To Use or Not to Use a Self-monitoring of Blood Glucose System? Real-world Flash Glucose Monitoring Patterns Using a Cluster Analysis of the FGM-Japan Study

Abstract

Objective

This study investigated self-monitoring of blood glucose (SMBG) adherence and flash glucose monitoring patterns using a cluster analysis in Japanese type 1 diabetes (T1D) patients with intermittently scanned continuous glucose monitoring (isCGM).

Methods

We measured SMBG adherence and performed a data-driven cluster analysis using a hierarchical clustering in T1D patients from Japan using the FreeStyle Libre system. Clusters were based on three variables (testing glucose frequency and referred Libre data for hyperglycemia or hypoglycemia).

Patients

We enrolled 209 participants. Inclusion criteria were patients with T1D, duration of isCGM use ≥3 months, age ≥20 years old, and regular attendance at the collaborating center.

Results

The rate of good adherence to SMBG recommended by a doctor was 85.0%. We identified three clusters: cluster 1 (low SMBG test frequency but high reference to Libre data, 17.7%), cluster 2 (high SMBG test frequency but low reference to Libre data, 34.0%), and cluster 3 (high SMBG test frequency and high reference to Libra data, 48.3%). Compared with other clusters, individuals in cluster 1 were younger, those in cluster 2 had a shorter Libre duration, and individuals in cluster 3 had lower time-in-range, higher severe diabetic distress, and high intake of snacks and sweetened beverages. There were no marked differences in the incidence of diabetic complications and rate of wearing the Libre sensor among the clusters.

Conclusion

We stratified the patients into three subgroups with varied clinical characteristics and CGM metrics. This new substratification might help tailor diabetes management of patients with T1D using isCGM.

Introduction

Currently, glucose monitoring is essential for diabetes management and for achieving fewer variations in glycemic control (1). At present, patients with type 1 diabetes (T1D) may choose two types of glucose monitoring systems: self-monitoring of blood glucose (SMBG), which measures glucose levels within the capillary blood, and continuous glucose monitoring (CGM), which measures glucose levels within the interstitial fluid.

Although SMBG and CGM systems offer different functionalities, both are intended to help patients improve their glycemic control. SMBG provides only a snapshot of the glucose levels, and the use of finger sticks is limited by cost. In contrast, the intermittently scanned CGM (isCGM) system can provide patients with up to 288 glucose values per day as well as temporal glucose trends and patterns. The IMPACT study indicated that SMBG testing frequency was reduced after introducing isCGM in European patients with T1D (2). The SMBG testing frequency varied after the introduction of real-time CGM in the DIAMOND trial, which included 99 T1D patients, although all patients were instructed to use SMBG results for diabetes management decisions (3). SMBG remains one of the most widely used methods to monitor glucose levels, given its relative accuracy, familiarity, and manageable costs (4-6).

In Japan, the Ministry of Health, Labour, and Welfare granted national reimbursement for the FreeStyle Libre glucose monitoring system in 2017. Recently, in April 2022, the Japan Diabetes Society, in their article titled, “Intermittently scanned continuous glucose monitoring system: the statement on FreeStyle version.4”, stated that blood glucose should be monitored using SMBG on an as-needed basis; however, the recommendation of SMBG adherence by a doctor is unknown.

Recently, a cluster analysis was performed in diabetes research (7). The present study therefore investigated the real-world SMBG adherence and flash glucose monitoring patterns using a cluster analysis in Japanese T1D patients using isCGM.

Materials and Methods

Study design

FGM-Japan was a cross-sectional study designed to evaluate adherence to SMBG and isCGM patterns in adults with T1D in a real-world setting in Japan. This study conformed to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for reporting cross-sectional studies (8).

The present study was approved by the ethics committee of the National Hospital Organization Kyoto Medical Center (No.19-072). Written informed consent was obtained from all the participants. The studies were registered in the UMIN Clinical Trials Registry (UMIN000039376).

Participants

Patients were recruited from a collaborating center between February 2020 and March 2022. Inclusion criteria were patients with T1D that was diagnosed by a diabetes specialist according to the diagnostic criteria of the Japan Diabetes Society Committee on Type 1 Diabetes Research (9-11), duration of isCGM use ≥3 months, age ≥20 years old, and regular attendance at any of 10 collaborating centers (Kyoto Medical Center, Kobe University Graduate School of Medicine, Tokyo Women's Medical University School of Medicine, Tokai University School of Medicine, National Cerebral and Cardiovascular Center, Nara Medical University, Okayama Saiseikai General Hospital, Kanda Naika Clinic, Sawaki Internal Medicine and Diabetes Clinic, and Matsuda Diabetes Clinic). Exclusion criteria were non-insulin therapy, anti-dementia drug use, and inappropriate cases judged by the research director or coordinators.

Clinical characteristics

Treatment for diabetic retinopathy, nephropathy, and peripheral neuropathy was performed by certified diabetologists according to the treatment guidelines for diabetes established between 2018 and 2019.

Diabetic retinopathy was assessed by an ophthalmologist using retinal photography. Retinopathy was classified as absent, simple, pre-proliferative, and proliferative. Nephropathy was classified as stages 1 to 5 based on the estimated glomerular filtration rate, albuminuria, or hemodialysis stage (12). Diabetic peripheral neuropathy (DPN) was diagnosed following specific criteria after patients were diagnosed with diabetes and excluded from having polyneuropathy except for diabetic polyneuropathy. DPN was determined to be positive in the presence of at least two of the following three criteria: 1) subjective symptoms (numbness, pain, or dysesthesia in the bilateral lower extremities), 2) decreased or absent bilateral Achilles tendon reflexes, and 3) diminished bilateral vibratory sensation at the malleolus medialis (＜10 s using a tuning fork at 128 Hz) (13).

The self-reported number of severe hypoglycemic episodes in the preceding year, defined as “a hypoglycemic episode that you were unable to treat yourself” was collected. The self-reported number of hypoglycemic episodes in the preceding month was recorded. Impaired awareness of hypoglycemia was determined using Gold's method. Data on hemoglobin A1c (HbA1c), liver enzymes (aspartate aminotransferase, alanine transaminase, and gamma-glutamyl transferase), and lipid profiles were collected from medical records. The neutrophil-to-lymphocyte ratio, defined as the neutrophil count divided by the lymphocyte count, and the platelet-to-lymphocyte ratio, defined as the platelet count divided by the lymphocyte count, were also recorded.

Lifestyle factors

Self-administered questionnaires regarding lifestyle behaviors (current smoking, regular exercise, dietary habits, drinking habits, and sleeping habits) were collected using a standardized questionnaire from the Specific Health Check and Guidance System (14). Exercise habits included three items: 1) regular exercise (≥2 times/week of exercise for ≥4 METs/h), 2) active physical activity (≥23 METs×h/week), and 3) walking pace (rapid or not rapid), an indicator of physical fitness. Excessive drinking was defined based on answers to the questions on drinking habits, i.e. both “occasionally or every day” and “≥180 mL of sake (equivalent to ≥20 g of alcohol)”. Sleep debt was defined as the difference between the self-reported total weekday and weekend sleep hours of at least 2 h (15). Healthy lifestyle behaviors included the intake of fruits, fish, and milk; exercise; avoidance of smoking; moderate alcohol intake; and moderate sleep duration (16).

Psychological distress

Fear of hypoglycemia was assessed using the Hypoglycemia Fear Survey (HFS) adapted for use in Japan (17,18). The HFS has two subscales: the HFS-B (behavior subscale) and HFS-W (worry subscale). A depressive state was assessed using the Patient Health Questionnaire-9 (PHQ-9) (19). Diabetes-related distress was assessed using the Problem Areas in Diabetes (PAID) questionnaire. Higher scores indicate greater diabetes-related distress, and a cut-off score of ≥40 indicates high distress (20). The health-related quality of life was assessed, and the utility index was calculated using the European Quality of Life-5-Dimension (EQ-5D) questionnaire (21).

CGM metrics

CGM metrics, including the average daily risk range (ADRR), average glucose, glucose management indicator, high blood glucose index, low blood glucose index, mean amplitude of glycemic excursion, mean daily difference for inter-day variation, standard deviation, time-in-range (TIR) 70-180 mg/dL, time-below-range ＜70 mg/dL or ＜54 mg/dL, coefficient of variation, and time-above-range ＞180 mg/dL or ＞250 mg/dL (22-24), were collected during the last 90 days using the FreeStyle Libre system (Abbott Diabetes Care, Alameda, USA). CGM metrics were calculated for adequate isCGM data (≥70%) (25). The ADRR was scored based on the following risk categories: low risk, 0-19; moderate risk, 20-40; and high risk, ≥41.

Diabetes self-management

Good adherence was defined as a number of SMBG procedures at or greater than the number recommended by the doctor, whereas poor adherence was defined as a number of SMBG procedures less than the number recommended by the doctor. Diabetes management behaviors included SMBG test frequency, referred to Libre data for hyperglycemia (0-100%), such as the correction of bolus insulin injections for hyperglycemia, referred to Libre data for hypoglycemia (0-100%), such as the consumption of sugar or snacks for hypoglycemia and driving decisions without SMBG. Five benefits and nine barriers for SMBG were evaluated using a self-administered questionnaire.

Statistical analyses

Cluster analyses using a hierarchical clustering with weighted Euclidean distance (26) were performed to stratify patients according to three variables (testing glucose frequency and referred Libre data for hyperglycemia or hypoglycemia). Clusters were defined and classified into three groups according to the weighted Euclidean distance. The research team then created names to describe each cluster. The results were analyzed using a one-way analysis of variance and Tukey's test at a 5% significance level. Qualitative variables were compared using Fisher's exact test, and Benjamini-Hochberg adjustment was used for multiple comparisons, such as the diabetic nephropathy and retinopathy categories.

The analysis was conducted using the R software program, version 4.1.2. Glycemic variability was calculated using the Gluvarpro and iglu R software packages.

Results

Clinical characteristics and diabetes management behavior

A total of 209 adults [85 men (40.7%) and 124 women (59.3%)] had a mean age of 50.9±15.2 years old, diabetes duration of 16.3±11.8 years, Libre usage duration of 2.1±1.0 years, and HbA1c of 7.6%±0.9%. The rate of good adherence to SMBG recommended by a doctor was 85.0%. The rates of using specific anatomic sites for placing the Libre system were 88.5%, 6.2%, 8.6%, and 2.9% for subcutaneous placement of the device on the upper arm, abdomen, upper thigh, and other sites, respectively. The participants perceived the following benefits of SMBG: checked the differences in glucose levels between SMBG and isCGM (66.0%), confirmed the glucose values by SMBG, and followed recommendations by the doctor (21.5%). However, they also perceived a few barriers to using SMBG, including expressing inconvenience (33.0%), referring to glucose levels based on isCGM data (29.7%), and high cost (9.1%).

Clustering and clinical characteristics

The patients were divided into 3 clusters: Libre-based users (n=37, 17.7%), SMBG-based users (n=71, 34.0%), and both device-based users (n=101, 48.3%). The first cluster was characterized by a high rate of diabetes management based on Libre data and a low SMBG test frequency (Table 1) and had poor adherence to SMBG recommended by a doctor. The second cluster was characterized by a low rate of diabetes management based on Libre data and a high SMBG test frequency. The third cluster had a high proportion of patients referring to Libre data and a high proportion of SMBG test frequency.

Table 1.

Clinical Characteristics of the Participants.

Variables	Cluster 1	Cluster 2	Cluster 3	p value
	Libre-based users	SMBG-based users	Both devices-based users
	(n=37)	(n=71)	(n=101)
Diabetes management behavior
SMBG test frequency, per day	0.7 (0.7)	5.3 (1.0)*1	4.8 (1.3)*1 *2	<0.001†
Referred to Libre data for hyperglycemia, %	69.0 (30.1)	18.5 (19.0)*1	79.1 (19.0)*1 *2	<0.001†
Referred to Libre data for hypoglycemia, %	67.9 (26.3)	32.6 (27.5)*1	74.0 (23.8)*2	<0.001†
Prescribed SMBG test frequency recommended by a doctor, per month	93.0 (35.9)	109.8 (26.8)	100.9 (29.1)	0.131
Good adherence to SMBG recommended by a doctor, %	37.8	95.7*1	95.0*1	<0.001†
Wearing rate of Libre sensor, %	89.7 (7.9)	90.6 (7.7)	88.9 (8.4)	0.409
Scan frequency, per day	10.3 (3.4)	10.7 (2.9)	10.3 (3.5)	0.671
Age, years	46.7 (13.4)	56.7 (14.7)*1	48.3 (15.0)*2	<0.001†
Male sex, %	29.7	43.7	42.6	0.334
Diabetes duration, years	19.6 (13.4)	15.3 (11.8)	15.7 (11.1)	0.166
Libre usage duration, years	2.3 (0.8)	1.8 (1.1)*1	2.3 (1.0)*2	0.005†
BMI, kg/m2	22.6 (3.0)	22.8 (3.7)	22.9 (3.6)	0.878
Diabetic complication, %
Retinopathy
NDR/SDR/PPDR/PDR	68.6/14.3/0/17.1	77.5/14.1/1.4/7.0	75.0/16.0/1.0/8.0	0.727
Photocoagulation	20	7	10	0.124
Nephropathy
1st/2nd/3rd/4th/5th	83.8/8.1/2.7/0/5.4	82.9/12.9/4.3/0/0	87.1/7.9/3.0/0/2.0	0.526
Peripheral neuropathy	18.9	29.6	18.2	0.197
Coronary artery disease, %	0	7	4	0.280
Cerebrovascular disease, %	2.7	5.6	5	0.914
Peripheral arterial disease, %	2.7	5.6	3	0.632
IAH, %	18.9	23.9	13	0.160
SH, %	10.8	14.3	9	0.585
Diabetes treatment, %
CSII, %	29.7	15.5	26.7	0.134
TDD, U/day	32.1 (11.9)	31.3 (16.3)	36.7 (15.7)	0.056
U/kg·day	0.53 (0.15)	0.53 (0.22)	0.60 (0.19)	0.038†
Basal, %	37.1 (14.9)	31.7 (15.3)	33.4 (12.5)	0.165
Neuropathy-related medicine, %
Mecobalamin	0	4.2	3.0	0.565
Epalrestat	0	1.4	1.0	>0.999
Pregabalin and duloxetine	0	0	3.0	0.292

Mean (standard deviation). ^*1p<0.05 (vs. Libre-based user group); ^*2p<0.05 (vs. SMBG-based user group). ^†p<0.05.

SMBG: self-monitoring of blood glucose, TDD: total daily dose of insulin, CSII: continuous subcutaneous insulin infusion, SH: severe hypoglycemia, IAH: impaired awareness of hypoglycemia, NDR: no diabetic retinopathy, SDR: simple diabetic retinopathy, PPDR: pre-proliferative diabetic retinopathy, PDR: proliferative diabetic retinopathy, BMI: body mass index, SPIDDM: slowly progressive insulin-dependent diabetes mellitus

The clinical characteristics according to the cluster are presented in Table 1. Patients in Cluster 1 were younger, and those in Cluster 2 were older than those in Cluster 3. Cluster 2 had a shorter Libre usage than Clusters 1 and 3. There were no marked differences in diabetic complications, hypoglycemia, or the rate of wearing a Libre sensor among the clusters. There were also no marked differences in the treatment of diabetes or hypertension among the clusters.

The prevalence of severe diabetic distress (PAID total score, ≥40) in Cluster 3 was higher than in other clusters. Patients in Cluster 1 reported “diabetes management depends on Libre data” as a barrier of SMBG (Table 2). Patients in Cluster 3 had a high intake of snacks and sweetened beverages (Table 3). Patients in Cluster 3 had a significantly higher rate of severe diabetes distress than patients in Cluster 1 (p=0.026). There were no marked differences in sleeping habits or laboratory data (Table 4) among the clusters. Patients in Cluster 3 had a lower TIR than those in other clusters (Table 5). There were no marked differences in scan frequency or rate of wearing a Libre sensor among the clusters.

Table 2.

Psychological Distress, Barriers, and Benefits of SMBG according to the Cluster.

Variables	Cluster 1	Cluster 2	Cluster 3	p value
	Libre-based users	SMBG-based users	Both devices-based users
Psychological distress, points
HFS-B	19.1 (5.1)	20.2 (6.0)	21.3 (5.4)	0.095
HFS-W	14.0 (11.4)	14.1 (9.8)	15.8 (10.6)	0.473
PHQ-9	4.1 (4.6)	3.4 (3.6)	4.3 (5.0)	0.447
PAID	27.8 (19.4)	26.5 (19.2)	32.9 (20.6)	0.099
≥40 points, %	18.9	23.9	40.6*1	0.016†
EQ-utility	0.91 (0.13)	0.92 (0.12)	0.90 (0.15)	0.602
Barriers of SMBG, %
1. high cost	0	15.5*1	7.9	0.018†
2. do not want to see hyperglycemia	5.4	2.8	3	0.767
3. do not want to see hypoglycemia	2.7	1.4	2	>0.999
4. inconvenient	45.9	28.2	31.7	0.174
5. cannot measure in workplace	37.8	16.9	24.8	0.058
6. not ready to measure	8.1	2.8	5.9	0.451
7. have no knowledge or confidence	0	1.4	0	0.517
8. depends on libre data	45.9	16.9*1	32.7	0.005†
9. others	18.9	15.5	11.9	0.526
None	8.1	25.4	21.8	0.088
Facilitators of SMBG, %
1. confirmed difference between libre and SMBG	67.6	71.8	61.4	0.351
2. confirm SMBG-based glucose data	45.9	66.2	49.5	0.052
3. recommended by a doctor	16.2	29.6	17.8	0.146
4. recommended by his/her family	5.4	2.8	3	0.767
5. others	10.8	16.9	10.9	0.514
None	0	1.4	1	>0.999

^*1p<0.05 (vs. Libre-based users). ^†p<0.05.

SMBG: self-monitoring of blood glucose, HFS-B: hypoglycemia fear survey (behavior subscale), HFS-W: hypoglycemia fear survey (worry subscale), PHQ-9: patient health questionnaire-9, PAID: problem areas in diabetes survey, EQ: European quality of life

Table 3.

Lifestyle Factors of the Participants according to the Cluster.

Variables	Cluster 1	Cluster 2	Cluster 3	p value
	Libre-based user	SMBG-based user	Both devices-based user
Eating habits, %
Skipping breakfast	16.2	5.6	12.9	0.141
Fast eating	21.6	29.6	37.6	0.191
Late-night dinner eating	35.1	16.9	30.7	0.059
Snack and sweetened beverage	80.6	55.7	77.2*1	0.004†
Fruits, ≥1 intake per day	24.3	29.6	32.7	0.643
Milk, ≥1 intake per day	62.2	74.6	63.4	0.241
Fish, ≥1 intake per day	13.5	11.3	11	0.918
Vegetables, ≥5 dishes intake per day	5.4	4.2	2	0.412
Exercise habit, %
Exercise	21.6	42.3	33.7	0.097
Physical activity	64.9	59.2	62.4	0.845
Fast walking	45.9	50.7	41.6	0.490
Lifestyle factors, %
Overwork	24.3	8.5	21.8	0.029
Current smoking	5.4	5.6	12.9	0.211
Drinking everyday	10.8	23.9	22.8	0.242
Excessive drinking	2.7	15.5	13.9	0.112
Healthy lifestyle score, points	4.8 (1.3)	4.7 (1.4)	4.6 (1.4)	0.682
Sleep habits, min
Average sleep time	394 (55)	395 (53)	404 (68)	0.550
Sleep time during a weekday	382 (59)	385 (57)	388 (74)	0.875
Sleep time during a weekend	424 (67)	418 (61)	442 (77)	0.081
Sleep debt	42 (63)	33 (54)	54 (73)	0.131
Non-restorative sleep, %	29.7	24.3	27.7	0.816

Mean (standard deviation). ^*1p<0.05 (vs. SMBG-based user group). ^†p<0.05.

SMBG: self-monitoring of blood glucose

Table 4.

Laboratory Data of the Study Participants.

Variables	Cluster 1	Cluster 2	Cluster 3	p value
	Libre-based users	SMBG-based users	Both devices-based users
TP, g/dL	7.0 (0.5)	6.9 (0.4)	7.0 (0.4)	0.745
Albumin, g/dL	4.2 (0.4)	4.1 (0.3)	4.2 (0.3)	0.392
AST, U/L	19.0 (5.5)	20.7 (6.4)	21.2 (9.4)	0.368
ALT, U/L	16.4 (8.4)	19.3 (10.7)	19.4 (10.8)	0.281
GGT, IU/L	15.5 (8.0)	19.8 (11.1)	23.4 (25.9)	0.139
BUN, mg/dL	16.6 (10.0)	17.0 (9.3)	15.0 (5.6)	0.247
Creatinine, mg/dL	1.2 (2.1)	0.8 (0.4)	0.8 (0.8)	0.161
eGFR, mL/min/1.73m2	79.3 (20.5)	75.0 (19.8)	79.8 (21.2)	0.297
HbA1c, %	7.4 (0.8)	7.4 (0.8)	7.7 (1.0)	0.191
LDL-C, mg/dL	111.6 (30.1)	101.4 (30.0)	107.4 (24.6)	0.166
HDL-C, mg/dL	80.3 (18.6)	78.9 (22.3)	74.2 (20.3)	0.208
TG, mg/dL	77.8 (36.7)	104.2 (79.7)	101.8 (60.8)	0.123
WBC, 103/μL	5.9 (2.8)	5.7 (1.8)	5.8 (1.6)	0.914
RBC, 106/μL	4.5 (0.5)	4.4 (0.5)	4.6 (0.5)	0.152
Hemoglobin, g/dL	13.6 (1.2)	13.5 (1.6)	13.6 (1.6)	0.915
Platelet, 109/L	242.9 (55.3)	238.7 (62.7)	245.9 (59.3)	0.744
NLR	2.1 (1.0)	2.7 (2.5)	2.0 (1.0)	0.188
PLR	8.4 (3.2)	9.7 (7.1)	8.5 (3.6)	0.522

Mean (standard deviation).

SMBG: self-monitoring of blood glucose, TP: total protein, Alb: albumin, ALT: alanine transaminase, AST: aspartate aminotransferase, GGT: gamma-glutamyl transferase, BUN: blood urea nitrogen, eGFR: estimated glomerular filtration rate, HbA1c: glycated hemoglobin, LDL-C: low-density lipoprotein cholesterol, HDL-C: high-density lipoprotein cholesterol, TG: triglyceride, WBC: white blood cell, RBC: red blood cell, NLR: neutrophil-to-lymphocyte ratio, PLR: platelet-to-lymphocyte ratio

Table 5.

CGM Metrics of the FGM-Japan Participants according to the Cluster.

Variables	Cluster 1	Cluster 2	Cluster 3	p value
	Libre-based users	SMBG-based users	Both devices-based users
Average glucose	142.8 (24.5)	146.6 (32.2)	152.6 (31.4)	0.187
GMI, %	6.7 (0.6)	6.8 (0.8)	7.0 (0.8)	0.187
TAR, %
>250 mg/dL	6.4 (7.4)	7.6 (9.4)	9.1 (9.1)	0.253
>180 mg/dL	24.5 (14.2)	26.6 (17.5)	31.1 (16.4)	0.059
TIR, %	65.4 (13.7)	64.2 (15.3)	59.0 (12.4)*1 *2	0.013†
TBR, %
<70 mg/dL	10.1 (7.0)	9.2 (9.7)	9.9 (9.8)	0.841
<54 mg/dL	4.4 (4.3)	3.8 (5.5)	4.3 (6.1)	0.784
Glycemic variability
ADRR	40.9 (11.7)	39.4 (12.1)	42.6 (9.3)	0.169
≥40 (high risk), %	62.2	47.9	55.4	0.356
CV	34.7 (6.8)	33.3 (6.2)	34.5 (5.6)	0.365
MAGE	142.0 (31.5)	136.5 (41.3)	148.3 (30.9)	0.094
HBGI	5.5 (3.6)	6.1 (4.6)	7.1 (4.4)	0.133
LBGI	2.6 (1.8)	2.3 (2.4)	2.5 (2.6)	0.897
SD	57.2 (13.5)	56.0 (16.4)	60.4 (13.2)	0.135
MODD	78.0 (28.3)	74.4 (32.2)	78.1 (23.8)	0.676

Mean (standard deviation). ^*1p<0.05 (vs. Libre-based user group); ^*2p<0.05 (vs. SMBG-based user group). ^†p<0.05.

SMBG: self-monitoring of blood glucose, GMI: glucose management indicator, TAR: time-above-range, TIR: time-in-range, TBR: time-below-range, ADRR: average daily risk range, CV: coefficient of variation, MAGE: mean amplitude of glycemic excursion, HBGI: high blood glucose index, LBGI: low blood glucose index, SD: standard deviation, MODD: mean daily difference for inter-day variation

Discussion

To our knowledge, this is the first study to examine adherence to SMBG and classify Japanese patients with T1D using either the Libre system or SMBG with isCGM data into three clusters.

Several real-life isCGM studies have been conducted in a real-world setting. FreeStyle Libre has several advantages: 1) finger sticks and blood samples are not required, 2) it is convenient for travel and work, 3) it is suitable for instances when a patient is on-the-go and needs to track their glucose trends, 4) it is more affordable than real-time CGM, and 5) its sensor are waterproof for short periods of swimming and bathing in a few feet of water. However, this system also has a few disadvantages: 1) irritation may be caused at the sensor insertion site, 2) the original version does not alert you if your blood glucose levels are altered until you actively check them, 3) results may not be as accurate as with a traditional blood calibration system, and 4) the sensor does not actually measure the glucose level in the blood (27,28).

The rate of adherence to SMBG recommended by a doctor was 85% in the present study, so careful attention should be paid to diabetes management in the residual 15% of patients. The least common sites for placing the Libre system device were subcutaneously on the abdomen (6.2%), upper thigh (8.6%), and at other sites, with the most common being the upper arm (88.5%). Diabetes-related health professionals should instruct patients to prefer placing the isCGM system on the upper arm because the accuracy of Libre sensors on the abdomen is poor (29).

Clinical implications

We stratified the patients into three subgroups (Libre-based, SMBG-based, and both device-based users) based on SMBG and Libre data. Libre-based users primarily referred to Libre data, although these users had a lower frequency of SMBG testing than others. SMBG should be recommended when 1) the glucose level is rapidly changing (at a rate ＞2 mg/dL per min) (30), 2) hypoglycemia is confirmed using SMBG, and 3) there is a discrepancy between hypo/hyperglycemic symptoms and glucose values based on the information available from the attached document of the pharmaceutical product. Specifically, discrepancies exist between Libre and SMBG data during exercise (31). Reddy et al. emphasized the need for continued SMBG testing among individuals at high risk for hypoglycemia who use isCGM (32). Diabetes-related health professionals should provide information based on the product's document and advise Libre-based users to use SMBG.

SMBG-based users had a high SMBG testing frequency but did not refer to Libre data. In addition, the SMBG-based users were older and had a shorter Libre usage duration than others. They might be not familiar with the use of the FreeStyle Libre system and how to read the results. Currently, two types of CGM systems are available: isCGM and real-time CGM. The real-time CGM systems available at present automatically transmit a continuous stream of glucose data to the user, provide alerts and active alarms, and transmit glucose-related data (trend and numerical) in real time to a receiver or smartphone (33,34). The ALERTT1 study indicated that switching from isCGM to real-time CGM improved the TIR and quality of life after six months of treatment in adults with T1D (35). Healthcare professionals should repeatedly explain the benefits and barriers of real-time CGM and recommend real-time CGM for tracking glucose levels accurately and conveniently, especially for older adults with T1D.

Both device-based users referred to users with both Libre data and SMBG data; however, these users had a lower TIR and a higher prevalence of severe diabetic distress than others. Consistent CGM use among young people with T1D is associated with treatment adherence and improved glycemic control without increasing their psychosocial distress (36). Many adults with T1D wish to discuss the emotional impact of diabetes with their diabetes-related health professionals (37). Appropriate education and regular support with isCGM by health professionals improve the quality of life, even in children with T1D (38).

The reason for the lower TIR in Cluster 3 than in the other clusters is unclear. Cluster 3 had higher rate of referencing Libre data for hyperglycemia or hypoglycemia than the other clusters. The elevated rate of correction of bolus insulin injections for hyperglycemia and/or immediate consumption of sugar or snacks for hypoglycemia might explain this discrepancy.

Careful attention should be paid to severe diabetic distress in adults with T1D. There was no marked difference in the TIR between the isCGM- and SMBG-based insulin dosing groups in children with T1D attending a summer camp (39) or in a real-world setting (40). This classification might help tailor diabetes management, but a further examination, including randomized controlled trials, will be required to confirm these issues.

Limitation of the study

Several limitations associated with the present study warrant mention, including the cross-sectional design, lack of laboratory data (islet-related autoantibodies and insulin secretion function test), and target population comprising only adult patients with T1D in this study. The cross-sectional nature of the study does not allow for inferences of causality to be made. We did not check for a history of diabetic ketoacidosis (DKA) associated with morbidity, psychological distress, mortality, and healthcare costs (41,42). The use of isCGM lowered the incidence of DKA in 47 T1D patients with recurrent DKA (43) as well as in a large database from the RELIEF study (44). Recently, with the introduction of the FreeStyle Libre 2 system, patients with T1D can utilize isCGM with the option of setting low and high glycemic thresholds and receiving alarms when these thresholds are crossed (45) Generalizability was limited because of the target population (Japanese adults with T1D).

Conclusion

We assessed adherence to SMBG and classified the patients into three groups using a cluster analysis. This information may help clinicians and patients achieve good adherence and diabetes management.

The authors state that they have no Conflict of Interest (COI).

Financial Support

The ISCHIA study was completed with funding from the Japan Agency for Medical Research and Development (AMED) (Grant number: 18ek0210104h0001, 19ek0210104h0002, 20ek0210104h0003) and Japan.