Cost-Utility of Real-Time Continuous Glucose Monitoring Versus Self-monitoring of Blood Glucose in People with Insulin-Treated Type 2 Diabetes in Sweden

Abstract

Introduction

The aim of this study was to investigate the cost-utility of real-time continuous glucose monitoring (rt-CGM) versus self-monitoring of blood glucose (SMBG) in people with insulin-treated type 2 diabetes (T2D) in Sweden.

Methods

The CORE Diabetes Model (CDM v10) was used for the analysis. Clinical effectiveness data were obtained from the Steno2Tech trial, an investigator-initiated, 12-month, single center randomized controlled trial based in Denmark. Adverse event rates were sourced from a large-scale observational study based in the USA. Costs were obtained from Swedish and European studies and inflated to 2023 Swedish Krona (SEK). The analysis adopted the perspective of the Swedish payer, and a remaining lifetime horizon was used in the base case. A discount rate of 3% was applied to future costs and outcomes on an annual basis. A commonly cited willingness-to-pay (WTP) threshold of SEK 500,000 was used.

Results

rt-CGM led to a gain in mean incremental survival by 0.082 years (11.529 life years for rt-CGM versus 11.447 life years for SMBG). Total mean incremental costs were SEK 138,448 higher with rt-CGM compared with SMBG (SEK 1,151,049 for rt-CGM versus SEK 1,012,601 for SMBG). However, rt-CGM incurred fewer overall diabetes-related complication costs than SMBG over the remaining lifetime horizon. Rt-CGM also yielded a gain in mean incremental quality-adjusted life years (QALYs) of 0.632 (8.608 QALYs for rt-CGM versus 7.976 QALYs for SMBG). The mean incremental cost-utility ratio (ICUR) for rt-CGM was SEK 219,063 per QALY gained, which showed rt-CGM to be cost-effective when compared with the WTP threshold of SEK 500,000. When various indirect cost estimates were incorporated, rt-CGM was consistently more cost-effective than in the base case analysis.

Conclusions

For individuals living in Sweden with T2D requiring insulin treatment, rt-CGM is a cost-effective management option relative to SMBG.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13300-025-01811-x.

Key Summary Points

Real-time continuous glucose monitoring (rt-CGM) may offer better glycemic control, reduce long-term micro and macrovascular complications, and economic advantages over self-monitoring of blood glucose (SMBG).

Cost-utility studies of rt-CGM versus SMBG across various settings, including USA, France, South Korea, Spain, Denmark, and Canada. In all these studies, rt-CGM was found to be a cost-effective glycemic management option. However, similar analyses have not been conducted within a Swedish setting.

This study estimated the cost-effectiveness of rt-CGM versus SMBG for people with insulin-treated type 2 diabetes (T2D) in Sweden, using the CORE Diabetes Model v10.

rt-CGM was associated with greater quality-adjusted life years (QALYs) and higher costs, yielding an incremental cost-utility ratio of SEK 219,063 per QALY gained, which was below the willingness-to-pay threshold of SEK 500,000.

These findings suggest that rt-CGM is a cost-effective management strategy for insulin-treated T2D in Sweden, in turn supporting wider adoption of the technology in routine clinical practice.

Introduction

The worldwide prevalence of diabetes has been growing for several decades and is projected to continue increasing [1, 2]. According to the International Diabetes Federation, in 2021, over 490,000 adults were living with diabetes in Sweden [3], with this number projected to increase to 940,000 by 2050 [1]. Type 2 diabetes (T2D) accounts for over 90% of all diabetes cases [4], which is associated with long-term chronic complications such as cardiovascular disease (CVD), retinopathy, nephropathy, and neuropathy [5], alongside a two-fold increase in mortality risk [6]. Achieving and maintaining normal physiological glucose levels is a key therapeutic goal for T2D, and aids in preventing the development of long-term diabetes-related complications [7]. However, in 2019, 45% of individuals living with T2D (and 30% of individuals recently diagnosed with the disease) in Sweden were found to have suboptimal glycemic control [8]. These clinical factors can lead to significant costs to the healthcare system. A Swedish study reported that patients living with T2D (without any cardiovascular complications) incurred average costs of Swedish Krona (SEK) 132,456 over a 5-year period [9]. Even beyond the direct healthcare costs, T2D can place a significant financial burden on both the patient population and the Swedish government. A study by Persson et al. found that T2D-related complications such as stroke, end-stage renal disease, and severe vision loss led to 102, 70, and 56 days of absence from work per individual per year, respectively, in Sweden [10]. Furthermore, a longitudinal analysis revealed that delayed control of glycated hemoglobin (HbA1c) over a 10-year period resulted in an estimated individual earnings loss of SEK 44,157. This reduction in income corresponded to a loss in tax revenue of SEK 24,287 per affected individual, highlighting the broader fiscal implications of suboptimal diabetes management [8].

Glycemic control through consistent glucose monitoring and treatment optimization can therefore lead to reduced financial burden on patients, the healthcare system, and governments, while improving patient quality of life (QoL) and overall survival [11, 12]. Recently, results from multiple randomized controlled trials (RCTs) have demonstrated that continuous glucose monitoring (CGM) systems, and in particular real-time CGM (rt-CGM) systems, can result in significant and sustained reductions in HbA1c as well as improvements in body mass index (BMI) and time in range (TIR), for patients living with T2D compared with self-monitoring of blood glucose (SMBG) [13–15]. The single-center, parallel and open-label Steno2Tech trial (which included adult participants living with poorly-controlled insulin-treated T2D in Denmark) showed that rt-CGM use led to between-group differences in HbA1c of −0.9% and BMI of −1.1 kg/m² over 12 months compared with SMBG [13]. The MOBILE RCT, which was conducted across 15 centers in the USA, found that rt-CGM use led to notable improvements in TIR (adjusted mean difference of + 15%) compared with SMBG over 8 months [14], with results sustaining over an additional 6 months provided that participants continued with their rt-CGM use [16]. These results align well with those from a large-scale, retrospective observational study, again conducted in the US, which found that for 36,080 patients living with T2D, using rt-CGM led to a statistically significant −0.56% adjusted reduction in HbA1c levels compared with SMBG [17].

Such clinical effects (particularly those from Karter et al. [17]) have already been incorporated within cost-utility studies of rt-CGM versus SMBG across various settings, including France, South Korea, Spain, Denmark, and Canada [18–22]. In all these studies, rt-CGM was found to be a cost-effective treatment option, leading to improved QoL for low additional costs; however, similar analyses have not been conducted within a Swedish setting.

The objective of this study, therefore, was to assess the long-term cost-utility of rt-CGM (specifically the Dexcom ONE+ system) versus SMBG in patients living with insulin-treated T2D in Sweden, using clinical effectiveness data from the Steno2Tech trial [13], adverse event rates from a large-scale US-based observational study [17], and direct cost inputs obtained from relevant cost-effectiveness studies [23–25]. A secondary objective was to consider the impact of indirect costs associated with diabetes-related complications on the cost-utility of rt-CGM.

Methods

Model Structure

The published and validated IQVIA CORE Diabetes Model (CDM, v10) was utilized under licensed agreement to assess the cost-utility of rt-CGM versus SMBG for people living with insulin-treated T2D in Sweden [26–28]. Both Palmer et al. 2004 [27] and McEwan et al. 2014 [26] outline key strengths and limitations of the model in more detail. The CDM comprises a series of interconnected Markov sub-models that simulate diabetes-related complications. The sub-models each comprise at least two different health states, and can interact with each other as required to estimate long-term health and cost outcomes for a specified healthcare technology used in either T1D or T2D [28]. Various equations (each derived from large-scale, landmark studies) that predict the progression of risk factors, including HbA1c, systolic and diastolic blood pressure, and the lipid profile of simulated patients, are built within the CDM. For CVD specifically, the CDM includes several distinct risk prediction equations to provide accurate and contemporary risk prediction tools that are relevant to a range of geographical settings [28]. The CDM also incorporates clinical data to inform the probabilities of developing other diabetes-related microvascular complications (e.g., retinopathy, nephropathy, and ulcers), as well as the likelihood of transitioning between different health states and from one complication to another. Given its previous validation, incorporation of robust clinical data inputs, flexibility in applying a range of risk equations, its broad utilization to estimate the cost-effectiveness of a variety of diabetes care technologies submitted to the UK National Institute for Health and Care Excellence (NICE) [29, 30], and in similar cost-utility analyses conducted across numerous settings [18–22], the CDM was considered an appropriate model for the present analysis.

CDM outputs that were recorded for the present analysis included standard economic evaluation metrics, such as life expectancy (LE), quality-adjusted life expectancy (measured in quality-adjusted life years [QALYs]), direct and indirect costs (expressed in 2023 SEK), and incremental cost-utility ratios (ICURs). Additional clinical outcomes were also reported, including the projected cumulative incidence and timing of onset for diabetes-related microvascular and macrovascular complications.

Baseline Cohort Characteristics

The analysis was based on previously conducted studies and did not contain any new studies involving human or animal participants that were performed by any of the authors. The baseline characteristics for the T2D population were obtained primarily from the Steno2Tech trial, an investigator-initiated 12-month single-center RCT based in Denmark [13], as well as a large-scale observational study based in the USA [17]. Both studies included participants living with insulin-treated T2D diabetes and investigated the impact of rt-CGM on glucose control versus SMBG. For any remaining characteristics not available in either of these studies, the default CDM values (which were based on the ACCORD trial [31, 32]) were used. The mean baseline HbA1c level for the simulated cohort was 8.6% (± 0.1%), mean age was 61 (± 8.3) years, and mean duration of diabetes was 18 (± 6.8) years (Table 1). Racial characteristics were sourced from a real-world, prospective study conducted across 28 centers in Sweden [33]. On the basis of these data, the racial distribution for the simulated cohort was 91.4% white, 1.1% Black, 5.3% Asian, with the remaining 2.2% of individuals assumed to be Hispanic (Supplementary Table S1) [33]. A summary of the additional baseline characteristics used in the present analysis can be found in Supplementary Table S1.

Table 1.

Baseline characteristics of the simulated patient cohort

Characteristic	Baseline value
Mean (SD) age, years	61 (8.3)
Mean (SD) duration of diabetes, years	18 (6.8)
Proportion male, %	62
Proportion female, %	38
Mean (SD) HbA1c, %	8.6 (0.1)

Values sourced from Lind et al. [13]

HbA1c glycated hemoglobin, SD standard deviation

Treatment Effects

Clinical efficacy data for rt-CGM versus SMBG in people living with insulin-treated T2D were obtained from the Steno2Tech trial [13]. The trial showed that compared with SMBG, rt-CGM was associated with favorable between-group differences in HbA1c levels (−0.9%) and BMI (−1.1 of kg/m²) [13]. The reduction in HbA1c effect was conservatively assumed to be sustained for three additional years (beyond the first year) in the rt-CGM arm, based on clinical evidence on glycemic outcomes published across multiple long-term studies [34–36]. Annual progression of HbA1c levels in the SMBG arm (after year one) and the rt-CGM arm (after year four) was calculated using the Swedish National Diabetes Register (SNDR) equation [37]. However, whilst the SNDR equation for HbA1c and BMI progression was developed using a Swedish population living with T2D, the limited follow-up period of four years did not account for lagged effects of covariates on risk factors [37]. Using the SNDR equation could therefore lead to inaccuracies in long-term risk factor progression estimates, subsequently impacting the accuracy of projected incidence of microvascular and macrovascular diabetes-related complications. The impact of using a range of alternative equations for HbA1c and BMI progression on the health economic results was therefore explored within the sensitivity analysis.

Adverse Events

The primary treatment-related adverse events considered in the analysis comprised severe hypoglycemic events (SHE) and severe hyperglycemic events, with the latter assumed to be diabetic ketoacidosis (DKA) events. Rates for both SHE and DKA events were sourced from emergency room visits or hospitalizations reported in Karter et al. [17]. Per 100 patient years, there were 0 SHE and 0 DKA events in the rt-CGM arm, and 4 SHE and 2.5 DKA events in the SMBG arm.

Utilities

At baseline, simulated patients living with T2D without any complications were assigned the CDM default utility value of 0.785. This value was sourced from a systematic literature review published by Beaudet et al. [38], which was also the main source for identifying disutility values for both diabetes-related complications and adverse effects stemming from diabetes treatment. A single disutility value associated with experiencing DKA events was sourced from Zhao et al. [39]. Two distinct utility benefits were also applied exclusively for patients in the rt-CGM arm. These included an assumed utility benefit of 0.03 due to the avoidance of fingerstick testing (AFS) [40], as well as an assumed 0.0155 utility gain owing to the avoidance of fear of hypoglycemia (FoH). The FoH utility value was calculated using disease-specific patient-reported outcomes (i.e., scores from the Hypoglycemia Fear Survey) mapped to the EQ-5D [41–43]. A full list of the utility and disutility values used in the present analysis can be found in Supplementary Table S2.

Discounting and Perspective

An annual discount rate of 3% was applied to future effects and costs, based on the general guidelines for economic evaluations report, published by the Swedish Pharmaceutical Benefits Board [44]. The analysis was conducted from the perspective of the Swedish payer.

Time Horizon and Willingness-To-Pay

A remaining lifetime horizon (of maximum 50 years) was chosen for the base case analysis, reflecting the long-term nature of T2D and the anticipated sustained effects of the interventions being evaluated [45]. As there is currently no official willingness-to-pay (WTP) threshold in Sweden, an informal, commonly-cited figure of SEK 500,000 per QALY was selected for this analysis [46–48].

Direct Costs

All costs included within the present analysis were expressed in SEK, and were inflated to 2023 values as necessary, using the Sweden Consumer Price Index for Health [49]. Direct costs for concomitant medications (i.e., aspirin, statins and angiotensin-converting enzyme inhibitors and angiotensin receptor blockers), screening tests (for ocular and renal disease), and diabetes-related complications were obtained from three previously published cost-effectiveness analyses [23–25], as outlined in Supplementary Table S3, with two of these conducted in Sweden. Data on accompanying medication use, as well as screening rates, were sourced from Karter et al. [17] and nationwide results (between 1996 and 2020) from the SNDR [50]. Supplementary Table S4 presents annual costs for the rt-CGM device (i.e., Dexcom ONE+) and accessories as well as costs for SMBG test strips. Annual costs were SEK 14,490 for rt-CGM and SEK 1,752 for SMBG, based on a cost of SEK 2.00 per strip [51] and an average use of 2.4 finger-sticks per day over 365 days, as reported in the Steno2Tech trial [13].

Indirect Costs

To assess the impact of productivity loss due to acute diabetes-related complications, indirect costs associated with long-term diabetes complications such as cardiovascular, renal, ocular, and peripheral complications (e.g., neuropathy, active ulcer, and amputation) were estimated. These costs were calculated using publicly-available absenteeism data, combined with the average annual salaries in Sweden for men (SEK 504,000) and women (SEK 453,600), according to 2023 data from the Swedish National Mediation Office (Medlingsinstitutet) [52]. Estimates of absenteeism due to the above-listed complications were sourced from two studies (i.e., Jendle et al.[23] and Persson et al. [10]), and can be found in Supplementary Table S5.

Sensitivity Analyses

For the 52 sensitivity analyses including only direct costs, a range of scenarios were explored, employing one-way, two-way, and three-way sensitivity analyses where appropriate. Here, selected parameter values were altered to assess their impact on the base case ICUR. One such parameter investigated was the comparative treatment efficacy of rt-CGM versus SMBG on HbA1c levels and, in some scenarios, BMI, using a range of alternative values sourced from numerous RCTs and an observational study [13–15, 17]. SHE and DKA rates within the SMBG arm were also altered across a range of scenarios, with these scenarios then supplemented with assumed changes to the FoH utility gain in the rt-CGM arm.

Owing to the availability of alternative equations for HbA1c, BMI progression and cardiovascular risk prediction, [53, 54], a range of three-way sensitivity analyses were conducted for these parameters.

Other parameters investigated within the sensitivity analyses included the: (1) daily frequency of SMBG testing, (2) time horizon, (3) duration of diabetes, (4) mean age of the patient cohort (alongside an assumed duration of diabetes in some cases), (5) discount rate, (6) treatment specific utilities, and (7) price of the rt-CGM system.

The impact of including indirect costs on the base case ICUR was examined in three distinct scenarios. In the first scenario, indirect cost data were taken from Jendle et al. [23]. The second scenario used data from Persson et al. [10] and assumed that costs in the year of the event were equivalent to those in subsequent years when cost data for the latter were unavailable. The third scenario also used data from Persson et al. [10] but included only the indirect costs incurred during the year of the event. A summary of the data input used for each of these scenarios can be found in Supplementary Table S5.

Projected Clinical Outcomes

As part of the base case analysis, the CDM estimated the projected cumulative incidence rate, which was used to calculate absolute risk reduction (ARR) and relative risk (RR) for a range of microvascular and macrovascular diabetes-related complications, based on treatment with rt-CGM versus SMBG. The base case analysis used the SNDR cardiovascular risk prediction equation, which was derived from a cohort of individuals with relatively well-controlled glycemia, with all individuals reporting no history of CVD. Therefore, it is likely the SNDR equation could potentially dilute the expected improvement in CVD risk when considering improvements in HbA1c levels. Additionally, the study that derived the SNDR cardiovascular risk prediction equation reported a mean follow-up period of 5.6 years, which could lead to an underestimation of cardiovascular event risk, as the likelihood of such events typically increases with the duration of diabetes and patient age [55]. Consequently, given that in a T2D cohort with underlying comorbidities and risk factors, the SNDR cardiovascular risk model may underestimate CVD risk and the potential risk reduction achieved through glycemic control, the projected risk of macrovascular complications, over a remaining lifetime horizon, were also estimated using the UKPDS 68 cardiovascular risk prediction equation.

Results

Base Case

The base case results are summarized in Table 2 and Supplementary Figure S1. rt-CGM use led to a gain in mean incremental survival by 0.082 years (11.529 life years for rt-CGM versus 11.447 life years for SMBG). Total mean incremental costs were SEK 138,448 higher with rt-CGM compared with SMBG (SEK 1,151,049 for rt-CGM versus SEK 1,012,601 for SMBG). However, rt-CGM incurred fewer overall diabetes-related complication costs than SMBG over the remaining lifetime horizon. In particular, rt-CGM led to total mean savings of SEK 5,446 and SEK 11,335 associated with renal complications and DKA, respectively.

Table 2.

Summary of base case findings

Base-case probabilistic results	rt-CGM	SMBG
Costs
Glucose monitoring	176,675.35	21,223.32
Management	24,517.62	24,157.76
Cardiovascular complications	460,995.60	460,186.60
Renal complications	227,697.48	233,143.45
Ulcer/amputation/neuropathy complications	246,564.90	244,780.30
Ocular complications	14,598.32	14,999.77
Severe hypoglycemia	0.00	2,775.26
Diabetic ketoacidosis	0.00	11,334.91
Total costs, SEK	1,151,049.38	1,012,601.44
Incremental cost (SEK) [CI Low–CI High]	138,447.91 [120,232.31–156,663.48]
Health outcomes
Life years (LYs)	11.529	11.447
Incremental LYs [CI Low-CI High]	0.082 [0.048–0.115]
Quality-adjusted life years (QALYs)	8.608	7.976
Incremental QALYs [CI Low–CI High]	0.632 [0.605–0.659]
Cost-effectiveness results
ICUR (SEK)	219,063
Probability of rt-CGM being cost-effective at a SEK 500,000 per QALY WTP threshold (%)	70%

CI confidence interval, ICUR incremental cost-utility ratio, rt-CGM real-time continuous glucose monitoring, SEK Swedish Krona, SMBG self-monitoring of blood glucose, WTP willingness-to-pay

rt-CGM yielded a gain in incremental QALYs of 0.632 (8.608 QALYs for rt-CGM versus 7.976 QALYs for SMBG). The ICUR for rt-CGM was SEK 219,063 per QALY gained, which showed rt-CGM to be cost-effective when compared with the WTP threshold of SEK 500,000 per QALY. The plotted cost-effectiveness acceptability data further supported this finding, showing a 70% likelihood of rt-CGM being cost-effective at the SEK 500,000 per QALY threshold (Supplementary Figure S2).

Sensitivity Analyses

Health economic results for each of the 52 scenarios that were investigated using only direct cost parameters are outlined in Table 3. One key trend observed was that the ICURs fell below the WTP threshold of SEK 500,000 per QALY in all cases, showing that rt-CGM consistently remained the cost-effective intervention when various cost and clinical parameter values were altered, in some cases quite significantly.

Table 3.

Summary findings of base case and sensitivity analyses for rt-CGM versus SMBG: direct costs only

Analysis	Total costs (SEK)			QALYs			ICUR (SEK/QALY) gained
	rt-CGM	SMBG	Difference	rt-CGM	SMBG	Difference
Base case (−0.9% HbA1c and −1.1 BMI)	1,151,049.38	1,012,601.44	138,447.91	8.608	7.976	0.632	219,063
HbA1c treatment effect
Diamond (−0.3%)	1,160,652.88	1,012,601.44	148,051.39	8.559	7.976	0.582	254,165
Mobile (−0.4%)	1,156,375.38	1,012,601.44	143,773.98	8.567	7.976	0.591	243,108
Kaiser (−0.56%)	1,159,100.38	1,012,601.44	146,499	8.587	7.976	0.611	239,612
Steno2Tech (−0.9%), No BMI change	1,144,724.25	1,012,601.44	132,122.80	8.582	7.976	0.606	217,881
Diamond (−0.3%), No BMI change	1,156,921.63	1,012,601.44	144,320.23	8.524	7.976	0.548	263,214
Mobile (−0.4%), No BMI change	1,155,601.75	1,012,601.44	143,000.27	8.536	7.976	0.560	255,312
Kaiser (−0.56%), no BMI change	1,149,571.25	1,012,601.44	136,969.83	8.559	7.976	0.582	235,141
+ 30% Steno2Tech effect (−1.17%)	1,144,075.50	1,012,601.44	131,474.02	8.634	7.976	0.658	199,748
-30% Steno2Tech effect (−0.63%)	1,154,470.63	1,012,601.44	141,869.20	8.583	7.976	0.607	233,568
Treatment specific utility
+ 50% AFS (0.045)	1,151,049.38	1,012,601.44	138,447.91	8.781	7.976	0.805	172,049
−50% AFS (0.015)	1,151,049.38	1,012,601.44	138,447.91	8.430	7.976	0.454	305,220
No AFS (0)	1,151,049.38	1,012,601.44	138,447.91	8.263	7.976	0.287	482,902
+ 50% FoH (0.023)	1,151,049.38	1,012,601.44	138,447.91	8.694	7.976	0.718	192,717
−50% FoH (0.008)	1,151,049.38	1,012,601.44	138,447.91	8.522	7.976	0.546	253,707
No FoH (0)	1,151,049.38	1,012,601.44	138,447.91	8.430	7.976	0.454	305,220
AFS (0.01) and FoH (0) and 10 SMBG tests per day	1,151,049.38	1,079,808.63	71,240.72	8.199	7.976	0.223	318,893
SHE/DKA rate
+ 50% SHE and DKA rate (6, 3.75)	1,151,049.38	1,013,984.69	137,064.66	8.608	7.944	0.665	206,236
−50% SHE and DKA rate (2, 1.25)	1,151,049.38	1,008,395.75	142,653.58	8.608	7.993	0.615	231,769
No SHE and DKA rate (0,0)	1,151,049.38	993,626.94	157,422.41	8.608	8.018	0.590	266,818
SHE/DKA rate and FoH utility
−50% SHE and DKA and FoH (2, 1.25, 0.008)	1,151,04(9.38	1,008,395.75	142,653.58	8.522	7.993	0.529	269,616
+ 50% SHE and DKA and FoH (6, 3.75, 0.023)	1,151,049.38	1,013,984.69	137,064.66	8.694	7.944	0.751	182,534
No SHE and DKA and FoH (0, 0, 0)	1,151,049.38	993,626.94	157,422.41	8.430	8.018	0.412	382,557
SMBG tests per day
1.5	1,151,049.38	1,004,642.69	146,406.64	8.608	7.976	0.632	231,656
4	1,151,049.38	1,024,981.69	126,067.63	8.608	7.976	0.632	199,474
5	1,151,049.38	1,035,593.38	115,455.97	8.608	7.976	0.632	182,683
10	1,151,049.38	1,079,808.63	71,240.72	8.608	7.976	0.632	112,723
Time horizon
1 year	80,179.18	69,369.47	10,809.71	0.726	0.679	0.047	230,484
5 years	361,514.03	314,106.81	47,407.22	3.198	2.983	0.214	221,013
10 years	640,978.56	559,750.13	81,228.45	5.444	5.064	0.380	213,815
30 years	1,119,876.88	990,161.81	129,715.03	8.481	7.866	0.615	210,987
Duration of diabetes
1 year	1,112,157.88	964,210.19	147,947.70	8.959	8.322	0.637	232,257
5 years	1,116,023.50	974,335.38	141,688.13	8.872	8.249	0.623	227,502
10 years	1,124,178.38	985,659.19	138,519.27	8.781	8.145	0.637	217,558
Mean age
30 years	2,165,056.75	1,958,504.50	206,552.33	14.660	13.656	1.004	205,770
40 years	1,853,132.25	1,659,784	193,348.17	13.131	12.222	0.908	212,892
50 years	1,513,783.88	1,345,193.63	168,590.27	11.167	10.376	0.791	213,055
Mean age and diabetes duration
Age 18 years and diabetes duration 3 years	2,219,429.75	1,990,559	228,870.63	16.182	15.100	1.082	211,506
Age 25 years and diabetes duration 5 years	2,161,011.75	1,943,151.88	217,859.73	15.559	14.523	1.036	210,371
Progression approach and cardiovascular risk prediction equation
HbA1c UKPDS 68, BMI UKPDS 90, CV RE UKPDS 68	1,518,027.13	1,417,236.50	100,790.63	10.033	9.212	0.821	122,706
HbA1c UKPDS 68, BMI UKPDS 90, CV RE UKPDS 82	1,112,069.75	1,016,315.06	95,754.70	8.458	7.786	0.672	142,556
HbA1c UKPDS 68, BMI UKPDS 90, CV RE SNDR	1,197,974.75	1,086,046.25	111,928.46	8.476	7.738	0.738	151,706
HbA1c and BMI clinical table, UKPDS 68 CV RE	1,538,170.38	1,607,152.38	−68,982.06	10.020	8.969	1.050	-65,678
HbA1c and BMI clinical table, SNDR CV RE	1,192,472.25	1,138,395	54,077.29	8.472	7.561	0.911	59,341
HbA1c and BMI SNDR, CV RE UKPDS 68	1,410,437.38	1,249,803.88	160,633.55	10.255	9.506	0.749	214,321
HbA1c and BMI SNDR, CV RE UKPDS 82	1,071,686.75	943,514.56	128,172.17	8.624	8.027	0.597	214,838
Discount rate for costs and effects
0%	1,597,838.13	1,408,258.38	189,579.70	11.377	10.519	0.859	220,801
1.5%	1,343,407.50	1,182,878.75	160,528.70	9.819	9.089	0.730	219,842
5%	960,579	844,132.88	116,446.08	7.373	6.839	0.534	218,146
Price of rt-CGM (SEK)
+ 10%	1,168,717	1,012,601	156,115	8.608	7.976	0.632	247,018
−10%	1,133,382	1,012,601	120,780	8.608	7.976	0.632	191,108

AFS avoidance of fingerstick testing, BMI body mass index, CV cardiovascular, DKA diabetic ketoacidosis, FoH fear of hypoglycemia, HbA1c glycated hemoglobin, ICUR incremental cost-utility ratio, QALY quality-adjusted life year, RE risk equation, rt-CGM real-time continuous glucose monitoring, SEK Swedish Krona, SHE severe hypoglycemic event, SMBG self-monitoring of blood glucose, SNDR Swedish National Diabetes Register, UKPDS United Kingdom Prospective Diabetes Study

Across all nine scenarios that investigated alternative HbA1c treatment effects rt-CGM remained cost-effective, with ICURs ranging from SEK 199,748 to SEK 263,214 per QALY gained. Similar trends were observed for both the time horizon and duration of diabetes, namely that as each increased, the ICUR value decreased, showing that rt-CGM became more cost-effective over time. This trend was also evident in the mean age scenarios, with younger simulated patient cohorts resulting in lower ICUR values compared with older cohorts.

The use of alternative HbA1c and BMI progression equations and CVD risk prediction equations led to some of the most favorable ICUR results for rt-CGM. One such scenario, where the clinical table was adopted for the HbA1c and BMI progression approach, alongside a UKPDS 68 CVD risk prediction equation, led to the lowest ICUR of SEK −65,678 per QALY gained, showing rt-CGM to be the dominant intervention.

Changes to the treatment-specific utility applied to the rt-CGM arm (i.e., FoH and AFS utilities) led to some of the greatest impacts on the ICUR. When the AFS utility was increased by 50% (utility value: 0.045), the ICUR decreased to SEK 172,049 per QALY gained. When the AFS benefit was removed entirely from the rt-CGM arm, the ICUR increased to the highest value observed across all scenarios explored, at SEK 482,902 per QALY gained.

SMBG testing frequency also led to notable changes in the base case ICUR. A daily frequency of 1.5 SMBG tests (according to data obtained from the MOBILE study [14]) led to the ICUR increasing to SEK 231,656 per QALY gained. However, when SMBG testing frequency was increased to 10 tests per day (sourced from a Swedish Dental and Pharmaceutical Benefits Agency health technology assessment of the FreeStyle Libre device [51]), the ICUR reduced to SEK 112,723 per QALY gained, the third lowest value across all 52 scenarios including direct costs.

The inclusion of indirect costs reduced the incremental costs associated with rt-CGM in all three scenarios when compared with the original base case, which in turn lowered the ICUR values. The lowest of these ICURs (SEK 170,686 per QALY gained) was observed in the scenario using indirect cost data from Jendle et al. [23] (Table 4).

Table 4.

Summary findings of sensitivity analyses for rt-CGM versus SMBG: including indirect costs

Analysis	Total costs (SEK)			QALYs			ICUR (SEK/QALY) gained
	rt-CGM	SMBG	Difference	rt-CGM	SMBG	Difference
Scenario analysis including indirect costs
Indirect costs from Jendle 2021	1,432,488	1,324,614	107,873	8.608	7.976	0.632	170,686
Indirect costs from Persson 2019 (event year + subsequent years)	1,527,731	1,401,816	125,915	8.608	7.976	0.632	199,233
Indirect costs from Persson 2019 (event year only)	1,192,037	1,061,158	130,879	8.608	7.976	0.632	207,086

ICUR incremental cost-utility ratio, QALY quality-adjusted life year, rt-CGM real-time continuous glucose monitoring, SEK Swedish Krona, SMBG self-monitoring of blood glucose

Projected Clinical Outcomes

Projected cumulative incidence of diabetes-related complication rates and the respective ARRs and RRs for rt-CGM versus SMBG are summarized in Table 5. With the exception of two CVD complications (angina [ARR:0.08] and stroke [ARR:0.00]), all ARR values were negative, demonstrating the comparative long-term clinical advantage of rt-CGM over SMBG, with neuropathy showing the lowest ARR of −1.91. Similarly, in the majority of cases, RR values were below 1.00, demonstrating a lower risk of complications with rt-CGM compared with SMBG, with proliferative diabetic retinopathy showing the lowest RR value of 0.93.

Table 5.

Projected diabetes complication rates for rt-CGM versus SMBG

Organ system	Complication	CI% [95% CI]		ARR (rt-CGM versus SMBG) % [95% CI]	RR (rt-CGM versus SMBG) [95% CI]
		rt-CGM	SMBG
Ophthalmic	Background diabetic retinopathy	19.03 [18.62, 19.45]	20.09 [19.66, 20.51]	−1.06 [−1.65, −0.47]	0.95 [0.92, 0.98]
	Proliferative diabetic retinopathy	7.16 [6.98, 7.34]	7.69 [7.51, 7.87]	−0.53 [−0.79, −0.27]	0.93 [0.90, 0.96]
	Macular edema	14.3 [13.97, 14.62]	15.06 [14.74, 15.39]	−0.76 [−1.22, −0.30]	0.95 [0.92, 0.98]
	Severe vision loss	12.47 [12.17, 12.77]	13.06 [12.76, 13.37]	−0.59 [−1.02, −0.16]	0.95 [0.92, 0.99]
	Cataract	8.34 [8.17, 8.50]	8.53 [8.36, 8.69]	−0.19 [−0.42, 0.04]	0.98 [0.95, 1.01]
Renal	Microalbuminuria	17.76 [17.36, 18.17]	18.73 [18.32, 19.13]	−0.97 [−1.54, −0.40]	0.95 [0.92, 0.98]
	Gross proteinuria	12.82 [12.52, 13.12]	13.41 [13.10, 13.71]	−0.59 [−1.02, −0.16]	0.96 [0.93, 0.99]
	End-stage renal disease	2.83 [2.73, 2.92]	2.94 [2.85, 3.04]	−0.11 [−0.25, 0.03]	0.96 [0.92, 1.01]
Cardiovascular	Congestive heart failure	28.07 [27.31, 28.82]	29.07 [28.30, 29.83]	−1.00 [−2.08, 0.08]	0.97 [0.93, 1.00]
	Congestive heart failure event fatality	16.59 [16.30, 16.88]	16.93 [16.64, 17.22]	−0.34 [−0.75, 0.07]	0.98 [0.96, 1.00]
	Peripheral vascular disease onset	11.82 [11.62, 12.02]	11.96 [11.76, 12.16]	−0.14 [−0.42, 0.14]	0.99 [0.97, 1.01]
	Angina	21.75 [21.25, 22.25]	21.67 [21.18, 22.17]	0.08 [−0.62, 0.78]	1.00 [0.97, 1.04]
	Stroke event	19.24 [18.93, 19.55]	19.24 [18.92, 19.56]	0.00 [−0.44, 0.44]	1.00 [0.98, 1.02]
	Stroke fatality	7.64 [7.53, 7.76]	7.68 [7.56, 7.80]	−0.04 [−0.20, 0.12]	0.99 [0.97, 1.02]
	Myocardial infarction event	47.61 [47.19, 48.04]	47.72 [47.30, 48.14]	−0.11 [−0.71, 0.49]	1.00 [0.99, 1.01]
	Myocardial infarction fatality	29.81 [29.46, 30.17]	29.87 [29.52, 30.22]	−0.06 [−0.56, 0.44]	1.00 [0.98, 1.01]
Extremities	Ulcer	6.03 [5.47, 6.59]	6.25 [5.66, 6.84]	−0.22 [−1.04, 0.60]	0.96 [0.84, 1.10]
	Recurring foot ulcer	16.37 [15.99, 16.74]	16.52 [16.12, 16.92]	−0.15 [−0.70, 0.40]	0.99 [0.96, 1.02]
	Amputation from foot ulcer	3.84 [3.57, 4.11]	3.89 [3.61, 4.17]	−0.05 [−0.44, 0.34]	0.99 [0.89, 1.09]
	Amputation from recurring foot ulcer	4.02 [3.86, 4.19]	4.03 [3.86, 4.21]	−0.01 [−0.25, 0.23]	1.00 [0.94, 1.06]
	Neuropathy	52.41 [51.68, 53.16]	54.32 [53.59, 55.05]	−1.91 [−2.95, −0.87]	0.96 [0.95, 0.98]

ARR absolute risk reduction, CI confidence interval, RR relative risk, rt-CGM real-time continuous glucose monitoring, SMBG self-monitoring of blood glucose

The projected ARRs, RRs and lifetime complications avoided per 1,000 patients for microvascular and macrovascular complications using alternative risk prediction and progression equations, are summarized in Table 6. Results followed a similar trend to those in Table 5; however, use of the clinical table approach, when compared with the SNDR and UKPDS 68, consistently led to the most favorable results for rt-CGM, showing the largest negative ARRs, the lowest RRs, and the greatest numbers of lifetime complications avoided per 1000 patients, across the majority of diabetes-related complications.

Table 6.

Comparison of projected diabetes-related complications for rt-CGM versus SMBG for each progression and CVD risk equation used

Complications	SNDR			UKPDS 68			Clinical table
	ARR (%) [95% CI]	RR (%) [95% CI]	LCs avoided *	ARR (%) [95% CI]	RR (%) [95% CI]	LCs avoided *	ARR (%) [95% CI]	RR (%) [95% CI]	LCs avoideda
Background diabetic retinopathy	−1.06 [−1.65, −0.47]	0.95 [0.92, 0.98]	11	−2.97 [−3.89, −2.05]	0.92 [0.89, 0.94]	30	−8.83 [−10.02, −7.64]	0.79 [0.77, 0.82]	88
Proliferative diabetic retinopathy	−0.53 [−0.79, −0.27]	0.93 [0.90, 0.96]	5	−1.88 [−2.37, −1.39]	0.88 [0.85, 0.91]	19	−6.69 [−7.46, −5.92]	0.70 [0.67, 0.73]	67
Macular edema	−0.76 [−1.22, −0.30]	0.95 [0.92, 0.98]	8	−2.39 [−3.15, −1.63]	0.91 [0.89, 0.94]	24	−7.99 [−9.03, −6.95]	0.77 [0.75, 0.80]	80
Severe vision loss	−0.59 [−1.02, −0.16]	0.95 [0.92, 0.99]	6	−1.75 [−2.51, −0.99]	0.92 [0.89, 0.96]	18	−5.6 [−6.57, −4.63]	0.80 [0.77, 0.83]	56
Cataract	−0.19 [−0.42, 0.04]	0.98 [0.95, 1.01]	2	−0.56 [−0.92, −0.20]	0.96 [0.93, 0.98]	6	−1.96 [−2.41, −1.51]	0.87 [0.84, 0.89]	20
Microalbuminuria	−0.97 [−1.54, −0.40]	0.95 [0.92, 0.98]	10	−3.07 [−3.99, −2.15]	0.91 [0.89, 0.94]	31	−9.75 [−10.99, −8.51]	0.77 [0.75, 0.80]	98
Gross proteinuria	−0.59 [−1.02, −0.16]	0.96 [0.93, 0.99]	6	−2.52 [−3.30, 1.74]	0.91 [0.88, 0.93]	25	−9.56 [−10.71, −8.41]	0.74 [0.71, 0.77]	96
End stage renal disease	−0.11 [−0.25, 0.03]	0.96 [0.92, 1.01]	1	−0.98 [−1.35, −0.61]	0.89 [0.85, 0.93]	10	−6.46 [−7.26, −5.66]	0.62 [0.58, 0.66]	65
Ulcer	−0.22 [−1.04, 0.60]	0.96 [0.84, 1.10]	2	−0.95 [−1.91, 0.01]	0.88 [0.78, 1.00]	10	−1.46 [−2.49, −0.43]	0.84 [0.74, 0.95]	15
Recurring foot ulcer	−0.15 [−0.70, 0.40]	0.99 [0.96, 1.02]	2	−0.62 [−1.34, 0.10]	0.97 [0.93, 1.01]	6	−1.04 [−1.81, −0.27]	0.95 [0.91, 0.99]	11
Amputation from foot ulcer	−0.05 [−0.44, 0.34]	0.99 [0.89, 1.09]	1	−0.37 [−0.97, 0.23]	0.95 [0.86, 1.04]	4	−0.46 [−1.10, 0.18]	0.94 [0.85, 1.03]	5
Recurring amputation	−0.01 [−0.25, 0.23]	1.00 [0.94, 1.06]	0	−0.19 [−0.61, 0.23]	0.97 [0.90, 1.04]	2	−0.51 [−0.96, −0.06]	0.92 [0.86, 0.99]	5
Neuropathy	−1.91 [−2.95, −0.87]	0.96 [0.95, 0.98]	19	−3.42 [−4.55, −2.29]	0.95 [0.94, 0.97]	34	−7.28 [−8.47, −6.09]	0.90 [0.89, 0.92]	73
Congestive heart failure	−1.00 [−2.08, 0.08]	0.97 [0.93, 1.00]	10	−1.08 [−2.16, 0.00]	0.94 [0.88, 1.00]	11	−1.88 [−2.99, −0.77]	0.90 [0.84, 0.96]	19
Congestive heart failure fatality	−0.34 [−0.75, 0.07]	0.98 [0.96, 1.00]	3	NA	NA	NA	NA	NA	NA
Peripheral vascular disease	−0.14 [−0.42, 0.14]	0.99 [0.97, 1.01]	1	−0.44 [−0.90, 0.02]	0.97 [0.95, 1.00]	4	−2.54 [−3.14, −1.94]	0.88 [0.85, 0.90]	25
Angina	0.08 [−0.62, 0.78]	1.00 [0.97, 1.04]	0	−0.72 [−1.37, −0.07]	0.96 [0.92, 1.00]	7	−1.70 [−2.38, −1.02]	0.91 [0.87, 0.94]	17
Stroke	0.00 [−0.44, 0.44]	1.00 [0.98, 1.02]	0	−0.41 [−1.32, 0.50]	0.97 [0.90, 1.04]	4	−1.32 [−2.24, −0.40]	0.91 [0.84, 0.97]	13
Stroke fatality	−0.04 [−0.20, 0.12]	0.99 [0.97, 1.02]	0	−1.53 [−2.49, −0.57]	0.94 [0.91, 0.98]	15	−4.84 [−5.80, −3.88]	0.84 [0.82, 0.87]	48
Myocardial Infarction	−0.11 [−0.71, 0.49]	1.00 [0.99, 1.01]	1	−1.29 [−2.45, −0.13]	0.95 [0.92, 1.00]	13	−2.87 [−4.05, −1.69]	0.91 [0.87, 0.94]	29
Myocardial Infarction fatality	−0.06 [−0.56, 0.44]	1.00 [0.98, 1.01]	1	NA	NA	NA	NA	NA	NA
Diabetes mortality	NA	NA	NA	−0.50 [−1.54, 0.54]	0.97 [0.92, 1.03]	5	−2.08 [−3.20, −0.96]	0.90 [0.85, 0.95]	21

^aPer 1000 patients

AR absolute risk reduction, CI confidence interval, LC lifetime complications, NA not applicable, RR relative risk, SNDR Swedish National Diabetes Register, UKPDS United Kingdom Prospective Diabetes Study

Discussion

The health economic analysis showed that for individuals living with insulin-treated T2D in Sweden, rt-CGM would be a cost-effective intervention when compared with SMBG, resulting in incremental costs of SEK 138,448, and an incremental gain of 0.632 QALYS. The ICUR of SEK 219,063 per QALY gained fell considerably below the adopted WTP threshold of SEK 500,000 per QALY. These findings were robust under a wide range of plausible assumptions around key input parameters. The results are also consistent with those from similar, previously conducted cost-utility analyses of rt-CGM versus SMBG set across France, South Korea, Spain, Denmark, and Canada [18–22].

Key findings from the sensitivity analyses showed that, where alternative HbA1c and BMI progression risk equations and CVD risk prediction equations were used in place of the SNDR equations, rt-CGM was shown to be even more cost-effective (Table 3). Notably, when the clinical table was adopted for the HbA1c and BMI progression approach, alongside the UKPDS 68 CVD risk prediction equation, rt-CGM became the dominant intervention when compared with SMBG. This finding further highlighted the previously discussed limitations of using the SNDR HbA1c and BMI progression approach to generate accurate long-term projections of risk factor progression and to predict the occurrence of diabetes-related complications, which stemmed from the relatively short mean follow-up period of four years in the original study. Additionally, the SNDR CVD risk prediction equation may not accurately estimate long-term cardiovascular risk, as it was derived from a Swedish participant sample with well-controlled HbA1c levels and no history of CVD, which may not be representative of the broader Swedish population living with T2D. On the basis of these points, as well as the results of the base case and sensitivity analyses, the use of the SNDR equations led to health economic and projected clinical outcomes that are likely to be conservative reflections of the cost-effectiveness and long-term clinical value of rt-CGM compared with SMBG. Nevertheless, our base case result demonstrates that there is still an incremental clinical and economic benefit of rt-CGM in people with T2D who may experience comparatively less severe disease progression and complications over their lifetime. By incorporating alternative risk equations from a range of sources and studies, the sensitivity analyses allowed for a thorough assessment of parameter value uncertainties and differences, providing a comprehensive evaluation of the clinical benefits associated with rt-CGM.

The HbA1c (and in some cases, BMI) treatment effect of rt-CGM was varied, using a range of values sourced from RCTs and a large-scale retrospective observational study [13–15, 17]. Not only did this allow for an assessment of the extent to which base case results would change based on the size and type of the treatment effect(s) investigated, but it also allowed for a comparison between the impact of RCT and real-world clinical data on the results obtained. Although RCTs allow for treatment effects to be better isolated from known confounding variables [56], the rigorous design of such studies could mean that the recruited sample may fail to be representative of the wider patient population of interest, whilst the treatment regimens (and subsequent efficacy) may also fail to reflect those observed in routine clinical practice [57]. Observational studies allow for additional factors such as patient adherence, a wider diversity in patient characteristics, and impacts of comorbidities to be considered when measuring treatment effects, and in some cases, also allow for these effects to be measured over a longer period of time than typically seen in RCTs [58]. In all these scenarios, regardless of the type of study used to source treatment effect data, the ICUR fell considerably below the WTP threshold of SEK 500,000 per QALY, showing that the cost-effectiveness of rt-CGM was robust across a range of evidence sources, whilst being reflective of both highly controlled and real-world clinical environments.

To the best of the authors’ knowledge, this is the first cost-utility analysis of rt-CGM versus SMBG to incorporate indirect costs associated with diabetes-related complications. Three distinct approaches to estimating these costs (each based on Swedish data sources) were applied separately to the base case. In all three instances, the resulting ICUR was lower than that of the base case, highlighting the substantial economic value of rt-CGM beyond direct healthcare savings. This is particularly relevant given that studies have shown sustained recent increases in both the prevalence of T2D and sickness-related absenteeism in Sweden [1, 59]. These trends demonstrate the importance of incorporating indirect costs within health economic evaluations, as effects of productivity losses experienced by patients can significantly influence the cost-effectiveness of an intervention, even without considering the subsequent impact to employers and governments. It is important to note, however, that the current analysis did not capture other forms of indirect costs (e.g., those incurred by caregivers) due to limited data availability. As a result, the cost-effectiveness of rt-CGM (from a societal perspective) could be underestimated in this evaluation.

Another strength of the study was the use of country-specific sources, where available, to identify complication and management costs for T2D, as well as screening rates and medication use. These sources included two cost-effectiveness studies set in Sweden and one in Europe [23–25], and the SNDR [50].The racial distribution of the simulated cohort was sourced from a real-world study conducted in Sweden. The SNDR equation was included in the base case, despite its known limitations, to estimate the progression of key risk factors (HbA1c and BMI) and the cardiovascular risk over a lifetime. As a result, the results presented here are likely to provide a plausible representation of outcomes observed in the Swedish population living with insulin-treated T2D.

Some potential limitations affecting the generalizability of the findings are that the treatment effects on HbA1c and BMI were derived from a single center RCT conducted in Denmark. Nevertheless, this study represented the most appropriate clinical evidence available at the time of the analysis, reflecting a Nordic healthcare setting. The rates of severe hypoglycemia and diabetic ketoacidosis (DKA) were sourced from a US-based study, as the Steno2Tech trial did not report significant differences in these events between intervention groups. It is important to note that hypoglycemia event rates reported in RCTs may be misleading since they often exclude individuals at higher risk (such as those with elevated baseline event rates or impaired awareness of hypoglycemia) potentially underestimating the true burden in real-world populations. Sensitivity analyses showed that even when the benefit of reduced hypoglycemic events was halved or removed entirely, rt-CGM remained cost-effective under the stated WTP. The costs of diabetes-related complications were derived from older, country-specific sources and subsequently adjusted to current values using inflation. While this method aligns with previous approaches, it may lead to under- or overestimation of the projected economic benefits, as healthcare costs and treatment pathways continue to evolve over time.

Another limitation is that the current analysis utilized rt-CGM cost data specific to the Dexcom ONE+ system, while clinical efficacy inputs derived from the Steno2tech study employed the Dexcom G6 system. There are some feature differences between the two systems, such as the inclusion of the optional "Urgent Low Soon" alert in the G6. Several real-world studies have demonstrated that these features (e.g., urgent low soon) contribute to improved glycemic outcomes [60, 61]. However, evidence from studies evaluating various Dexcom rt-CGM systems suggests that glycemic outcomes are generally comparable across different models in diverse patient populations and real-world settings [17, 62]. A prospective observational study conducted in the United Kingdom reported a significant 1.7% reduction in mean HbA1c after 6 months of Dexcom ONE use in a cohort of individuals with type 2 diabetes (T2D) treated with insulin [62]. Notably, this cohort had experienced prolonged suboptimal glycemic control despite treatment with two or more insulin injections, with or without additional antidiabetic medications or prior use of intermittent scanning CGM. The study also reported improvements in diabetes-related distress and other patient-reported outcomes following the initiation of Dexcom ONE in a socioeconomically diverse T2D population [62]. These findings support the conclusion that Dexcom ONE and ONE+ systems offer comparable glycemic benefits while providing optimal features for effective glucose monitoring.

Conclusions

For individuals living with insulin-treated T2D in Sweden, rt-CGM is likely to be a cost-effective management option relative to SMBG, resulting in improved glycemic control, reduced onset of diabetes-related complications, and improved survival, with these findings remaining consistent across a range of scenario analyses. From a societal perspective, rt-CGM could aid in minimizing productivity losses for patients owing to diabetes-related complications, thereby alleviating broader economic burdens incurred by employers and governments.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Johan Jendle, Hamza Alshannaq. and Gregory.J.Norman. conceived of the study. Sabrina Ilham. designed the analyses and conducted the analyses. Jessica Y.Matuoka. designed and conducted scenario analyses including indirect cost data. Waqas Ahmed and Richard F Pollock drafted the manuscript. All authors contributed to data interpretation and all authors reviewed and edited the manuscript. All authors approved the final version of the manuscript.

Funding

Funding for the analysis, manuscript preparation, and the journal’s Rapid Service Fee was provided by Dexcom.

Data Availability

The present study did not report original data. Data used for modeling were derived from public sources and have been reported in full in the paper and the accompanying online-only supplemental material.

Declarations

Conflict of Interest

Richard F Pollock and Waqas Ahmed are full-time employees, and Richard F Pollock is a director and shareholder in, Covalence Research Ltd., which has received consulting fees from Dexcom for preparing this manuscript and from Dexcom outside the submitted work. Sabrina Ilham, Jessica Y. Matuoka, and Gregory J Norman are current employees of Dexcom. Hamza Alshannaq is a former employee of Dexcom, currently an employee of Neurocrine Biosciences Gregory J Norman and Jessica Y. Matuoka hold stock or stock options in Dexcom. Johan Jendle has received remuneration for participation in Advisory Boards or invited lectures from, Abbott Diabetes Care, Dexcom, Eli Lilly, Medtronic, NovoNordisk, Ypsomed, and Tandem. He also has received research support to his institution from Novo Nordisk and Medtronic.

Ethical Approval

The analysis was based on previously conducted studies and did not contain any new studies involving human or animal participants that were performed by any of the authors.