Evaluating the Cost-Effectiveness of Real-Time Continuous Glucose Monitoring Versus Self-Monitoring of Blood Glucose in the Treatment of Patients with Insulin-Treated Type 2 Diabetes in Australia

Abstract

Introduction

Type 2 diabetes (T2D) is a major public health concern in Australia, associated with substantial clinical, humanistic, and economic burden. The condition is linked to high rates of cardiovascular and microvascular complications, premature mortality, and reduced quality of life. Effective glycemic management is central to reducing these adverse outcomes. Real-time continuous glucose monitoring (RT-CGM) has been shown to improve glycemic control in insulin-treated T2D compared with self-monitoring of blood glucose (SMBG). However, evidence of its cost-effectiveness in the Australian setting is limited. This study aimed to evaluate the cost-effectiveness of Dexcom ONE+ RT-CGM versus SMBG in adults with insulin-treated T2D in Australia.

Methods

A lifetime economic evaluation was conducted using version 10 of the IQVIA CORE Diabetes Model. The analysis simulated clinical and economic outcomes for two subgroups: those on intensive insulin therapy (IIT) and non-intensive insulin therapy (NIIT). Treatment effects were sourced from clinical trials and real-world evidence. Outcomes included life years, quality-adjusted life years (QALYs), and direct healthcare costs. Incremental cost-effectiveness ratios (ICERs) were calculated as cost per QALY gained. Scenario and sensitivity analyses tested robustness.

Results

RT-CGM was dominant compared to SMBG in both IIT and NIIT subgroups. In IIT, RT-CGM yielded 0.567 additional QALYs and cost savings of AUD 9869. In NIIT, it yielded 0.319 additional QALYs and savings of AUD 5253. Results were robust across sensitivity analyses. Health equity considerations were also identified, particularly for Indigenous populations and those with youth-onset T2D.

Conclusions

RT-CGM was dominant in both insulin-treated subgroups, improving patient outcomes while reducing healthcare costs. These findings highlight the potential value of RT-CGM for broad reimbursement in Australia and the importance of addressing inequities in glycemic management, particularly among Indigenous Australians and younger individuals with T2D.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12325-025-03430-1.

Key Summary Points

Why carry out this study?

Real-time continuous glucose monitoring (RT-CGM) was dominant compared with self-monitoring of blood glucose (SMBG) in insulin-treated type 2 diabetes in Australia, producing both health gains and cost savings in intensive and non-intensive insulin therapy subgroups.

The analysis was based on a validated diabetes model with inputs from peer-reviewed literature, clinical studies, and national data sources, ensuring methodological rigor.

What was learned from the study?

Results were robust across a range of sensitivity and scenario analyses, including variation of discount rate, baseline hemoglobin A1c (HbA1c), age and diabetes duration, treatment effect, utility values for finger-stick avoidance and fear of hypoglycemia, SMBG and RT-CGM costs, racial distribution, and cardiovascular risk.

Scenario analyses highlighted particular value for high-need populations, including Indigenous Australians and younger adults, supporting potential equity benefits of RT-CGM.

Limitations include reliance on non-Australian baseline characteristics and census data that lack ethnicity detail, use of some older cost inputs, assumptions regarding HbA1c persistence and severe hypoglycemia risk, limited evidence for non-intensive insulin therapy, and exclusion of intermittently scanned CGM (IS-CGM/Flash) as a comparator

Introduction

Type 2 diabetes (T2D) affects many Australians. According to the Australian Institute of Health and Welfare (AIHW), just over 1.3 million were diagnosed between 2000 and 2021, equating to approximately 60,000 new cases per year [1]. T2D impairs insulin function, causing hyperglycemia that, if unmanaged, leads to severe complications including micro- and macrovascular damage [2]. Such complications are common, with 40.3% of adults with T2D also having cardiovascular disease (CVD), as reported in the CAPTURE cross-sectional study [3]. Additionally, age-adjusted hospitalization rates for myocardial infarction, stroke, and heart failure for Australians with T2D between 2018 and 2019 were reported as 0.78%, 0.54%, and 1.30%, respectively [4]. Microvascular complications are also prevalent, with an Australian primary care study revealing that chronic kidney disease (CKD) affects nearly half of patients with T2D consulting their GP [5]. Similarly, diabetic retinopathy was found to have a prevalence of 21.9% among Australians with diagnosed T2D [6].

These complications contribute to premature mortality, with an estimated T2D age-standardized mortality rate of 4122 per 100,000 [7]. T2D also imposes a substantial humanistic burden, with a diagnosis linked to a fivefold increase in the odds of Australians reporting a significant decline in quality of life (QoL) [8]. The economic burden is considerable, with annual direct healthcare costs for Australians with T2D estimated at AUD 1359, 1328, and 1676 for individuals with normal weight, overweight, and obesity, respectively [9]. Projections suggest that CVD-related healthcare costs in patients with T2D will reach AUD 9.59 billion between 2022 and 2031 [10].

Glycemic management is a key determinant in the development and progression of complications. A 1% reduction in mean glycated hemoglobin (HbA1c) is associated with a 14% lower risk of myocardial infarction and a 37% lower risk of microvascular complications [11]. Treatments that improve glycemic control can reduce complication incidence, mortality, and healthcare expenditure.

Real-time continuous glucose monitoring (RT-CGM) systems measure and display interstitial glucose concentrations on a receiver or mobile device. Dexcom RT-CGM has been shown to significantly improve glycemic control in patients with T2D on multiple daily injections, as evidenced by reductions in HbA1c [12–15]. However, evidence of its cost-effectiveness in this population is also necessary to support broader adoption within the Australian healthcare system. Evidence of RT-CGM cost-effectiveness from the Australian payer perspective is currently limited. This study aimed to address this gap by evaluating the cost-effectiveness of Dexcom ONE+ RT-CGM versus self-monitoring of blood glucose (SMBG).

Methods

A widely used and validated model—version 10 of the IQVIA CORE Diabetes Model (CDM)—was used to assess the cost-effectiveness of RT-CGM versus SMBG for people with T2D using insulin therapy in Australia. The analysis examined costs, quality-adjusted life years (QALYs), and life years (LYs) over a 50-year lifetime horizon, with a 5% annual discount rate applied to reflect the diminishing value of future outcomes [16]. The 50-year time horizon allows for all people modeled in the microsimulations to transition to death, encapsulating costs and disease complications to end-of-life. As an economic modelling analysis based on published literature, the present study did not enroll human participants or make useof human biological samples or individually identifi able data, and ethics approval was therefore not required.

Model Structure

The CDM is a microsimulation model that uses defined health states to project the progression of diabetes and its complications. A detailed description of the model structure and the validation process has been provided by Palmer et al., and validations of later model iterations have been reported by McEwan et al. [17, 18]. The model captures a broad range of T2D complications, including microvascular (e.g., retinopathy, nephropathy, neuropathy) and macrovascular (e.g., CVD) complications, projecting long-term clinical and economic outcomes. The CDM outputs key metrics, including life expectancy (LE), QALYs, direct and indirect costs, cumulative incidence and time to onset of diabetic complications, and incremental cost-effectiveness ratios (ICERs) per additional LY or QALY gained. In scenarios where dominance of RT-CGM was observed, meaning both more effective and less costly than SMBG, net monetary benefit (NMB) was used in place of ICERs, with a larger NMB indicating a more dominant intervention.

Model Inputs

Base Case Scenarios

The analysis included base case scenarios for two population subgroups: patients with T2D on basal or non-intensive insulin therapy (categorized as patients on NIIT) and patients with T2D on intensive insulin therapy (IIT). NIIT refers to once-daily, long-acting insulin. When NIIT insufficiently controls blood glucose, patients typically progress to IIT, which involves multiple doses of both short-acting and long-acting insulin per day to maintain glycemia [19]. The study assumed that baseline characteristics for both populations were the same, while clinical effectiveness and utility values differed.

Baseline Characteristics

The data for the baseline characteristics were assumed to be the same for both NIIT and IIT T2D subpopulations, and were sourced from the Kaiser Permanente (KP) retrospective cohort study, with additional risk factors sourced from the ACCORD trial [14, 20, 21]. The key baseline characteristics included a baseline mean HbA1c of 8.3%, a mean age of 64.5 years, and a mean duration of diabetes of 16 years, respectively. Publicly available data from the Australian Bureau of Statistics served as a proxy for ethnic distribution in the analysis [22] (Supplementary Table 1).

Clinical Effectiveness

The clinical effectiveness of RT-CGM systems versus SMBG for the IIT population was measured through the HbA1c effect, rates of severe hypoglycemic events (SHE), and severe hyperglycemia, with separate inputs for the NIIT and IIT subpopulations.

T2D IIT Population

The clinical effectiveness of RT-CGM systems versus SMBG for the IIT population was obtained from the KP retrospective cohort study [14]. A reduction in HbA1c of 0.56% was assumed for RT-CGM during the initial year of treatment based on the adjusted mean difference between RT-CGM and SMBG. As supported by longitudinal studies showing glycemic improvements for up to 10 years, this effect was sustained for an additional 2 years in the RT-CGM group [23–25]. In line with the CDM default clinical table, HbA1c began increasing by 0.15% per year after 3 years for RT-CGM, whereas for SMBG, this increase occurred after 1 year.

SHE were recorded as emergency room visits or hospitalizations and obtained from the Kaiser retrospective cohort study [14]. Based on the mean adjusted difference in the percentage of patients experiencing one or more hypoglycemic events during the 1-year follow-up, the model used annual probabilities of 0% SHE for RT-CGM and 4% SHE for SMBG.

Finally, severe hyperglycemic events were also recorded as emergency room visits or hospitalizations. Based on the mean adjusted difference in the percentage of patients experiencing one or more hyperglycemic events during the 1-year follow-up from the Kaiser retrospective cohort study, annual probabilities of severe hyperglycemia of 0% and 2.5% were applied in the RT-CGM and SMBG arms, respectively [14].

T2D NIIT Population

The clinical effectiveness of RT-CGM systems versus SMBG on HbA1c in the NIIT population was estimated using a random-effects meta-analysis, which combined the adjusted mean difference in HbA1c from two studies. One was the Steno2Tech study, a single-center, parallel, open-label randomized controlled trial comparing CGM with blood glucose monitoring in adults with inadequately controlled, insulin-treated T2D [15]. The other was the MOBILE trial, a randomized controlled trial assessing the effect of CGM on glycemic control in patients with T2D treated with basal insulin [13]. The pooled analysis showed a reduction in HbA1c of 0.60% (95% CI  − 1.08 to  − 0.12) in favor of RT-CGM. Based on longitudinal studies demonstrating glycemic improvements for up to 10 years, this effect was assumed to be sustained for 3 years in the RT-CGM group [23–25]. In line with the CDM default clinical table, HbA1c levels began increasing by 0.15% per year after 3 years for RT-CGM, whereas for SMBG, this increase occurred after 1 year.

Limited data on the frequency of hypoglycemic episodes in patients with T2D on NIIT insulin therapy were available. One observational cross-sectional study in the USA examined patient and physician perspectives on basal insulin titration in T2D [26]. Among patients who initiated basal insulin within 1 month of the titration period, 49% experienced at least one episode of hypoglycemia, including 19% who experienced severe hypoglycemia. While 33% reported a single episode, the remaining patients experienced two or more events. Additionally, a narrative review of treat-to-target randomized controlled trials comparing basal insulin analogues in T2D found that IIT patients had twice the risk of hypoglycemic events compared with NIIT patients [27]. Based on these studies, SHE rates for the NIIT population were assumed to be half of those reported for the IIT subpopulation in the KP retrospective study—0 and 2 events per 100 patient-years for RT-CGM and SMBG, respectively [14].

Cardiovascular Risk Equations

The Fremantle cardiovascular risk equation, developed using data from the Fremantle Diabetes Study cohort to reflect the Australian context and cultural background, demonstrated good calibration and discrimination [28]. As a result, it was considered an accurate tool for predicting CVD (both stroke and coronary heart disease) in Australian patients with T2D and used for both the IIT and NIIT populations.

Mortality Approach

Mortality in the model was estimated using the Western Australian equation, which is based on an administrative dataset containing hospitalization and death records for patients with T2D. The equation includes two age- and gender-specific risk models to estimate the risk of death following a major diabetes complication. These risk estimates were adapted by combining them with the UKPDS Outcomes Model, which accounts for the risk of death from causes unrelated to diabetes, providing a more comprehensive overall mortality estimate [29].

Quality of Life

Quality of life was partially captured through health state utilities and event-related disutilities, which were assigned to all possible disease states and acute events and were assumed to be the same across IIT and NIIT populations. The values were applied annually and accumulated over time. The health state utility for a person with T2D without complications was 0.785 [30].

A utility gain was incorporated to account for the indirect impacts of quality of life resulting from a reduction in hypoglycemic events, with different utility values assigned to the IIT and NIIT populations. NICE highlights severe hypoglycemic events as a key outcome in diabetes management, not only due to the immediate reduction in QoL for patients directly experiencing such events but also due to the broader reduction in QoL driven by the fear of future events [31]. This QoL impact extends beyond physiological distress, as patients often modify their insulin doses to mitigate the risk of hypoglycemia, which can further impair QoL. For example, a study conducted on patients with T2D reported that 29.9% experienced increased fear of future episodes following mild-to-moderate hypoglycemia, and 43.3% reported modifying their insulin doses [32].

A utility gain was also incorporated to capture the elimination of the need for daily and frequent finger-prick tests for RT-CGM due to its continuous glucose monitoring capability [33]. This utility benefit was applied additively to the utility benefit associated with reduced fear of hypoglycemia (FOH) and differed for the IIT and NIIT populations.

T2D IIT Quality of Life

FOH data for the IIT population were sourced from a secondary analysis of patients with T2D in the COACH Study, a prospective observational cohort study that assessed FOH using the Hypoglycemia Fear Survey (HFS) in patients withboth T1D and T2D on insulin therapy [34, 35]. This secondary analysis estimated the mean change in HFS from baseline (while patients were on SMBG) to the study end (after 6 months on RT-CGM) at 1.932, reflecting a reduction in the likelihood of severe hypoglycemia for patients using RT-CGM. Using findings from Currie et al. [36], where a 1-unit change in HFS corresponded to a 0.008-unit change in the EQ-5D index score, this translated to a calculated FOH utility gain of 0.0155 (1.932 × 0.008) for the RT-CGM group. No FOH utility benefit was assumed for patients in the SMBG arm. Patients using RT-CGM also experienced an additional 0.03 utility benefit due to the system eliminating the need for daily and frequent finger pricks [33].

T2D NIIT Quality of Life

Limited data were available to inform the utility impact of reducing FOH in patients on NIIT. Therefore, it was assumed that the utility benefit would be half that estimated for IIT patients—0.00775 for RT-CGM and 0 for SMBG. Similarly, due to limited data on the utility gain from avoiding finger-prick testing, it was assumed to be half of that for IIT—0.015 for RT-CGM and 0 for SMBG.

Costs

Annual intervention costs for RT-CGM were based on the current list price of Dexcom ONE+, which was AUD 2340, and included 36 sensors per year. For SMBG, daily testing assumptions were derived from clinical trials: 3.8 tests per day for the IIT population (based on the DIAMOND trial in T2D IIT), resulting in an annual cost of AUD 208, and 2.0 tests per day for the NIIT population (from the MOBILE trial in patients on basal insulin therapy), leading to an annual cost of AUD 110 [12, 13].

Costs of management, screening, and complications were divided into first-year and subsequent-year costs where applicable and inflated to 2024 values as needed (Supplementary Table 2). The percentage of patients requiring each type of management was stratified by primary and secondary prevention, with additional inputs for eye and renal disease screening (Supplementary Table 3). Types of management included angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, statins, and aspirin.

Model Outputs

Deterministic Results

The CDM reports LYs, QALYs, direct and indirect costs, cumulative incidence and time to onset of diabetes complications, and ICERs per additional LY or QALY gained. ICERs were assessed against a willingness-to-pay (WTP) threshold of 50,000 AUD [37]. In scenarios where RT-CGM was dominant over SMBG (i.e., more effective and less costly), the NMB was used to compare scenarios.

Sensitivity Analysis

A range of sensitivity analyses were conducted to explore the impact of varying key assumptions and inputs. These included adjustments to the FOH utility benefit, HbA1c effects, baseline HbA1c, and probabilities of severe hyperglycemia and SHE, as well as variations in SMBG testing frequency, RT-CGM price, time horizon, cohort characteristics, CVD risk equations, and discount rates.

Results

Base Case: IIT

For the IIT subpopulation, treatment with RT-CGM resulted in an increase of 0.567 QALYs and 0.098 LYs compared with SMBG, along with a total direct cost saving of AUD 9869 (Table 1). Accordingly, RT-CGM was found to be dominant compared with SMBG for the IIT population.

Table 1.

Base case cost-effectiveness results for RT-CGM versus SMBG in the IIT and NIIT populations

Outcomes	RT-CGM	SMBG	Incremental
IIT population
Costs, AUD	263,069	272,938	− 9869
QALYs	7.765	7.198	0.567
LYs	10.487	10.389	0.098
ICER (cost/QALY)	–	–	RT-CGM dominant
NIIT subpopulation
Costs, AUD	263,570	268,823	− 5253
QALYs	7.531	7.211	0.319
LYs	10.484	10.391	0.093
ICER (cost/QALY)	–	–	RT-CGM dominant

ICER incremental cost-effectiveness ratio, LYs life years, QALYs quality-adjusted life years, RT-CGM real-time continuous glucose monitoring, SMBG self-monitoring of blood glucose

Patients with IIT treated with RT-CGM experienced a reduced incidence of complications across all categories, primarily due to its sustained reduction in HbA1c (Table 2). Sizeable reductions were observed in the incidence of eye and renal disease. Specifically, the cumulative incidences of background diabetic retinopathy, proliferative diabetic retinopathy, macular edema, microalbuminuria, and gross proteinuria were reduced by 6.75, 5.11, 6.01, 6.77, and 6.54 percentage points, respectively, compared with SMBG. Key reductions were also observed in CV complications incidence, with decreases of 1.74, 2.88, 0.84, 1.03, 1.31, and 1.35 percentage points for congestive heart failure events, peripheral vascular disease onset, angina, diabetes-related mortality, stroke events, and myocardial infarction events, respectively, for RT-CGM compared with SMBG.

Table 2.

Incidence and relative risk of diabetes-related complications IIT and NIIT populations

Organ system	Complication	IIT			NIIT
		Cumulative incidence ± SE		Relative risk (95% CI) (RT-CGM vs SMBG)	Cumulative incidence ± SE		Relative risk (95% CI) (RT-CGM vs SMBG)
		RT-CGM	SMBG		RT-CGM	SMBG
Ophthalmic	Background diabetic retinopathy	32.49 ± 0.61	39.24 ± 0.65	0.83 (0.79–0.87)	31.95 ± 0.61	39.21 ± 0.65	0.81 (0.78–0.86)
	Proliferative diabetic retinopathy	15.43 ± 0.40	20.54 ± 0.47	0.75 (0.70–0.80)	15.25 ± 0.40	20.54 ± 0.47	0.74 (0.69–0.79)
	Macular edema	26.25 ± 0.53	32.26 ± 0.59	0.81 (0.77–0.86)	25.90 ± 0.53	32.25 ± 0.59	0.80 (0.76–0.85)
	Severe visual loss	22.62 ± 0.46	27.16 ± 0.53	0.83 (0.79–0.88)	22.44 ± 0.46	27.19 ± 0.53	0.83 (0.78–0.87)
	Cataract	12.12 ± 0.21	13.91 ± 0.24	0.87 (0.83–0.91)	12.00 ± 0.21	13.91 ± 0.24	0.86 (0.82–0.91)
Renal	Microalbuminuria	31.24 ± 0.61	38.01 ± 0.66	0.82 (0.78–0.87)	30.76 ± 0.61	38.17 ± 0.66	0.81 (0.76–0.85)
	Gross proteinuria	25.02 ± 0.56	31.56 ± 0.62	0.79 (0.75–0.84)	24.75 ± 0.56	31.59 ± 0.62	0.78 (0.74–0.83)
	End-stage renal disease	10.32 ± 0.35	14.76 ± 0.45	0.70 (0.64–0.77)	10.16 ± 0.35	14.76 ± 0.44	0.69 (0.63–0.76)
	Nephropathy death	0.00 ± 0.00	0.00 ± 0.00	0.00 (0.00–0.00)	0.00 ± 0.00	0.00 ± 0.00	0.00 (0.00–0.00)
Extremities	Ulcer	9.29 ± 0.42	10.40 ± 0.45	0.89 (0.79–1.01)	9.22 ± 0.41	10.40 ± 0.46	0.89 (0.78–1.00)
	Recurrent ulcer	18.98 ± 0.31	19.77 ± 0.33	0.96 (0.92–1.01)	18.96 ± 0.30	19.80 ± 0.33	0.96 (0.92–1.00)
	First amputation	6.48 ± 0.24	6.96 ± 0.26	0.93 (0.84–1.03)	6.40 ± 0.23	6.98 ± 0.26	0.92 (0.83–1.01)
	Second amputation	6.02 ± 0.17	6.48 ± 0.19	0.93 (0.86–1.01)	5.99 ± 0.16	6.55 ± 0.19	0.91 (0.85–0.99)
	Neuropathy	61.63 ± 0.70	67.97 ± 0.66	0.91 (0.88–0.93)	61.21 ± 0.71	67.95 ± 0.67	0.90 (0.87–0.93)
CV	Congestive heart failure event	43.85 ± 0.47	45.59 ± 0.47	0.96 (0.93–0.99)	44.05 ± 0.47	45.64 ± 0.48	0.97 (0.94–0.99)
	Peripheral vascular disease onset	20.64 ± 0.29	23.52 ± 0.31	0.88 (0.84–0.91)	20.39 ± 0.29	23.61 ± 0.32	0.86 (0.83–0.90)
	Angina	25.08 ± 0.36	25.92 ± 0.37	0.97 (0.93–1.01)	25.08 ± 0.37	25.91 ± 0.37	0.97 (0.93–1.01)
	Diabetes mortality	42.42 ± 0.22	43.45 ± 0.24	0.98 (0.96–0.99)	42.38 ± 0.23	43.44 ± 0.24	0.98 (0.96–0.99)
	Stroke event	29.18 ± 0.24	30.49 ± 0.25	0.96 (0.94–0.98)	29.04 ± 0.24	30.42 ± 0.25	0.95 (0.93–0.98)
	Myocardial infarction event	51.87 ± 0.27	53.22 ± 0.27	0.97 (0.96–0.99)	51.88 ± 0.28	53.26 ± 0.27	0.97 (0.96–0.99)

CI confidence interval, CV cardiovascular, IIT intensive insulin therapy, NIIT non-intensive insulin therapy, RT-CGM real-time continuous glucose monitoring, SMBG self-monitoring of blood glucose, SE standard error

Despite the higher glucose monitoring costs associated with RT-CGM as a result of its annual cost (AUD 2340) being substantially higher than that of SMBG (AUD 208), RT-CGM led to a reduction in total costs driven by reduced incidence of diabetes complications (Table 3). Reductions were observed across all categories of complications, including cardiovascular complications, renal complications, ulcer/amputation/neuropathy complications, ophthalmic complications, and severe hypoglycemia requiring medical assistance.

Table 3.

Breakdown of costs by outcome for the IIT and NIIT populations

Outcomes	Costs, AUD
	RT-CGM	SMBG	Incremental
IIT population
Glucose monitoring costs, AUD	25,709	2266	23,442
Management (preventative screening, medication) costs, AUD	3028	2983	45
CV complications, AUD	141,042	146,680	− 5638
Renal complications, AUD	48,494	62,795	− 14,301
Ulcer/amputation/neuropathy complications, AUD	12,640	13,883	− 1243
Ophthalmic complications, AUD	32,155	37,234	− 5078
Severe hypoglycemia (requiring medical assistance), AUD	0.00	3873	− 3873
Severe hyperglycemia, AUD	0.00	3224	− 3224
NIIT subpopulation
Glucose monitoring costs, AUD	25,702	1193	24,509
Management (preventative screening, medication) costs, AUD	3043	2985	58
CV complications, AUD	141,603	146,714	− 5111
Renal complications, AUD	48,169	62,816	− 14,647
Ulcer/amputation/neuropathy complications, AUD	12,678	13,875	− 1197
Ophthalmic complications, AUD	32,376	37,156	− 4780
Severe hypoglycemia (requiring medical assistance), AUD	0.00	1938	− 1938
Severe hyperglycemia, AUD	0.00	2147	− 2147

Values are shown in Australian dollars (AUD) for RT-CGM, SMBG, and the incremental difference (RT-CGM minus SMBG)

CV cardiovascular, RT-CGM real-time continuous glucose monitoring, SMBG self-monitoring of blood glucose

Base Case: NIIT

Similarly to the IIT population, RT-CGM was dominant when compared with SMBG for the NIIT population, with an increase in QALYs and LYs of 0.319 and 0.093, respectively, and a reduction in direct costs of AUD 5253 (Table 1).

The dominance of RT-CGM in the NIIT population resulted primarily from the sustained reduction in HbA1c for RT-CGM, reducing the incidence of T2D-related complications (Table 2). The most sizable reductions in complications were observed in background diabetic retinopathy and macular edema, with reductions in cumulative incidence of 7.26 and 6.35 percentage points, respectively, with RT-CGM compared to SMBG. Moreover, key reductions were observed across CV-related events, with a total reduction in incidence across all events of 9.46 percentage points for RT-CGM compared to SMBG. Again, in this population, the reduction in complication costs was enough to offset the higher treatment costs associated with RT-CGM, resulting in overall cost savings (Table 3).

Sensitivity Analyses

A range of sensitivity analyses were conducted to explore the impact of varying key assumptions and inputs, with the results demonstrating that the findings from the base case analysis of the cost-effectiveness of RT-CGM were mostly robust for both base case scenarios (Figs. 1, 2). The NMB was utilized to allow comparisons across dominant scenarios, with the NMB representing the monetary value of the health benefits gained, plus any associated cost savings; in this analysis, a higher NMB indicates increased dominance of RT-CGM.

Fig. 1.

Grouped one-way sensitivity analysis showing the effect of varying key model parameters on net monetary benefit (NMB) in the IIT population. Bars represent the NMB under each parameter scenario. AFS anxiety and fear of severe hypoglycemia, AUD Australian dollar, CGM continuous glucose monitoring, COACH COACH study (source for fear of hypoglycemia utility data) [34], DIAMOND DIAMOND study (source for HbA1c effect) [28], Hypo hypoglycemia event, FoH fear of hypoglycemia, HbA1c hemoglobin A1c (glycated hemoglobin), RT-CGM real-time continuous glucose monitoring, SHE severe hyperglycemic event, SMBG self-monitoring of blood glucose, UKPDS United Kingdom Prospective Diabetes Study [29, 38]

Fig. 2.

Grouped one-way sensitivity analysis showing the effect of varying key model parameters on net monetary benefit (NMB) in the NIIT population. Bars represent the NMB under each parameter scenario. AFS anxiety and fear of severe hypoglycemia, AUD Australian dollar, CGM continuous glucose monitoring, COACH COACH study (source for fear of hypoglycemia utility data) [34], FoH fear of hypoglycemia, HbA1c hemoglobin A1c (glycated hemoglobin), Hypo hypoglycemia event, Karter Karter et al. [14], MOBILE Mobile Trial (source for HbA1c effect) [13], RT-CGM real-time continuous glucose monitoring, SHE severe hyperglycemic event, SMBG self-monitoring of blood glucose, Steno2Tech Steno2Tech study (source for HbA1c effect) [15], UKPDS United Kingdom Prospective Diabetes Study [29, 38]

Sensitivity Analysis: IIT

Across sensitivity analyses, RT-CGM remained cost-effective for the IIT population, with dominant outcomes or ICERs below an AUD 50,000/QALY WTP threshold. When the time horizon was shortened, RT-CGM was cost-effective but not dominant, yielding ICERs of AUD 12,512 (10 years), AUD 21,045 (5 years), and AUD 26,571 (1 year) per QALY gained. Using the UKPDS 82 Cardiovascular Risk Prediction/UKPDS mortality equation also resulted in a cost-effective ICER of AUD 10,507 per QALY gained [29, 38].

The base-case NMB was AUD 38,219, with several scenarios notably impacting NMB and dominance. Altering baseline age showed that treating younger cohorts increased RT-CGM’s dominance; modeling a mean age of 35 years with a T2D duration of 1 year, a mean age of 45 years with a T2D duration of 5 years, and a mean age of 55 years with a T2D duration of 5 years increased the NMB to AUD 74,739, AUD 65,380, and AUD 52,139, respectively. Reducing the HbA1c treatment effect of Dexcom RT-CGM to 0.3 (from the DIAMOND T2D RCT) and 0.392 (a 30% reduction) lowered NMB to AUD 29,624 and AUD 32,538, respectively, while increasing it to 0.728 (30% increase) raised NMB to AUD 42,799 [28]. Thus, RT-CGM remained dominant, with dominance varying proportionally to treatment effect. Changes to FOH utility benefit also significantly impacted results. Removing the benefit reduced incremental QALYs by 0.156, lowering NMB to AUD 30,419. Increasing the FOH utility benefit to 0.02336 (per DIAMOND T2D trial) raised incremental QALYs by 0.099 and NMB to AUD 43,169. While racial distribution changes did not significantly affect RT-CGM dominance, notable disparities emerged across populations. Under conventional treatment, quality-adjusted life expectancy in the 100% Indigenous T2D population was 6.92 QALYs versus 7.77 QALYs in the whole Australian population, compared with 7.76 QALYs in a 100% white population, and 8.18 QALYs in a 100% Southern European population. Costs were also substantially higher for Indigenous populations by AUD 63,290, AUD 62,906, and AUD 88,851 versus whole Australian, 100% white and 100% Southern European populations, respectively.

Sensitivity Analysis: NIIT

RT-CGM was cost-effective or dominant in most NIIT scenarios. Only two scenarios—5- and 1-year time horizons—yielded ICERs above the AUD 50,000 WTP threshold (AUD 59,518 and AUD 71,862). Higher ICERs also occurred with shorter time horizons (10, 20 years), reduced HbA1c effect (0.4%), and use of UKPDS 82 equations, but all remained below the threshold, with all other scenarios remaining dominant. The base-case NMB was AUD 21,203, with the modeled HbA1c treatment effect driving relatively large changes in the NMB; increasing the HbA1c effect to 0.9% (Steno2Tech study) increased the NMB to AUD 29,554 [15]. Conversely, reducing the HbA1c benefit to 0.4% in line with the MOBILE Trial resulted in an NMB of AUD 15,046 [13].

Lower mean ages at baseline increased dominance: mean age of 35 years with a T2D duration of 1 year, a mean age of 45 years with a T2D duration of 5 years, and a mean age of 55 years with a T2D duration of 5 years increased NMB to AUD 50,938, AUD 39,800, and AUD 30,060, respectively.

Alterations to the FOH utility also had significant impacts on RT-CGM’s dominance. Removing the FOH utility decrement resulted in reduced dominance, with an NMB of AUD 17,353, while increasing the FOH utility benefit in line with the COACH Study led to increased dominance, with an NMB of AUD 24,853 [34].

Similar to the IIT population, disparities in health and cost outcomes were evident for the 100% Aboriginal and Torres Strait Islander population. Under conventional treatment, quality-adjusted life expectancy was 6.72 QALYs (Indigenous), 7.53 QALYs (whole Australian), 7.52 QALYs (100% white), and 7.93 QALYs (Southern European). Costs were AUD 64,403, AUD 64,176, and AUD 90,511 higher in Indigenous populations versus these respective populations.

Discussion

The analyses consistently showed RT-CGM to be cost-effective, with dominance observed in both IIT and NIIT populations. The sensitivity analysis demonstrated that these findings were mostly robust to variations in key inputs while highlighting the importance of assumptions related to age, time horizon, and clinical parameters. Therefore, this study concludes that RT-CGM represents a good use of healthcare resources in Australia.

The cost-effectiveness of RT-CGM was driven primarily by its favorable efficacy profile, modeled based on a sustained reduction in HbA1c compared with SMBG, and translating to reduced incidence of long-term complications. Many complications of T2D have a substantial detrimental effect on QoL and survival, making their reduction a key contributor to the increased QALYs modeled in patients using RT-CGM in this analysis [39, 40]. This relationship was highlighted in the sensitivity analysis for both the IIT and NIIT populations, where reductions in the modeled HbA1c treatment effect reduced the dominance of RT-CGM, while increases strengthened it.

In addition to the direct HbA1c effect of treatment, RT-CGM provided a utility benefit through the reduced FOH, with sensitivity analyses quantifying the impact of this benefit. Removing the FOH utility benefit was associated with a reduction in incremental QALYs gained from the base cases, resulting in reduced dominance. Conversely, increasing the utility benefit resulted in sizable increases in the incremental QALYs gained with RT-CGM.

As highlighted in the sensitivity analysis, considerable disparities were observed when modeling specific demographic groups. Specifically, both for patients with T2D on IIT and NIIT, when modeling a 100% Indigenous population, costs on conventional treatment were substantially higher, and QALYs substantially lower, compared with the whole Australian population (base case), 100% white population, and 100% European population. While there was a limited difference in the dominance of RT-CGM across these populations, these cost and QALY disparities represent the substantial value and need of RT-CGM for Indigenous populations with T2D. Further to this, evidence also suggests that there is a disproportionate burden of T2D complications in Aboriginal and Torres Strait Islander peoples, including CVD, nephropathy, end-stage renal disease, retinopathy, neuropathy and lower extremity amputations, in addition to higher rates of T2D-related hospitalizations and mortality [41]. This highlights the heightened need for a system better equipped to maintain glycemic control within the Indigenous T2D Australian population compared with the general T2D population in Australia.

The prevalence of diabetes among Indigenous Australians is significantly higher than in non-Indigenous Australians. According to the 2018–19 National Aboriginal and Torres Strait Islander Health Survey, around 8% of Indigenous Australians have diabetes, rising to 12% in remote areas [42]. In comparison, the national prevalence was 5.3% in 2022 [43]. Although not stratified by diabetes type, obesity—a major risk factor for T2D—is notably more common in Indigenous communities, with adults 1.5 times more likely to be obese [44]. These disparities underscore the need for targeted interventions like RT-CGM, especially in remote areas where the burden is greatest. However, access to CGM technology in Australia has historically been limited to specific groups, with the National Diabetes Services Scheme subsidies currently focused on people with type 1 diabetes [45]. Diabetes Australia and others have advocated for expanded funding to include insulin-treated T2D and Indigenous populations, recognizing the significant equity gap [46].

In addition to variation across demographic groups, younger adults with T2D represent another high-need population [47]. The prevalence of T2D among younger adults is rising sharply, particularly among those from minority ethnic backgrounds and socially deprived areas [48, 49]. Youth-onset T2D is associated with a more aggressive disease course, including greater insulin resistance, more rapid β cell decline, and earlier and more severe complications compared with late-onset T2D [48, 49]. These factors make T2D in younger adults harder to manage and lead to markedly worse long-term outcomes, including higher rates of morbidity and mortality [48, 49]. Consistent with this, the present analysis showed that the value of RT-CGM was greater when treating younger adults in both IIT and NIIT subgroups, with substantially higher NMB observed in lower mean age groups. These findings reinforce the potential importance of improving glycemic control through RT-CGM for this challenging and growing population.

However, the analysis presented does have some limitations. Firstly, the findings are specific to the modeled population, which limits their generalizability to alternative populations with different patient characteristics. Related to this, while baseline characteristics were sourced from a US cohort (Kaiser Permanente) and adjusted to reflect Australian demographics, residual differences in comorbidity profiles, treatment adherence, and patterns of healthcare utilization between the two settings should be considered when interpreting the findings in the Australian context. The Australian census also lacks the granularity needed to provide the demographic data required in the IQVIA CDM model. Specifically, the data records ancestry but does not include details on race or ethnicity. Since the CDM requires the user to populate multiple ethnicity categories, which intersect with disease epidemiology, health equity, and healthcare access, this presents a limitation of the analysis.

In addition, some sources used in the analysis were several years old. While costs were adjusted for inflation, they may not account for newer, potentially more expensive treatments or diagnostic procedures introduced since the study. The model assumed Dexcom ONE+ as the CGM device, although other CGMs may incur different daily costs; this variation was addressed through sensitivity analyses.

Furthermore, although longitudinal studies support sustained improvements in glycemic control with CGM for up to 10 years, the model conservatively assumed persistence of HbA1c benefit for only 3 years. Nonetheless, in real-world settings, variation in adherence and potential discontinuation of CGM use could attenuate the observed benefit over time. Likewise, while the assumption of no SHEs for RT-CGM users was informed by the Kaiser retrospective cohort study based on the adjusted difference between RT-CGM and SMBG, it is unlikely that the true risk is zero in broader clinical practice, and this may therefore overstate reductions in SHEs [14].

An additional limitation was the relative lack of clinical trial and real-world data for people with T2D using NIIT. Although the analysis was based on the best available evidence, the limited data in this subgroup reduces certainty around the projected cost-effectiveness of RT-CGM compared with the more robust evidence base for IIT.

Moreover, the analysis compared RT-CGM with SMBG only, without including intermittently scanned continuous glucose monitoring (IS-CGM, also known as Flash) as a comparator. This decision reflected the primary reimbursement question in Australia, where SMBG remains the standard of care for many people with insulin-treated T2D, as well as the limited availability of robust long-term data on IS-CGM in this population. Nevertheless, comparing RT-CGM with IS-CGM is an important area for future research.

Finally, as with all health economic models, outcomes are projections based on assumptions and data inputs from external sources rather than trial-based endpoints, and results should therefore be interpreted as estimates rather than direct clinical observations.

The results of the analysis align with previous studies that have demonstrated the cost-effectiveness of RT-CGM systems versus SMBG in patients with T2D, such as those conducted in the UK and Spain [50, 51]. However, this study expands on this evidence by exploring an Australian population, where the cost-effectiveness of Dexcom RT-CGM systems have been less explored. Akin to this analysis, key drivers across the previous studies included the reduction in complications due to the reduction in HbA1c, time horizon, and the utility benefits associated with FOH and finger pricks.

Conclusions

The present analysis found RT-CGM to be dominant compared with SMBG for patients with T2D on either IIT and NIIT. From a policy perspective, the findings suggest that integrating RT-CGM into the treatment pathway could lead to improved patient outcomes while promoting reductions in diabetes costs for the Australian healthcare system, representing an efficient allocation of healthcare resources. The analysis also underscores the need for improved glycemic management among Indigenous Australians, given the higher prevalence of T2D and related complications in this population—particularly in remote areas—and among younger individuals in whom the prevalence of T2D is high and the disease trajectory is less favorable.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Dr Hamza Alshannaq contributed to supervision, conceptualization, data curation, methodology, and formal analysis. Prof David Simmons contributed to manuscript review and editing. Ms Jessica Matuoka and Mr Greg Norman contributed to data curation, methodology, formal analysis, and manuscript review and editing. Mr Matthew Tucker and Mr Richard Pollock-Wilkins contributed to data visualization, original draft writing, and manuscript review and editing. Mr Moin Ahmed contributed to manuscript review and editing. All authors approved the final version of the manuscript.

Funding

This work and the Rapid Service and Open Access Fees were supported by Dexcom.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflicts of Interest

Jessica Matuoka and Greg Norman are employees of Dexcom and hold stock in Dexcom. Hamza Alshannaq is an employee of Neurocrine Biosciences and holds stock in Dexcom. Moin Uddin Ahmed declares no conflicts of interest. David Simmons is a Distinguished Professor at the Western Sydney University School of Medicine, Deputy Head of Department at the Macarthur Diabetes Endocrinology and Metabolism Services (MDEMS), and Chief Medical Officer of Diabetes Australia. David Simmons reports receipt of education grants from Abbott, Ascensia, and Boehringer Ingelheim; devices for research and clinical service development from Dexcom and Abbott; speaker fees from Novo Nordisk and Sanofi; and research support from Hitachi. Richard Pollock and Matthew Tucker are full-time employees of Covalence Research Ltd, which received consultancy fees from Dexcom to prepare the manuscript.

Ethical Approval

As an economic modelling analysis based on published literature, the present study did not enroll human participants or make use of human biological samples or individually identifiable data, and ethics approval was therefore not required.