DIABETICMedicine DOI: 10.1111/dme.12661
Research: Health Economics Health-economic analysis of real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze1, R. Saunders2, A.-S. Brandt3, S. de Portu4, N. L. Papo4 and J. Jendle5,6 1 HEVA HEOR, Lyon, France, 2Ossian Health Economics and Communications, Basel, 3Medtronic Danmark A/S, Copenhagen, Denmark, 4Medtronic International € € University, Orebro, Trading Sarl, Tolochenaz, 5Endocrine and Diabetes Centre, Karlstad Hospital, Karlstad, and 6Faculty of Health Sciences and Medicine, Orebro Sweden
Accepted 2 December 2014
Abstract Aim To evaluate the clinical benefits and cost-effectiveness of the sensor-augmented pump compared with self– monitoring of plasma glucose plus continuous subcutaneous insulin infusion in people with Type 1 diabetes. Methods The CORE Diabetes Model was used to simulate disease progression in a cohort of people with baseline characteristics taken from a published meta-analysis. Direct and indirect costs for 2010–2011 were calculated from a societal payer perspective, with cost-effectiveness calculated over the patient’s lifetime. Discount rates of 3% per annum were applied to the costs and the clinical outcomes.
Use of the sensor-augmented pump was associated with an increase in mean discounted, quality–adjusted life expectancy of 0.76 quality–adjusted life years compared with continuous subcutaneous insulin infusion (13.05 0.12 quality–adjusted life years vs 12.29 0.12 quality–adjusted life years, respectively). Undiscounted life expectancy increased by 1.03 years for the sensor-augmented pump compared with continuous subcutaneous insulin infusion. In addition, the onset of complications was delayed (by a mean of 1.15 years) with use of the sensor-augmented pump. This analysis resulted in an incremental cost–effectiveness ratio of 367,571 SEK per quality–adjusted life year gained with the sensor-augmented pump. The additional treatment costs related to the use of the sensor-augmented pump were partially offset by the savings attributable to the reduction in diabetes-related complications and the lower frequency of self-monitoring of plasma glucose.
Results
Conclusions Analysis using the CORE Diabetes Model showed that improvements in glycaemic control associated with sensor-augmented pump use led to a reduced incidence of diabetes-related complications and a longer life expectancy. Use of the sensor-augmented pump was associated with an incremental cost–effectiveness ratio of 367,571 SEK per quality–adjusted life year gained, which is likely to represent good value for money in the treatment of Type 1 diabetes in Sweden.
Diabet. Med. 00, 000–000 (2015)
Introduction Type 1 diabetes is characterized by deficient insulin production, but its cause remains unknown and the condition is not preventable. Treatment of Type 1 diabetes requires daily administration of insulin, which aims to normalize glucose metabolism. Diabetes is associated with an increased risk of numerous serious complications and comorbidities, although the risk is reduced if diabetes is well controlled. Two diabetes-related comorbidities with an important impact on life expectancy are cardiovascular disease and renal failure [1,2]. In addition to their quality-of-life implications, diabeCorrespondence to: Stephane Roze. E-mail: [email protected]
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
tes-related complications have a significant cost impact on healthcare provision [1,3]. Diabetes imposes an increasing economic burden on the Swedish healthcare system [4]. The Type 1 diabetes incidence rate in Sweden has increased rapidly since the 1990s and in 2012 the Swedish National Diabetes Register reported that there were 43 393 people with Type 1 diabetes [5]. Although it has a high Type 1 diabetes burden, Sweden was ranked in first place for diabetes treatment at the recent European Association for the Study of Diabetes 2014 congress. For this reason, and the availability of comprehensive Type 1 diabetes data, evaluation of diabetes practices in Sweden offers the opportunity to inform other healthcare systems and may provide a framework for further analyses in other settings.
1
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
What’s new? • This study builds on a recent meta-analysis to provide further insights into the clinical and safety aspects of real-time continuous glucose monitoring. • Simulations of disease progression show how these aspects translate into long-term patient benefits and healthcare costs. • Simulations indicate that continuous glucose monitoring leads to later onset and reduced incidence of acute and long-term diabetes-related complications. • The findings may have an important influence on informing treatment choices with respect to providing value both for patients and healthcare payers. Active self-management of diabetes can help people to achieve glycaemic control and reduce the risk of complications. An important aspect of self-management is self-monitoring of plasma glucose (SMPG), although advantages may be gained through the use of continuous glucose monitoring (CGM). In 2006, Deiss et al. [6] reported the clinical benefit of CGM compared with conventional SMPG for the reduction of HbA1c levels. Additionally, CGM can determine the frequency and severity of hypoglycaemic episodes and provides maximum information about changing blood glucose levels throughout the day to help optimize treatment patterns [7]. Currently, CGM requires approximately two SMPG tests per day for calibration. With regard to delivering insulin, continuous subcutaneous insulin infusion (CSII) is associated with improved glycaemic control and fewer hypoglycaemic events compared with multiple daily injections [8,9]. Hypoglycaemic events can still occur, however, with CSII. Sensor–augmented pump devices, which combine CSII and CGM, have been introduced, and one of them includes the facility to suspend delivery of insulin at times of low blood glucose. The low glucose suspension feature aims to prevent severe hypoglycaemic episodes by automatically suspending insulin delivery for 2 h when the hypoglycaemic threshold is crossed based on CGM readings. Automatic resumption of insulin delivery may also help avoid hyperglycaemic rebound. The clinical benefit of this sensor-augmented pump + low glucose suspension system was demonstrated by Ly et al. [10], who reported a significant reduction in the rate of symptomatic and severe hypoglycaemia. To date, multiple randomized clinical trials have shown the clinical effectiveness and safety of sensor-augmented pump systems (with or without the low glucose suspension feature) compared with CSII alone [10,11]. In 2012, it was reported that 18.7% of people with Type 1 diabetes in Sweden were using CSII [12], and may benefit from CGM. Given Swedish recommendations, it is assumed that these patients represent a subset of the Type 1 diabetes
2
population who experience recurrent hypoglycaemic and hyperglycaemic events [13]. In the present study, we assess the cost-effectiveness and long-term clinical benefits of the sensor-augmented pump compared with CSII alone in people with Type 1 diabetes in Sweden. Estimation of the current and future economic burden of disease on healthcare systems, and how this burden is affected by different therapies, can assist decision-makers in optimizing resource allocation [4].
Patients and methods CORE Diabetes Model
The CORE Diabetes Model (see Supporting Information, File S1) simulates diabetes progression and the development of diabetes-related complications.This analysis used a second order Monte-Carlo simulation with 1000 patients and 1000 iterations with event probabilities derived from published sources. The development of complications, life expectancy, quality-adjusted life expectancy measured in quality-adjusted life years (QALYs), direct medical costs and indirect costs were modelled.
Clinical benefits
The base case used a rate of 2.6 severe hypoglycaemic events/ 100 patient years. The clinical impact of sensor-augmented pump therapy used in the present study was derived from the only patient-level meta-analysis in this therapy area published to date. The meta-analysis reported the clinical benefits of CGM compared with SMPG (e.g. as used with CSII) [8]. The analysis evaluated six randomized controlled trials, totalling 892 people, published between 2006 and 2009. CGM was associated with a mean reduction in HbA1c level compared with SMPG of 3 mmol/mol ( 0.3%, 95% CI 0.4 to 0.2%). Furthermore, for every extra day of sensor use per week, the absolute impact of CGM increased by 2 mmol/mol ( 0.15%, 95% credible interval 0.19 to 0.11%). CGM also reduced exposure to hypoglycaemia by 23% compared with SMPG. This reduction in hypoglycaemia was assessed via sensitivity analyses. A Swedish observational study of routine care reported that the daily frequency of SMPG was reduced when using a sensor-augmented pump compared with the use of CSII alone [14]. Based on these data, the base case included 4.4 SMPG tests/day for sensor-augmented pump therapy and 7.1 SMPG tests/day for CSII alone.
Simulation cohorts
The cohort definition (Table 1) for age, diabetes duration, gender proportion and HbA1c level was based on mean population characteristics from the meta-analysis [8]. The mean cohort characteristics at baseline were: age 27 years; duration of diabetes 13 years; and HbA1c 70 mmol/mol
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 1 Characteristics of the base case Cohort characteristic Men, % Age, years Duration of diabetes, years BMI, kg/m² Risk factors HbA1c mmol/mol % Systolic blood pressure, mmHg Total cholesterol, mmol/l HDL cholesterol, mmol/l LDL cholesterol, mmol/l Triglycerides, mmol/l Other* Sensor used (CGM + CSII), weeks/years Sensor used (CGM + CSII), days/week Number of SMPG measurements/day (CGM + CSII) Number of SMPG measurements/day (SMPG + CSII) Ketoacidosis event rate, episodes/100 patients-years Mean severe hypoglycaemic event rate, episodes/100 patients-years Fear of hypoglycaemic event utility Time horizon, years Discount rates of both costs and clinical outcomes, %
Mean 45.5 27 13 23.75
Reference Pickup Pickup Pickup Pickup
et et et et
al. al. al. al.
[8] [8] [8] [8]
Pickup et al. [8] 71 8.6 115 4.57 1.26 2.87 0.98
Utility values Nathan et al. [15] Nathan Nathan Nathan Nathan
et et et et
al. al. al. al.
[15] [15] [15] [15]
48 5.52 4.35
7.11
0 2.6
0.0184 70 3
*No related references. CSII, continuous subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
(8.6%). The meta-analysis did not report all patient characteristics required within the CORE Diabetes Model, and for this reason the remaining characteristics were derived from the Diabetes Control and Complications Trial secondary cohort [15], which was well matched for age, HbA1c, duration of diabetes, gender proportion, systolic blood pressure, lipids and BMI.
Costs
A Swedish societal perspective was taken, with all costs expressed in 2011 Swedish Krona (SEK). Costs refer to intervention, complication and indirect costs. Intervention costs were calculated from Swedish Pharmaceutical Benefits Board (TLV) data and are presented in Table 2. A total of 48 sensors/year were used in the base case, a CGM frequency of 5.5 days/week. Other costs (Table 2) were obtained from published and official sources and were, if required, inflated to 2011 prices
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
using the Swedish consumer price index. Swedish guidelines recommend the inclusion of ‘production loss’ resulting from illness, which should be based on the human capital approach, in the analysis of costs [16]. For this reason, indirect costs were estimated using this method and included the average annual salary and days-off work because of diabetes-related complications.
Health state utilities were obtained from published sources. If a patient experienced multiple complications, the lowest utility value was applied. Yeh et al. [17] reported that fear of hypoglycaemia was reduced with CGM compared with SMPG. Evaluating the change in the Hypoglycaemia Fear Survey scores from baseline, the mean group difference (CGM – SMPG) was found to be: 2.3 (95% CI 8.2 to 3.6), a value supported by Beck et al. [18]. Given that a 1-unit increase in the Hypoglycaemia Fear Survey score results in a 0.008-unit decrease in the Euro-Qol five-dimension health questionnaire index [19], CGM is associated with a 0.0184-unit increase (2.3 9 0.008) in this index, reflecting the patient benefit derived from reduced fear of hypoglycaemia. For myocardial infarction, hypoglycaemic, stroke and amputation events, a one-off disutility was subtracted. The disutility for a major hypoglycaemic event was applied to ketoacidosis because of a lack of specific data in the literature. In the base case, however, ketoacidosis was not considered in either arm because no studies were identified to inform the model.
Discounting and time horizon
According to the Swedish guidelines, a 3% per annum discount rate was applied to both costs and clinical outcomes [16]. As recommended in the guidelines, simulations used a patient lifetime time horizon, in which the health economic analysis continued until 100% mortality was reached.
Sensitivity analyses
One-way sensitivity analyses were used to assess the robustness of the results. The first analysis increased the frequency of CGM from 48 to 51 (5.9 days/week) or 60.8 (7 days/ week) sensors/year. The number of SMPG tests per day with use of the sensor-augmented pump varied from 7.1 (as with CSII) to 2.1 SMPG tests/day. The baseline HbA1c level varied between 55 and 75 mmol/mol (7.2 and 9.0%). The impact of severe hypoglycaemia was evaluated using data from Beck et al. [18], in which a high-risk population experienced severe hypoglycaemia at a rate of 15 events/100 patient-years (sensor-augmented pump) and 27.7 events/100 patient-years (CSII). The fear of hypoglycaemia utility ranged from 0 (no improvement) up to 0.0552, representing a 6.9-unit decrease
3
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
Table 2 Intervention, direct and indirect costs used in the analysis 2011 Cost, SEK Intervention costs Enlite sensor, pharmacy wholesale price Serter, pharmacy wholesale price Batteries, three in pack, pharmacy wholesale price Minilink, pharmacy wholesale price Lancets, 50 in pack, pharmacy sales price Test strips, 50 in pack, pharmacy sales price Direct costs Cardiovascular complications Myocardial infarction: 1st year Myocardial infarction: 2nd+ years Angina: 1st year Angina: 2nd+ years Congestive heart failure: 1st year Congestive heart failure: 2nd+ years Stroke: 1st year Stroke: 2nd+ years Stroke death within 30 days Peripheral vascular disease: 1st year Peripheral vascular disease: 2nd+ years Renal complications Haemodialysis: 1st year Haemodialysis: 2nd+ years Peritoneal dialysis: 1st year Peritoneal dialysis: 2nd+ years Renal transplant: 1st year Renal transplant: 2nd+ years Acute events Hypoglycaemia with medical assistance Minor hypoglycaemia Eye disease Laser treatment Cataract surgery Following cataract surgery Other Neuropathy Amputation Amputation prosthesis Gangrene treatment After healed ulcer Infected ulcer Standard uninfected ulcer treat Indirect costs Annual salary Men, SEK Women, SEK
2011 Cost, EUR
Reference/source
387 600 38 5,200 45 169
43 66 4 576 5 19
41,138 8,446 48,045 8,446 58,514 8,446 45,951 8,446 888 8,446 8,446
4,554 935 5,319 935 6,477 935 5,088 935 98 935 935
Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson
et et et et et et et et et et et
al. al. al. al. al. al. al. al. al. al. al.
[28] [28] [28] [28] [28] [28] [28] [28] [28] [28] [28]
520,673 520,673 520,673 520,673 285,430 48,265
57,639 57,639 57,639 57,639 31,597 5,343
Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson
et et et et et et
al. al. al. al. al. al.
[28] [28] [28] [28] [28] [28]
3,016 0
334 0
Henriksson et al. [28] Henriksson et al. [28]
6,648 16,278 0
735 1,802 0
Henriksson et al. [28] Henriksson et al. [28] Henriksson et al. [28]
38,848 95,724 19,404 25,030 0 16,741 14,566
4,300 10,597 2,148 2,771 0 1,853 1,611
Valentine et al. [29] Henriksson et al. [28] Ghatnekar et al. [30] Ghatnekar et al. [30] Assumed Ghatnekar et al. [30] Ghatnekar et al. [30]
374,400 321,600
41,446 35,601
Statistics Centre Sweden (2012)* Statistics Centre Sweden (2012)*
*Wage distribution by sector and sex in 2011, accessed June 2012 and available at http://www.scb.se/Pages/TableAndChart____149077.aspx. The Swedish Consumer Price Index was sourced from 2012 was last accessed on 3 June 2013 and is available at: http://www.scb.se/Pages/ TableAndChart____272152.aspx
in the Hypoglycaemia Fear Survey score [20]. Sensitivity to ketoacidosis was assessed using an event rate of 2.74 and 13.1 events/100 patient-years for the sensor-augmented pump and CSII, respectively (SWITCH Safety report, Medtronic data on file). As per Swedish guidelines, discount rates varied between 0 and 5% for both costs and clinical benefits. Complications costs varied by 10%. Monte Carlo simulations were used to derive the acceptability curve, showing the likelihood for the intervention to be cost-effective according to various willingness–to–pay
4
thresholds. A first-order simulation accounted for patientlevel uncertainty, whereas a second-order simulation assessed parameter-level uncertainty.
Results Life expectancy, quality-adjusted life expectancy and incidence of complications
In the base case, use of the sensor-augmented pump improved discounted life expectancy by 0.43 years compared
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 3 Summary results of the base case CGM + CSII, mean ( Discounted life expectancy, years Undiscounted life expectancy, years Quality-adjusted life expectancy, QALY Total direct costs, SEK Total combined costs (direct + indirect), SEK D direct costs/D quality-adjusted life expectancy, SEK per QALY gained D combined costs/D quality-adjusted life expectancy, SEK per QALY gained D direct costs/D quality-adjusted life expectancy, EUR per QALY gained* D combined costs/D quality-adjusted life expectancy, EUR per QALY gained*
18.54 29.7 13.05 868,897 2,872,525
( ( ( ( (
0.18) 0.40) 0.12) 13,362) 64,226)
SD)
SMPG + CSII, mean ( 18.11 28.65 12.29 453,791 2,592,563
( ( ( ( (
0.18) 0.39) 0.12) 12729) 63127)
SD)
Difference 0.42 1.03 0.76 415,106 279,962 545,005 367,571 60,332 40,690
*Mean exchange rate (1:0.1107) for 2011 used for conversion (http://www.x-rates.com). QALY, quality-adjusted life years; CGM, continuous glucose monitoring; CSII, continuous subcutaneous insulin infusion; SMPG, self-monitoring of plasma glucose.
with CSII (Table 3). Mean SD discounted life expectancy was 18.54 0.18 years with use of the sensor-augmented pump compared with 18.11 0.18 years with CSII. The quality-adjusted life expectancy associated with use of the sensor-augmented pump was 0.76 QALYs higher than it was for CSII (13.05 0.12 vs 12.29 0.12 QALYs). All diabetes-related complications occurred, on average, later with use of the sensor-augmented pump than with CSII (Fig. 1). Considering all complications, use of the sensor-augmented pump increased the mean time to complication onset by 1.15 years compared with CSII. On average, each myocardial infarction occurred 1.05 years later with the sensor-augmented pump than with CSII. Furthermore, the cumulative incidence of complications was lower in people using the sensor-augmented pump vs those receiving CSII. For microvascular complications, the cumulative incidence of end-stage renal disease was 18.2 and 20.4% (Fig. 2) and severe vision loss was 38.5 and 39.5% for the sensor-augmented pump and CSII, respectively. With regard to macrovascular complications, the cumulative incidence of myocardial infarction was 31.5 and 32.2% for the sensor-augmented pump and CSII, respectively. The cumulative incidence of stroke was, however, increased with the sensor-augmented pump compared with CSII, being 27.3 and 25.2%, respectively. Analysis indicates that the increase in stroke was probably attributable to the survival paradox (See Supporting Information File S1).
Cost-effectiveness base case
The mean SD lifetime costs (both direct and indirect) associated with the sensor-augmented pump were projected to be 2,872,525 SEK 64,226 (Table 3), compared with a value of 2,592,563SEK 63,127 for CSII. The incremental cost of the sensor-augmented pump over patient lifetimes was therefore 279,962 SEK. Given the incremental life expectancy (0.42 years) with use of the sensor-augmented ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
pump, the cost per life year gained was 666,576 SEK. Evaluating cost-effectiveness from the perspective of quality-adjusted life expectancy, the gain of 0.76 QALYs resulted in the sensor-augmented pump having an estimated incremental cost-effectiveness ratio (ICER) of 367,571 SEK per QALY gained (Table 3). The acceptability curve showed that the likelihood for the sensor-augmented pump to be cost-effective at a willingness–to–pay threshold of 500,000 SEK per QALY gained was > 70%. A breakdown of total direct costs showed that the greatest component of direct lifetime costs for the sensor-augmented pump was treatment costs (73.2%). For CSII, the cost associated with complications (51.9%) was greater than that for treatment (45.6%), probably showing the impact of reduced incidence and later onset of complications with the sensor-augmented pump compared with CSII. Total complication costs with use of sensor-augmented pump were, however, higher than with CSII, potentially because of the additional costs associated with increased life expectancy (the survival paradox) with the use of the sensor-augmented pump. Incremental complication costs for the sensor-augmented pump were 14,029 SEK over the course of the patient’s lifetime.
Sensitivity analyses
Increasing the CGM sensor use to 51 sensors/year (5.87 days/week) in the sensor-augmented pump arm resulted in a 1.06% increase in the ICER to 371,482 SEK per QALY gained. If, however, sensor usage was increased to 60.83 sensors/year (7 days/week) the ICER decreased by 3.11% to 356,132 SEK per QALY gained. Analyses showed that self-management requirements influenced the ICER. Varying the number of SMPG tests/day for the sensor-augmented pump from 7.1, though 6.1, to 2.1 resulted in ICERs of 475,467 SEK, 436,374 SEK and 280,004 SEK per QALY gained, respectively. Patient characteristics were also
5
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
FIGURE 1 Graph showing that use of the sensor-augmented pump is associated with a delay in the onset of key diabetes–related complications. CSII, continuous-subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
FIGURE 2 Graph showing that use of the sensor-augmented pump is associated with a reduced incidence of end-stage renal disease. CSII, continuous-subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
important drivers of outcomes. Altering the baseline HbA1c value from 11 mmol/mol (8.6%) to 55 mmol/mol (7.2%) and 60 mmol/mol (7.6%) increased the ICER to 593,655 SEK and 500,378 SEK per QALY gained, respectively. By
6
contrast, increasing HbA1c to 75 mmol/mol (9%) decreased the ICER to 343,637 SEK per QALY gained. Complications are known to be a major driver of diabetes-related healthcare costs. Sensitivity analysis that
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
increased the rate of severe hypoglycaemic events resulted in a reduced ICER (296,633 SEK per QALY gained). Likewise, applying a ketoacidosis event rate in both arms reduced the ICER to 45,406 SEK per QALY gained. Patient perception of complications was also found to influence the ICER. Assuming that use of the sensor-augmented pump had no impact on fear of hypoglycaemia, the ICER reached 665,742 SEK per QALY gained, whereas if the fear of hypoglycaemia utility was 0.00552 the ICER decreased to 193,892 SEK per QALY gained. Evaluating the impact of discount rates, a 0% discount decreased the ICER to 294,794 SEK per QALY gained. Conversely, increasing the discount rate to 5% increased the ICER to 418,287 SEK per QALY gained. Varying complication costs by 10% led to a 0.50% variation in the ICER. Sensitivity analysis results are shown in Fig. 3.
Discussion Previous publications based in the Swedish setting have used a cost per QALY of ~500,000 SEK (~55,000 EUR) to define the cost-effectiveness threshold [21,22]. This is in line with estimates for Swedish cost-effectiveness analyses that also considered factors such as human dignity, need and solidarity [14]. In this context, the present health economic analysis clearly shows that the sensor-augmented pump is cost-effective compared with CSII for the treatment of Type 1 diabetes in Sweden. Treatment with the sensor-augmented pump was associated with gains in undiscounted life expectancy (+1.03 years) and quality-adjusted life expectancy (+0.76
DIABETICMedicine
QALYs) compared with treatment with CSII. Despite their increased life expectancy, people using the sensor-augmented pump experienced fewer diabetes-related complications and incurred lower complication costs: an incremental difference compared with CSII of 14,029 SEK per patient. Overall, the incremental lifetime cost of the sensor-augmented pump was 279,962 SEK, providing an estimated ICER of 367,571 SEK per QALY gained for the sensor-augmented pump compared with CSII. Assuming a willingness–to-pay threshold of 500,000 SEK, the ICER represents good value for money in the Swedish setting. The findings of the present study are in line with previous estimates made with regard to the US setting. McQueen et al. [23] estimated that, compared with SMPG, the use of CGM with intensive insulin treatment for the treatment of Type 1 diabetes resulted in improvement in effectiveness (0.52 QALYs gained) and a cost-effective ICER (~32,500 EUR per QALY gained). Similarly, Huang et al. [24] estimated that the ICER of CGM vs standard glucose monitoring in Type 1 diabetes was ~56,800 EUR per QALY gained. The meta-analysis by Pickup et al. [8] showed that CGM, when compared with SMPG, has a beneficial impact on HbA1c and the authors concluded that CGM was likely to be cost-effective in the treatment of Type 1 diabetes, particularly in those with poor glycaemic control. We used the CORE Diabetes Model in the present analysis to project the impact of interventions on the progression of diabetes. The high number of previous analyses using the CORE Diabetes Model for various types of interventions, country settings and perspectives, indicates its application
FIGURE 3 Incremental cost-effectiveness ratio (ICER) difference between the sensitivity analyses and the base case (SEK/quality-adjusted life year). SMPG: self-monitoring of plasma glucose.
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
7
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
and reliability through external validation. As with all health economic models, however, the CORE Diabetes Model uses short-term data to project long-term outcomes. As previously reported, data used to design the CORE Diabetes Model are primarily from clinical studies and few real-life factors, such as compliance, effectiveness and treatment drop out, were considered [9]. To minimize the potential limitations in the present study, an extensive literature search was performed to inform the modelling process. In particular, the patient population was taken from literature specific to CGM usage, estimates of sensor usage were used as a proxy for treatment compliance and a fear of hypoglycaemia utility was incorporated to extend the application of the CORE Diabetes Model to CGM. It is assumed that the analysis in the present study represents an appropriate estimate of the Swedish setting. The influence of assumptions used in the model was tested and model outcomes were found to be robust under a wide range of plausible assumptions. The results were most sensitive to variation in the fear of hypoglycaemia utility, whereby the ICER ranged from 193,892 to 665,742 SEK per QALY gained. Also important was the HbA1c baseline value and/or its associated reduction; variation in these values resulted in an ICER in the range of 343,637 to 593,655 SEK per QALY gained. The base–case hypothesis regarding the fear of hypoglycaemia is conservative when compared with the findings of Norgaard et al. [20], who reported a much greater reduction in fear of hypoglycaemia using CGM than did Yeh et al. [17], whose reported value was used in the base case. Hypoglycaemia is an important complication associated with diabetes and the use of insulin products. In addition to its impact on patient quality of life, hypoglycaemia is a major driver of costs. In a recent study in Germany, France and the UK, 10% of people with insulin–treated diabetes missed ≥ 1 day of work in the previous year as a result of hypoglycaemia [25]. In the assessment of non-severe hypoglycaemia, it was determined that each event resulted in 8.3–15.9 h of lost productivity per month [26]. Even if a non-severe hypoglycaemic event occurred outside of work time, 22.7% of participants were late to or did not attend work [26]. A prospective, 12-month study in the USA, however, found that mild hypoglycaemia resulted in missing 5–30 min of work [27], indicating that, if diabetes is well controlled, hypoglycaemia is not detrimental to patient productivity. When asked about self-management, 80% of respondents reported that they would value an indicator warning if their plasma glucose was getting high/low [25]. Hypoglycaemia has a range of consequences both in terms of economic cost to society and the healthcare systems as well as patient glycaemic control, safety and treatment satisfaction. The low glucose suspension function has been shown to substantially reduce the rate of hypoglycaemia, and this may translate into improved patient satisfaction and reduced healthcare costs.
8
The inclusion of low glucose suspension in the present analysis may reduce the ICER still further. Assumptions made in the model did not alter the health economic conclusion of the study. Compared with CSII, the sensor-augmented pump provided clinical benefit at an ICER below the cost–effectiveness threshold. The sensor-augmented pump is, therefore, expected to be a beneficial and cost–effective option for the treatment of Type 1 diabetes in Sweden. In conclusion, the sensor-augmented pump represents a cost–effective option in the treatment of Type 1 diabetes in Sweden compared with CSII alone. Use of the sensor-augmented pump is associated with substantial clinical benefits and notably increases the time to onset of all major diabetes-related complications considered in this analysis.
Funding sources
This study was supported by an unrestricted grant from Medtronic.
Competing interests
S.R. and R.S. are employees of health economics consultancies that received fees for participation in this research. A.S.B., S.D.P. and N.P. are full-time employees of Medtronic or its affiliates. N.P. holds Medtronic stock options. J.J. has received honoraria for work undertaken with Medtronic, but no remuneration was made with respect to this research project or the manuscript development.
References 1 Chevreul K, Berg Brigham K, Bouche C. The burden and treatment of diabetes in France. Global Health 2014; 10: 6. 2 Hill CJ, Cardwell CR, Patterson CC, Maxwell AP, Magee GM, Young RJ et al. Chronic kidney disease and diabetes in the National Health Service: a cross-sectional survey of the U.K. National Diabetes Audit. Diabet Med 2014; 31: 448–454. 3 Ward A, Alvarez P, Vo L, Martin S. Direct medical costs of complications of diabetes in the United States: estimates for event-year and annual state costs (USD 2012). J Med Econ 2014; 17: 176–183. 4 Zhang P, Zhang X, Brown J, Vistisen D, Sicree R, Shaw J et al. Global healthcare expenditure on diabetes for 2010 and 2030. Diabetes Res Clin Pract 2010; 87: 293–301. 5 Swedish National Diabetes Register. Swedish National Diabetes Register Annual Report 2012. Centre of Registers Region V€ astra G€ otaland, Swedish National Diabetes Register, 413 45 Gothenburg, Sweden. 2012. 6 Deiss D, Bolinder J, Riveline J-P, Battelino T, Bosi E, Tubiana-Rufi N et al. Improved glycemic control in poorly controlled patients with Type 1 diabetes using real-time continuous glucose monitoring. Diabetes Care 2006; 29: 2730–2732. 7 American Diabetes Association. Executive summary: Standards of medical care in diabetes–2013. Diabetes Care 2013; 36(Suppl. 1): S4–S10. 8 Pickup JC, Freeman SC, Sutton AJ. Glycaemic control in type 1 diabetes during real time continuous glucose monitoring compared
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
9
10
11
12
13
14
15
16
17
18
19
20
with self monitoring of blood glucose: meta-analysis of randomised controlled trials using individual patient data. BMJ 2011; 343: d3805. Roze S, Valentine WJ, Zakrzewska KE, Palmer AJ. Health-economic comparison of continuous subcutaneous insulin infusion with multiple daily injection for the treatment of Type 1 diabetes in the UK. Diabet Med 2005; 22: 1239–1245. Ly TT, Nicholas JA, Retterath A, Lim EM, Davis EA, Jones TW. Effect of sensor-augmented insulin pump therapy and automated insulin suspension vs standard insulin pump therapy on hypoglycemia in patients with type 1 diabetes: a randomized clinical trial. JAMA 2013; 310: 1240–1247. Garg SK, Voelmle MK, Beatson CR, Miller HA, Crew LB, Freson BJ et al. Use of continuous glucose monitoring in subjects with Type 1 diabetes on multiple daily injections versus continuous subcutaneous insulin infusion therapy: a prospective 6-month study. Diabetes Care 2011; 34: 574–579. Carlsson BM, Andersson PN, Alnervik J, Carstensen J, Lind M. Availability of insulin pump therapy in clinical practice. Diabet Med 2012; 29: 1055–1059. Swedish National Board of Health and Welfare. Swedish Guidelines for Diabetes Care 2010 [In Swedish: Nationella riktlinjer f€ or diabetesv arden 2010: St€ od f€ or styrning och ledning]. Stockholm: Socialstyrelsen, 2010. Persson U, Glenngard AH, Hjortsberg C. Estimating the willingness to pay for a QALY in Sweden - a pilot study. In: Scalone L, Mantovani LG (eds), 25th Meeting of the EuroQol Group, 2010, pp. 121–132. Nathan DM, Cleary PA, Backlund J-YC, Genuth SM, Lachin JM, Orchard TJ et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med 2005; 353: 2643–2653. The Pharmaceutical Benefits Board. General guidelines for economic evaluations from the Pharmaceutical Benefits Board. The Dental and Pharmaceutical Benefits Agency, TLV, 2003. Available at http://www.tlv.se. Last accessed 30 September 2014. Yeh H-C, Brown TT, Maruthur N, Ranasinghe P, Berger Z, Suh YD et al. Comparative effectiveness and safety of methods of insulin delivery and glucose monitoring for diabetes mellitus: a systematic review and meta-analysis. Ann Intern Med 2012; 157: 336–347. Beck RW, Lawrence JM, Laffel L, Wysocki T, Xing D, Huang ES et al. Quality-of-life measures in children and adults with type 1 diabetes: Juvenile Diabetes Research Foundation Continuous Glucose Monitoring randomized trial. Diabetes Care 2010; 33: 2175–2177. Currie CJ, Morgan CL, Poole CD, Sharplin P, Lammert M, McEwan P. Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes. Curr Med Res Opin 2006; 22: 1523–1534. Norgaard K, Scaramuzza A, Bratina N, Lalic NM, Jarosz-Chobot P, Kocsis G et al. Routine sensor-augmented pump therapy in type
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
21
22
23
24
25
26
27
28
29
30
1 diabetes: the INTERPRET study. Diabetes Technol Ther 2013; 15: 273–280. Persson U, Hjelmgren J. H€ also-och sjukv arden beh€ over kunskap om hur befolkningen v€ arderar h€ alsan [Article in Swedish: Health services need knowledge about how people value health]. L€ akartidningen 2003; 100: 3436–3437. Socialstyrelsen. Nationella riktlinjer f€ or prostatacancersjukv ard Medicinskt och h€ alsoekonomiskt faktadokument [Report in Swedish: National guidelines for prostate health - Medical and health economic facts document]. Stockholm, Sweden: Socialstyrelsen, 2007. McQueen RB, Ellis SL, Campbell JD, Nair KV, Sullivan PW. Cost-effectiveness of continuous glucose monitoring and intensive insulin therapy for type 1 diabetes. Cost Eff Resour Alloc 2011; 9: 13. Huang ES, O’Grady M, Basu A, Winn A, John P, Lee J et al. The cost-effectiveness of continuous glucose monitoring in type 1 diabetes. Diabetes Care 2010; 33: 1269–1274. Willis WD, Diago-Cabezudo JI, Madec-Hily A, Aslam A. Medical resource use, disturbance of daily life and burden of hypoglycemia in insulin-treated patients with diabetes: results from a European online survey. Expert Rev Pharmacoecon Outcomes Res 2013; 13: 123–130. Brod M, Christensen T, Thomsen TL, Bushnell DM. The impact of non-severe hypoglycemic events on work productivity and diabetes management. Value Health 2011; 14: 665–671. Leckie AM, Graham MK, Grant JB, Ritchie PJ, Frier BM. Frequency, severity, and morbidity of hypoglycemia occurring in the workplace in people with insulin-treated diabetes. Diabetes Care 2005; 28: 1333–1338. Henriksson F. Applications of economic models in healthcare: the introduction of pioglitazone in Sweden. Pharmacoeconomics 2002; 20(Suppl. 1): 43–53. Valentine WJ, Aagren M, Haglund M, Ericsson A, Gschwend MH. Evaluation of the long-term cost-effectiveness of insulin detemir compared with neutral protamine hagedorn insulin in patients with type 1 diabetes using a basal-bolus regimen in Sweden. Scand J Pub Heatlh 2011; 39: 79–87. Ghatnekar O, Persson U, Willis M, Odegaard K. The cost effectiveness of Becaplermin (REGRANEX) in the treatment of diabetic foot ulcers in four European countries. Pharmacoeconomics 2000; 19: 767–778.
Supporting Information Additional Supporting Information may be found in the online version of this article: File S1. Extended methods and results.
9
Research: Health Economics Health-economic analysis of real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze1, R. Saunders2, A.-S. Brandt3, S. de Portu4, N. L. Papo4 and J. Jendle5,6 1 HEVA HEOR, Lyon, France, 2Ossian Health Economics and Communications, Basel, 3Medtronic Danmark A/S, Copenhagen, Denmark, 4Medtronic International € € University, Orebro, Trading Sarl, Tolochenaz, 5Endocrine and Diabetes Centre, Karlstad Hospital, Karlstad, and 6Faculty of Health Sciences and Medicine, Orebro Sweden
Accepted 2 December 2014
Abstract Aim To evaluate the clinical benefits and cost-effectiveness of the sensor-augmented pump compared with self– monitoring of plasma glucose plus continuous subcutaneous insulin infusion in people with Type 1 diabetes. Methods The CORE Diabetes Model was used to simulate disease progression in a cohort of people with baseline characteristics taken from a published meta-analysis. Direct and indirect costs for 2010–2011 were calculated from a societal payer perspective, with cost-effectiveness calculated over the patient’s lifetime. Discount rates of 3% per annum were applied to the costs and the clinical outcomes.
Use of the sensor-augmented pump was associated with an increase in mean discounted, quality–adjusted life expectancy of 0.76 quality–adjusted life years compared with continuous subcutaneous insulin infusion (13.05 0.12 quality–adjusted life years vs 12.29 0.12 quality–adjusted life years, respectively). Undiscounted life expectancy increased by 1.03 years for the sensor-augmented pump compared with continuous subcutaneous insulin infusion. In addition, the onset of complications was delayed (by a mean of 1.15 years) with use of the sensor-augmented pump. This analysis resulted in an incremental cost–effectiveness ratio of 367,571 SEK per quality–adjusted life year gained with the sensor-augmented pump. The additional treatment costs related to the use of the sensor-augmented pump were partially offset by the savings attributable to the reduction in diabetes-related complications and the lower frequency of self-monitoring of plasma glucose.
Results
Conclusions Analysis using the CORE Diabetes Model showed that improvements in glycaemic control associated with sensor-augmented pump use led to a reduced incidence of diabetes-related complications and a longer life expectancy. Use of the sensor-augmented pump was associated with an incremental cost–effectiveness ratio of 367,571 SEK per quality–adjusted life year gained, which is likely to represent good value for money in the treatment of Type 1 diabetes in Sweden.
Diabet. Med. 00, 000–000 (2015)
Introduction Type 1 diabetes is characterized by deficient insulin production, but its cause remains unknown and the condition is not preventable. Treatment of Type 1 diabetes requires daily administration of insulin, which aims to normalize glucose metabolism. Diabetes is associated with an increased risk of numerous serious complications and comorbidities, although the risk is reduced if diabetes is well controlled. Two diabetes-related comorbidities with an important impact on life expectancy are cardiovascular disease and renal failure [1,2]. In addition to their quality-of-life implications, diabeCorrespondence to: Stephane Roze. E-mail: [email protected]
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
tes-related complications have a significant cost impact on healthcare provision [1,3]. Diabetes imposes an increasing economic burden on the Swedish healthcare system [4]. The Type 1 diabetes incidence rate in Sweden has increased rapidly since the 1990s and in 2012 the Swedish National Diabetes Register reported that there were 43 393 people with Type 1 diabetes [5]. Although it has a high Type 1 diabetes burden, Sweden was ranked in first place for diabetes treatment at the recent European Association for the Study of Diabetes 2014 congress. For this reason, and the availability of comprehensive Type 1 diabetes data, evaluation of diabetes practices in Sweden offers the opportunity to inform other healthcare systems and may provide a framework for further analyses in other settings.
1
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
What’s new? • This study builds on a recent meta-analysis to provide further insights into the clinical and safety aspects of real-time continuous glucose monitoring. • Simulations of disease progression show how these aspects translate into long-term patient benefits and healthcare costs. • Simulations indicate that continuous glucose monitoring leads to later onset and reduced incidence of acute and long-term diabetes-related complications. • The findings may have an important influence on informing treatment choices with respect to providing value both for patients and healthcare payers. Active self-management of diabetes can help people to achieve glycaemic control and reduce the risk of complications. An important aspect of self-management is self-monitoring of plasma glucose (SMPG), although advantages may be gained through the use of continuous glucose monitoring (CGM). In 2006, Deiss et al. [6] reported the clinical benefit of CGM compared with conventional SMPG for the reduction of HbA1c levels. Additionally, CGM can determine the frequency and severity of hypoglycaemic episodes and provides maximum information about changing blood glucose levels throughout the day to help optimize treatment patterns [7]. Currently, CGM requires approximately two SMPG tests per day for calibration. With regard to delivering insulin, continuous subcutaneous insulin infusion (CSII) is associated with improved glycaemic control and fewer hypoglycaemic events compared with multiple daily injections [8,9]. Hypoglycaemic events can still occur, however, with CSII. Sensor–augmented pump devices, which combine CSII and CGM, have been introduced, and one of them includes the facility to suspend delivery of insulin at times of low blood glucose. The low glucose suspension feature aims to prevent severe hypoglycaemic episodes by automatically suspending insulin delivery for 2 h when the hypoglycaemic threshold is crossed based on CGM readings. Automatic resumption of insulin delivery may also help avoid hyperglycaemic rebound. The clinical benefit of this sensor-augmented pump + low glucose suspension system was demonstrated by Ly et al. [10], who reported a significant reduction in the rate of symptomatic and severe hypoglycaemia. To date, multiple randomized clinical trials have shown the clinical effectiveness and safety of sensor-augmented pump systems (with or without the low glucose suspension feature) compared with CSII alone [10,11]. In 2012, it was reported that 18.7% of people with Type 1 diabetes in Sweden were using CSII [12], and may benefit from CGM. Given Swedish recommendations, it is assumed that these patients represent a subset of the Type 1 diabetes
2
population who experience recurrent hypoglycaemic and hyperglycaemic events [13]. In the present study, we assess the cost-effectiveness and long-term clinical benefits of the sensor-augmented pump compared with CSII alone in people with Type 1 diabetes in Sweden. Estimation of the current and future economic burden of disease on healthcare systems, and how this burden is affected by different therapies, can assist decision-makers in optimizing resource allocation [4].
Patients and methods CORE Diabetes Model
The CORE Diabetes Model (see Supporting Information, File S1) simulates diabetes progression and the development of diabetes-related complications.This analysis used a second order Monte-Carlo simulation with 1000 patients and 1000 iterations with event probabilities derived from published sources. The development of complications, life expectancy, quality-adjusted life expectancy measured in quality-adjusted life years (QALYs), direct medical costs and indirect costs were modelled.
Clinical benefits
The base case used a rate of 2.6 severe hypoglycaemic events/ 100 patient years. The clinical impact of sensor-augmented pump therapy used in the present study was derived from the only patient-level meta-analysis in this therapy area published to date. The meta-analysis reported the clinical benefits of CGM compared with SMPG (e.g. as used with CSII) [8]. The analysis evaluated six randomized controlled trials, totalling 892 people, published between 2006 and 2009. CGM was associated with a mean reduction in HbA1c level compared with SMPG of 3 mmol/mol ( 0.3%, 95% CI 0.4 to 0.2%). Furthermore, for every extra day of sensor use per week, the absolute impact of CGM increased by 2 mmol/mol ( 0.15%, 95% credible interval 0.19 to 0.11%). CGM also reduced exposure to hypoglycaemia by 23% compared with SMPG. This reduction in hypoglycaemia was assessed via sensitivity analyses. A Swedish observational study of routine care reported that the daily frequency of SMPG was reduced when using a sensor-augmented pump compared with the use of CSII alone [14]. Based on these data, the base case included 4.4 SMPG tests/day for sensor-augmented pump therapy and 7.1 SMPG tests/day for CSII alone.
Simulation cohorts
The cohort definition (Table 1) for age, diabetes duration, gender proportion and HbA1c level was based on mean population characteristics from the meta-analysis [8]. The mean cohort characteristics at baseline were: age 27 years; duration of diabetes 13 years; and HbA1c 70 mmol/mol
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 1 Characteristics of the base case Cohort characteristic Men, % Age, years Duration of diabetes, years BMI, kg/m² Risk factors HbA1c mmol/mol % Systolic blood pressure, mmHg Total cholesterol, mmol/l HDL cholesterol, mmol/l LDL cholesterol, mmol/l Triglycerides, mmol/l Other* Sensor used (CGM + CSII), weeks/years Sensor used (CGM + CSII), days/week Number of SMPG measurements/day (CGM + CSII) Number of SMPG measurements/day (SMPG + CSII) Ketoacidosis event rate, episodes/100 patients-years Mean severe hypoglycaemic event rate, episodes/100 patients-years Fear of hypoglycaemic event utility Time horizon, years Discount rates of both costs and clinical outcomes, %
Mean 45.5 27 13 23.75
Reference Pickup Pickup Pickup Pickup
et et et et
al. al. al. al.
[8] [8] [8] [8]
Pickup et al. [8] 71 8.6 115 4.57 1.26 2.87 0.98
Utility values Nathan et al. [15] Nathan Nathan Nathan Nathan
et et et et
al. al. al. al.
[15] [15] [15] [15]
48 5.52 4.35
7.11
0 2.6
0.0184 70 3
*No related references. CSII, continuous subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
(8.6%). The meta-analysis did not report all patient characteristics required within the CORE Diabetes Model, and for this reason the remaining characteristics were derived from the Diabetes Control and Complications Trial secondary cohort [15], which was well matched for age, HbA1c, duration of diabetes, gender proportion, systolic blood pressure, lipids and BMI.
Costs
A Swedish societal perspective was taken, with all costs expressed in 2011 Swedish Krona (SEK). Costs refer to intervention, complication and indirect costs. Intervention costs were calculated from Swedish Pharmaceutical Benefits Board (TLV) data and are presented in Table 2. A total of 48 sensors/year were used in the base case, a CGM frequency of 5.5 days/week. Other costs (Table 2) were obtained from published and official sources and were, if required, inflated to 2011 prices
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
using the Swedish consumer price index. Swedish guidelines recommend the inclusion of ‘production loss’ resulting from illness, which should be based on the human capital approach, in the analysis of costs [16]. For this reason, indirect costs were estimated using this method and included the average annual salary and days-off work because of diabetes-related complications.
Health state utilities were obtained from published sources. If a patient experienced multiple complications, the lowest utility value was applied. Yeh et al. [17] reported that fear of hypoglycaemia was reduced with CGM compared with SMPG. Evaluating the change in the Hypoglycaemia Fear Survey scores from baseline, the mean group difference (CGM – SMPG) was found to be: 2.3 (95% CI 8.2 to 3.6), a value supported by Beck et al. [18]. Given that a 1-unit increase in the Hypoglycaemia Fear Survey score results in a 0.008-unit decrease in the Euro-Qol five-dimension health questionnaire index [19], CGM is associated with a 0.0184-unit increase (2.3 9 0.008) in this index, reflecting the patient benefit derived from reduced fear of hypoglycaemia. For myocardial infarction, hypoglycaemic, stroke and amputation events, a one-off disutility was subtracted. The disutility for a major hypoglycaemic event was applied to ketoacidosis because of a lack of specific data in the literature. In the base case, however, ketoacidosis was not considered in either arm because no studies were identified to inform the model.
Discounting and time horizon
According to the Swedish guidelines, a 3% per annum discount rate was applied to both costs and clinical outcomes [16]. As recommended in the guidelines, simulations used a patient lifetime time horizon, in which the health economic analysis continued until 100% mortality was reached.
Sensitivity analyses
One-way sensitivity analyses were used to assess the robustness of the results. The first analysis increased the frequency of CGM from 48 to 51 (5.9 days/week) or 60.8 (7 days/ week) sensors/year. The number of SMPG tests per day with use of the sensor-augmented pump varied from 7.1 (as with CSII) to 2.1 SMPG tests/day. The baseline HbA1c level varied between 55 and 75 mmol/mol (7.2 and 9.0%). The impact of severe hypoglycaemia was evaluated using data from Beck et al. [18], in which a high-risk population experienced severe hypoglycaemia at a rate of 15 events/100 patient-years (sensor-augmented pump) and 27.7 events/100 patient-years (CSII). The fear of hypoglycaemia utility ranged from 0 (no improvement) up to 0.0552, representing a 6.9-unit decrease
3
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
Table 2 Intervention, direct and indirect costs used in the analysis 2011 Cost, SEK Intervention costs Enlite sensor, pharmacy wholesale price Serter, pharmacy wholesale price Batteries, three in pack, pharmacy wholesale price Minilink, pharmacy wholesale price Lancets, 50 in pack, pharmacy sales price Test strips, 50 in pack, pharmacy sales price Direct costs Cardiovascular complications Myocardial infarction: 1st year Myocardial infarction: 2nd+ years Angina: 1st year Angina: 2nd+ years Congestive heart failure: 1st year Congestive heart failure: 2nd+ years Stroke: 1st year Stroke: 2nd+ years Stroke death within 30 days Peripheral vascular disease: 1st year Peripheral vascular disease: 2nd+ years Renal complications Haemodialysis: 1st year Haemodialysis: 2nd+ years Peritoneal dialysis: 1st year Peritoneal dialysis: 2nd+ years Renal transplant: 1st year Renal transplant: 2nd+ years Acute events Hypoglycaemia with medical assistance Minor hypoglycaemia Eye disease Laser treatment Cataract surgery Following cataract surgery Other Neuropathy Amputation Amputation prosthesis Gangrene treatment After healed ulcer Infected ulcer Standard uninfected ulcer treat Indirect costs Annual salary Men, SEK Women, SEK
2011 Cost, EUR
Reference/source
387 600 38 5,200 45 169
43 66 4 576 5 19
41,138 8,446 48,045 8,446 58,514 8,446 45,951 8,446 888 8,446 8,446
4,554 935 5,319 935 6,477 935 5,088 935 98 935 935
Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson
et et et et et et et et et et et
al. al. al. al. al. al. al. al. al. al. al.
[28] [28] [28] [28] [28] [28] [28] [28] [28] [28] [28]
520,673 520,673 520,673 520,673 285,430 48,265
57,639 57,639 57,639 57,639 31,597 5,343
Henriksson Henriksson Henriksson Henriksson Henriksson Henriksson
et et et et et et
al. al. al. al. al. al.
[28] [28] [28] [28] [28] [28]
3,016 0
334 0
Henriksson et al. [28] Henriksson et al. [28]
6,648 16,278 0
735 1,802 0
Henriksson et al. [28] Henriksson et al. [28] Henriksson et al. [28]
38,848 95,724 19,404 25,030 0 16,741 14,566
4,300 10,597 2,148 2,771 0 1,853 1,611
Valentine et al. [29] Henriksson et al. [28] Ghatnekar et al. [30] Ghatnekar et al. [30] Assumed Ghatnekar et al. [30] Ghatnekar et al. [30]
374,400 321,600
41,446 35,601
Statistics Centre Sweden (2012)* Statistics Centre Sweden (2012)*
*Wage distribution by sector and sex in 2011, accessed June 2012 and available at http://www.scb.se/Pages/TableAndChart____149077.aspx. The Swedish Consumer Price Index was sourced from 2012 was last accessed on 3 June 2013 and is available at: http://www.scb.se/Pages/ TableAndChart____272152.aspx
in the Hypoglycaemia Fear Survey score [20]. Sensitivity to ketoacidosis was assessed using an event rate of 2.74 and 13.1 events/100 patient-years for the sensor-augmented pump and CSII, respectively (SWITCH Safety report, Medtronic data on file). As per Swedish guidelines, discount rates varied between 0 and 5% for both costs and clinical benefits. Complications costs varied by 10%. Monte Carlo simulations were used to derive the acceptability curve, showing the likelihood for the intervention to be cost-effective according to various willingness–to–pay
4
thresholds. A first-order simulation accounted for patientlevel uncertainty, whereas a second-order simulation assessed parameter-level uncertainty.
Results Life expectancy, quality-adjusted life expectancy and incidence of complications
In the base case, use of the sensor-augmented pump improved discounted life expectancy by 0.43 years compared
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 3 Summary results of the base case CGM + CSII, mean ( Discounted life expectancy, years Undiscounted life expectancy, years Quality-adjusted life expectancy, QALY Total direct costs, SEK Total combined costs (direct + indirect), SEK D direct costs/D quality-adjusted life expectancy, SEK per QALY gained D combined costs/D quality-adjusted life expectancy, SEK per QALY gained D direct costs/D quality-adjusted life expectancy, EUR per QALY gained* D combined costs/D quality-adjusted life expectancy, EUR per QALY gained*
18.54 29.7 13.05 868,897 2,872,525
( ( ( ( (
0.18) 0.40) 0.12) 13,362) 64,226)
SD)
SMPG + CSII, mean ( 18.11 28.65 12.29 453,791 2,592,563
( ( ( ( (
0.18) 0.39) 0.12) 12729) 63127)
SD)
Difference 0.42 1.03 0.76 415,106 279,962 545,005 367,571 60,332 40,690
*Mean exchange rate (1:0.1107) for 2011 used for conversion (http://www.x-rates.com). QALY, quality-adjusted life years; CGM, continuous glucose monitoring; CSII, continuous subcutaneous insulin infusion; SMPG, self-monitoring of plasma glucose.
with CSII (Table 3). Mean SD discounted life expectancy was 18.54 0.18 years with use of the sensor-augmented pump compared with 18.11 0.18 years with CSII. The quality-adjusted life expectancy associated with use of the sensor-augmented pump was 0.76 QALYs higher than it was for CSII (13.05 0.12 vs 12.29 0.12 QALYs). All diabetes-related complications occurred, on average, later with use of the sensor-augmented pump than with CSII (Fig. 1). Considering all complications, use of the sensor-augmented pump increased the mean time to complication onset by 1.15 years compared with CSII. On average, each myocardial infarction occurred 1.05 years later with the sensor-augmented pump than with CSII. Furthermore, the cumulative incidence of complications was lower in people using the sensor-augmented pump vs those receiving CSII. For microvascular complications, the cumulative incidence of end-stage renal disease was 18.2 and 20.4% (Fig. 2) and severe vision loss was 38.5 and 39.5% for the sensor-augmented pump and CSII, respectively. With regard to macrovascular complications, the cumulative incidence of myocardial infarction was 31.5 and 32.2% for the sensor-augmented pump and CSII, respectively. The cumulative incidence of stroke was, however, increased with the sensor-augmented pump compared with CSII, being 27.3 and 25.2%, respectively. Analysis indicates that the increase in stroke was probably attributable to the survival paradox (See Supporting Information File S1).
Cost-effectiveness base case
The mean SD lifetime costs (both direct and indirect) associated with the sensor-augmented pump were projected to be 2,872,525 SEK 64,226 (Table 3), compared with a value of 2,592,563SEK 63,127 for CSII. The incremental cost of the sensor-augmented pump over patient lifetimes was therefore 279,962 SEK. Given the incremental life expectancy (0.42 years) with use of the sensor-augmented ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
pump, the cost per life year gained was 666,576 SEK. Evaluating cost-effectiveness from the perspective of quality-adjusted life expectancy, the gain of 0.76 QALYs resulted in the sensor-augmented pump having an estimated incremental cost-effectiveness ratio (ICER) of 367,571 SEK per QALY gained (Table 3). The acceptability curve showed that the likelihood for the sensor-augmented pump to be cost-effective at a willingness–to–pay threshold of 500,000 SEK per QALY gained was > 70%. A breakdown of total direct costs showed that the greatest component of direct lifetime costs for the sensor-augmented pump was treatment costs (73.2%). For CSII, the cost associated with complications (51.9%) was greater than that for treatment (45.6%), probably showing the impact of reduced incidence and later onset of complications with the sensor-augmented pump compared with CSII. Total complication costs with use of sensor-augmented pump were, however, higher than with CSII, potentially because of the additional costs associated with increased life expectancy (the survival paradox) with the use of the sensor-augmented pump. Incremental complication costs for the sensor-augmented pump were 14,029 SEK over the course of the patient’s lifetime.
Sensitivity analyses
Increasing the CGM sensor use to 51 sensors/year (5.87 days/week) in the sensor-augmented pump arm resulted in a 1.06% increase in the ICER to 371,482 SEK per QALY gained. If, however, sensor usage was increased to 60.83 sensors/year (7 days/week) the ICER decreased by 3.11% to 356,132 SEK per QALY gained. Analyses showed that self-management requirements influenced the ICER. Varying the number of SMPG tests/day for the sensor-augmented pump from 7.1, though 6.1, to 2.1 resulted in ICERs of 475,467 SEK, 436,374 SEK and 280,004 SEK per QALY gained, respectively. Patient characteristics were also
5
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
FIGURE 1 Graph showing that use of the sensor-augmented pump is associated with a delay in the onset of key diabetes–related complications. CSII, continuous-subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
FIGURE 2 Graph showing that use of the sensor-augmented pump is associated with a reduced incidence of end-stage renal disease. CSII, continuous-subcutaneous insulin infusion; CGM, continuous glucose monitoring; SMPG, self-monitoring of plasma glucose.
important drivers of outcomes. Altering the baseline HbA1c value from 11 mmol/mol (8.6%) to 55 mmol/mol (7.2%) and 60 mmol/mol (7.6%) increased the ICER to 593,655 SEK and 500,378 SEK per QALY gained, respectively. By
6
contrast, increasing HbA1c to 75 mmol/mol (9%) decreased the ICER to 343,637 SEK per QALY gained. Complications are known to be a major driver of diabetes-related healthcare costs. Sensitivity analysis that
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
increased the rate of severe hypoglycaemic events resulted in a reduced ICER (296,633 SEK per QALY gained). Likewise, applying a ketoacidosis event rate in both arms reduced the ICER to 45,406 SEK per QALY gained. Patient perception of complications was also found to influence the ICER. Assuming that use of the sensor-augmented pump had no impact on fear of hypoglycaemia, the ICER reached 665,742 SEK per QALY gained, whereas if the fear of hypoglycaemia utility was 0.00552 the ICER decreased to 193,892 SEK per QALY gained. Evaluating the impact of discount rates, a 0% discount decreased the ICER to 294,794 SEK per QALY gained. Conversely, increasing the discount rate to 5% increased the ICER to 418,287 SEK per QALY gained. Varying complication costs by 10% led to a 0.50% variation in the ICER. Sensitivity analysis results are shown in Fig. 3.
Discussion Previous publications based in the Swedish setting have used a cost per QALY of ~500,000 SEK (~55,000 EUR) to define the cost-effectiveness threshold [21,22]. This is in line with estimates for Swedish cost-effectiveness analyses that also considered factors such as human dignity, need and solidarity [14]. In this context, the present health economic analysis clearly shows that the sensor-augmented pump is cost-effective compared with CSII for the treatment of Type 1 diabetes in Sweden. Treatment with the sensor-augmented pump was associated with gains in undiscounted life expectancy (+1.03 years) and quality-adjusted life expectancy (+0.76
DIABETICMedicine
QALYs) compared with treatment with CSII. Despite their increased life expectancy, people using the sensor-augmented pump experienced fewer diabetes-related complications and incurred lower complication costs: an incremental difference compared with CSII of 14,029 SEK per patient. Overall, the incremental lifetime cost of the sensor-augmented pump was 279,962 SEK, providing an estimated ICER of 367,571 SEK per QALY gained for the sensor-augmented pump compared with CSII. Assuming a willingness–to-pay threshold of 500,000 SEK, the ICER represents good value for money in the Swedish setting. The findings of the present study are in line with previous estimates made with regard to the US setting. McQueen et al. [23] estimated that, compared with SMPG, the use of CGM with intensive insulin treatment for the treatment of Type 1 diabetes resulted in improvement in effectiveness (0.52 QALYs gained) and a cost-effective ICER (~32,500 EUR per QALY gained). Similarly, Huang et al. [24] estimated that the ICER of CGM vs standard glucose monitoring in Type 1 diabetes was ~56,800 EUR per QALY gained. The meta-analysis by Pickup et al. [8] showed that CGM, when compared with SMPG, has a beneficial impact on HbA1c and the authors concluded that CGM was likely to be cost-effective in the treatment of Type 1 diabetes, particularly in those with poor glycaemic control. We used the CORE Diabetes Model in the present analysis to project the impact of interventions on the progression of diabetes. The high number of previous analyses using the CORE Diabetes Model for various types of interventions, country settings and perspectives, indicates its application
FIGURE 3 Incremental cost-effectiveness ratio (ICER) difference between the sensitivity analyses and the base case (SEK/quality-adjusted life year). SMPG: self-monitoring of plasma glucose.
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
7
DIABETICMedicine
Real-time continuous glucose monitoring in people with Type 1 diabetes S. Roze et al.
and reliability through external validation. As with all health economic models, however, the CORE Diabetes Model uses short-term data to project long-term outcomes. As previously reported, data used to design the CORE Diabetes Model are primarily from clinical studies and few real-life factors, such as compliance, effectiveness and treatment drop out, were considered [9]. To minimize the potential limitations in the present study, an extensive literature search was performed to inform the modelling process. In particular, the patient population was taken from literature specific to CGM usage, estimates of sensor usage were used as a proxy for treatment compliance and a fear of hypoglycaemia utility was incorporated to extend the application of the CORE Diabetes Model to CGM. It is assumed that the analysis in the present study represents an appropriate estimate of the Swedish setting. The influence of assumptions used in the model was tested and model outcomes were found to be robust under a wide range of plausible assumptions. The results were most sensitive to variation in the fear of hypoglycaemia utility, whereby the ICER ranged from 193,892 to 665,742 SEK per QALY gained. Also important was the HbA1c baseline value and/or its associated reduction; variation in these values resulted in an ICER in the range of 343,637 to 593,655 SEK per QALY gained. The base–case hypothesis regarding the fear of hypoglycaemia is conservative when compared with the findings of Norgaard et al. [20], who reported a much greater reduction in fear of hypoglycaemia using CGM than did Yeh et al. [17], whose reported value was used in the base case. Hypoglycaemia is an important complication associated with diabetes and the use of insulin products. In addition to its impact on patient quality of life, hypoglycaemia is a major driver of costs. In a recent study in Germany, France and the UK, 10% of people with insulin–treated diabetes missed ≥ 1 day of work in the previous year as a result of hypoglycaemia [25]. In the assessment of non-severe hypoglycaemia, it was determined that each event resulted in 8.3–15.9 h of lost productivity per month [26]. Even if a non-severe hypoglycaemic event occurred outside of work time, 22.7% of participants were late to or did not attend work [26]. A prospective, 12-month study in the USA, however, found that mild hypoglycaemia resulted in missing 5–30 min of work [27], indicating that, if diabetes is well controlled, hypoglycaemia is not detrimental to patient productivity. When asked about self-management, 80% of respondents reported that they would value an indicator warning if their plasma glucose was getting high/low [25]. Hypoglycaemia has a range of consequences both in terms of economic cost to society and the healthcare systems as well as patient glycaemic control, safety and treatment satisfaction. The low glucose suspension function has been shown to substantially reduce the rate of hypoglycaemia, and this may translate into improved patient satisfaction and reduced healthcare costs.
8
The inclusion of low glucose suspension in the present analysis may reduce the ICER still further. Assumptions made in the model did not alter the health economic conclusion of the study. Compared with CSII, the sensor-augmented pump provided clinical benefit at an ICER below the cost–effectiveness threshold. The sensor-augmented pump is, therefore, expected to be a beneficial and cost–effective option for the treatment of Type 1 diabetes in Sweden. In conclusion, the sensor-augmented pump represents a cost–effective option in the treatment of Type 1 diabetes in Sweden compared with CSII alone. Use of the sensor-augmented pump is associated with substantial clinical benefits and notably increases the time to onset of all major diabetes-related complications considered in this analysis.
Funding sources
This study was supported by an unrestricted grant from Medtronic.
Competing interests
S.R. and R.S. are employees of health economics consultancies that received fees for participation in this research. A.S.B., S.D.P. and N.P. are full-time employees of Medtronic or its affiliates. N.P. holds Medtronic stock options. J.J. has received honoraria for work undertaken with Medtronic, but no remuneration was made with respect to this research project or the manuscript development.
References 1 Chevreul K, Berg Brigham K, Bouche C. The burden and treatment of diabetes in France. Global Health 2014; 10: 6. 2 Hill CJ, Cardwell CR, Patterson CC, Maxwell AP, Magee GM, Young RJ et al. Chronic kidney disease and diabetes in the National Health Service: a cross-sectional survey of the U.K. National Diabetes Audit. Diabet Med 2014; 31: 448–454. 3 Ward A, Alvarez P, Vo L, Martin S. Direct medical costs of complications of diabetes in the United States: estimates for event-year and annual state costs (USD 2012). J Med Econ 2014; 17: 176–183. 4 Zhang P, Zhang X, Brown J, Vistisen D, Sicree R, Shaw J et al. Global healthcare expenditure on diabetes for 2010 and 2030. Diabetes Res Clin Pract 2010; 87: 293–301. 5 Swedish National Diabetes Register. Swedish National Diabetes Register Annual Report 2012. Centre of Registers Region V€ astra G€ otaland, Swedish National Diabetes Register, 413 45 Gothenburg, Sweden. 2012. 6 Deiss D, Bolinder J, Riveline J-P, Battelino T, Bosi E, Tubiana-Rufi N et al. Improved glycemic control in poorly controlled patients with Type 1 diabetes using real-time continuous glucose monitoring. Diabetes Care 2006; 29: 2730–2732. 7 American Diabetes Association. Executive summary: Standards of medical care in diabetes–2013. Diabetes Care 2013; 36(Suppl. 1): S4–S10. 8 Pickup JC, Freeman SC, Sutton AJ. Glycaemic control in type 1 diabetes during real time continuous glucose monitoring compared
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
Research article
9
10
11
12
13
14
15
16
17
18
19
20
with self monitoring of blood glucose: meta-analysis of randomised controlled trials using individual patient data. BMJ 2011; 343: d3805. Roze S, Valentine WJ, Zakrzewska KE, Palmer AJ. Health-economic comparison of continuous subcutaneous insulin infusion with multiple daily injection for the treatment of Type 1 diabetes in the UK. Diabet Med 2005; 22: 1239–1245. Ly TT, Nicholas JA, Retterath A, Lim EM, Davis EA, Jones TW. Effect of sensor-augmented insulin pump therapy and automated insulin suspension vs standard insulin pump therapy on hypoglycemia in patients with type 1 diabetes: a randomized clinical trial. JAMA 2013; 310: 1240–1247. Garg SK, Voelmle MK, Beatson CR, Miller HA, Crew LB, Freson BJ et al. Use of continuous glucose monitoring in subjects with Type 1 diabetes on multiple daily injections versus continuous subcutaneous insulin infusion therapy: a prospective 6-month study. Diabetes Care 2011; 34: 574–579. Carlsson BM, Andersson PN, Alnervik J, Carstensen J, Lind M. Availability of insulin pump therapy in clinical practice. Diabet Med 2012; 29: 1055–1059. Swedish National Board of Health and Welfare. Swedish Guidelines for Diabetes Care 2010 [In Swedish: Nationella riktlinjer f€ or diabetesv arden 2010: St€ od f€ or styrning och ledning]. Stockholm: Socialstyrelsen, 2010. Persson U, Glenngard AH, Hjortsberg C. Estimating the willingness to pay for a QALY in Sweden - a pilot study. In: Scalone L, Mantovani LG (eds), 25th Meeting of the EuroQol Group, 2010, pp. 121–132. Nathan DM, Cleary PA, Backlund J-YC, Genuth SM, Lachin JM, Orchard TJ et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med 2005; 353: 2643–2653. The Pharmaceutical Benefits Board. General guidelines for economic evaluations from the Pharmaceutical Benefits Board. The Dental and Pharmaceutical Benefits Agency, TLV, 2003. Available at http://www.tlv.se. Last accessed 30 September 2014. Yeh H-C, Brown TT, Maruthur N, Ranasinghe P, Berger Z, Suh YD et al. Comparative effectiveness and safety of methods of insulin delivery and glucose monitoring for diabetes mellitus: a systematic review and meta-analysis. Ann Intern Med 2012; 157: 336–347. Beck RW, Lawrence JM, Laffel L, Wysocki T, Xing D, Huang ES et al. Quality-of-life measures in children and adults with type 1 diabetes: Juvenile Diabetes Research Foundation Continuous Glucose Monitoring randomized trial. Diabetes Care 2010; 33: 2175–2177. Currie CJ, Morgan CL, Poole CD, Sharplin P, Lammert M, McEwan P. Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes. Curr Med Res Opin 2006; 22: 1523–1534. Norgaard K, Scaramuzza A, Bratina N, Lalic NM, Jarosz-Chobot P, Kocsis G et al. Routine sensor-augmented pump therapy in type
ª 2014 The Authors. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
21
22
23
24
25
26
27
28
29
30
1 diabetes: the INTERPRET study. Diabetes Technol Ther 2013; 15: 273–280. Persson U, Hjelmgren J. H€ also-och sjukv arden beh€ over kunskap om hur befolkningen v€ arderar h€ alsan [Article in Swedish: Health services need knowledge about how people value health]. L€ akartidningen 2003; 100: 3436–3437. Socialstyrelsen. Nationella riktlinjer f€ or prostatacancersjukv ard Medicinskt och h€ alsoekonomiskt faktadokument [Report in Swedish: National guidelines for prostate health - Medical and health economic facts document]. Stockholm, Sweden: Socialstyrelsen, 2007. McQueen RB, Ellis SL, Campbell JD, Nair KV, Sullivan PW. Cost-effectiveness of continuous glucose monitoring and intensive insulin therapy for type 1 diabetes. Cost Eff Resour Alloc 2011; 9: 13. Huang ES, O’Grady M, Basu A, Winn A, John P, Lee J et al. The cost-effectiveness of continuous glucose monitoring in type 1 diabetes. Diabetes Care 2010; 33: 1269–1274. Willis WD, Diago-Cabezudo JI, Madec-Hily A, Aslam A. Medical resource use, disturbance of daily life and burden of hypoglycemia in insulin-treated patients with diabetes: results from a European online survey. Expert Rev Pharmacoecon Outcomes Res 2013; 13: 123–130. Brod M, Christensen T, Thomsen TL, Bushnell DM. The impact of non-severe hypoglycemic events on work productivity and diabetes management. Value Health 2011; 14: 665–671. Leckie AM, Graham MK, Grant JB, Ritchie PJ, Frier BM. Frequency, severity, and morbidity of hypoglycemia occurring in the workplace in people with insulin-treated diabetes. Diabetes Care 2005; 28: 1333–1338. Henriksson F. Applications of economic models in healthcare: the introduction of pioglitazone in Sweden. Pharmacoeconomics 2002; 20(Suppl. 1): 43–53. Valentine WJ, Aagren M, Haglund M, Ericsson A, Gschwend MH. Evaluation of the long-term cost-effectiveness of insulin detemir compared with neutral protamine hagedorn insulin in patients with type 1 diabetes using a basal-bolus regimen in Sweden. Scand J Pub Heatlh 2011; 39: 79–87. Ghatnekar O, Persson U, Willis M, Odegaard K. The cost effectiveness of Becaplermin (REGRANEX) in the treatment of diabetic foot ulcers in four European countries. Pharmacoeconomics 2000; 19: 767–778.
Supporting Information Additional Supporting Information may be found in the online version of this article: File S1. Extended methods and results.
9
