Comparison of Real-Time and Intermittently-Scanned Continuous Glucose Monitoring for Glycemic Control in Type 1 Diabetes Mellitus: Nationwide Cohort Study
Article information
Abstract
Background
This study compares the association between real-time continuous glucose monitoring (rtCGM) and intermittently-scanned CGM (isCGM) and glycemic control in individuals with type 1 diabetes mellitus (T1DM) in a real-world setting.
Methods
Using data from the Korean National Health Insurance Service Cohort, individuals with T1DM managed by intensive insulin therapy were followed at 3-month intervals for 2 years after the initiation of CGM. The glycosylated hemoglobin (HbA1c) levels and coefficients of variation (CVs) of rtCGM and isCGM users were compared using independent two-sample t-test and a linear mixed model.
Results
The analyses considered 7,786 individuals (5,875 adults aged ≥19 years and 1,911 children and adolescents aged <19 years). Overall, a significant reduction in HbA1c level was observed after 3 months of CGM, and the effect was sustained for 2 years. The mean HbA1c level at baseline was higher in rtCGM users than in isCGM users (8.9%±2.7% vs. 8.6%±2.2%, P<0.001). However, from 3 to 24 months, rtCGM users had lower HbA1c levels than isCGM users at every time point (7.1%±1.2% vs. 7.5%±1.3% at 24 months, P<0.001 for all time points). In both adults and children, the greater reduction in HbA1c with rtCGM remained significant after adjusting for the baseline characteristics of the users. The CV also showed greater decrease with rtCGM than with isCGM.
Conclusion
In this large nationwide cohort study, the use of rtCGM was associated with a greater improvement in glycemic control, including HbA1c reduction, than the use of isCGM in both adults and children with T1DM.
Highlights
• In a nationwide cohort study, 7,786 CGM users with T1DM were followed for 2 years.
• HbA1c reduction was greater in rtCGM users than in isCGM users.
• CV also showed a greater decrease with rtCGM than with isCGM.
• The results were consistent in both adults and children.
• RtCGM use was associated with better glycemic control than isCGM use.
INTRODUCTION
Continuous glucose monitoring (CGM), which enables users to monitor their glucose levels constantly without the need for fingerstick blood glucose testing, has demonstrated its benefits for glycemic control in individuals with type 1 diabetes mellitus (T1DM) [1-8]. Thus, current guidelines recommend that individuals with T1DM use CGM constantly [9-11]. There are two types of CGM systems: intermittently-scanned CGM (isCGM) and real-time CGM (rtCGM) [12]. Users of isCGM can check their glucose values by scanning the sensor, and users of rtCGM automatically receive their glucose values every 5 minutes through a transmitter. In addition, rtCGM provides predictive alerts for high and low blood glucose levels [13].
Several randomized clinical trials (RCTs) have compared the efficacy of rtCGM and isCGM for glycemic control [14-16]. Representative RCTs, the Comparing Continuous with Flash Glucose Monitoring in Adults With Type 1 Diabetes (ALERTT1) trial [14] and Comparison of CGM in Randomized Study of Real-time and Intermittently Scanned Systems in type 1 diabetes With Normal Awareness of Hypoglycemia (CORRIDA) study [15], demonstrated that rtCGM was superior to isCGM for glycemic control. However, the evidence supporting the superiority of rtCGM over isCGM in the pediatric population is insufficient because the RCTs were conducted in adults. Although the representative isCGM device, FreeStyle Libre 1 (Abbott Diabetes Care, Witney, Oxon, UK), has developed into an rtCGM device (FreeStyle Libre 3), only the FreeStyle Libre 1 is currently available in several countries. Thus, comparing the effectiveness of isCGM with that of rtCGM is an important question. No previous real-world studies with a large population and long-term follow-up have compared the effectiveness of rtCGM over isCGM in glycemic control.
This study compares the associations of rtCGM and isCGM with glycemic control in individuals with T1DM in a real-world setting. In South Korea, rtCGM and isCGM sensors and rtCGM transmitters for individuals with T1DM have been partially reimbursed by the Korean National Health Insurance Service (NHIS) since 2019 and 2020, respectively [17]. We used nationwide cohort data from the NHIS to compare the glycemic control status of rtCGM users and isCGM users, both adults and children.
METHODS
Data source
We used the customized Korean NHIS database (2016–2022), which contains data for nearly the entire South Korean population [18]. This database contains longitudinal information on demographics, medical conditions (recorded as International Classification of Disease, 10th Revision [ICD-10] codes), and treatments. Prescription records (2019–2022) for CGM devices, including the start date; duration; and metrics of glycosylated hemoglobin (HbA1c), mean glucose, and coefficient of variation (CV) or standard deviation (SD), were used in our analyses.
The Institutional Review Board of Samsung Medical Center approved this study (approval no. SMC 2023-04-090). All procedures involving human individuals were conducted in accordance with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The requirement for informed consent was waived because the data were publicly available and de-identified.
Study population
The flow chart of patient inclusion is presented in Supplementary Fig. 1. From the NHIS database, we selected insulin users with T1DM who used a CGM sensor at least once between 2019 and 2022 (n=11,271). We included insulin users with T1DM who fulfilled one of the following criteria because the CGM devices were reimbursed only for this population in South Korea: (1) fasting C-peptide level ≤0.6 ng/mL (0.2 nmol/L); (2) stimulated C-peptide level ≤1.8 ng/mL (0.6 nmol/L); (3) positive result for glutamic-acid-decarboxylase or other autoantibodies; (4) urine C-peptide level less than 30 μg per day; or (5) history of diabetic ketoacidosis at the time of diabetes diagnosis [19]. The index date was set as the date of the first prescription for a CGM sensor. Next, we selected individuals who had a minimum of three prescriptions for rapid-acting insulin after the index date as those receiving intensive insulin therapy (n=10,816). Finally, individuals with HbA1c levels available at baseline and at least one subsequent follow-up time or CGM metrics at 3 months and at least one subsequent follow-up time were included (n=7,786; 5,875 adults [≥19 years] and 1,911 children and adolescents [<19 years]). Because some prescription records included either HbA1c levels or CGM metrics, we constructed separate datasets: the HbA1c dataset and the CGM dataset. Individuals who had HbA1c levels at both the initial assessment and at least one subsequent follow-up time were included in the HbA1c dataset (n=4,333; 3,148 adults and 1,185 children and adolescents). Individuals who had CGM metrics at both 3 months after the index date and at least one subsequent follow-up time were included in the CGM dataset (n=6,257; 4,731 adults and 1,526 children and adolescents). A total of 2,804 individuals (2,004 adults and 800 children and adolescents) were included in both the HbA1c and CGM datasets. HbA1c levels were followed from the initial assessment (baseline) to 24 months, but CGM metrics were followed from 3 months after CGM initiation to 24 months because baseline CGM metrics were not available.
Outcome measures
We compared the glycemic parameters between users of rtCGM and isCGM. The CGM type was categorized as rtCGM (Dexcom G5, Dexcom G6 [San Diego, CA, USA], and Medtronic Guardian Sensor 3 [Minneapolis, MN, USA]) and isCGM (FreeStyle Libre 1). When individuals changed their type of device during follow-up, the noted type was that used for the longest amount of time.
The primary outcome was the change in HbA1c levels. We also estimated the achievement rate of the HbA1c target (<7%). Secondary outcomes included the change in CV and the proportion of participants achieving CV <36%. Additional secondary outcomes included the proportion of participants simultaneously achieving HbA1c <7%, CV <36%, and a glucose management indicator (GMI) <7%.
HbA1c, CV, and GMI were followed every 3 months from baseline (HbA1c) or from 3 months (CV and GMI) to 24 months. Because mean glucose and either CV or SD values were available, the GMI was calculated as 3.31+0.02392×mean glucose (mg/dL) [20]. When SD values were provided, CVs were calculated as 100×SD/mean glucose.
Other variables
To compare the participants’ baseline characteristics, we investigated their medical history, institution type, and income status. The medical history of individuals was identified using ICD-10 codes and prescription records (details are described in Supplementary Table 1). Institutions were categorized as primary, secondary, and tertiary based on location of insulin prescription during the course of 1 year. If a patient received insulin from a higher institution at least once, their medical institution was classified as the higher institution. Individual income status was classified as <20th percentile or ≥20th percentile depending on their medical insurance premium, which was directly proportional to income.
As the database did not contain the information on the percentage of sensor wear time, we investigated the adherence to CGM, defined as the sum of the duration of CGM prescription divided by the duration of the observational period.
Statistical analysis
Percentages are used to report all category variables. Continuous variables are represented as mean and SD. The baseline characteristics of participants were compared according to the types of CGM devices used, with stratification by an age cutoff of 19 years. We used the independent two-sample t-test to compare continuous measurements and either the chi-square or Fisher’s exact test to compare categorical variables. We compared the HbA1c and CV of rtCGM users with those of isCGM users at 3-month intervals using an independent two-sample t-test at each time point. From the initial HbA1c measurements, we also determined the mean and SD of changes in subsequent observations relative to baseline. Using the independent two-sample t-test, we examined whether the change in HbA1c levels differed significantly between rtCGM and isCGM users. We determined the target achievement rate by setting a goal of 7% for HbA1c and GMI and 36% for CV. The chi-square test was used to determine the significance of differences between two groups. We used only the available data rather than imputing missing values.
Additionally, from 3 to 21 months after initiation of CGM, we used a repeated measures analysis in a linear mixed effect model to handle missing data and incorporate all available data. Data at 24 months were excluded because of the high rate of missing data. Three models with different levels of adjustment were fitted to ensure that any baseline confounding variables were considered. Model 1 was unadjusted. Model 2 had adjustments for age and sex. Model 3 further considered income status, institution, and baseline HbA1c (in the HbA1c dataset) or CGM metrics at 3 months (in the CGM dataset). Finally, to confirm the quality of the CGM metrics data, we identified the relationship between HbA1c and GMI at all available follow-up months.
Statistical analyses were conducted using R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and SAS Enterprise Guide version 7.1 for Windows (SAS Institute, Cary, NC, USA). A significance level of P<0.05 was applied to each analysis.
RESULTS
Baseline characteristics of the study population
The baseline features of individuals who had HbA1c data at baseline and follow-up are presented in Table 1 (n=4,333). The mean age of the study individuals was 34.5±19.8 years (adults, 43.3±16.0 years; children and adolescents, 11.4±4.2 years); 54% of them were female (adults, 52%; children and adolescents, 58%); and 35.2% were rtCGM users. While 26.1% of adults received rtCGM, 59.3% of children and adolescents received rtCGM. In both adults and children, compared with isCGM users, rtCGM users were younger and more likely to receive treatment at a tertiary hospital. In adults, rtCGM users were significantly more likely to have a history of hypoglycemia than isCGM users (11.8% vs. 8.8%, P=0.016), but this difference was not significant in children and adolescents. In adults, isCGM users had more comorbidities such as hypertension, dyslipidemia, and cancer. Individuals with income status ≥20th percentile used rtCGM more than those with low income status, especially among children and adolescents (89.8% vs. 85.3%, P=0.025). The baseline features of individuals who had available CGM data (n=6,257) (Supplementary Table 2) were similar to those in the HbA1c dataset (Table 1). The mean adherence to CGM was 80% in both adults and children (82% for isCGM users and 80% for rtCGM users).
Baseline characteristics of individuals in the HbA1c dataset (n=4,333)
Changes in HbA1c
The changes in HbA1c levels from the initial measurement to subsequent assessments are illustrated in Fig. 1, Supplementary Table 3. In the total population, rtCGM users had higher baseline HbA1c than isCGM users (8.9%±2.7% vs. 8.6%±2.2%, P<0.001), but at 3 months, the HbA1c of rtCGM users was significantly lower than that of isCGM users (7.1%±1.2% vs. 7.5%±1.3%, P<0.001) (Fig. 1A). The HbA1c of rtCGM users was consistently lower than that of isCGM users at all specific time points from 3 to 24 months (P<0.001 for all).
Glycosylated hemoglobin (HbA1c) levels from baseline to 24 months in (A) the total population, (B) adults, and (C) children and adolescents. Mean±standard error is shown. rtCGM, real-time continuous glucose monitoring; isCGM, intermittently-scanned continuous glucose monitoring. aP<0.05, bP<0.01, cP<0.001.
In adults, rtCGM users had lower HbA1c levels than isCGM users at baseline (7.9%±1.8% vs. 8.4%±1.9%, P<0.001) and throughout the follow-up period (7.1%±1.2% vs. 7.5%±1.2% at 3 months, P<0.001; 7.0% ±1.1% vs. 7.3% ±1.2% at 24 months, P=0.016) (Fig. 1B). The difference between the two groups in terms of the change in HbA1c, compared with HbA1c levels at 3 months, was found to be significant at both 18 and 24 months (P=0.041 from 3 to 18 months, P=0.027 from 3 to 24 months) (Supplementary Table 3).
In children and adolescents, rtCGM users had higher HbA1c than isCGM users at baseline (10.1%±3.0% vs. 9.7%±2.9%, P=0.016) but lower HbA1c from 3 to 24 months (7.0%±1.1% vs. 7.4%±1.4% at 3 months, P<0.001; 7.2%±1.2% vs. 7.8%±1.6% at 24 months, P=0.002) (Fig. 1C). The HbA1c level in rtCGM and isCGM users was lowest at 3 months.
In the linear mixed effect model, every 3 months from 3 to 21 months, rtCGM use was associated with a greater reduction in HbA1c compared to isCGM use in both adults (P=0.034) and children and adolescents (P=0.037), after adjustment for age, sex, income status, institution, and baseline HbA1c level (model 3) (Supplementary Table 4).
Fig. 2 presents the change in HbA1c target (<7%) achievement rate. In both the total population and adults, rtCGM users had a higher HbA1c target achievement rate than isCGM users during the entire follow-up period. In adults, the HbA1c target achievement rate of rtCGM users increased steadily from 31.1% at baseline to 58.1% at 24 months, whereas that of isCGM users increased from 21.3% to 43.4%. In children and adolescents, rtCGM users had significantly higher HbA1c target achievement rates at 3, 6, 9, and 24 months.
Glycosylated hemoglobin (HbA1c) target (<7%) achievement rate from baseline to 24 months in (A) the total population, (B) adults, and (C) children and adolescents. rtCGM, real-time continuous glucose monitoring; isCGM, intermittently-scanned continuous glucose monitoring. aP<0.05, bP<0.01, cP<0.001.
Changes in CV
CGM metrics were followed from 3 to 24 months (Fig. 3, Supplementary Table 5). In both adults and children, rtCGM users had consistently lower CVs than isCGM users during follow-up (Fig. 3A-C). In adults, the CV of rtCGM users decreased from 34.1%±12.1% at 3 months to 32.4%±14.1% at 24 months, whereas the CV of isCGM users decreased from 38.8%±11.4% to 37.3%±10.0%. The change in CV from 3 to 24 months was greater in rtCGM users (3.2%±16.0% vs. 0.6%±11.2%, P=0.004). In children and adolescents, CV did not decrease significantly during follow-up in either rtCGM or isCGM users. Instead, CV increased in isCGM users from 34.8%±13.6% at 9 months to 36.9%±12.6% at 21 months, whereas it remained stable in rtCGM users.
Coefficient of variation (CV) of (A) the total population, (B) adults, and (C) children and adolescents, and the proportion of people achieving a CV of <36% in (D) the total population, (E) adults, and (F) children and adolescents, from 3 to 24 months. Mean±standard error is shown in Fig. 3A-C. rtCGM, real-time continuous glucose monitoring; isCGM, intermittently-scanned continuous glucose monitoring. aP<0.05, bP<0.01, cP<0.001.
In the linear mixed model, the average difference between the mean reduction in CV of rtCGM users and that of isCGM users every 3 months from 3 to 21 months in adults and children was −0.310% (95% confidence interval [CI], −0.445 to −0.175; P<0.001) and −0.291% (95% CI, −0.519 to −0.063; P=0.012), respectively, after adjustment for age, sex, income status, institution, and metrics at 3 months (model 3) (Supplementary Table 6).
The proportion of participants achieving CV <36% was higher among rtCGM users, both adults and children, at every time point (Fig. 3D-F).
The proportion of participants achieving HbA1c <7%, CV <36%, and GMI <7% simultaneously
We additionally investigated the proportion of participants simultaneously achieving HbA1c <7%, CV <36%, and GMI <7% (n=2,804). This rate was higher in rtCGM users than in isCGM users throughout follow-up in the total population and in adults (Supplementary Fig. 2). In adults, the rate increased from 21.0% at 3 months to 28.3% at 24 months in rtCGM users and from 12.6% to 16.4% in isCGM users (Supplementary Fig. 2B). In children and adolescents, rtCGM users also showed higher rates than isCGM users, but the rate was highest at 3 months (Supplementary Fig. 2C).
DISCUSSION
In this nationwide cohort study, the use of rtCGM was associated with a greater reduction in HbA1c levels than the use of isCGM in individuals with T1DM. Even though the baseline HbA1c level of rtCGM users was higher than that of isCGM users, the HbA1c levels of rtCGM users from 3 to 24 months were consistently lower than those of isCGM users.
Previous RCTs demonstrated the superiority of rtCGM over isCGM in glycemic control [14-16]. The CORRIDA trial demonstrated that rtCGM was superior to isCGM in reducing time below range and improving time in range for 60 adults with T1DM during 4 days of physical activity and a subsequent 4 weeks at home [15]. The ALERTT1 trial demonstrated that switching from isCGM to rtCGM significantly improved time in range after 6 months of treatment in 254 adults with T1DM [14], and its extension study observed that glycemic control improved for up to 24 months after switching from isCGM to rtCGM [21]. However, to the best of our knowledge, no previous large-scale study compared rtCGM and isCGM in a real-world setting. The evidence supporting the superiority of rtCGM over isCGM in the pediatric population was especially insufficient because the RCTs were conducted in adults [14-16,22]. Our study included 1,911 children and adolescents, and they showed a greater reduction in HbA1c levels with rtCGM than with isCGM. In children and adolescents, the mean HbA1c level decreased from 10.1% to 7.2% in rtCGM users, whereas it decreased from 9.7% to 7.8% in isCGM users.
In adults, the superiority of rtCGM over isCGM in improving HbA1c levels was less evident than in children and adolescents because the mean HbA1c level of rtCGM users was lower than that of isCGM users at baseline. However, the proportion of people achieving the target HbA1c level (<7%) showed a larger increase with rtCGM (from 31.1% to 58.1%) than with isCGM (from 21.3% to 43.4%). Additionally, the linear mixed model showed that the HbA1c levels decreased slightly more with rtCGM than with isCGM after adjusting for baseline HbA1c levels and other confounding factors. Nevertheless, selection bias, where well-controlled and highly motivated adults might tend to use rtCGM devices more often than isCGM devices, might have affected this finding. Our study demonstrates the real-world practice in South Korea: poorly controlled adults received isCGM devices rather than rtCGM devices, and they had consistently higher HbA1c levels than the well-controlled individuals who received rtCGM devices. Notably, only one-fourth of adults with T1DM used rtCGM, with the rest using isCGM. The low prescription rate of rtCGM might be related to its price and the additional process of attaching a transmitter to the sensor.
In a previous hospital-based cohort study, we observed a large CV decrease between 3 and 12 months among persons with T1DM who used CGM of any kind [23]. Our ability to evaluate the association between CV improvement and use of rtCGM or isCGM in the current study was limited because the database did not contain baseline CV data. However, from 3 to 24 months, CV showed a larger decrease with rtCGM than with isCGM, consistent with the results of a previous RCT [14]. At every time point from 3 to 24 months, the mean CV of rtCGM users was lower than that of isCGM users. The distinct characteristics of rtCGM (the presence of alarms and the availability of real-time glucose values without scanning) probably contribute to the greater improvement in CV and HbA1c levels.
Another noteworthy aspect of our study is that a significant improvement in HbA1c levels was observed after 3 months of using CGM, and those decreased HbA1c levels remained for 2 years. After 3 months, the reduction in HbA1c levels was more than 1%, which is impressive because CGM was associated with a modest reduction in HbA1c of 0.17% in a meta-analysis of RCTs in T1DM and type 2 diabetes mellitus patients [24]. Our finding might be related to a nationwide education program for T1DM, the Korean national home care pilot program for T1DM, which provides repeated systematic education about insulin dose adjustment, CGM data interpretation, and carbohydrate counting [25]. Especially in adults, we observed a gradual increase in the number of people achieving the HbA1c <7% target, along with HbA1c <7%, GMI <7%, and CV <36% simultaneously. Still, less than one-third of people achieved all three criteria simultaneously.
In children and adolescents, the mean HbA1c was lowest at 3 months, and the achievement rate for the HbA1c <7% target gradually decreased after 3 months. The proportion of participants achieving CV <36% was also highest at 3 months. This trend may reflect the tendency of patients and caregivers to manage blood glucose rigorously after initiating CGM, but becoming less attentive over time, especially in this age group. Especially, children and adolescents using isCGM exhibited a gradual increase in mean HbA1c after 3 months and an increase in mean CV after 9 months, whereas the mean HbA1c and CV of those using rtCGM remained relatively stable during the same period. This could be because glycemic control with isCGM might be more influenced by patient behaviors, such as the frequency of scanning. Given the gradual worsening of glycemic parameters in children and adolescents after 3 months, additional efforts at the societal level may be needed to maintain optimal glycemic control in this population.
A key strength of our study is that we studied follow-up data for all individuals with T1DM receiving CGM in South Korea. We analyzed data for 7,786 individuals, including 1,911 children and adolescents. Our findings support the superiority of rtCGM over isCGM in glycemic control and have important implications for clinical decision making. No previous nationwide study has reported the glycemic status of CGM users in South Korea. Our study offers insight into the current glycemic status of CGM users with T1DM in South Korea.
It is important to note that the glycemic profiles of our study participants reflect CGM users in South Korea, rather than the entire T1DM population. During the study period, we estimated that the T1DM population receiving intensive insulin therapy in South Korea was 56,908. Among them, only 7,786 used CGM at least twice and included in our analyses. Therefore, our study population may represent early CGM technology adopters who were highly motivated to control their glucose levels. These characteristics of the study population may have partly contributed to the significant HbA1c reduction observed after initiating CGM.
Our study has several limitations. First, data on CGM metrics were analyzed only for mean glucose, CV, and GMI. Other core CGM metrics, such as time in range and time below range, and baseline CGM metrics were unavailable. The percentage of sensor wear time, the exact number of days the CGM device was worn, and the frequency of isCGM scanning were also unavailable. Second, information regarding whether the participants received education combined with CGM was not analyzed due to a lack of data. In RCTs, structured education combined with CGM has been shown to result in better glycemic control than CGM alone [26,27]. We believe that the significant HbA1c reduction observed upon initiating CGM is partly attributed to the education; however, we could not analyze the glycemic profiles according to education status. Third, we collected the HbA1c level and CGM metrics from prescription records, so we cannot exclude the possibility of inaccuracy. However, we believe that the prescription records are reliable because the mean GMI was 7.3%±1.0%, the mean HbA1c was 7.3%±1.2%, and the mean difference between GMI and HbA1c was 0.1%±0.8%, in the 13,558 readings that measured HbA1c and CGM metrics simultaneously in a specific follow-up month (Supplementary Fig. 3). Fourth, a relatively small number of people were followed for more than 12 months, and some participants had missing data at certain time points. This could bias the results, as those without missing data may have been more motivated to manage their glycemic control. Fifth, the majority of the study participants were adults. To address this limitation, we also conducted separate analyses for adults and children, respectively. Sixth, there is a possibility that individuals with non-T1DM who experienced diabetic ketoacidosis at the time of diabetes diagnosis were included in the current analyses. Finally, due to the retrospective study design, we cannot establish causal relationships. As mentioned earlier, our results might be affected by selection bias because well-controlled and highly motivated adults might tend to use rtCGM devices more than isCGM devices.
In this large nationwide cohort study, the use of rtCGM was associated with a greater improvement in glycemic control, including HbA1c reduction, than the use of isCGM in both adults and children with T1DM. Clinicians should consider this finding when prescribing CGM devices.
SUPPLEMENTARY MATERIALS
Supplementary materials related to this article can be found online at https://doi.org/10.4093/dmj.2024.0160.
Identification of medical history
Baseline characteristics of individuals in the CGM dataset (n=6,257)
HbA1c levels from baseline to 24 months (n=4,333)
Average difference in the HbA1c (%) level every 3 months (from 3 to 21 months) as estimated by a repeated measures analysis in a linear mixed effect model (n=4,333)
Coefficient of variation from 3 to 24 months (n=6,257)
Average difference in CV (%) every 3 months (from 3 to 21) as estimated by a repeated measure analysis in a linear mixed effect model (n=6,257)
Flow diagram of the study subject inclusion. CGM, continuous glucose monitoring; HbA1c, glycosylated hemoglobin.
The rate at which (A) the total population, (B) adults, and (C) children and adolescents achieved glycosylated hemoglobin (HbA1c) <7%, coefficient of variation (CV) <36%, and glucose management indicator (GMI) <7% simultaneously, from 3 to 24 months. rtCGM, real-time continuous glucose monitoring; isCGM, intermittently-scanned continuous glucose monitoring. aP<0.05, bP<0.01, cP<0.001.
Difference between the glucose management indicator (GMI) and glycosylated hemoglobin (HbA1c) (13,558 readings of simultaneously measured HbA1c and CGM metrics). The mean GMI was 7.3%±1.0%, the mean HbA1c was 7.3%±1.2%, and the mean difference between GMI and HbA1c was 0.1%±0.8% (HbA1c=−0.064+1.016×GMI).
Notes
CONFLICTS OF INTEREST
No potential conflict of interest relevant to this article was reported.
AUTHOR CONTRIBUTIONS
Conception or design: all authors.
Acquisition, analysis, or interpretation of data: all authors.
Drafting the work or revising: all authors.
Final approval of the manuscript: all authors.
FUNDING
None
ACKNOWLEDGMENTS
This study was conducted as part of a special project by the National Health Insurance Service and the Korean Diabetes Association. We thank the National Health Insurance Service for developing this database.