Rapid improvements in glycemic management with use of continuous glucose monitoring in adults with type 2 diabetes treated with basal insulin: 3-month analysis of the MOBILE study

Abstract

Introduction

This analysis investigated whether use of real-time continuous glucose monitoring (CGM) compared with blood glucose monitoring (BGM) results in rapidly improved glycemic management in adults with type 2 diabetes (T2D) treated with basal insulin.

Research design and methods

Using data from the MOBILE study where adults (n=175) with T2D treated with basal insulin without prandial insulin were randomized (2:1) to either CGM (n=116) or BGM (n=59), the treatment effect on glycemic management was determined over 3 months. The main outcome was a between-group difference in hemoglobin A_1c (HbA_1c) at 3 months adjusted for baseline value. Other outcomes included changes in CGM-derived glucose metrics and hypoglycemic events.

Results

After 3 months, there was a greater reduction from baseline in mean HbA_1c in the CGM group compared with the BGM group, from 9.1±1.0% (76±11 mmol/mol) to 8.0±1.2% (64±13 mmol/mol) in the CGM group and from 9.0±0.9% (75±10 mmol/mol) to 8.5±1.5% (69±16 mmol/mol) in the BGM group (adjusted difference, −0.6% (95% CI –0.9% to −0.3%); −6.6 mmol/mol (95% CI –10.2 to –2.9), p<0.001). Mean time spent in range 70–180 mg/dL (3.9–10.0 mmol/L) increased significantly more in the CGM group than the BGM group (adjusted difference, +9.3% (95% CI 2.1% to 16.4%), p<0.001). There also was a greater reduction in mean time >250 mg/dL (>13.9 mmol/L) with CGM (adjusted difference, −5.8% (95% CI −10.4% to −1.2%), p<0.001) without an increase in time <70 mg/dL (<3.9 mmol/L). Mean weekly hypoglycemic event rate was lower with CGM than BGM (adjusted difference, −0.2 events per week (95% CI −0.4 to –0.1), p<0.001). Further, in the CGM group, significant improvements in CGM metrics were observed during the first 7 days of CGM use.

Conclusions

In adults with basal insulin-treated T2D, use of CGM compared with BGM resulted in rapidly improved glycemic management, with a substantial reduction in HbA_1c over 3 months.

Trial registration number

NCT03566693.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• In people with type 2 diabetes (T2D) treated with basal insulin, continuous glucose monitoring (CGM) has been shown to improve glycemic management after 8 months of use compared with blood glucose monitoring (BGM). However, it is unknown how quickly glycemic improvements occur after CGM initiation in this population.

WHAT THIS STUDY ADDS

• Adults with T2D treated with basal insulin experienced rapid improvements in hemoglobin A_1c and CGM metrics 3 months after real-time CGM initiation compared with BGM use.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Use of real-time CGM can improve glycemic outcomes sooner after initiation than previously reported in basal insulin-treated T2D users, supporting the use of real-time CGM to manage T2D.

Introduction

Diabetes affects ~422 million adults worldwide, with ~90% diagnosed with type 2 diabetes (T2D).¹ Glucose management using blood glucose monitoring (BGM) has been the standard for glucose monitoring in T2D.² However, testing limitations and poor adherence to BGM due to inconvenience and intermittent measurement frequency exist, resulting in failure to detect important glycemic fluctuations.³ Indeed, population-level data indicate that suboptimal glycemic management is a continued concern; a significant percentage of people with T2D on multiple daily injections (MDI) or on basal insulin do not meet target hemoglobin A_1c (HbA_1c).^{4 5} Inadequate glycemic management significantly increases the risk of both acute⁶ and long-term⁷ complications with accompanying increased costs.^{8 9}

Real-time continuous glucose monitoring (CGM) has emerged as a valuable tool to assess glycemic management and has several advantages over BGM.¹⁰ Notably, CGM provides a continuous and extensive picture of glucose dynamics and optional alerts that inform users of hypoglycemic and hyperglycemic glucose values and trends.^{10 11} Randomized-controlled trials (RCTs) have consistently shown use of CGM to be superior to BGM in glycemic management in individuals with type 1 diabetes (T1D) and T2D.12,19 The MOBILE study was an RCT that investigated the impact of CGM in adults with T2D treated with basal insulin without prandial insulin. The study results demonstrated use of CGM compared with BGM over 8 months resulted in greater decreases in HbA_1c (−0.4% (−4.4 mmol/mol)) and hyperglycemia (measured by time >250 mg/dL (>13.9 mmol/L) (−16%)), and a 15% greater increase in time in range (TIR) 70–180 mg/dL (3.9–10.0 mmol/L).¹⁸ Similar findings were also reported in a subanalysis of the MOBILE study in older adults (≥65 years).²⁰ Use of CGM is endorsed by the European Association for the Study of Diabetes (EASD),⁹ the American Diabetes Association (ADA),²¹ and the Asia-Pacific consensus panel on CGM²² in all people with T2D using insulin. Availability of CGM is growing as insurance and healthcare systems expand coverage for people with diabetes. However, cost and coverage eligibility restrictions continue to constrain access to real-time CGM technology, while the National Institute for Health and Care Excellence (NICE) guidelines still only support CGM use for adults with T2D treated with MDI who meet specific additional criteria.²³

As real-time CGM use expands to more people living with diabetes, there is growing interest in understanding how quickly glycemic management improves with use of CGM in those with T2D. Although currently published data from the MOBILE study demonstrated effective long-term use of CGM on glycemic management in participants with T2D using basal insulin without prandial insulin, no RCT has specifically explored whether continuous CGM use results in improved glycemic management over a shorter time frame (ie, ≤3 months) compared with BGM in this population. Furthermore, while data are available highlighting that participants with T2D using insulin observe clinically meaningful improvements in CGM-derived glucose metrics in as little as ~7–14 days, which are sustained over 13 weeks,²⁴ little is known on the time frame of improvements in the T2D basal insulin-only population. The aim of this analysis was to use data collected from the MOBILE study to investigate the effect of CGM use compared with BGM on glycemic management over 3 months and to determine the time frame of clinically relevant glycemic improvements in people with T2D treated with basal insulin without prandial insulin.

Research design and methods

Study design

The MOBILE study was a 14-month, two-phase, multicenter, randomized, open-label, parallel-group trial carried out at 15 centers in the USA between July 2018 and January 2021. Comprehensive information on the protocol and procedures has previously been published¹⁸; pertinent features are outlined below. The present article reports previously unpublished analysis of the first 3 months of the original 8-month RCT which compared CGM and BGM use.

Participants

Adults aged ≥30 years with T2D using basal insulin without prandial insulin were recruited from primary care practices and could not currently be receiving care from an endocrinologist. Full criteria for inclusion and exclusion have already been published.¹⁸ Participants were allowed to be taking any other non-insulin glucose-lowering medication in addition to basal insulin as long as their regimen had been stable for ≥3 months prior to enrolling in the study.

Procedures

Prior to randomization (referred to as baseline), all participants wore a blinded Dexcom G6 Pro sensor (Dexcom, Inc., San Diego, California, USA) for up to 10 days. Participants were randomized (2:1) to either the CGM or BGM group. Both groups were provided a Bluetooth-enabled blood glucose meter (OneTouch Verio Flex; LifeScan, Malvern, Pennsylvania, USA). The BGM group was asked to perform fasting and postprandial BGM testing one to three times per day.

Both groups attended a visit at 3 months for HbA_1c testing, during which the BGM group had a blinded CGM sensor placed for 10 days. HbA_1c was measured at a central laboratory (University of Minnesota Advanced Research Diagnostic Laboratory) using the Tosoh G8 HPLC system (TOSOH Biosciences). Medication changes were recorded, if applicable.

Outcome measurements

The prespecified primary, secondary, and exploratory endpoints of the MOBILE study were previously published.¹⁸ This post-hoc exploratory analysis assessed mean HbA_1c at 3 months, adjusted for the baseline value. The percentage of participants achieving specific HbA_1c targets at 3 months and the percentage of participants achieving specific absolute and relative HbA_1c reductions from baseline to 3 months were also analyzed. Other outcomes included changes from baseline to 3 months in CGM-derived glucose metrics, including CGM-measured weekly hypoglycemia event rate. CGM-measured hypoglycemia events were defined as at least two sensor values <54 mg/dL (<3.0 mmol/L) that were ≥15 min apart plus no intervening values >54 mg/dL (>3.0 mmol/L). At least two sensor values >70 mg/dL (>3.9 mmol/L) that were ≥30 min apart with no intervening values ≤70 mg/dL (≤3.9 mmol/L) defined the end of a hypoglycemic event. CGM data were processed consistently with the primary MOBILE study analysis.¹⁸ Table 1 provides a detailed list of the outcome measurements analyzed in this post-hoc analysis.

Table 1. Post-hoc analysis outcome measurements (baseline to 3 months).

Change in mean HbA1c	% (mmol/mol)
Change in mean HbA1c stratified by baseline HbA1c	≥8.5% (≥69 mmol/mol), % (mmol/mol)
	≥9.0% (≥75 mmol/mol), % (mmol/mol)
	≥9.5% (≥80 mmol/mol), % (mmol/mol)
	≥10.0% (≥86 mmol/mol), % (mmol/mol)
Participants achieving specific HbA1c targets	<7.0% (<53 mmol/mol), n (%)
	<7.5% (<58 mmol/mol), n (%)
	<8.0% (<64 mmol/mol), n (%)
Participants achieving specific magnitudes of reduction in HbA1c	Absolute reduction ≥0.5% (≥5.5 mmol/mol), n (%)
	Absolute reduction ≥1.0% (≥11 mmol/mol), n (%)
	Relative reduction ≥10%, n (%)
Change in CGM-derived glucose metrics	Time in range (TIR) 70–180 mg/dL (3.9–10.0 mmol/L), %
	Time >180 mg/dL (>10.0 mmol/L), %
	Time >250 mg/dL (>13.9 mmol/L), %
	Time >300 mg/dL (>16.7 mmol/L), %
	Time <70 mg/dL (<3.9 mmol/L), %
	Time <54 mg/dL (<3.0 mmol/L), %
	Mean glucose, mg/dL (mmol/L)
	Glucose management indicator (GMI), %
	Glucose variability (coefficient of variation), %
	Area under curve of 180 mg/dL
	CGM-measured weekly hypoglycemia event rate
Change in insulin use and diabetes medications	Total daily insulin, U and U/kg
	Basal daily insulin, U and U/kg
	Addition of ≥1 diabetes medication after randomization, n (%)
	Removal of ≥1 diabetes medication in use at the time of randomization, n (%)
	Addition of prandial insulin, n (%)
	Addition of GLP-1 RA, n (%)
	Addition of SGLT-2 inhibitor, n (%)

CGM-measured hypoglycemia events were defined as at least two sensor values <54 mg/dL (<3.0 mmol/L) ≥15 min apart plus no intervening values >54 mg/dL (>3.0 mmol/L). At least two sensor values >70 mg/dL (>3.9 mmol/L) that were ≥30 min apart with no intervening values ≤70 mg/dL (≤3.9 mmol/L) defined the end of a hypoglycemic event.

CGM, continuous glucose monitoring; HbA_1c, hemoglobin A_1c; GLP-1 RA, glucagon-like peptide-1 receptor agonist; SGLT-2, sodium-glucose cotransporter-2.

Statistical analysis

The change in HbA_1c from baseline to 3 months and HbA_1c binary outcomes were analyzed using mixed-effects linear regression model and mixed effects logistic regression models. Treatment and source of HbA_1c (central or local laboratory, ie, local HbA_1c level was included as an auxiliary variable) were included as fixed effects. The same model was applied to explore the effects in cohorts with different baseline HbA_1c. Additionally, for each CGM metric, a mixed-effects model was fitted with change from baseline to 3 months as response, with treatment as a fixed effect. In each model, clinical site was included as a random effect while baseline HbA_1c or baseline CGM metrics were adjusted for as covariates. Linear regression models (for insulin doses) adjusting for the baseline value and a random site effect and logistic regression (change in diabetes medicine) with random site effect were used to evaluate potential changes in diabetes medications from baseline to 3 months. Furthermore, for the CGM group, change from baseline for CGM-derived metrics at each follow-up week (defined as the 7-day interval) was derived and modeled using a mixed model with repeated measures to evaluate the first onset of treatment effect. Statisticians were not blinded to participant grouping.

All p values were two-sided. Statistical significance was defined as p<0.05. For CGM metrics and HbA_1c binary outcomes, group-specific Benjamini-Hochberg procedure was used to adjust for false discovery rate (FDR) from multiple comparisons. The models for HbA_1c and CGM metrics handled missing data using the direct likelihood method.²⁵

Results

Participants

A total of 176 adults were recruited from primary care practices and were randomly assigned to the CGM group (n=117) or the BGM group (n=59). Among them, one randomized participant in the CGM group was subsequently diagnosed with T1D and excluded from the analyses. The total number of participants included in the analyses was therefore 175 (n=116 in the CGM group and n=59 in the BGM group). The flow of participants throughout the study, reasons for withdrawal, and baseline characteristics (including insulin use and other diabetes medications) have been detailed elsewhere.¹⁸ In summary, participants’ mean±SD age was 57±9 years (range 33–79 years), mean body mass index was 33.9±6.6 kg/m², 88 (50%) were women, and 92 (53%) were of minority race or ethnicity. Most participants (91%) were taking at least one non-insulin glucose-lowering medication, in addition to basal insulin. Of those taking only one medication, metformin was the most commonly used (n=31 (27%) in the CGM group and n=13 (22%) in the BGM group). Among participants taking two medications, the most common combinations were metformin with a sulfonylurea (n=31 (27%) in the CGM group and n=21 (36%) in the BGM group) or metformin with a glucagon-like peptide-1 receptor agonist (GLP-1 RA) (n=19 (16%) in the CGM group and n=3 (5%) in the BGM group). Overall, 20% of participants (n=27 (23%) in the CGM group and n=8 (14%) in the BGM group) were taking a GLP-1 RA and 7% of participants (n=9 (8%) in the CGM group and n=4 (7%) in the BGM group) were taking a sodium-glucose cotransporter-2 (SGLT-2) inhibitor. Mean HbA_1c at screening was 9.2±1.0% (77±11 mmol/mol; range 7.8–11.4% (62–101 mmol/mol)). In the CGM group, median CGM use was 6.3 days per week (IQR, 5.4–6.7 days per week) over the 3-month period.

HbA_1c outcomes

Mean HbA_1c decreased from 9.1±1.0% at baseline (76±11 mmol/mol) to 8.0±1.2% (64±13 mmol/mol) at 3 months in the CGM group and from 9.0±0.9% (75±10 mmol/mol) to 8.5±1.5% (69±16 mmol/mol) in the BGM group (adjusted difference, −0.6% (95% CI –0.9% to −0.3%); −6.6 mmol/mol (95% CI −10.2 to –2.9), p<0.001, see table 2).

Table 2. HbA_1C outcomes.

		Baseline (mean±SD)		3 months (mean±SD)		Change from baseline at 3 months (mean±SD)		Model adjusted between group difference or OR at 3 months (95% CI) (p value)
		CGM	BGM	CGM	BGM	CGM	BGM
HbA1C	n	115	58	109	58	108	57
	%	9.1±1.0	9.0±0.9	8.0±1.2	8.5±1.5	−1.2±1.2	−0.6±1.2	−0.6 (−0.9 to −0.3) (<0.001)
	mmol/mol	76±11	75±10	63±13	69±16	−13±13	−6.4±13	−6.6 (−10.2 to −2.9) (<0.001)
HbA1C outcomes stratified by baseline HbA1C
≥8.5% (≥69 mmol/mol)	n	84	41	78	41	78	41
	%	9.6±0.8	9.4±0.7	8.2±1.1	8.8±1.6	−1.4±1.1	−0.6±1.4	−0.8 (−1.2 to -0.4) (<0.001)
	mmol/mol	81±9	80±8	66±13	73±17	−16±12	−7±15	−8.7 (−13.0 to -4.3) (<0.001)
≥9.0% (≥75 mmol/mol)	n	59	30	57	30	57	30
	%	9.9±0.6	9.7±0.6	8.4±1.2	9.0±1.7	−1.5±1.2	−0.7±1.5	−0.9 (−1.4 to -0.4) (<0.001)
	mmol/mol	85±7	83±7	68±13	75±18	−17±13	−7±17	−9.6 (−15.2 to -4.1) (<0.001)
≥9.5% (≥80 mmol/mol)	n	44	15	42	15	42	15
	%	10.2±0.5	10.2±0.5	8.5±1.2	9.9±1.8	−1.7±1.1	−0.4±1.8	−1.2 (−2.0 to -0.5) (0.002)
	mmol/mol	88±6	88±6	69±13	84±20	−19±13	−4±20	−13.5 (−21.8 to -5.2) (0.002)
≥10.0% (≥86 mmol/mol)	n	26	9	24	9	24	9
	%	10.5±0.4	10.6±0.4	8.4±1.3	10.5±1.8	−2.1±1.1	−0.1±2.0	−1.7 (−2.8 to -0.7) (0.002)
	mmol/mol	91±4	92±4	69±14	91±19	−23±12	−1±21	−19.0 (−30.6 to -7.4) (0.002)
No. (%) participants achieving
n		–	–	109	58
HbA1c <7.0% (<53 mmol/mol)		–	–	19 (17)	4 (7)	–	–	3.5 (0.9 to 12.8) (0.06)
HbA1c <7.5% (<58 mmol/mol)		–	–	44 (40)	21 (36)	–	–	1.5 (0.7 to 3.2) (0.40)
HbA1c <8.0% (<64 mmol/mol)		–	–	64 (59)	27 (47)	–	–	2.3 (0.9 to 5.7) (0.06)
Absolute HbA1c reduction ≥0.5% (≥5.5 mmol/mol)		–	–	83 (76)	37 (64)	–	–	2.1 (0.9 to 4.9) (0.08)
Absolute HbA1c reduction ≥1% (≥11 mmol/mol)		–	–	68 (62)	21 (36)	–	–	3.9 (1.3 to 11.5) (0.004)
Relative HbA1c reduction ≥10%		–	–	70 (64)	23 (40)	–	–	3.6 (1.4 to 9.4) (0.004)

p values for binary HbA_1c outcomes were adjusted for multiple comparisons using the Benjamini-Hochberg method. CIs for continuous HbA_1c were reported at a significant level of 0.05 (ie, 95% CI). CIs for binary HbA_1c were corrected using the Benjamini-Hochberg adjusted significance level.

BGM, blood glucose monitoring; CGM, continuous glucose monitoring; HbA_1c, hemoglobin A_1c.

Statistically significant between-group differences in favor of the CGM group were also observed for change in HbA_1c from baseline to 3 months stratified by baseline HbA_1c of ≥8.5% (≥69 mmol/mol), ≥9.0% (≥75 mmol/mol), ≥9.5% (≥80 mmol/mol), and ≥10.0% (≥86 mmol/mol) (p≤0.002 in each stratum). The percentage of participants achieving a HbA_1c level <7.0% (<53 mmol/mol), <7.5% (<58 mmol/mol), and <8.0% (<64 mmol/mol) at 3 months were all higher in the CGM group, though not statistically significant after adjusting for FDR (p≥0.06). The odds of achieving an absolute HbA_1c reduction of ≥1% (≥11 mmol/mol) and a relative reduction of ≥10% were significantly higher in the CGM group compared with the BGM group (odds ratio (OR), 3.9, p=0.004 for absolute reduction of ≥1% (≥11 mmol/mol) and 3.6, p=0.004 for relative reduction of ≥10%).

CGM-derived outcomes

Statistically significant between-group differences were observed across most CGM metrics (see table 3 and figure 1) after controlling for multiple comparisons. There was a greater increase in mean % TIR from baseline to 3 months in the CGM group compared with the BGM group (adjusted difference, 9.3%, p=0.02). Mean percent time >180 mg/dL (>10.0 mmol/L) (p=0.03), >250 mg/dL (>13.9 mmol/L) (p=0.02), >300 mg/dL (>16.7 mmol/L) (p=0.009), mean area under curve 180 mg/dL (p=0.01), and mean coefficient of variation (p=0.001) all showed greater reductions in the CGM group compared with the BGM group. Significant between-group differences favoring the CGM group were also detected for mean percent time <70 mg/dL (<3.9 mmol/L) (p<0.001), <54 mg/dL (<3.0 mmol/L) (p<0.001), and mean hypoglycemia event rate (p<0.001). Greater reductions were observed in the CGM group compared with the BGM group for both mean glucose and GMI, though not statistically significant (p=0.07 for both). There was very little hypoglycemia in either the CGM or BGM groups.

Table 3. CGM-derived outcomes.

		Baseline (mean±SD)		3 months (mean±SD)		Change from baseline at 3 months (mean±SD)		Model adjusted between group difference at 3 months (95% CI) (p value)
		CGM	BGM	CGM	BGM	CGM	BGM
n		114	59	109	54	107	54
% Time in range 70–180 mg/dL (3.9–10.0 mmol/L)		40±26	40±25	58±26	49±24	17±28	10±28	9.3 (1.5 to 17) (0.02)
% Time >180 mg/dL (>10.0 mmol/L)		59±26	59±26	42±26	49±25	−17±29	−10±29	−8.4 (−15.9 to −0.8) (0.03)
% Time >250 mg/dL (>13.9 mmol/L)		27±22	25±21	13±15	19±18	−13±21	−7±21	−5.8 (−10.6 to −0.9) (0.02)
% Time >300 mg/dL (>16.7 mmol/L)		12±13	11±12	5±7	8±11	−7±12	−3±13	−3.5 (−6.3 to −0.8) (0.009)
% Time <70 mg/dL (<3.9 mmol/L)		0.3±0.5	0.3±0.6	0.3±0.4	0.9±1.3	−0.02±0.55	0.56±1.22	−0.6 (−0.9 to −0.2) (<0.001)
% Time <54 mg/dL (<3.0 mmol/L)		0.03±0.08	0.06±0.11	0.04±0.08	0.26±0.50	0.01±0.10	0.20±0.47	−0.2 (−0.3 to −0.1) (<0.001)
Area under curve 180 mg/dL		44.6±31.6	42.0±29.6	24.4±23.1	33.0±27.4	−19.7±30.8	−9.7±31.4	−9.7 (−17.6 to −1.8) (0.01)
Mean glucose	mg/dL	209±49	206±45	181±45	191±45	−27±48	−16±51	−11.7 (−24.3 to 1.0) (0.07)
	mmol/L	11.6±2.7	11.4±2.5	10.1±2.5	10.6±2.5	−1.5±2.7	−0.9±2.8	−0.6 (−1.4 to 0.1) (0.07)
Glucose management indicator (GMI), %		8.3±1.2	8.2±1.1	7.6±1.1	7.9±1.1	−0.6±1.2	−0.4±1.2	−0.3 (−0.6 to 0) (0.07)
Coefficient of variation, %		28±7	28±7	27±7	30±7	−0.4±5.5	2.6±7.0	−2.9 (−4.9 to −0.9) (0.001)
Hypoglycemia event rate per week		0.1±0.2	0.1±0.3	0.1±0.2	0.3±0.6	0.0±0.3	0.2±0.6	−0.2 (−0.4 to −0.1) (0.001)

% Time >250 mg/dL (>13.9 mmol/L), % Time >300 mg/dL (>16.7 mmol/L), % Time <70 mg/dL (<3.9 mmol/L), % Time <54 mg/dL (<3.0 mmol/L), Area under curve 180 mg/dL, and hypoglycemia event rate per week were winsorized at the 10th and 90th percentiles prior to reporting summary statistics to account for outliers. p values for CGM-metrics were adjusted for multiple comparisons using Benjamini-Hochberg method. CIs for CGM-metrics were corrected using the Benjamini-Hochberg adjusted significance level. GMI = (3.31 + 0.02392 × mean glucose).

BGM, blood glucose monitoring; CGM, continuous glucose monitoring.

Figure 1. CGM-derived times (%) in glucose ranges at baseline and 3 months. Data may not total 100% due to rounding. p values indicate between-group difference at 3 months (model adjusted mean difference). At the bottom of each bar, the data label on the left of the slash denotes the time <70 mg/dL (<3.9 mmol/L) and the data label on the right denotes the time <54 mg/dL (3.0 mmol/L). BGM, blood glucose monitoring; CGM, continuous glucose monitoring.

Weekly change of CGM metrics

Significant improvements from baseline were observed after 7 days of CGM wear in the CGM group for mean TIR (+10.4%, p<0.001), mean percent time >180 mg/dL (>10.0 mmol/L) (−10.3%, p<0.001), and mean percent time >250 mg/dL (>13.9 mmol/L) (−8.7%, p<0.001, figure 2). The significant effects from baseline were consistent over the course of the 3 months.

Figure 2. Weekly CGM-derived metrics for the CGM group. (A) shows estimated change from baseline (%) using mixed models with repeated measurements and (B) shows absolute weekly values (%). Data in (B) may not total 100% due to rounding. In (B), at the bottom of each bar, the data label on the left of the slash denotes the time <70 mg/dL (<3.9 mmol/L) and the data label on the right denotes the time <54 mg/dL (3.0 mmol/L). CGM, continuous glucose monitoring.

Diabetes treatments

There were no significant statistical differences between groups in terms of total daily insulin or basal insulin daily doses, nor in the addition or discontinuation of non-insulin diabetes medications (p≥0.38, online supplemental tables 1 and 2).

Conclusions

This analysis of data from the MOBILE study demonstrates that use of CGM results in a rapid improvement in TIR and other CGM metrics reflective of hyperglycemia in adults with T2D on basal insulin without prandial insulin. These improvements were sustained over 3 months and resulted in clinically significant improvements in glycemic outcomes with CGM use compared with BGM use. HbA_1c decreased from baseline by 1.2% (13 mmol/mol) at 3 months and this decrease was sustained at 8 months (−1.1% (−12 mmol/mol)) as previously reported in the primary analysis of the MOBILE study,¹⁸ and also at 14 months in a subset of participants who continued to use CGM (−1.0% (−11 mmol/mol)).²⁶

It is well established that a reduction in HbA_1c of 0.4–0.5% (4.4–5.5 mmol/mol) is clinically meaningful^{27 28} and strongly correlated with a reduced risk of chronic diabetes-related vascular complications in both people with T1D^{29 30} and T2D.^{31 32} Similarly, TIR is also strongly associated with chronic diabetes-related complications.³³ Cross-sectional studies, along with secondary analyses of data from the Diabetes Control and Complications Trial (DCCT), have documented this association,33,35 with one analysis by Beck et al³⁵ reporting reduced risk of retinopathy by 22% with every 5% increase in TIR. Thus, the degree of improvement in HbA_1c and TIR found in the current analysis and previously published MOBILE study results may suggest benefits for reduced risk on incidence of long-term diabetes-related complications. While the adjusted between-group difference in HbA_1c between CGM and BGM decreased non-significantly (data not reported) from 3 months, as presented in this analysis, to 8 months, as reported in the primary analysis of the MOBILE study¹⁸ (from −0.6% (−6.6 mmol/mol) to −0.4% (-4.4 mmol/mol)), this occurred due to modest changes (±0.1% (±1.1 mmol/mol)) in HbA_1c in both groups between these time points. Therefore, the present 3-month analysis combined with the previously published 8-month¹⁸ and 14-month²⁶ results of the MOBILE study suggests that continued CGM use results in rapid and sustained improvements in glycemic management in participants with T2D treated with only basal insulin.

The value of CGM in adults with T2D on insulin was also demonstrated in the Steno2tech study. In this 12-month RCT, 83% of participants used basal insulin without prandial insulin and 17% of participants were on MDI. The study results demonstrated a significant 0.8% (8.7 mmol/mol) and 0.9% (9.8 mmol/mol) greater reduction in HbA_1c and a 12.4% and 15.2% greater increase in TIR, at 6 and 12 months, respectively, in those on basal insulin and MDI following use of CGM compared with BGM.¹⁹ The Multiple Daily Injections and Continuous Glucose Monitoring in Diabetes (DIAMOND) study, a 6-month RCT in people with T2D on MDI demonstrated a greater reduction in HbA_1c by 0.3% (3.3 mmol/mol) with CGM use compared with BGM use.¹⁴ Together, the current evidence base highlights the importance of CGM use in managing glycemia in all people with insulin-treated T2D. Expanding CGM availability to those with insulin-treated T2D may be appropriate, considering the clinical benefit. However, it is important to note that there are a number of factors that influence access to CGM beyond clinical benefit, including cost and patient quality of life. Improved glycemic outcomes are associated with fewer emergency department visits and hospitalizations from diabetes-related complications,³⁶ with related reductions in healthcare costs.^{37 38} The ADA Standards of Care guidelines currently recommend CGM use for all people with insulin-treated T2D and suggest consideration of CGM use for those not treated with insulin.²¹ Future work may show the benefit of CGM use in non-insulin-treated populations,³⁹ expanding the potential for CGM use along the full spectrum of T2D progression and management.

A noteworthy outcome of the present analysis in T2D participants on basal insulin without prandial insulin is the rapid time frame in which clinically relevant improvements in CGM-derived glycemic metrics occurred in the CGM group. In as little as 7 days, significant improvements in TIR, and time >180 mg/dL (>10.0 mmol/L) and >250 mg/dL (>13.9 mmol/L) were observed, which were sustained throughout the 3-month analysis period. These data are consistent with a pooled cohort of T2D individuals on either MDI or basal insulin without prandial insulin, which demonstrated an 11% increase in TIR and decreases in time >180 mg/dL (>10.0 mmol/L) and >250 mg/dL (>13.9 mmol/L) of −10% and −7%, respectively, in 7 days.²⁴ These findings suggest that individuals with T2D at different stages of insulin therapy can experience glycemic benefits very soon after initiating CGM use. The improvement in glycemic outcomes observed in the current study over 3 months is meaningful, since increased TIR is associated with corresponding reductions in HbA_1c.¹¹

A strength of this analysis is that it included a population of individuals with T2D receiving diabetes management in a primary care setting who were racially and socioeconomically diverse. As a result, the findings should be generalizable to many individuals with T2D who use basal insulin without prandial insulin. A limitation of the study is that participants had more contact with clinic staff than they would have as part of standard care. This particularly affected the BGM group, which received significantly more interaction than expected in standard care, likely contributing to better outcomes in this study cohort. Additionally, this trial was only conducted within the context of a US-based healthcare system. Nevertheless, other studies exploring real-time CGM use in cohorts from Europe¹⁹ and Australia⁴⁰ demonstrate glycemic benefit for adults with T2D. It is also important to note as a limitation that this post-hoc analysis was not prespecified and exploratory in nature. Due to the lack of difference in changes in daily insulin delivery and diabetes medications between groups, it may be hypothesized that the rapid glycemic improvements in this analysis occurred due to changes in participant behavior. Patient-reported outcomes were not assessed in the first 3 months of the MOBILE study; however, this limited our ability to reveal the mechanisms of rapid glycemic improvements in CGM users and the magnitude of benefits related to changes in diet, physical activity, or medication adherence. Nevertheless, this presents an opportunity for future research in this population.

In conclusion, the results of this 3-month post-hoc analysis of the MOBILE study demonstrate that among adults with T2D with suboptimal glucose levels, CGM use is associated with rapid glycemic improvements in as little as 7 days, with significant improvements compared with BGM without an increase in hypoglycemia at 3 months. These findings add to the growing body of evidence that demonstrates the value of CGM to reduce hyperglycemia in adults with T2D on basal insulin.

Representatives of the sponsor, Dexcom, Inc., participated in developing the design of the study but were not involved in the conduct of the study other than participating in site monitoring; and were not involved in the data collection or management. Representatives of the sponsor were involved in the interpretation and analysis of the data, and the preparation of the manuscript. The decision to submit the manuscript for publication was a joint agreement between the sponsor and the coauthors not employed by the sponsor. The sponsor did not influence the results/outcomes of the study despite author affiliations with the sponsor.

Data availability statement

Data are available upon reasonable request.