased visit frequency, glucose and GMI were unchanged (188 mg/dL [10.4 mmol/L] and 7.8% [62 mmol/mol], respectively, P = 0.86 for both).

Figure 1.

Mean diabetes visits by baseline glycated hemoglobin A1c group. Error bars are standard deviations. Prepandemic period is defined as April 1, 2019-March 15, 2020, and during pandemic is defined as April 1, 2020-March 15, 2021. P < 0.01 was considered significant; these values are in bold. Both telemedicine and face-to-face visits were included. HbA1c, hemoglobin A1c. HbA1c groups: <7%, <53 mmol/mol, 7%-9%, 53-75 mmol/mol, >9%, >75 mmol/mol.

Table 2.

CGM outcomes by change in medical visit frequency from prepandemic to pandemic period

	Same or increased visit frequency (n = 437, 79%)			Decreased visit frequency (n = 118, 21%)
	Prepandemic	During pandemic	P	Prepandemic	During pandemic	P
Mean glucose, mg/dL (mmol/L)	194 (10.8)	187 (10.4)	<0.001	187 (10.4)	187 (10.4)	0.86
SD glucose, mg/dL (mmol/L)	73 (4.06)	69 (3.83)	<0.001	71 (3.94)	69 (3.83)	0.03
Glucose CV, %	37.6	37.0	0.003	38.0	37.2	0.01
GMI, % (mmol/mol)	8.0 (64)	7.8 (62)	<0.001	7.8 (62)	7.8 (62)	0.86

P < 0.01 was considered significant; these values are in bold.

Abbreviations: CGM, continuous glucose monitor; CV, coefficient of variation; GMI, glucose management indicator.

When we divided patients into categories of glycemic risk according to prepandemic HbA1c levels, we found that the majority 392 (71%) had HbA1c 7% to 9% (53-75 mmol/mol) (Table 1), with 16% having values < 7% (<53 mmol/mol) and 13% >9% (>75 mmol/mol). There were no significant differences in baseline characteristics across the 3 HbA1c groups (Table 1). GMI was similar pre- and during the pandemic in those with HbA1c < 7% (<53 mmol/mol) (GMI 7.0% to 6.9% [53 to 52 mmol/mol], P = 0.13), and appeared clinically similar in those with HbA1c 7% to 9% (53-75 mmol/mol) (GMI 7.9% to 7.8% [63 to 62 mmol/mol], P < 0.001). For those in the highest prepandemic HbA1c group, GMI improved significantly from 9.0% to 8.6% (75 to 70 mmol/mol) (P = 0.002) (Table 1). Increased visit frequency during the pandemic was observed in the 2 higher baseline HbA1c groups, 7% to 9% (53-75 mmol/mol) and > 9% (>75 mmol/mol) (Fig. 1). Visit frequency increased from 3.7 to 4.2 visits per person per year (P < 0.001) in those with prepandemic HbA1c 7% to 9% (53-75 mmol/mol) and from 4.8 to 5.6 visits per person per year (P = 0.04) in those with HbA1c > 9% (>75 mmol/mol). Visit frequency in the lowest baseline HbA1c group < 7% (<53 mmol/mol) was similar in the 2 periods (3.5 and 3.6 visits per person per year, P = 0.18).

Discussion

This retrospective follow-up study included a paired analysis of children and adolescents with T1D using CGM before and during the COVID-19 pandemic. Prepandemic, all visits were in-person, whereas during the pandemic, the overwhelming majority of visits used telemedicine. Those who had similar or increased visit frequency during the pandemic demonstrated a significant improvement in CGM-based GMI, whereas those who had a decrease in visit frequency did not. Those children and adolescents meeting the American Diabetes Association-recommended HbA1c goal of < 7% (<53 mmol/mol) maintained their glycemic control and visit frequency, whereas those with suboptimal glycemic control increased their visit frequency and improved their CGM-based GMI.

The transition to virtual care coincided with the State of Emergency and federal recommendations for quarantine. At our institution, virtual care for pediatric diabetes patients continued in this manner as the predominant modality for diabetes care delivery throughout the first year of the pandemic, likely for multiple reasons. First, given the urgent need for virtual care, telephone-based care was rapidly transitioned to a Health Insurance Portability and Accountability Act-compliant video platform. Next, systems were immediately implemented in our clinic for the remote transfer of pediatric diabetes device data, including CGM, insulin pumps, glucose meters, smart insulin pens, with support staff to assist patients/families in the uploading process. Last, the ongoing utilization of telehealth care delivery reflected the apparent high acceptability of this model of care by pediatric patients and their families.

Our results document the facility with which telemedicine preserved visit frequency in pediatric CGM users with established T1D during the COVID-19 pandemic because 79% had similar or increased visits comparing the prepandemic with the pandemic period. We noted that visit frequency of 4 or more annually was preserved in accordance with American Diabetes Association guidelines (34). Furthermore, telehealth supported an increase in visit frequency for those with suboptimal control, increasing their touchpoints with the health care team. Before the pandemic, other pediatric centers had already begun testing a telehealth model to reach remote areas without pediatric diabetes expertise (15, 35). During the pandemic, a few centers reported observed benefits of telemedicine based on CGM measures from smaller samples (4, 5). The current analysis adds to these earlier observations using a relatively large cohort of pediatric patients followed both before the pandemic (using in-person visits) and during the pandemic (primarily virtual visits), permitting a paired comparison of visit frequency and CGM-based GMI.

In addition to the use of telemedicine, there are other potential explanations for the observed increase in visit frequency and improved GMI. First, remote school and work may have allowed more flexibility within families to increase visit frequency. Remote home activities likely allowed for more regimented eating patterns, increased diabetes self-care behaviors (2), and more parental oversight of children (6). Additionally, the second major hybrid closed-loop system, the Tandem T-slim X:2 with Control IQ, received Food and Drug Administration approval at the end of 2019, becoming available in early 2020 at the pandemic’s start. Utilization of this system may have contributed to some of the observed glycemic benefits, given its documented benefit (36). On the other hand, there were other circumstances during the pandemic that were counter to optimal glycemic control in youth with T1D. Pediatric obesity increased (37), as did stress and anxiety (38), both of which would likely worsen glycemic control (39).

There are limitations to this retrospective analysis. First, this study included only CGM users. Many have noted the importance of CGM use during the pandemic because the cloud-based data storage allowed treatment teams to assess glycemic control remotely (1-8, 24-26, 33). Furthermore, Danne et al. demonstrated increased adoption of CGM during the fall 2020 wave of COVID-19 in various countries (33). However, we recognize the need to understand glycemic patterns of all children; a recent study of pediatric patients with T1D reported glycemic benefit in CGM users but not in non-CGM users (40). Next, our analysis was limited by the availability of electronically extractable data from our EHR, which did not include CGM percent time-in-range, an important marker of glycemic control. Degree of remote schooling necessitated by the COVID-19 pandemic varied across regions served by our center and, thus, cannot be reported. Further, our outpatient EHR is not linked to a hospital system so does not automatically track emergency room visits; this information tends to be recorded narratively based on family or outside provider report. Similarly, for the purposes of this analysis, we were unable to automatically extract exactly when those youth who initiated hybrid closed-loop therapy during the pandemic did so; this change in management may have contributed to the observed improvements. Next, data on race, ethnicity, and socioeconomic factors were not always routinely collected as part of clinical care, and as such could not be reliably reported, although our sample does not appear to be very diverse. Future research of telemedicine care delivery can include more diverse samples as far as race, ethnicity, and technology use, helping researchers and clinicians to better understand salient factors associated with the increased frequency of visits and the observed changes in glycemia.

Utilization of telemedicine during the pandemic and the attendant observations of preserved or increased visit frequency along with improvements in GMI compared with the prepandemic period is encouraging. However, it is essential to recognize limitations of telemedicine, including the inability to monitor physical findings such as pubertal status, height, weight, and blood pressure; delayed routine screening of T1D-related comorbidities; lack of timely laboratory data; and potential deficient care for patients experiencing challenges with diabetes management or use of diabetes technologies (41). After the strict, early restrictions of the pandemic subsided, some visits (7.8%) began to be conducted in-person, likely including families experiencing challenges using diabetes technologies and those at high risk for comorbidities or other medical concerns.

In consideration of the potential benefits and limitations of telemedicine, we suggest a hybrid model of care for pediatric patients with established T1D, where face-to-face visits alternate with telemedicine at the discretion of providers and patients/families. It is reassuring that visit frequency and CGM-based GMI were preserved or improved in our cohort of youth with T1D treated with CGM and receiving primarily telehealth care. Future analyses of visits and glycemic control in diverse pediatric samples with and without technology use can add to our understanding of the benefits of a future hybrid telemedicine care model for pediatric T1D.

Glossary

Abbreviations

CGM

continuous glucose monitoring

EHR

electronic health record

GMI

glucose management indicator

HbA1c

glycated hemoglobin

T1D

type 1 diabetes

Financial Support

This work was supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (P30DK036836 [to Joslin Diabetes Center] and K12DK094721 [to L.L. and T.K.]) of the National Institutes of Health; the Katherine Adler Astrove Youth Education Fund; the Maria Griffin Drury Pediatric Fund; and the Eleanor Chesterman Beatson Fund.

Disclosure Statement

The authors have nothing to disclose.

Conflict of Interest

LML has provided advisory/consulting services unrelated to the manuscript to the following: Boehringer Ingelheim, Eli Lilly, Dexcom, Dompe, Janssen, Insulet, Medtronic, Provention, and Roche.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.