Cost–Utility Analysis of FreeStyle Libre Systems in People with Type 2 Diabetes Mellitus on Treatment with Basal Insulin and Poor Glycemic Control in Spain

Abstract

Introduction

FreeStyle Libre (FSL) systems are effective and user-friendly glucose monitoring devices. This cost-effectiveness analysis compared FSL vs. self-blood glucose monitoring (SBGM) in patients with poorly controlled [hemoglobin A1c (HbA1c) > 8%] type 2 diabetes (T2DM) on basal insulin, from the Spanish National Health System perspective.

Methods

The DEDUCE model, which simulated 10,000 patients with T2DM over a 50 years’ time horizon (annual discount rate = 3.00%), was adapted to the Spanish setting. The population characteristics, frequency of acute and chronic diabetic complications, costs (€, 2025) and utilities/disutilities proceeded from scientific literature and were validated by national multidisciplinary experts. The annual probabilities of acute events associated with SBGM were 17.02% for non-severe hypoglycemia (SHE) (€3.92; disutility = – 0.0016), 2.50% for SHE (€1031.69; disutility = – 0.0470) and 0.25% for ketoacidosis (DKA) (€2523.93; disutility = – 0.0470). The RECODe risk engine was used to model chronic diabetic complications (myocardial infarction [€1248.44–€31,013.22; disutility = – 0.0550]; heart failure [€1523.14–6505.08; disutility = – 0.1080]; stroke [€3187.92–7849.48; disutility = – 0.1640]; blindness [€2943.37; disutility = – 0.0740]; renal failure [€4057.05–42,757.39; disutility = – 0.2040]). According to the Spanish recommendations, a patient with SBGM required 2.5 reactive strips/day and 2.5 lancets/day (€0.57/strip; €0.14/lancet; VAT included). FSL (26 sensors/year; €3.00/day; VAT included) was associated with reductions of 58% in hypoglycemia, 68% in DKA, 83% in the use of strips/lancets, and an absolute decrease of 1.1% in HbA1c. Deterministic and probabilistic sensitivity analyses (SAs) were conducted.

Results

While SBGM yielded 9.18 quality-adjusted life years (QALYs) and total costs of €77,092 (glucose monitoring = €17,080; diabetic complications = €68,272), FSL yielded 9.98 QALYs and total costs of €61,447 (glucose monitoring = €8820; diabetic complications = €44,367). Compared with SBGM, FSL produced total cost savings of €15,645 and 0.80 additional QALYs per patient, being a dominant alternative compared to SBGM. FSL was found to be dominant in all SAs.

Conclusions

This analysis suggests that FSL, which provides better clinical outcomes at a lower overall cost, is a preferable alternative to SBGM among people with poorly controlled T2DM on basal insulin.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13300-025-01816-6.

Key Summary Points

Why carry out this study?

People with poorly controlled (hemoglobin A1c (HbA1c) > 8%) type 2 diabetes mellitus (T2DM) on basal insulin represent a prevalent population leading to remarkable costs to the Spanish National Health System (NHS).

Compared with self-blood glucose monitoring (SBGM), the FreeStyle Libre (FSL) systems are an effective and convenient alternative for optimizing glucose monitoring in this population.

This study aimed to provide economic evidence from the perspective of the Spanish NHS regarding the cost-effectiveness of FSL systems compared with SBGM in people with poorly controlled T2DM on basal insulin.

What has been learned from the study?

In patients with poorly controlled T2DM on basal insulin in Spain, FSL compared with SBGM yielded more quality-adjusted life years (+ 0.80) and cost savings (€15,645 per patient’s lifetime) through reductions in the management of acute (€19,837) and chronic diabetic complications (€4069).

FSL represents a cost-effective or dominant alternative for glucose monitoring in patients with poorly controlled (HbA1c > 8%) T2DM on basal insulin, for a time horizon encompassing the whole patient’s lifetime.

This study provides key evidence to clinicians and supports informed decision-making by healthcare institutions.

Introduction

Improving the glycemic control of people living with type 2 diabetes mellitus (T2DM) remains as the main target in clinical practice, as achieving optimal blood glucose levels reduces the occurrence of acute events in the short term, including hypoglycemia and diabetic ketoacidosis (DKA) [1]. Among others, glycemic control is one of the important factors connected to the risk of suffering chronic diabetic micro and macrovascular complications in the long term [2, 3].

Self-blood glucose monitoring (SBGM) represents the traditional monitoring method used to measure capillary glucose levels. However, recent international clinical practice guidelines recommend the use of continuous glucose monitoring (CGM) systems [4]. These innovative devices address most of the limitations associated with SBGM [5], including pain, discomfort and low adherence to the recommendation of scientific societies regarding the frequency of glucose level measurements [6].

Among the wide variety of CGM devices, while real-time CGM (rt-CGM) transmit glucose measurements automatically, on-demand measuring systems, also called intermittently scanned or flash monitoring devices, provide on-demand measures of blood glucose levels [7]. Specifically, FreeStyle Libre systems (FSL) (Abbott Diabetes Care), an interstitial fluid glucose monitoring system available as flash glucose monitoring or rt-CGM, has proved to be effective in several populations with type 1 diabetes mellitus (T1DM) [8, 9] and T2DM [10–12]. Recently, the RELIEF study provided real-world data (RWD) on the reductions in both severe hypoglycemic events (SHEs) and DKAs, as well as improvements in targeting adequate blood glucose levels measured by HbA1c, in patients with T2DM receiving basal insulin [12].

This real-world evidence (RWE) study [12] led to the reimbursement of FSL for patients with T2DM in several regions worldwide [13–15]. In Spain, reimbursement for these devices is available for patients with T1DM [16] and patients with T2DM receiving multiple doses of insulin (MDI), including basal-bolus regimens [17]. In addition, flash glucose monitoring with FSL represents the unique alternative that is reimbursed for pediatric patients [18], and for insulin-dependent patients that are different from those with T1DM or T2DM (e.g., monogenic diabetes, cystic fibrosis, hemochromatosis, etc.) [19]. However, flash glucose monitoring systems are still not reimbursed for those patients with T2DM receiving basal insulin.

Among the population with T2DM on basal insulin, those patients exhibiting poor glycemic control represent a priority group for several reasons. First, the number of potential patients reveals the epidemiological burden in Spain, where it is estimated that 7.8% of the population suffers from T2DM [20], which is higher than the prevalence estimated in other European countries [21]. Among these patients, a total of 81.6% receive pharmaceutical treatment, with 21.3% on insulin-based regimens [22], and approximately 61.8% of those regimens consisting of basal insulin administration [23]. In addition to the volume of patients concerning this subgroup, a study conducted in 2022–2023 highlighted its economic burden by estimating that a patient with T2DM produced a public expenditure of €5171 due to hospital admissions (€2228), treatments (€1703) and assistance and complementary tests (€1240) [24]. Finally, the increased risk of comorbidities [2, 3] and health-related quality of life (HRQoL) deterioration [25] due to poor glycemic control in this subgroup reveals the unmet clinical needs of patients with poorly controlled T2DM receiving basal insulin.

In this context, FSL could be a suitable alternative for those patients, which has also been shown to produce cost savings in other populations in Spain, including people with T1DM [26], and T2DM receiving both MDI [27] and basal insulin [28, 29]. However, it is still uncertain whether the reimbursement of FSL in this population is a cost-effective strategy in Spain. Thus, this study aimed to conduct a cost–utility analysis (CUA) of FSL compared with SBGM in adults with T2DM receiving basal insulin therapy and exhibiting poor glycemic control, from the perspective of the Spanish National Health System.

Methods

Model Description

The DEDUCE (Determination of Diabetes Utilities, Costs and Effects), a recently developed and validated patient-level microsimulation model [30], was adapted to the Spanish setting for this study.

An in-depth description of the model design and workflow is available elsewhere [30]. Briefly, the model first creates a hypothetical patient and establishes baseline demographics and clinical characteristics. These features are randomly assigned based on probability distributions determined by either the mean and standard deviation (SD) for quantitative variables or the relative frequency in the case of qualitative variables. This patient is subsequently assigned to each intervention before the simulation begins. In 1-year cycles, the model determines whether a patient suffers any diabetic complications, including acute events such as SHE and DKA; and chronic diabetic complications such as myocardial infarction (MI), congestive heart failure (HF), stroke, blindness, and renal failure. The frequency of acute events is captured by their respective reported rates, which vary between interventions. In contrast, chronic complications were modeled via the RECODe risk engine [31], which yields the risk of these complications and mortality according to the baseline patient characteristics and to the levels of glycated HbA1c. Each acute or chronic diabetic complication is associated with management costs and a health disutility value, both of which are updated in each model cycle to account for the cost derived from the management of the patient and the impact on patient HRQoL caused by the disease. The model simulates the patient’s evolution until the time horizon is reached, or the patient dies; then, the process is repeated with a new hypothetical patient. The overall results are estimated as a mean value of the health results and costs achieved by all 10,000 patients in the hypothetical cohort.

The parameters considered in the analysis were agreed upon by a multidisciplinary board including experts specializing in Endocrinology, Internal Medicine, Primary Care and Health Economics. All the evidence available was gathered via a literature review and presented to the multidisciplinary board in a structured survey, which was fulfilled individually by the clinicians. Their answers were revised by the whole group in a face-to-face meeting where the most adequate sources and values were validated by consensus according to the clinical practice in Spain.

This manuscript was conceptualized on the basis of the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist published by the International Society for Pharmacoeconomics & Outcomes Research (ISPOR) in 2022 (Supplementary Table 1) [32]. This CUA is based on previously conducted studies and does not contain any new studies with human participants or animals.

Study Population

The target population of this CUA consisted of patients with T2DM who were receiving basal insulin and exhibited poor glycemic control. Table 1 summarizes all the baseline demographic and clinical characteristics of these patients [33–38].

Table 1.

Baseline demographic and clinical characteristics of the target population

Baseline characteristic	Mean value	SD
Age (years) [33]	62.9	12.8
Percentage of women (%) [33]	41.0%	–
Baseline level of glycated HbA1c (%) [34]	9.2%	1.0%
SBP (mmHg) [33]	139.9	28.4
HDL cholesterol (mg/dl) [33]	46.8	9.6
Total cholesterol (mg/dl) [33]	201.8	41.2
Serum creatinine (mg/dl) [35]	0.86	0.2
UACR (mg/g) [36]	99.2	359.4
Percentage of active smokers (%) [33]	27.8%	–
Percentage of patients with CVD (%) [36]	35.6%	–
Percentage of patients treated with statins (%) [33]	36.6%	–
Percentage of patients treated with antihypertensives (%) [33]	44.3%	–
Percentage of patients treated with anticoagulants (%) [35]	39.5%	–
Percentage of patients treated with oral antidiabetics (%) [37]	84.4%	–
Baseline health utility value (HRQoL) [38]	0.7840	–

CVD cardiovascular disease, HDL high-density lipoprotein, HRQoL health-related quality of life, SBP systolic blood pressure, SD standard deviation, UACR urinary albumin–creatinine ratio

The experts involved in this study suggested defining poor glycemic control as HbA1c levels greater than 8%. Accordingly, the baseline HbA1c level (9.2% ± 1.0%) was selected based on expert criteria [34].

Clinical Inputs

Both acute and chronic diabetic complications were considered in the analysis. In terms of acute events, the occurrence of non-severe hypoglycemic events (NSHEs), SHEs and/or DKA were determined by the annual probability of suffering each event, and the events/person-year, both of which were derived from scientific literature.

According to the clinicians’ criteria, the most representative source to apply to Spanish clinical practice in relation to the frequency of mild–moderate and severe hypoglycemia in patients receiving SBGM was the Hypoglycemia Assessment Tool (HAT) [39]. Up to 24 countries participated in this multicentric study, which included 27,585 patients, 19,563 of whom presented with T2DM and received insulin therapy. According to the HAT study, 8.9% of patients suffered SHEs annually, with a mean of 2.5 events/person-year [39]. In relation to the NSHE, the model was adapted considering that all patients suffer these events; thus, the probability was 100% and the annual frequency was 17.04 events/person-year as observed in the HAT study [39].

The occurrence of DKA in those patients with SBGM was derived from a retrospective study conducted in Andalusia (Spain) [40], which gathered 2484 medical records of patients with diabetes suffering hyperglycemic episodes defined by the International Classification of Disease (ICD) codes. Although data on the hyperosmolar hyperglycemic state (ICD-9: 250.2) were available, the rate of DKA (ICD-9: 250.1) was exclusively selected (2.5 events/1000 people-year) for this economic evaluation [40]. Given that the data were adjusted to all the participants in the study, an annual probability of 0.25% [40] and 1 event/person-year were considered for the analysis.

The frequency of these acute events in patients receiving FSL was determined by a reduction in the occurrence of both hypoglycemia and DKA events, associated with the use of FSL. The reduction associated with the use of FSL proceeded from the RELIEF [12], a RWE study conducted in France, which included 5933 patients with T2DM on basal insulin [12]. After 12 months of replacing SBGM with FSL, both hypoglycemic events ( – 58.0%) and DKA ( – 68.0%) were reduced [12].

To account for the excess risk of death specific to those acute events, mortality rates were derived from the scientific literature and were validated by clinicians. The mortality rate associated with SHE (0.32%) proceeded from the PAUEPAD project [41], a retrospective observational study using data from the 1,137,738 emergency calls to the Public Company of Health Emergencies in Andalusia (Spain). Among those emergency calls, 8.683 had a primary diagnosis of hypoglycemia [41]. The DKA mortality rate (0.40%) was obtained from the previously described retrospective study conducted by using medical records in Andalusia (Spain) [40].

The risk of chronic diabetic complications and death was determined by the RECODe motor [31], which relies on data from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study [36, 42–44], and was subsequently validated with data from the Diabetes Prevention Program Outcomes Study (DPPOS) [45] and Action for Health in Diabetes (Look AHEAD) [46]. This risk engine considers the baseline demographic and clinical characteristics of patients, and their levels of glycated HbA1c (Supplementary Table 2) [31]. Except for age and HbA1c levels, the remaining variables were modeled with a constant effect on the risk of chronic diabetic complications defined by their respective baseline values. In contrast, the impact of age on the risk of suffering chronic diabetic complications was modeled every 10 years. Finally, the average HbA1c level at baseline was established for patients with SBGM (9.2% ± 1.0%) [34]. A one-time absolute reduction of 1.1% [34] in HbA1c was applied to the patients receiving FSL. Both parameters were derived from an RWE study including the clinical records of 191 patients from the United States and Canada, who were diagnosed with T2DM and received a basal insulin regimen [34].

Health Care Resource Consumption and Costs

This CUA was conducted from the Spanish National Health System perspective, and included the cost of direct health resource consumption for glucose monitoring (sensors, reactive strips and lancets), and the cost associated with the management of acute and chronic events.

The consumption of strips and lancets in those patients receiving SBGM was determined by the Spanish Diabetes Society (Sociedad Española de Diabetes [SED]) guidelines [47], which recommend daily consumption of 2.5 strips and 2.5 lancets for those patients with poorly controlled T2DM [47]. In order to estimate the daily consumption of strips and lancets per patient receiving FSL, a reduction of 83% [12] was considered based on the findings of the RELIEF study [12]. The unitary cost of strips (€0.57/unit) [48] and lancets (€0.14/unit) [48] was derived from an economic evaluation of flash glucose monitoring systems in patients with T1DM, conducted and published by a Health Technology Assessment (HTA) Agency in Spain [48]. The daily cost of FSL was €3.00. VAT was applied to all costs associated with those resources used to monitor blood glucose levels.

The costs per acute event were estimated based on the consumption required for their management, including self-management, specialists’ consultations, and hospital assistance with or without admission. With respect to NSHE, the unitary cost associated with hypoglycemia requiring management by a specialist (€17.89) [49] was applied to 21.9% of the cases [50]. The remaining 78.1% were considered to be self-managed, and no cost was applied to those events. The proportion of NSHE requiring specialist care was derived from a national study that enrolled 630 patients with both T1DM and T2DM aged 15 years or older [50]. Additionally, SHEs were first divided according to whether the event required hospital assistance (26.1%) [51] or not (73.9%) [51]. The cost associated with those events not requiring hospital assistance was €461.48 [52]. Among the SHEs requiring assistance, 21.7% [51] required hospital admission (€4512.09 [53]) and the remaining 78.3% did not require admission (€2129.08 [41]). The distribution of SHEs across treatment pathways was derived from a cross-sectional survey in Spain and seven other countries, which included 193 and 184 participants with T1DM and T2DM, respectively [51]. Finally, DKA included those events requiring hospital assistance in 87.1% [40] of the cases (€2897.74 [53]), and the remaining 12.9% of cases were considered self-assisted and, therefore, did not incur additional costs. The distribution was also reported in the regional study previously depicted [40].

In order to determine the most homogeneous costs associated with the management of chronic diabetic complications, the costs and sources [54–56] were derived from the evaluation of CGM devices in patients with T2DM, published by the HTA Agency in Canarias (Spain) [38], whose evaluations represent the main evidence that supports decision-making by the Spanish Ministry of Health. This approach was validated by experts and allowed the alignment of the present CUA with similar published economic evaluations conducted by the Canarian HTA Agency [38].

All costs were included in euros (€), and as stated by national recommendations for the development of economic evaluations [57], the VAT was included when applicable. All costs were inflated to 2025 when applicable, according to variations in the consumer price index (CPI) [58].

Health Utility/Disutility Values

To model the impact of the disease on HRQoL and estimate quality-adjusted life years (QALYs), a baseline utility of 0.7840 [38] was set for each patient with T2DM. Every event had an associated disutility (Table 2) [38, 59, 60], which was applied to the baseline utility to model the reduction in HRQoL in each model cycle, being additive when various health events were present in the same cycle. The general population utility and disutility values associated with chronic diabetic complications were derived from the previously mentioned economic evaluation of CGM devices in people with T2DM conducted by a Spanish HTA Agency [38]. The disutility values of the acute events were derived from other independent CUA involving patients with T2DM in Spain [59, 60].

Table 2.

Clinical inputs, costs, and health disutility values associated with health events

Model input	SBGM	FSL	Cost per event (€, 2025)		Disutility (first and subsequent years)
	Annual probability (%) (events per patient-year)	Reduction of the frequency (%)	First year	Subsequent years
NSHE	100% (17.02) [39]	– 58.0% [12]	€3.92a	€3.92b	– 0.0016 [59]
SHE	8.9% (2.5) [39]	– 58.0% [12]	€1031.69b	€1031.69a	– 0.0470 [60]
DKA	0.25% (1.0) [40]	– 68.0% [12]	€2523.93c	€2523.93c	– 0.0470 [60]
MI	RECODe Risk Engine [31] (baseline HbA1c: 9.2% [34])	RECODe Risk Engine [31] (reduction HbA1c: – 1.1% [34])	€31,013.22 [38, 54]	€1248.44 [38, 54]	– 0.0550 [38]
CHF	RECODe Risk Engine [31] (baseline HbA1c: 9.2% [34])	RECODe Risk Engine [31] (reduction HbA1c: – 1.1% [34])	€6505.08 [38, 55]	€1523.14 [38, 55]	– 0.1080 [38]
Stroke	RECODe Risk Engine [31] (baseline HbA1c: 9.2% [34])	RECODe Risk Engine [31] (reduction HbA1c: – 1.1% [34])	€7849.48 [38, 56]	€3187.92 [38, 56]	– 0.1640 [38]
Blindness	RECODe Risk Engine [31] (baseline HbA1c: 9.2% [34])	RECODe Risk Engine [31] (reduction HbA1c: – 1.1% [34])	€2943.37 [38]	€2943.37 [38]	– 0.0740 [38]
Renal failure	RECODe Risk Engine [31] (baseline HbA1c: 9.2% [34])	RECODe Risk Engine [31] (reduction HbA1c: – 1.1% [34])	€4057.05 [38, 54]	€42,757.39 [38, 54]	– 0.2040 [38]

CHF congestive heart failure, DKA diabetic ketoacidosis, FSL FreeStyle Libre systems, MI myocardial infarction, NSHE non-severe hypoglycemic event, SBGM self-blood glucose monitoring, SHE severe hypoglycemic event

^aCost estimated as a weighted average of specialist-managed event (21.9% [50]; €17.89 [49]) and patient-managed event (complementary data: 78.10%; €0.00)

^bCost estimated as a weighted average of SHE that requires hospital assistance (26.10% [51]; hospital assistance with admission [21.7% [51]; €4512.09 [53], and without admission [complementary data: 78.3%; €2129.08 [41]) and not requiring hospital assistance (complementary data: 78.1%; €461.48 [52])

^cCost estimated as a weighted average of DKA requiring hospital assistance (87.1% [40]; €2897.74 [53]) and not requiring hospital assistance (complementary data: 12.9%; €0.00)

Finally, an additional disutility of – 0.0310 [61] was included to capture the impact of finger sticks. This input was sourced from a study that aimed to estimate health utility values associated with glucose monitoring devices [61].

Base-Case and Sensitivity Analyses

In this analysis, the model simulated a cohort of 10,000 patients during a lifetime horizon (50 years). Both half-cycle correction and a discount rate of 3.00% for costs and health outcomes were applied according to the national guidelines [57]. Given that the base case consisted of a first-order Monte Carlo simulation, the model used the mean values of all inputs, except for the baseline demographic and clinical characteristics. For the quantitative parameters, values were assigned based on the normal distribution defined by their mean and SD. Those qualitative inputs were determined by a random number generated by the model, which ranged from 0 to 1. Baseline characteristics were present in the hypothetical patient when the random number was lower than the proportion established for each input. The model estimated life years gained (LYG), QALYs and costs per patient. The incremental cost-effectiveness ratio (ICER) was subsequently estimated by comparing FSL versus SBGM.

In order to test the robustness of the model and results, both alternative scenarios (AS) and probabilistic sensitivity analyses (PSA) were conducted.

Concerning the AS, the key model inputs informed by the clinicians consulted were modified based on alternative inputs observed in other studies. These key parameters included the following: AS1) annual frequency of SHEs (1 event/person-year [conservative assumption]); AS2) annual SHE probability and frequency (14.8% and 1.4 events/person-year, respectively) [62]; AS3) reduction in both mild–moderate and severe hypoglycemia associated with the use of FSL (29.0%) [11]; AS4) reduction in DKA associated with the use of FSL (52.1%) [23]; AS5) unitary cost per strip and lancet (€0.00 [conservative assumption]); AS6) cost per NSHE (€64.91) [48]; AS7) cost per NSHE (€0.00 [conservative assumption]); and AS8) reduction in HbA1c levels associated with FSL (0.6%) [63].

The PSA consisted of 1000 second order Monte Carlo simulations of the whole cohort (N ~ 10,000 hypothetical patients), which assigned a value for each model input according to several probability distributions (normal, log-normal, beta, and gamma). The mean values and their respective SDs were used to define the parameters of these probability distributions, except for those inputs lacking a dispersion measure in the original publication, for which a SD of ± 20% from the mean value was considered (Supplementary Table 3). The 1000 simulations were presented in a scatterplot and averaged to determine the result of the PSA. The standard error (SE) of the average results was also estimated.

Results

Base-Case Results

In the base case, the model estimated that a patient on basal insulin and with poorly controlled T2DM lived 9.18 QALYs. Compared with patients receiving FSL, who lived 9.98 QALYs, the device produced 0.80 additional QALYs per patient (Fig. 1).

Fig. 1.

Visual summary of the study methods and results. Cho cholesterol, DEDUCE Determination of Diabetes Utilities, Costs and Effects, DKA diabetic ketoacidosis, FSL FreeStyle Libre systems, HbA1c hemoglobin A1c, HRQoL health-related quality of life, NSHE non-severe hypoglycemic event, QALY quality-adjusted life year, SBP systolic blood pressure, SBGM self-blood glucose monitoring, SHE severe hypoglycemic event, T2DM type 2 diabetes mellitus

Additionally, the total costs for a lifetime horizon were €77,092 per patient receiving SBGM and €61,447 per patient receiving FSL. Thus, the device also produced overall cost savings of almost €15,645 per patient’s lifetime. Although FSL increased the costs per patient associated with glucose monitoring by €8260 compared with SBGM, FSL reduced the costs associated with the management of all acute events (€19,837) and chronic diabetic complications (€4069).

In summary, FSL represented a dominant intervention compared with SBGM, as the devices increased the health benefits in terms of QALYs and reduced the overall costs (Table 3).

Table 3.

Base-case results

Base-case model output	Results per patient (lifetime)
	SBGM	FSL	Incremental (FSL vs. SBGM)
Health results
LYG	13.26	13.86	+ 0.60
QALYs	9.18	9.98	+ 0.80
Total costs (€, 2025)	€77,092	€61,447	– €15,645
Glucose monitoring costsa	€8,820	€17,080	+ €8,260
Cost of management of acute events
NSHE	€865	€380	– €485
SHE	€33,459	€14,165	– €19,294
DKA	€86	€28	– €58
Cost of management of chronic diabetic complications
MI	€15,951	€14,057	– €1894
CHF	€4109	€3669	– €441
Stroke	€4103	€3148	– €956
Blindness	€4372	€4110	– €262
Renal failure	€5328	€4811	– €516
ICER (€/QALY)			Dominantb (– 19,504 €/QALY)

CHF congestive heart failure, DKA diabetic ketoacidosis, FSL FreeStyle Libre systems, ICER incremental cost-effectiveness ratio, LYG life year gained, MI myocardial infarction, NSHE non-severe hypoglycemic event, QALY quality-adjusted life year, SBGM self-blood glucose monitoring, SHE severe hypoglycemic event

^aThe glucose monitoring costs include the costs associated with the acquisition of reactive strips, lancets and sensors (in the case of patients receiving FreeStyle Libre)

^bFreeStyle Libre is dominant compared to SBGM: increases QALYs (+ 0.80) and produce cost savings (– €14,797). For that reason, the estimated ICUR is negative

Alternative Scenarios Results

Compared with SBGM, FSL was found to be dominant in all AS (Table 4).

Table 4.

Results of the alternative scenarios

Scenarios	Parameters		Results per patient (lifetime)
	Base-case value	Sensitivity analysis value	Δ Costs (FSL vs. SBGM)	Δ QALYs (FSL vs. SBGM)	ICER (€/QALY) (FSL vs. SBGM)
Base case	–	–	– €15,645	+ 0.80	Dominant
AS1	Probability of SHEs [39] = 8.9% Frequency of SHEs [39] = 2.5 events/person-year	Probability of SHEs [39] = 8.9% Frequency of SHEs A = 1 event/person-year	– €9149	+ 0.80	Dominant
AS2	Probability of SHEs [39] = 8.9% Frequency of SHEs [39] = 2.5 events/person-year	Probability of SHEs [62] = 14.8% Frequency of SHEs = 1.4 events/person-year	– €3605	+ 0.83	Dominant
AS3	Reduction of hypoglycemia (FSL) [12] = – 58.0%	Reduction of hypoglycemia (FSL) [11] = – 29.0%	– €4620	+ 0.68	Dominant
AS4	Reduction of DKA (FSL) [12] = – 68.0%	Reduction of DKA (FSL) = – 52.4%	– €15,632	+ 0.80	Dominant
AS5	Unitary cost per strip [48] = €0.57 Unitary cost per lancet [48] = €0.14	Unitary cost per strip A = €0.00 Unitary cost per lancet A = €0.00	– €8390	+ 0.80	Dominant
AS6	Cost of NSHE B = €3.92	Cost of NSHE event [48] = €66.73	– €23,415	+ 0.80	Dominant
AS7	Cost of NSHE B = €3.92	Cost of NSHE A = €0.00	– €15,161	+ 0.80	Dominant
AS8	Reduction of HbA1c levels associated with FSL [34] = 1.1%	Reduction of HbA1c levels associated with FSL [63] = 0.6%	– €14,508	+ 0.56	Dominant

AS alternative scenario, DKA diabetic ketoacidosis, FSL FreeStyle Libre systems, ICER incremental cost-effectiveness ratio, NSHE non-severe hypoglycemic event, QALY quality-adjusted life year, SBGM self-blood glucose monitoring, SHE severe hypoglycemic event

^aConservative assumption

^bCost estimated as a weighted average of specialist-managed event (21.9% [50]; €17.89 [49]) and patient-managed event (complementary data: 78.10%; €0.00)

Although no remarkable variations were observed in the cost-effectiveness results, AS1, AS2 and AS3 revealed that hypoglycemia-related inputs represented the most sensitive inputs of the model. Alternative probabilities and frequencies of SHEs (base-case: probability of SHE = 8.9% [39] [2.5 events/person-year] [39]; AS1: 8.9% [39] [1 event/person-year]; AS2: 14.8% [62] [1.4 events/person-year] [62]) produced differences in incremental costs (AS1: – €9149; AS2: – €3605) compared with the base-case savings per patient (– €15,645). Regarding the reduction in SHEs (base case: – 58.0% [12], AS3: – 29.0% [11]) associated with FSL, both incremental QALYs (base case = 0.80; AS3 = 0.68) and costs (base case = – €15,645; AS3 = – €4620) varied considerably. Nonetheless, FSL remained a dominant alternative compared with SBGM.

In contrast, considering alternative reductions in DKA (base case: – 68.0% [12], AS4: – 52.4% [23]) associated with FSL did not produce important variations compared with the base case.

The results of the AS5 suggested that FSL would be a dominant intervention compared with SBGM, even when both strips and lancets are free. In this scenario, lifetime cost savings per patient receiving FSL vs. SBGM were estimated at €8390.

In addition, the cost associated with the management of NSHE was studied in AS6 and AS7. No relevant variations in the cost savings were observed when the cost per NSHE was set at €64.91 [48] in AS6 (incremental costs = – €23,415) or €0.00 in AS7 (incremental costs = – €15,161).

Finally, as studied in AS8, the reduction in HbA1c levels associated with FSL provided a substantial decrease in QALYs (0.56 QALYs). However, there were still cost savings of €14,508 per patient.

Probabilistic Sensitivity Analyses Results

All second order Monte Carlo simulations resulted in FSL being dominant compared to SBGM (Fig. 2).

Fig. 2.

PSA results: Scatterplot of the 1000 second order Monte Carlo simulation. PSA probabilistic sensitivity analysis, QALY quality-adjusted life year, WTP willingness-to-pay

The 1000 simulations of approximately 10,000 patients in the PSA estimated an average QALY gain of 0.79 (SE = 0.02) per patient receiving FSL (mean QALYs = 9.92; SE = 0.31) compared with SBGM (mean QALYs = 9,12; SE = 0.29). In economic terms, the overall costs per patient on FSL and SBGM were €60,895 (SE = 617.02) and €76,417 (SE = 524.56), respectively. Thus, the cost savings achieved due to FSL devices in the PSA resulted in €15,523 (SE = 557.09).

Discussion

Recently, several RWE studies have demonstrated the effectiveness of FSL in people living with T2DM treated with basal insulin [10–12]. Additionally, recent findings from the REFLECT study revealed that FSL not only reduces hypoglycemia and DKA but also lowers HbA1c levels and the incidence of diabetic complications [64, 65]. In this context, assessing the cost-effectiveness of FSL in this population is crucial to support informed decision-making by physicians and healthcare institutions.

This CUA demonstrated that, compared with SBGM, FSL is a cost-effective glucose monitoring alternative in Spain for people with T2DM treated with basal insulin and poor glycemic control, considering the €22,000/QALY threshold used in HTAs conducted by RedETS in Spain [66]. Nonetheless, no explicit cost-effectiveness threshold exists in Spain and alternative thresholds of €27,000–€34,000/QALY have recently been proposed based on a fixed-effect econometric approach, under which FSL still represents a cost-effective alternative [67]. PSA showed that FSL provided cost savings and achieved more QALYs in all Monte Carlo simulations, resulting in a dominant strategy.

To the authors’ knowledge, this is the first analysis conducted in Spain to evaluate the cost-effectiveness of FSL in patients with T2DM receiving basal insulin and exhibiting poor glycemic control. However, previous studies have assessed the cost-effectiveness of FSL in other countries and populations. In people with T2DM treated with basal insulin, FSL has been found to be a cost-effective alternative in other settings. A recently published CUA carried out in Italy, demonstrated that despite the increase in costs associated with FSL (€5338), these devices provided additional QALYs (0.51), resulting in an ICER of €10,556/QALY versus SBGM [68]. Moreover, another recent CUA conducted from the Canadian private-payer perspective concluded that FSL, compared with SBGM, represents a dominant glucose monitoring strategy, achieving more QALYs (+ 0.480) and providing cost savings (CAD $8091; reference year 2022) [69].

In contrast to the Italian CUA mentioned above, the Spanish adaptation included several inputs such as the cost of the NSHE and a higher rate of SHE, which may influence these economic outcomes. However, the model inputs were validated by a multidisciplinary board of seven experts, and the results were tested through various sensitivity analyses, which demonstrated consistency across all the scenarios assessed.

One of the most controversial inputs is the incidence of mild to moderate and severe hypoglycemia. Clinicians agreed to select one of the highest reported rates from the scientific literature [39]. In addition, none of the participants enrolled in the HAT study were from Spain, and owing to the study design, it was not possible to differentiate between basal insulin and MDI users [39]. Nevertheless, based on methodological criteria, the source was considered one of the most robust studies. Likewise, when the incidence of SHE was varied in sensitivity analyses, FSL remained a cost-effective strategy.

Furthermore, the usage and cost of strips and lancets could be considered high in comparison with those in clinical practice. However, FSL demonstrated to produce cost savings and increase QALYs, even when considering a cost of €0.00 per unit of both strips and lancets.

Finally, another source of uncertainty in the model is the effectiveness of FSL, which is measured as a reduction in acute diabetic events and HbA1c levels. Regarding the reductions in hypoglycemia and DKA events, a RWE study executed in France [12] in patients with T2DM on basal insulin treatment was selected as the most faithful reflection of clinical practice in Spain. According to the sensitivity analysis, FSL reduced costs and increased QALYs. On the other hand, the reduction in HbA1c levels associated with FSL was obtained from a meta-analysis conducted in Canada and the United States [34]. Recently, new European data have been published [63], and despite its limitations, sensitivity analysis showed that FSL were dominant compared to SBGM.

In addition, the use of FSL has several implications in clinical terms. This technology reduces the frequency of glycemic SBGM measurements, improving patients´ quality of life and lowering the consumption of strips and lancets. Moreover, several studies in Europe demonstrated that patients had strong concerns about the need for less frequent glucose self-monitoring measurements [70] and that they preferred glucose-monitoring devices over finger-prick tests [71]. In addition, a study conducted in Germany [72] revealed that patients monitored with FSL presented greater satisfaction (57%) than those monitored with other CGM devices (43%) because of the ease of use, ease of interpreting results, alarm for high/low sugar, size and appearance and customer service/support. Thus, these additional advantages should be considered when making reimbursement decisions, because they translate into potential benefits for patients and healthcare systems [72].

Treatment guidelines for T2DM emphasize that patient improvement is achieved through a triad of therapeutic objectives: improving HbA1c levels, limiting glycemic variability and preventing hypoglycemia [73]. Although FSL improves outcomes in patients with T2DM patients treated with basal insulin in line with these guidelines, the device is not yet reimbursed for this patient group in Spain. Therefore, this economic analysis suggests that FSL could enhance treatment outcomes in patients with T2DM exhibiting poor glycemic control and receiving basal insulin in a cost-effective manner.

This study has both strengths and limitations. The main limitation is the use of multiple data sources from different settings or populations due to the limited availability of local data. However, this limitation was partially addressed by validating the inputs through a multidisciplinary expert board and testing them in the sensitivity analyses presented in this study. Further research is needed to assess the clinical effectiveness of FSL in Spain. In contrast, one of the main strengths is that the DEDUCE microsimulation model has already been validated and that it uses updated risk equations [30, 31]. Additionally, as a microsimulation model, DEDUCE captures patient heterogeneity and variability, providing more realistic results than deterministic models do. Another strength is the model’s adaptation via recently published RWE studies, which better reflect clinical practice.

Conclusions

In conclusion, the use of FSL provides cost savings from the Spanish healthcare system perspective through a reduction in acute diabetic events and long-term complications. This analysis suggests that FSL, by delivering better clinical outcomes at a lower overall cost, is a preferable alternative to SBGM for patients with T2DM on basal insulin with poor glycemic control.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Fernando Gómez-Peralta, Virginia Bellido, Francisco Javier Ampudia-Blasco, Juana Carretero Gómez, Ana María Cebrián-Cuenca and Pedro Mezquita-Raya participated in the validation of the model inputs, interpretation of the results, and drafting and validation of the final version of the manuscript. Mireya Robles-Plaza and Alberto de la Cuadra-Grande participated in the conceptualization of the study, development of the economic model, interpretation of the results and manuscript drafting.

Funding

This study was funded by Abbott Diabetes Care. The funding source is the owner of the DEDUCE model, which has provided the model for conducting this project. The sponsor also funded the medical writing and the journal’s Rapid Service Fee.

Data Availability

All data generated or analyzed during this study are included in this published article/as supplementary information files.

Declarations

Conflict of Interest

Mireya Robles-Plaza and Alberto de la Cuadra-Grande are employees of PORIB, a consultant company specializing in economic evaluation of health care interventions and outcomes research, which received fees for providing methodological support to this project. Fernando Gómez-Peralta has taken part in advisory panels for Abbott, Insulcloud S.L., Sanofi and Novo Nordisk; has received research support from Sanofi, Novo Nordisk, Boehringer Ingelheim Pharmaceuticals and Lilly; and has acted as a speaker for Abbott, Sanofi, Novo Nordisk, Boehringer Ingelheim Pharmaceuticals, AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb Co. and Lilly. Virginia Bellido has served as a consultant/advisor for Abbott Diabetes Care, AstraZeneca, Eli Lilly, Merck, Novo Nordisk, and Sanofi and as a speaker for Abbott Diabetes Care, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Merck, Novo Nordisk, and Sanofi and has received grant support from Abbott Diabetes Care, Eli Lilly, Novo Nordisk and Sanofi. Francisco Javier Ampudia-Blasco has served as a consultant/advisor for Abbott Diabetes Care, AstraZeneca, Boehringer Ingelheim, Eli Lilly, GlaxoSmithKline, LifeScan, MannKind Co., Medtronic, Menarini, Merck, Novartis, Novo Nordisk, and Sanofi and as a speaker for Abbott Diabetes Care, AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, LifeScan, Eli Lilly, Madaus, Medtronic, Menarini, Merck, Novartis, Novo Nordisk, and Sanofi and has received grant support from Novo Nordisk and Sanofi. Juana Carretero Gómez reported receiving consulting fees and speaking from AstraZeneca, Boehringer Ingelheim, Lilly, Abbot Diabetes Care and Novo Nordisk. Ana María Cebrián-Cuenca has served as a consultant/advisor for AstraZeneca, Boehringer Ingelheim, Eli Lilly, Menarini, Merck, Novo Nordisk, and Sanofi and as a speaker for Abbott Diabetes Care, AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Eli Lilly, Menarini, Merck, Novartis, Novo Nordisk, and Sanofi and has received grant support from Merck, Novo Nordisk and Sanofi. Pedro Mezquita-Raya has served as a consultant/advisor on scientific boards for Abbott, AstraZeneca, FAES and Novo Nordisk; as a speaker for AstraZeneca, Eli Lilly, FAES, Fresenius and Novo Nordisk; and in research activities for Eli Lilly & Company and Novo Nordisk. None of the authors received fees for being listed in this manuscript’s authorship.

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.