Long-term health economic benefits of sensor-augmented pump therapy vs continuous subcutaneous insulin infusion alone in type 1 diabetes: a UK perspective

Abstract

Aims/hypothesis:

Continuous subcutaneous insulin infusion (CSII) is an important treatment option for type 1 diabetes patients unable to achieve adequate glycemic control with multiple daily injections (MDI). Combining CSII with continuous glucose monitoring (CGM) in sensor-augmented pump therapy (SAP) with a low glucose-suspend (LGS) feature may further improve glycemic control and reduce the frequency of hypoglycemia. A cost-effectiveness analysis of SAP + LGS vs CSII plus self-monitoring of blood glucose (SMBG) was performed to determine the health economic benefits of SAP + LGS in type 1 diabetes patients using CSII in the UK.

Methods:

Cost-effectiveness analysis was performed using the CORE diabetes model. Treatment effects were sourced from the literature, where SAP + LGS was associated with a projected HbA_1c reduction of −1.49% vs −0.62% for CSII, and a reduced frequency of severe hypoglycemia. The time horizon was that of patient lifetimes; future costs and clinical outcomes were discounted at 3.5% and 1.5% per annum, respectively.

Results:

Projected outcomes showed that SAP + LGS was associated with higher mean quality-adjusted life expectancy (17.9 vs 14.9 quality-adjusted life years [QALYs], SAP + LGS vs CSII), and higher life expectancy (23.8 vs 21.9 years), but higher mean lifetime direct costs (GBP 125,559 vs GBP 88,991), leading to an incremental cost-effectiveness ratio (ICER) of GBP 12,233 per QALY gained for SAP + LGS vs CSII. Findings of the base-case analysis remained robust in sensitivity analyses.

Conclusions/interpretation:

For UK-based type 1 diabetes patients with poor glycemic control, the use of SAP + LGS is likely to be cost-effective compared with CSII plus SMBG.

Keywords::

• Continuous subcutaneous insulin infusion
• Continuous glucose monitoring
• Sensor-augmented pump therapy
• Type 1 diabetes
• Cost-effectiveness

Introduction

Recent estimates suggest that in the UK there are just under a quarter of a million people living with type 1 diabetes1, with the annual direct costs of treatment in 2010/2011 estimated at ∼GBP 1 billion2. An estimated 70% of total direct costs are due to long-term complications of the disease, which include cardiovascular disease, renal disease, and ophthalmic problems2. The chronic nature of the disease, combined with high direct costs, make long-term cost-effectiveness analyses pivotal in informing payer decisions regarding the reimbursement/funding of new therapeutic technologies.

Patients with type 1 diabetes require lifetime treatment with exogenous insulin for survival, and landmark studies such as the Diabetes Control and Complications Trial have shown that good glycemic control is fundamental in reducing the risk of developing long-term diabetes-related complications3. Current optimized or intensive insulin administration regimens that aim to achieve strict glycemic control are multiple daily injections (MDI) and continuous subcutaneous insulin infusion (CSII, insulin pump therapy). Guidance from the UK National Institute for Health and Care Excellence (NICE) states the CSII is indicated for patients with type 1 diabetes (adults and children aged ≥12 years) if, despite optimization of MDI, HbA_1c remains ≥8.5% (69 mmol/mol), or when reaching HbA_1c targets with MDI results in disabling hypoglycemia4. NICE also recommends CSII in children younger than 12 years if MDI is considered impractical or inappropriate. In the UK, insulin pumps were used by an estimated 6% of adult and 19% of pediatric type 1 diabetes patients in a survey performed in 20125. As such, the usage of insulin pumps in the UK is considerably below that reported in many other European settings and in the US, where an estimated 40% of people with type 1 diabetes use an insulin pump6.

Insulin pumps have been used in type 1 diabetes since the 1970s7; however, a more recent advance in insulin pump technology and its clinical use is sensor-augmented pump (SAP) therapy, where CSII is used in conjunction with data from continuous glucose monitoring (CGM), either with the signal from a subcutaneously implanted glucose sensor wirelessly transmitted to a wearable or hand-held receiver or to a compatible insulin pump. This allows patients themselves to use CGM data as an aid to further improve glycemic control with CSII. Some SAP systems also employ pumps with a low-glucose insulin-suspend (LGS) feature, whereby the delivery of basal insulin is suspended for up to 2 h when CGM-measured low glucose levels are detected. SAP with LGS have been shown to reduce both HbA_1c8–10 and the frequency of hypoglycemia10–14 in patients with type 1 diabetes.

In spite of evidence for its clinical efficacy, SAP is not widely used, and one barrier to uptake may be cost. Previous analyses have shown that, in several settings, CSII is cost-effective relative to MDI in type 1 diabetes15–17. The cost-effectiveness of SAP is not well studied, although recent reports showed SAP to be cost-effective relative to CSII alone in the Swedish18 and Australian setting19. Given the clinical benefits associated with SAP + LGS, a long-term cost-effectiveness analysis of SAP + LGS vs CSII plus self-monitoring of blood glucose (SMBG) was performed to establish if SAP + LGS is cost-effective in the UK setting for type 1 diabetes patients.

Methods

Simulation cohort and treatment effects

A summary of baseline characteristics of the simulation cohort is provided in Table 1. The mean age of the cohort was 27 years, mean duration of diabetes 13 years, and mean HbA_1c at baseline 10% (86 mmol/mol). The additional glycemic (HbA_1c lowering) effect of CGM with CSII vs CSII with SMBG was sourced from an individual patient data meta-analysis20. From this, everyday use of SAP was projected to lead to an HbA_1c reduction of −1.49% vs −0.62% for CSII alone. Use of SMBG strips in the SAP + LGS arm was assumed to be lower than for CSII alone, based on findings from an observational study conducted in Sweden21, where CGM reduced SMBG use from a mean 7.11 SMBG strips and lancets per day to 4.35 per day. SAP + LGS vs CSII alone was also projected to lead to a reduction in the frequency of severe hypoglycemic events, based on a randomized controlled trial of SAP with LGS vs CSII conducted in type 1 diabetes patients in Australia14. In this study, the rate of severe hypoglycemic events in the SAP + LGS group was 0 per 100 patient months compared with 2.2 per 100 patient months in the CSII-alone group.

Table 1. Baseline cohort characteristics and treatment effects.

Parameter	Value
Mean age, years	27
Male, %	48.5
HbA1c, % (mmol/mol)	10 (86)
Duration of diabetes, years	13
Treatment effect, reduction in HbA1c, %
SAP + LGS	−1.49
CSII	−0.62

CSII, continuous subcutaneous insulin infusion; HbA_1c, glycated hemoglobin; SAP, sensor-augmented pump therapy.

The impact of reduced fear of hypoglycemia in the SAP + LGS group was also included in the analysis, based on the findings of Nørgaard et al.22, who reported that the use of SAP vs CSII alone was associated with a reduction of 6.9 units in the Hypoglycemia Fear Survey (HFS) score. A 1-unit increase in HFS score has been shown to correspond to a 0.008 unit decrease in the EQ-5D index score, a measure of health-related quality-of-life23; the 6.9 unit difference in HFS score between the two arms was, therefore, translated into a utility benefit of 0.0552, which was applied to the SAP + LGS arm.

Model description

The analysis was performed using the CORE Diabetes Model (CDM; IMS Health, Basel, Switzerland). The CDM is a validated non-product-specific policy analysis tool for cost-effectiveness analysis in both type 1 and type 2 diabetes; a detailed description of the model architecture (including schematic diagrams) and validation is available in publications by Palmer et al.24^,25 and, more recently, McEwan et al.26. In summary, the model is based on a series of inter-dependent sub-models that simulate both acute and long-term diabetes-related complications (angina, myocardial infarction, congestive heart failure, stroke, peripheral vascular disease, diabetic retinopathy, macula edema, cataract, hypoglycemia, ketoacidosis, lactic acidosis, depression, edema, nephropathy and end-stage renal disease, neuropathy, foot ulcer and amputation, and non-specific mortality). The sub-models have a semi-Markov structure and use time, state, time-in-state, and diabetes type-dependent probabilities derived from published sources to simulate disease progression. Monte Carlo simulation using tracker variables is used to overcome the memory-less properties of the standard Markov model and allows for interconnectivity and interaction between individual sub-models.

Costs and utilities

The base-case analysis was performed from the perspective of the UK National Health Service (a secondary analysis was also performed using a societal perspective). Direct costs were sourced from published literature and, where necessary, were inflated to 2013 GBP27–37. In the sensitivity analysis using the societal perspective, indirect costs were calculated based on the human capital approach, using average annual salaries for the UK. Days off work were sourced from a study conducted in the Netherlands (Medtronic International Trading Sarl, Tolochenaz, Switzerland; data on file, available on request).

For treatment costs, only the incremental costs between the two arms were considered, namely the different frequency of use of SMBG strips and lancets between CSII alone and SAP + LGS therapy, and use of the CGM transmitter and glucose sensors. Average UK costs for glucose sensors and sensor transmitters were taken to be costs of Enlite sensors and MiniLink transmitters (Medtronic) and the costs of SMBG strips and lancets were sourced from the British National Formulary 63 (March 2012)38.

Health state utility values were taken from published literature39 and references therein.

Statistical approach and other model settings

For each simulation, a simulated cohort of 1000 patients was run through the model 1000 times using first-order Monte Carlo simulation. Long-term outcomes included total direct costs, life expectancy, quality-adjusted life expectancy, and time to onset of complications. Future costs were discounted at a rate of 3.5% per annum and clinical outcomes were discounted at a rate of 1.5% per annum in line with NICE guidance on long-term conditions, where treatment effects are sustained over a prolonged period of time40. The mean values from the simulation (a total of 1000 mean values, each from a cohort of 1000 patients run through the model) were then used to generate scatterplots of incremental costs vs incremental effectiveness (quality-adjusted life years [QALYs]) for SAP + LGS vs CSII. Data from the scatterplot were then used to generate a cost-effectiveness acceptability curve.

Sensitivity analysis

In order to explore the robustness of the base-case findings and establish the key drivers of results, a series of one-way and two-way sensitivity analyses were performed around key input parameters: the utility benefit associated with SAP + LGS due to reduced fear of hypoglycemia was negated; the SAP + LGS -associated benefit from reduced severe hypoglycemia was negated (an event rate of 2.6 events per 100 patient years was applied in both treatment arms); the baseline HbA_1c was set to 9.0% (75 mmol/mol) and 8.1% (65 mmol/mol); a discount rate of 3.5% per annum was applied to both future costs and clinical outcomes; and an analysis was conducted from the societal perspective. A two-way sensitivity analyses was also performed in which both the benefits from reduced fear of hypoglycemia and reduced frequency of severe hypoglycemia with SAP + LGS therapy were negated.

Results

Cost-effectiveness of SAP vs CSII in the base case

In the base-case analysis (payer perspective) the use of SAP + LGS therapy was associated with an improvement in life expectancy of 1.9 years (23.8 years vs 21.9 years) and an improvement in quality-adjusted life expectancy of 3.0 QALYs) (17.9 QALYs vs 14.9 QALYs), compared with CSII alone (Table 2). Total direct lifetime costs were GBP 36,568 higher for the SAP + LGS group than for the CSII group (GBP 125,559 vs GBP 88,991), resulting in an incremental cost-effectiveness ratio (ICER) of GBP 12,233 per QALY gained for SAP + LGS vs CSII alone.

Table 2. Summary of base-case results.

	SAP + LGS	CSII alone	Difference
Life expectancy, years	23.79	21.91	1.88
Undiscounted life expectancy, years	30.86	27.83	3.03
Quality-adjusted life expectancy, QALYs	17.88	14.89	2.99
Total direct costs, GBP	125,559	88,991	36,568
Treatment costs, GBP	59,561	14,238	45,323
Cardiovascular complication costs, GBP	4674	4346	328
Renal complication costs, GBP	8270	11,536	−3266
Ulcer/amputation/neuropathy costs, GBP	33,505	38,456	−4951
Ophthalmic complication costs, GBP	17,793	17,285	508
ICER (95% CI), GBP per QALY gained	12,233 (12,225–12,351)

CI, confidence interval; CSII, continuous subcutaneous insulin infusion; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year; SAP, sensor-augmented pump therapy; LGS, low glucose suspend.

All costs are presented in 2013 GBP.

The use of SAP + LGS relative to CSII was also projected to delay the mean onset of diabetes-related complications (Table 3). The most pronounced delays were seen in the projected onset of first ulcer and gross proteinuria, the onset of which was delayed by an average of over 3.5 years in the SAP + LGS group compared with the CSII-alone group.

Table 3. Delay in onset of diabetes-related complications with SAP + LGS vs CSII alone.

	Time alive and free of complication, years
	SAP + LGS	CSII alone	Difference
Microalbuminuria	11.6	9.1	2.5
Gross proteinuria	26.7	22.9	3.8
End-stage renal disease	30.2	27.0	3.2
First ulcer	22.7	19.1	3.6
Amputation	28.3	25.1	3.2
Neuropathy	14.6	11.2	3.4
Peripheral vascular disease	29.4	26.5	2.9
Congestive heart failure	29.5	26.6	3.0
Angina	29.2	26.3	2.8
Myocardial infarction	29.7	26.8	2.9
Stroke	29.3	26.7	2.7
Cataract	27.0	24.6	2.4
Macular edema	19.0	16.8	2.2
Severe vision loss	23.2	20.8	2.3
Proliferative retinopathy	26.0	22.5	3.5

CSII, continuous subcutaneous insulin infusion; SAP, sensor-augmented pump therapy; LGS, low glucose suspend.

Sensitivity analyses

In the sensitivity analysis conducted from the societal perspective, which captures indirect costs due to lost productivity, the ICER was lower at GBP 3359 per QALY gained for SAP + LGS vs CSII. The higher overall direct costs were driven mainly by the higher incremental treatment costs for SAP + LGS (GBP 59,561 for SAP vs GBP 14,238 for CSII); however, this was partially offset by lower complication costs, particularly renal and ulcer/amputation/nephropathy-related costs, due to improved glycemic control. Indeed, total mean per patient lifetime costs for ulcer/amputation/nephropathy were GBP 4951 lower with SAP + LGS than with CSII (Table 2). It is important to note that, as projected life expectancy was 1.9 years higher in the SAP group than the CSII group, the time spent at risk for complications was higher for the SAP group, which will be a contributing factor in the higher overall lifetime costs in the SAP group.

The robustness of the base-case results was also explored using a series of one-way and two-way sensitivity analyses (Table 4), where the results of the analyses remained robust despite changes in key input parameters. In the analysis where both SAP-related benefits associated with reduced fear of hypoglycemia and reduced incidence of severe hypoglycemia were negated, the ICER increased to GBP 26,600 per QALY gained. In one-way sensitivity analyses in which baseline HbA_1c was decreased to 9.0% (75 mmol/mol) and 8.1% (65 mmol/mol) (compared with 10.0% [86 mmol/mol] in the base case), the ICER increased to GBP 13,840 and GBP 15,392 per QALY gained, respectively. Sensitivity analysis around discount rates influenced the ICER as expected (use of undiscounted results led to an increased ICER; data not shown). A sensitivity analysis was also performed using a discount rate of 3.5% per annum for both future costs and clinical outcomes. In this scenario the ICER was increased to GBP 18,371 per QALY gained.

Table 4. Summary findings of sensitivity analyses.

Scenario	Quality-adjusted life expectancy, QALYs			Costs, GBP			ICER, GBP per QALY gained
	SAP  + LGS	CSII	Difference	SAP  + LGS	CSII	Difference
Base case	17.86	14.87	2.99	125,559	88,991	36,568	12,233
Societal perspective	17.86	14.87	2.99	250,651	240,609	10,042	3359
No FoH utility (base case 0.0552 benefit for SAP)	16.56	14.87	1.68	125,559	88,991	36,568	21,816
No severe hypoglycemia benefit	17.81	15.03	2.73	125,513	87,831	37,682	13,818
No FoH utility and no severe hypoglycemia benefit	16.50	15.08	1.42	125,513	87,831	37,682	26,600
Baseline HbA1c 9.0% (75 mmol/mol) (base case 10%)	18.91	16.08	2.83	122,536	83,387	39,149	13,839
Baseline HbA1c 8.1% (65 mmol/mol) (base case 10%)	19.81	17.12	2.69	119,690	78,348	41,342	15,392
Discount rates 3.5% per annum costs and outcomes (base case 3.5% for costs and 1.5% for clinical outcomes)	13.43	11.44	1.99	125,559	88,991	36,568	18,371

CSII, continuous subcutaneous insulin infusion; ICER, incremental cost-effectiveness ratio; FoH, fear of hypoglycemia; QALY, quality-adjusted life year; SAP, sensor augmented pump therapy; SMBG, self-monitoring of blood glucose; LGS, low glucose suspend.

Cost-effectiveness and willingness-to-pay

Analysis of the cost-effectiveness acceptability curve showed that, at a willingness-to-pay threshold of between GBP 20,000–30,000 per QALY gained, SAP + LGS is likely to be considered cost-effective compared with CSII alone in the UK (Figure 1).

Figure 1. Cost-effectiveness acceptability curve for SAP + LGS vs CSII. CSII, continuous subcutaneous insulin infusion; QALY, quality-adjusted life year; SAP, sensor-augmented pump.

Discussion

Our long-term cost-effectiveness analysis suggests that, with the commonly used willingness-to-pay threshold of between GBP 20,000–30,000 per QALY gained, SAP + LGS is likely to be cost-effective vs CSII alone for patients with poorly-controlled type 1 diabetes in the UK. The ICER was GBP 12,233 per QALY gained with a baseline HbA_1c of 10% (86 mmol/mol), GBP 13,840 per QALY gained for an HbA_1c of 9% (75 mmol/mol), and GBP 15,392 per QALY gained at an HbA_1c of 8.1% (65 mmol/mol).

From the perspective of the payer, although SAP + LGS was associated with higher direct treatment costs vs CSII alone, this is at least partially offset by reduced long-term complication costs due to improved glycemic control, as well as a reduction in the incidence of severe hypoglycemic events. Indeed, the improved glycemic control resulting from the use of SAP + LGS delayed the onset of all complications considered in the analysis by at least 2 years. From the perspective of the patient, SAP + LGS is associated with improved quality-of-life due to reduced risk of tissue complications, reduced frequency of severe hypoglycemic events, and a reduced fear of hypoglycemia.

The present analysis is among the first to capture the health economic benefits of SAP + LGS in terms of reduced fear of hypoglycemia. A substantial proportion of type 1 diabetes patients, particularly those with past severe hypoglycemic events, report having fear of hypoglycemia. This can negatively impact health-related quality-of-life41^,42, and can result in detrimental behavior patterns such as deliberately decreasing insulin dose to maintain a “safety net” in terms of blood glucose levels and thereby reducing the likelihood of a hypoglycemic event41^,42. The security provided by SAP, and in particular the LGS feature, may give patients the confidence to try to achieve lower glycemic targets than they would otherwise, and this improved glycemic control would in turn reduce the risk of progressing to late stage diabetes-related complications. However, sensitivity analysis showed that, even when the benefit of reduced fear of hypoglycemia was negated, SAP + LGS is likely to remain cost-effective relative to CSII alone in patients with poorly controlled disease. In addition to a quality-of-life benefit associated with reduced fear of hypoglycaemia, a recent study by McQueen et al.43 showed that, in type 1 diabetes, improved HbA_1c was also associated with improved quality-of-life, independent of other factors, with a 1% increase in HbA1c corresponding to a −0.027 (95% CI = −0.049–0.006) decrease in utility. This potential benefit was not captured in the current analysis, thereby potentially making the findings of the analysis conservative.

A limitation of our study is that evidence for the effectiveness of SAP at reducing severe hypoglycemia is relatively limited: a randomized controlled trial14 used SAP with LGS (not standard SAP) vs CSII, although the marked effectiveness of both standard SAP and SAP plus LGS at reducing severe hypoglycemia was confirmed in an observational study of patients with problematic severe hypoglycemia, despite structured diabetes education and insulin pump therapy12. A further limitation allied to this is that the paucity of clinical data comparing SAP with CSII in type 1 diabetes meant that it was necessary to use multiple input sources for clinical data: Treatment effects relating to HbA_1c were sourced from an individual patient level meta-analysis20 and clinical data relating to hypoglycemia event rates were taken from a randomized controlled trial conducted in type 1 diabetes patients in Australia14.

In conclusion, the findings of long-term cost-effectiveness analysis suggest that, for patients with poorly controlled type 1 diabetes already using CSII, the switch to SAP + LGS is likely to be cost-effective in the UK setting, largely due to improved glycemic control, which in turn leads to a lower incidence of long-term complications, together with reduced fear of hypoglycemia and a reduced incidence of actual severe hypoglycemic events. Moreover, sensitivity analyses showed that the clinical and economic benefits of SAP + LGS relative to CSII alone are largest in patients with poorer levels of baseline glycemic control.

Abbreviations
CDM	=	CORE Diabetes Model
CGM	=	Continuous glucose monitoring
CSII	=	Continuous subcutaneous insulin infusion
GBP	=	Great Britain pound
HFS	=	Hypoglycemia Fear Survey
ICER	=	Incremental cost-effectiveness ratio
LGS	=	Low-glucose insulin suspend
MDI	=	Multiple daily injections
NICE	=	National Institute for Health and Care Excellence
QALY	=	Quality-adjusted life year
SAP	=	Sensor augmented pump therapy
SMBG	=	Self-monitoring of blood glucose

Transparency

Declaration of funding

This study was supported by funding from Medtronic International Trading Sàrl.

Declaration of financial/other relationships

SR is a current employee of HEVA HEOR, which has received consulting fees from Medtronic. JSP and WV are current employees of Ossian Health Economics and Communications, which has received consulting fees from Medtronic. MC and MJ are current employees of Medtronic UK. SP is a current employee of Medtronic International Trading Sàrl. JCP has received speaker and/or consultancy fees from Medtronic, Roche, Cequr and Cellnovo, manufacturers of insulin pumps and/or CGM systems. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.