Economic Value of Improved Accuracy for Self-Monitoring of Blood Glucose Devices for Type 1 and Type 2 Diabetes in England

Abstract

Objective:

The objective was to model clinical and economic outcomes of self-monitoring blood glucose (SMBG) devices with varying error ranges and strip prices for type 1 and insulin-treated type 2 diabetes patients in England.

Methods:

We programmed a simulation model that included separate risk and complication estimates by type of diabetes and evidence from in silico modeling validated by the Food and Drug Administration. Changes in SMBG error were associated with changes in hemoglobin A1c (HbA1c) and separately, changes in hypoglycemia. Markov cohort simulation estimated clinical and economic outcomes. A SMBG device with 8.4% error and strip price of £0.30 (exceeding accuracy requirements by International Organization for Standardization [ISO] 15197:2013/EN ISO 15197:2015) was compared to a device with 15% error (accuracy meeting ISO 15197:2013/EN ISO 15197:2015) and price of £0.20. Outcomes were lifetime costs, quality-adjusted life years (QALYs) and incremental cost-effectiveness ratios (ICERs).

Results:

With SMBG errors associated with changes in HbA1c only, the ICER was £3064 per QALY in type 1 diabetes and £264 668 per QALY in insulin-treated type 2 diabetes for an SMBG device with 8.4% versus 15% error. With SMBG errors associated with hypoglycemic events only, the device exceeding accuracy requirements was cost-saving and more effective in insulin-treated type 1 and type 2 diabetes.

Conclusions:

Investment in devices with higher strip prices but improved accuracy (less error) appears to be an efficient strategy for insulin-treated diabetes patients at high risk of severe hypoglycemia.

Keywords

cost-effectiveness self-monitoring of blood glucose accuracy glycemic control simulation

In England, significant costs are devoted to treating and managing potentially avoidable diabetes complications. The most comprehensive estimate of total annual cost of type 1 and type 2 diabetes to the National Health Service (NHS) was £9.8 billion in 2010/11;¹ inflating this cost gives an estimated direct cost for 2016/2017 of £11.5 billion. Avoiding short- and long-term complications requires frequent hemoglobin A1c (HbA1c) testing and daily self-monitoring of blood glucose (SMBG) for type 1 and insulin-treated type 2 diabetes patients.¹ SMBG encompasses a system of strips paired with a device that tracks glucose concentrations multiple times per day for patients treated with intensive insulin therapy. Frequent SMBG improves HbA1c and assists in avoiding diabetes-related complications.^1,2 Moreover, past evidence has suggested an association between improved glycemic control and cost savings as well as improved quality of life.^3-5

Not all SMBG devices have the same accuracy which can impact HbA1c levels and the risk of hypoglycemia from inaccurate insulin dosing. Clinical trials comparing accuracy of different SMBG devices are difficult, if not impossible, to conduct given ethical considerations. Previous evidence employed a realistic computer simulation to provide evidence on the relationship between SMBG accuracy and glycemic control.⁶ Specifically, the authors found that as SMBG errors increase, HbA1c increases or separately, the risk of hypoglycemic episodes increase; these findings were later reinforced from clinical observations in German/Austrian diabetes registries.^7,8 Strip prices also vary by SMBG device and tend to be more expensive per strip for devices with less error.⁹ Previous evidence suggested investing in SMBG devices with higher strip prices and less error may offset downstream costs associated with complications.¹⁰

Previous analyses associating changes in SMBG error to economic outcomes have been applied to the Canadian and German setting in type 1 diabetes patients only; however, evidence is lacking in England and among insulin-treated type 2 diabetes patients.^10,11 The objective of this study was to expand these previous analyses by estimating the cost-effectiveness of blood glucose monitoring devices with varying SMBG error (ie, SMBG devices meeting and exceeding standards allowed by ISO 15197:2013/EN ISO 15197:2015) and strip price among type 1 and insulin-treated type 2 diabetes (hereto referred as type 2 diabetes) patients from the English NHS perspective.

Methods

Motivation for Simulation Modeling

When comparing accuracy of SMBG devices, randomized clinical trials are not possible given ethical considerations. We rely on previous in silico (computer) modeling evidence validated by the Food and Drug Administration to inform the association between SMBG accuracy and short-term clinical outcomes. The short-term clinical outcomes described in the next section are used as inputs to a cost-effectiveness simulation model designed to reveal the relationship between inputs (eg, disease progression of diabetes, risk reduction of complications from reduced HbA1c) and outputs (eg, survival, quality of life, and costs) that are relevant to decision makers.¹² Simulation models are frameworks for synthesizing the best available evidence from longitudinal studies, randomized trials, and public health statistics, among other sources.^12,13 Furthermore, simulation models may forecast this synthesized evidence into relevant outcomes for decision making. Policy decision making for SMBG devices has occurred despite a lack of comparative clinical or economic evidence. Motivation for the use of simulation modeling evidence is that policy decision makers will be better able to have explicit and evidence-based discussions about important attributes and their uncertainty.

Changes in SMBG Accuracy Associated With Short-Term Changes in HBA1c and Hypoglycemia

Breton and Kovatchev investigated the impact of errors from SMBG devices on HbA1c and hypoglycemia using in silico (computer) modeling validated by the Food and Drug Administration.⁶ Specifically, a 5% error resulted in a 0.01% absolute increase in HbA1c; 10% error resulted in a 0.12% absolute increase in HbA1c; and 15% error resulted in a 0.26% absolute increase in HbA1c. In a separate simulation experiment, the authors found increasing SMBG error resulted in an increased percentage of hypoglycemic episodes to 15.2% at 5% error, 18.8% at 10% error, 22% at 15% error, and 25.6% at 20% error. Errors from self-monitoring of blood glucose devices were associated with changes in hypoglycemic events defined between 50 mg/dl and 70 mg/dl.⁶ We used the information from Breton and Kovatchev to generate a relative risk between devices with different SMBG errors and applied those relative risks to a baseline number of events per year for each device. The baseline number of events per year were derived from a population-based study of health service resource use for patients with type 1 and type 2 diabetes from the NHS perspective.¹⁴ Leese and colleagues estimated the number of hypoglycemic episodes over a one year time period and split them by inpatient and accident and emergency/outpatient resource use. Using the number of events and patients over the one-year time period, we calculated a baseline number of hypoglycemic episodes for use in the model.

Markov Model Structure

We followed good research practices for development and reporting of our cost-effectiveness simulation modeling analysis.^12,15 The Markov cohort simulation model (Figure 1) estimates long-run clinical and economic outcomes for the average adult type 1 and type 2 diabetes patient at an average age of 35 years and mean duration of diabetes of 9 years for type 1 diabetes and 60 years with no duration of diabetes for type 2 diabetes (Table 1). These values were based on the mean age of diagnosis and disease duration from the Scottish Diabetes Survey (SDS) (2015) of 26.1 years and 59.7 years for type 1 and type 2, respectively. The baseline HbA1c was set to 8.47% for type 1 diabetes and 7.44% for type 2 diabetes.¹⁶ These values were calculated using data from the SDS (2015)¹⁶ and are consistent with mean values from the National Diabetes Inpatient Audit 2015.¹⁷ The populations in England with type 1 and insulin dependent type 2 diabetes are from the National Diabetes Audit (NDA).¹⁸

Figure 1.

A Markov modeling framework for type 1 and type 2 diabetes disease progression. The population includes adult type 1 and type 2 diabetes patients who are on two distinct treatment paths. An example one year cycle has patients starting at no complications and moving into microvascular disease states (eg, retinopathy, neuropathy, and nephropathy) or macrovascular disease states (eg, myocardial infarction, congestive heart failure, ischemic heart disease, and stroke). Patients can die from all cause death or complication-related death.

Table 1.

Key Model-Wide Inputs.

	Type 1 diabetes	Type 2 diabetes	Source
Baseline HbA1c (mean)	8.47%	7.44%	NHS Scotland¹⁶
HbA1c for SMBG with 15% error^a	8.79%	7.75%	Breton and Kovatchev⁶
HbA1c for SMBG with 8.4% error^a	8.65%	7.61%	Breton and Kovatchev⁶
Utilization of strips per day (mean)	4		NICE¹⁹
Price per strip for SMBG with 15% error	0.20		National Health Service Business Services Authority⁹
Price per strip for SMBG with 8.4% error	0.30		National Health Service Business Services Authority⁹
Prevalence of people with insulin dependent diabetes in England	272 682	417 733	National Health Service¹⁸
Mean age in years	35	60	NHS Scotland¹⁶
Duration of diabetes in years	9	0	NHS Scotland¹⁶
Discount rate	3.5%		NICE²⁰

HbA1c values increased from baseline HbA1c values using error calculations from Breton et al.⁶

The model reflects the biological process of type 1 and 2 diabetes and is applicable to a wide range of treatment settings (Figure 1). The model structure for type 1 diabetes includes the major diabetes complication states categorized by the American Diabetes Association (ADA): coronary heart disease (CHD), nephropathy, neuropathy, retinopathy, and end-stage renal disease. The model structure for type 2 diabetes includes major complications from the United Kingdom Prospective Diabetes Study (UKPDS): blindness, renal failure, amputation, ulcer, myocardial infarction, congestive heart failure, ischemic heart disease, and stroke.

Across both type 1 and type 2, a similar structure of disease progression has been used in many other validated diabetes modeling studies.^21-25 The model structures use a cycle length of one year with varying time horizons based on each scenario. Lifetime costs and outcomes are compared across SMBG error levels and changes in HBA1c (Scenario 1) for type 1 and type 2 diabetes. A three year time horizon was used in Scenario 2 to compare the difference in costs and outcomes across SMBG error levels and changes in hypoglycemia for type 1 and type 2 diabetes over a shorter time horizon. A shorter time horizon of three years was used because acute short-term events do not impact long-run complications and their associated costs. The Markov model outputted cumulative incidence estimates, quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratios (ICERs) for the purposes of estimating the efficiency of SMBG devices with better accuracy (ie, less error) but higher strip prices. The QALY is a combined metric that represents both quality and quantity of life; therefore a QALY of 1.0 represents one year of life lived in perfect health. The ICER is interpreted as the average additional cost required to achieve an outcome within a treated population; when QALYs are the health outcomes observed in the ICER, the interpretation is the average added cost required to achieve one additional year of perfect health in a treated population when comparing the health technology in question to its alternative. Therefore, the ICER addresses an efficiency research question (ie, what we gain in terms of health for an average added expenditure within a specified population), also referred to as value for money. For more model details including validation procedures, please see the technical appendix.

Clinical Inputs

The model approach links HbA1c to the risk of long-term diabetes complications. Specifically, annual transition probabilities were derived from original Diabetes Control and Complications Trial (DCCT) prediction models.^26-30 For microvascular intermediate disease states and end-stage complications, we used hazard rates from the DCCT calculated by Eastman et al.^31,32 Due to a lack of data in type 1 diabetes, for CHD, we used hazard rates derived from the UKPDS population of type 2 diabetes patients.³³ Transition probabilities for type 2 diabetes were derived from the UKPDS Outcomes Model 2, an event-based model that predicts outcomes from the 30-year follow-up of type 2 diabetes patients enrolled in the UKPDS.³⁴ Transition probabilities for different SMBG error rates are impacted by the changes in HbA1c from Breton et al.⁶ These transitions represent the movement into intermediate and end stage diabetes-related complications.

Utility Inputs

Health state utilities for type 1 and type 2 diabetes adopted values used by the National Institute of Clinical Excellence (NICE) to inform its clinical guidelines (Table 2).^19,35 For type 1 diabetes, the analyses informing the NICE Guideline²⁹ had a health state for proteinuria but not for nephropathy. We applied the proteinuria utility value to both proteinuria and nephropathy. This approach is consistent with the previous version of this model used in the Canadian setting which adopted equivalent utility values for the two health states.¹⁰ For type 2 diabetes, the economic model informing the NICE Guideline³⁰ was based on the same UKPDS model and used the same health states as the current model.

Table 2.

Utility Inputs.^a

State/event (annual cycle length)	Type 1 diabetes^b
Diabetes with no complications	0.814
Background diabetic retinopathy	0.780
Macular edema	0.745
Proliferative retinopathy	0.715
Blindness	0.711
Neuropathy	0.701
Lower extremity amputation	0.534
Subsequent lower extremity amputation	0.534
Nephropathy	0.737
Proteinuria	0.737
End-stage renal disease	0.621
Coronary heart disease	0.759
Disutility of hypoglycemic event	−0.012
	Type 2 diabetes^c
Diabetes with no complications	0.785
Blindness utility	0.711
Lower extremity amputation	0.505
Subsequent lower extremity amputation	0.505
Renal failure	0.522
Congestive heart failure	0.677
Ischemic heart disease	0.695
Myocardial infarction	0.730
Stroke	0.621
Ulcer	0.615

Values are varied in sensitivity analyses using lower and upper values of the 2.5 and 97.5 percentiles of beta distributions.

Adopted values from NICE guidance.³⁶

Adopted values from UKPDS model that informed NICE guidance.¹⁹

Health State Cost Inputs

Health state costs for type 1 diabetes, where possible, were derived from the evidence base which informed the NICE Guideline for management of type 1 diabetes (Table 3).³⁵ The evidence did not have costs for macular oedema, diabetic retinopathy, background retinopathy, nephropathy and proteinuria. Costs for retinopathy and macular oedema were estimated by costing the resources required within the NICE treatment pathway³⁷ and applying relevant drug costs,³⁸ dosing regimens,³⁹ and staff costs.⁴⁰ For nephropathy and proteinuria we used the average inpatient costs for patients with chronic kidney disease,³⁵ annual medication comprising losartan, a statin, and a beta blocker³² plus two outpatient attendances per year.³⁵

Table 3.

Cost Inputs in 2017 Pound Sterling.^a

State/event	Type 1 diabetes^b
State/event	First year	Subsequent year
Diabetes with no complications	777	777
Background diabetic retinopathy	298	298
Macular edema	2963	2963
Proliferative retinopathy	1015	1015
Blindness	5774	5578
Neuropathy	386	386
Lower extremity amputation	15 765	1832
Subsequent lower extremity amputation	15 765	1832
Nephropathy	3890	3890
Proteinuria	3890	3890
End-stage renal disease	31 509	31 509
Coronary heart disease	3970	814
Severe hypoglycemic event requiring inpatient visit	944	944
Hypoglycemia event requiring outpatient visit	276	276
	Type 2 diabetes^c
Diabetes with no complications	544	544
Congestive heart failure	3923	1506
Ischemic heart disease	11 805	883
Myocardial infarction	6907	1180
Stroke	10 044	1150
Blindness	1385	463
Ulcer	5699	0
Amputation	10 661	1832
Renal failure	31 509	31 509
Severe hypoglycemic event requiring inpatient visit	959	959
Hypoglycemia event requiring outpatient visit	257	257

Values are varied in sensitivity analyses using lower and upper values of the 2.5 and 97.5 percentiles of gamma distributions.

Derived from evidence base informing NICE guidelines for management of type 1 diabetes.^35-38,41,42

Derived from UKPDS evidence and NICE guidance on management of type 2 diabetes.^19,41,43

The NICE Guideline for type 2 diabetes³⁰ noted new costs from UKPDS were pending at publication. These have now been published by Alva et al;⁴³ this source has been used where possible. Alva et al⁴³ did not include costs for ulcers of the lower limb or for renal failure. The value for renal failure informing the NICE type 2 diabetes guideline used a source year of 1996 and did not have a value for ulcer of the lower limb. Hence for these health states the costs from the NICE Guideline for type 1 diabetes were used.

Costs for inpatient and outpatient hypoglycemic events for both type 1 and type 2 diabetes were taken from Hammer et al.⁴¹

All costs have been inflated to 2016 costs using relevant inflation indices.⁴⁰ Costs and outcomes were discounted at 3.5% per annum.

Intervention Cost Inputs

The published price for the strip exceeding ISO standards (ie, more accurate) was £0.30p, with the price for the strip meeting ISO standards (ie, less accurate) was £0.20p (Table 1).⁹ A NICE guideline recommends testing at least four times per day, including before each meal and before bed.³⁶ Based on this guideline we have used four strips per day as our base-case estimate.

Scenario and Sensitivity Analyses

First, clinical event model validation scenarios were conducted by norming the HbA1c and age values to compare the ten-year clinical event findings from our model to that of the UKPDS Outcomes Model 2.³⁴ Cumulative expected clinical events over ten years were considered replicated when within one absolute percentage point of the UKPDS Outcomes Model 2 (please see technical appendix for additional validation procedures). Univariate sensitivity analyses were used to vary one input across a possible range, while holding all other inputs constant, to assess the impact of input variation and uncertainty on the overall results. Furthermore, we varied strip price until costs were equivalent between comparator SMBG devices and we assumed the same strip price to display the impact on Scenario 1 and Scenario 2 results. To further assess the robustness of the results and conclusions, we also performed multivariate probabilistic sensitivity analyses (PSA) by varying all input parameters simultaneously over their possible ranges and plotting simulation values on a cost-effectiveness plane after assigning evidence-based probability distributions.⁴⁴

Results

Clinical Results

Decreases in cumulative incidence of complications were observed when SMBG error increased (Table 4). For example, an SMBG error of 8.4% was associated with a relative percentage decrease in blindness of -6.94% versus an SMBG device with 15% error over a lifetime horizon in type 1 diabetes patients (Scenario 1). In type 2 diabetes patients, an SMBG error of 8.4% was associated with a relative percentage decrease in amputation of –2.69% versus an SMBG device with 15%. Changes in HbA1c did not impact incidence of complications such as renal failure, congestive heart failure, and ischemic heart disease in type 2 diabetes as derived from UKPDS evidence.³⁴

Table 4.

Estimates of Clinical Diabetes-Related Complications.

		8.4% error (0.14 increase in HbA1c)	15% error (0.25 increase in HbA1c)	Relative change (8.4% vs 15% error)
Scenario 1 (lifetime time horizon). % error associated with HbA1c but NOT directly associated with hypoglycemic events^a
			Type 1 diabetes	Retinopathy^a
	Background diabetic retinopathy	56.78%	61.03%	−6.96%
	Macular edema	61.73%	65.70%	−6.03%
	Blindness	32.41%	34.83%	−6.94%
Nephropathy^a
	Microalbuminuria	55.85%	57.43%	−2.76%
	Renal failure	2.67%	2.94%	−7.49%
Neuropathy^a
	Neuropathy	27.57%	29.36%	−6.09%
	Amputation	14.07%	14.96%	−5.96%
Coronary heart disease		43.78%	44.53%	−1.67%
			Type 2 diabetes	Microvascular
	Blindness	4.06%	4.16%	−2.29%
	Renal failure	0.72%	0.72%	0.00%
	Amputation	2.99%	3.08%	−2.69%
	Ulcer	3.01%	3.07%	−2.15%
Macrovascular
	Myocardial infarction	15.87%	16.08%	−1.25%
	Congestive heart failure	4.76%	4.76%	0.00%
	Ischemic heart disease	16.18%	16.18%	0.00%
	Stroke	7.31%	7.40%	−1.20%
Scenario 2 (3-year time horizon). % error associated with hypoglycemic events but NOT directly associated with changes in HbA1c^b
			Type 1 diabetes
Severe hypoglycemic events requiring inpatient visit		0.79	1.02	−23%
Hypoglycemic event requiring outpatient visit		2.95	3.83	−23%
			Type 2 diabetes
Severe hypoglycemic events requiring inpatient visit		0.70	0.91	−23%
Hypoglycemic event requiring outpatient visit		2.58	3.34	−23%

Assuming SMBG errors are only associated with changes in HbA1c.

Assuming SMBG errors are only associated with inpatient or outpatient hypoglycemic events.

A significant reduction in risk of three year hypoglycemic events requiring inpatient or outpatient visits (–23%) was also observed when comparing an SMBG device of 8.4% error versus a device with 15% error (Scenario 2).

Cost-Utility Results

With SMBG errors associated with changes in HbA1c but not directly associated with hypoglycemic events (Scenario 1), the incremental cost-effectiveness ratio was £3064 per QALY in type 1 diabetes and £264 668 per QALY in type 2 diabetes for an SMBG device with 8.4% error and strip price of £0.30 versus an SMBG device with 15% error and strip price of £0.20 (Table 5). In other words, the cost to achieve one additional year of life in perfect health using devices with less error (more accuracy) will cost approximately £3064 and £264 668 for type 1 diabetes patients and type 2 diabetes patients, respectively. The higher ICER among type 2 diabetes is a result of less QALYs gained (0.006 QALY gain) as compared to the type 1 diabetes scenario (0.122 QALY gain). With SMBG errors associated with hypoglycemic events but not directly associated with changes in HbA1c (Scenario 2), an SMBG device with 8.4% error and strip price of £0.30 is less costly and more effective than an SMBG device with 15% error and strip price of £0.20 in type 1 and type 2 diabetes.

Table 5.

Cost-Utility Results.

	Costs (2016 pound sterling)	QALYs	Combined hypoglycemic events^a	Incremental cost-effectiveness ratio (pound sterling per QALY)
Scenario 1 (lifetime time horizon). % error associated with HbA1c but NOT directly associated with hypoglycemic events
		Type 1 diabetes
SMBG device with 15% error, £0.20 per strip	£81 809	15.292	—	—
SMBG device with 8.4% error, £0.30 per strip	£82 182	15.414	—	£3064/QALY
		Type 2 Diabetes
SMBG device with 15% error, £0.20 per strip	£39 523	9.743	—	—
SMBG device with 8.4% error, £0.30 per strip	£41 176	9.750	—	£264 668/QALY
Scenario 2 (3-year time horizon). % error associated with hypoglycemic events but NOT directly associated with changes in HbA1c
		Type 1 diabetes
SMBG device with 15% error, £0.20 per strip	£7296	2.142	4.85	—
SMBG device with 8.4% error, £0.30 per strip	£7191	2.155	3.74	Dominant^b
		Type 2 diabetes
SMBG device with 15% error, £0.20 per strip	£5223	2.166	4.25	—
SMBG device with 8.4% error, £0.30 per strip	£5174	2.178	3.28	Dominant^b

All severe hypoglycemic events requiring inpatient visits and hypoglycemic events requiring outpatient visits.

Less costly and more effective.

Univariate Sensitivity Analyses

For results specific to SMBG errors associated with changes in HbA1c (Table 6), assuming the same strip price between cohorts resulted in a cost saving scenario (8.4% SMBG error less costly, more effective than 15% SMBG error) for both type 1 and type 2 diabetes. Incremental costs equated to £0 for an 8.4% error device with strip price of £0.28 in type 1 diabetes and £0.21 in type 2 diabetes while holding strip price constant for the comparator SMBG device at £0.20. For results specific to SMBG errors associated with changes in hypoglycemia, assuming the same strip price resulted in a cost savings scenario for both type 1 and type 2 diabetes. Incremental costs equated to £0 when changing the strip price for an SMBG device with 8.4% error device to £0.33 and £0.31 in type 1 and type 2 diabetes, respectively, while holding strip price constant for the comparator SMBG device at £0.20.

Table 6.

Univariate Sensitivity Analysis.

Sensitivity analyses	Outcome of sensitivity analysis
Scenario 1 (lifetime time horizon). % error associated with HbA1c but NOT directly associated with hypoglycemic events
	Type 1 diabetes	Type 2 diabetes
Assume same strip price (£0.20)	Dominant: less costly and more effective—cost saving scenario	Dominant: less costly and more effective—cost saving scenario
At what strip price for the device with 8.4% error would incremental costs = £0?	At £0.28 incremental costs = 0	At £0.21 incremental costs = 0
Scenario 2 (3-year time horizon) % error associated with hypoglycemic events but NOT directly associated with changes in HbA1c
	Type 1 diabetes	Type 2 diabetes
Assume same strip price (£0.20)	Dominant: less costly and more effective—cost saving scenario	Dominant: less costly and more effective—cost saving scenario
At what strip price for the device with 8.4% error would incremental costs = £0?	At £0.33 incremental costs = 0	At £0.31 incremental costs = 0

Probabilistic Sensitivity Analyses

After varying all inputs simultaneously in Scenario 1 within the type 1 diabetes subgroup, 64% of simulations were cost-effective at a willingness-to-pay of £30 000/QALY for an SMBG device with 8.4% error and strip price of £0.30 versus an SMBG device with 15% error and strip price of £0.20 (Figure 2A). In Scenario 1 within the type 2 diabetes subgroup, 38% of simulations were cost-effective at a willingness-to-pay £30 000/QALY for an SMBG device with 8.4% error and strip price of £0.30 versus an SMBG device with 15% error and strip price of £0.20 (Figure 2B).

Figure 2.

Incremental cost-effectiveness scatterplot for Scenario 1: SMBG error of 8.4% with strip price at £0.30 versus SMBG error of 15% with strip price at £0.20 assuming percentage error associated with HbA1c but NOT directly associated with hypoglycemic events. Figure 2A and 2B display results for type 1 and type 2 diabetes, respectively. Each point represents one Monte Carlo simulation. (A) Type 1 diabetes. (B) Type 2 diabetes.

Discussion

We estimated scenarios related to changes in blood glucose control (as measured by HbA1c or separately by inpatient or outpatient hypoglycemic events) resulting from differences in SMBG device error. Both scenarios resulted in higher average effectiveness in type 1 and type 2 diabetes and in most cases, higher costs. In Scenario 1, while holding hypoglycemia constant, the mean findings were cost-effective with 64% and 38% likelihood for type 1 and type 2 diabetes, respectively, at a willingness-to-pay of £30 000 per QALY. In Scenario 2, while holding changes in HbA1c constant, an SMBG device with 8.4% error was less costly and more effective than an SMBG device with 15% error in type 1 diabetes and type 2 diabetes.

Our modeling findings are similar to other modeling results associating SMBG accuracy standards with clinical and economic outcomes.^10,45 Our previous modeling findings found improved accuracy of SMBG devices in type 1 diabetes resulted in efficient and affordable findings from a Canadian payer perspective. Our current analysis differs with the addition of insulin-treated type 2 diabetes patients and the application to an English NHS perspective which has a different willingness-to-pay threshold than Canada. While there are differences between these modeling analyses, findings are similar for Scenario 2, where the value of investing in SMBG devices with less error is greatest among type 1 and insulin-treated type 2 diabetes patients at risk of hypoglycemic events. Additionally, investing in SMBG devices with less error may also be an efficient strategy over the longer term for type 1 diabetes patients.

Previous research using clinical practice or claims data has also linked better glycemic control to significant cost-savings.^3,4,46,47 Two recent retrospective analyses found better glycemic control was associated with reduced health care utilization and medical costs.^3,46 While we use simulation modeling methods, our study adds to the body of literature suggesting better glycemic control is associated with reduced health care utilization and costs.

We performed an external model validation exercise to compare our model cost estimates from an English NHS perspective with estimates produced by the IMS CORE model.⁵ For example, after 10 years, a reduction in HbA1c from 9% to 8% resulted in a cost savings per patient of £607 and £523 from the IMS CORE and our modeling analysis, respectively. Our model estimates are in very similar ranges to the IMS CORE model estimates when HbA1c reductions are larger (eg, reduction from 9% to 8%). However, our model underestimates the cost savings with smaller reductions in HbA1c as compared to the IMS CORE model (eg, reduction from 8% to 7.5%) and therefore the cost savings estimates may be considered as conservative. Even with the different approach taken by the IMS CORE and McQueen et al model, the estimates are in reasonable ranges that move in the expected direction of higher differences in HbA1c equating to higher differences in complications and costs. For the purposes of estimating cost savings from HbA1c improvements, one could interpret the cost estimates projected by our model as being conservative and therefore have a higher likelihood of being realized in real-world practice.

There is uncertainty in how error in SMBG impacts HbA1c or hypoglycemia, and in how HbA1c or hypoglycemia impacts long-term costs and outcomes. The limitation of error in SMBG impacting HbA1c or hypoglycemia is of particular relevance to type 2 diabetes patients since the in silico simulation model used to inform our cost-effectiveness analysis was focused on type 1 diabetes patients.⁶ However, uncertainty in model inputs (ie, impact of SMBG accuracy on changes in HbA1c, risk of diabetes complications, etc) is propagated through the cost-effectiveness model and the resulting uncertainty is encompassed in one-way and multivariate sensitivity analyses. That is, any uncertainty around our assumptions of the impact of SMBG accuracy changes on HbA1c for type 2 diabetes patients is encompassed in simulations shown in Figure 2. Alternative approaches such as using real-world data or using an alternative modeling method may produce different findings than produced in this analysis. Furthermore, the relationship between HbA1c and complications does not capture all of the benefit that an SMBG device can provide to patients.

Conclusions

At higher strip prices, improved accuracy of SMBG ranges from good value for money (Scenario 1) to cost saving in the short-term (Scenario 2). Investment in devices with higher strip prices but improved accuracy (less error) appears to be an efficient strategy for insulin-treated diabetes patients at high risk of severe hypoglycemia.

Supplemental Material

Technical_Appendix – Supplemental material for Economic Value of Improved Accuracy for Self-Monitoring of Blood Glucose Devices for Type 1 and Type 2 Diabetes in England

Supplemental material, Technical_Appendix for Economic Value of Improved Accuracy for Self-Monitoring of Blood Glucose Devices for Type 1 and Type 2 Diabetes in England by Robert Brett McQueen, Marc D. Breton, Joyce Craig, Hayden Holmes, Melanie D. Whittington, Markus A. Ott and Jonathan D. Campbell in Journal of Diabetes Science and Technology

Footnotes

Abbreviations

ADA, American Diabetes Association; CHD, coronary heart disease; DCCT, Diabetes Control and Complications Trial; ICERs, incremental cost-effectiveness ratios; NDA, National Diabetes Audit; NHS, National Health Service; NICE, National Institute of Clinical Excellence; PSA, probabilistic sensitivity analyses; QALYs, quality-adjusted life years; SDS, Scottish Diabetes Survey; SMBG, self-monitoring blood glucose; UKPDS, United Kingdom Prospective Diabetes Study.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RBM has received research support from Ascensia Diabetes Care and served as an advisory board participant. MDB has received research support and honorarium from Ascensia Diabetes Care. MAO was an employee of Ascensia Diabetes Care at the time this work was conducted. JDC received research support from Ascensia Diabetes Care. YHEC has received payment from Ascensia Diabetes Care for the input of JC and HH on this project. MAO was an employee of Ascensia Diabetes Care at the time this work was conducted.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Ascensia Diabetes Care provided funding for this study.

Supplemental material

Supplementary material for this article is available online.

References

Otto

Nitsche

Jung

et al . Determinants of neointimal proliferation and stent coverage after intracoronary therapy with drug-eluting devices in stable coronary artery disease: role of endothelial progenitor cells and interleukin-1 family cytokines. J Invasive Cardiol. 2014;26:648-653.

Miller

Beck

Bergenstal

et al . Evidence of a strong association between frequency of self-monitoring of blood glucose and hemoglobin A1c levels in T1D exchange clinic registry participants. Diabetes Care. 2013;36:2009-2014.

Aagren

Luo

Association between glycemic control and short-term healthcare costs among commercially insured diabetes patients in the United States. J Med Econ. 2011;14:108-114.

McQueen

Ellis

Maahs

et al . Association between glycated hemoglobin and health utility for type 1 diabetes. Patient. 2014;7:197-205.

Baxter

Hudson

Mahon

et al . Estimating the impact of better management of glycaemic control in adults with type 1 and type 2 diabetes on the number of clinical complications and the associated financial benefit. Diabet Med. 2016;33:1575-1581.

Breton

Kovatchev

BP.

Impact of blood glucose self-monitoring errors on glucose variability, risk for hypoglycemia, and average glucose control in type 1 diabetes: an in silico study. J Diabetes Sci Technol. 2010;4:562-570.

Boettcher

Dost

Wudy

et al . Accuracy of blood glucose meters for self-monitoring affects glucose control and hypoglycemia rate in children and adolescents with type 1 diabetes. Diabetes Technol Ther. 2015;17:275-282.

Boettcher

Wudy

Dost

et al . Self-measured blood glucose values vs. simultaneous laboratory-obtained glucose-concentrations in children and adolescents with type 1 diabetes—an analysis of glucose meter accuracy and its impact on metabolic control based on a German/Austrian pediatric diabetes registry. Pediatr Diabetes. 2013;14:30.

National Health Service Business Services Authority. NHS Electronic Drug Tariff Part IXR. 2017 Available at: http://www.drugtariff.nhsbsa.nhs.uk/#/00475250-DA_1/DA00474778/Part%20IXR%20-%20Chemical%20Reagents%20. Accessed August 2017.

10.

McQueen

Breton

Ott

et al . Economic value of improved accuracy for self-monitoring of blood glucose devices for type 1 diabetes in Canada. J Diabetes Sci Technol. 2015;10:366-377.

11.

Schnell

Erbach

Wintergerst

Higher accuracy of self-monitoring of blood glucose in insulin-treated patients in Germany: clinical and economical aspects. J Diabetes Sci Technol. 2013;7:904-912.

12.

Weinstein

O’Brien

Hornberger

et al . Principles of good practice for decision analytic modeling in health-care evaluation: report of the ISPOR Task Force on Good Research Practices—Modeling Studies. Value Health. 2003;6:9-17.

13.

Stahl

JE.

Modelling methods for pharmacoeconomics and health technology assessment: an overview and guide. Pharmacoeconomics. 2008;26:131-148.

14.

Leese

Wang

Broomhall

et al . Frequency of severe hypoglycemia requiring emergency treatment in type 1 and type 2 diabetes: a population-based study of health service resource use. Diabetes Care. 2003;26:1176-1180.

15.

Husereau

Drummond

Petrou

et al . Consolidated Health Economic Evaluation Reporting Standards (CHEERS)—explanation and elaboration: a report of the ISPOR health economic evaluation publication guidelines good reporting practices task force. Value Health. 2013;16:231-250.

16.

NHS Scotland. Scottish Diabetes Survey 2015. 2015. Available at: http://diabetesinscotland.org.uk/Publications/SDS2015.pdf. Accessed December 6, 2016.

17.

NHS Digital. National Diabetes Inpatient Audit (NaDIA)—2015. 2016. Available at: http://content.digital.nhs.uk/searchcatalogue?productid=20443&q=national+diabetes+inpatient+audit&sort=Relevance&size=10&page=1#top. Accessed December 15, 2016.

18.

National Health Service. National Diabetes Audit. 2016. Available at: http://content.digital.nhs.uk/nda. Accessed August, 2017.

19.

NICE. Type 2 diabetes in adults: management, NICE guideline NG28, evidence, appendix F: Full health economics report. 2015. Available at: https://www.nice.org.uk/guidance/ng28/evidence/appendix-f-full-health-economics-report-pdf-2185320355. Accessed August 22, 2017.

20.

NICE. Guide to the methods of technology appraisal. 2013. Available at: https://www.nice.org.uk/process/pmg9/chapter/foreword. 2017. Accessed August, 2017.

21.

Grima

Thompson

Sauriol

Modelling cost effectiveness of insulin glargine for the treatment of type 1 and 2 diabetes in Canada. Pharmacoeconomics. 2007;25:253-266.

22.

Huang

O’Grady

Basu

et al . The cost-effectiveness of continuous glucose monitoring in type 1 diabetes. Diabetes Care. 2010;33:1269-1274. [Erratum appears in Diabetes Care. 2010;33:2129].

23.

McQueen

Ellis

Campbell

Nair

Sullivan

PW.

Cost-effectiveness of continuous glucose monitoring and intensive insulin therapy for type 1 diabetes. Cost Eff Resour Alloc. 2011;9:13.

24.

Thokala

Kruger

Brennan

et al . Assessing the cost-effectiveness of type 1 diabetes interventions: the Sheffield type 1 diabetes policy model. Diabet Med. 2014;31:477-486.

25.

Hayes

Leal

Gray

Holman

Clarke

PM.

UKPDS outcomes model 2: a new version of a model to simulate lifetime health outcomes of patients with type 2 diabetes mellitus using data from the 30 year United Kingdom Prospective Diabetes Study: UKPDS 82. Diabetologia. 2013;56:1925-1933.

26.

Cusick

Meleth

Agrón

et al . Associations of mortality and diabetes complications in patients with type 1 and type 2 diabetes: early treatment diabetic retinopathy study report no. 27. Diabetes Care. 2005;28:617-625.

27.

Diabetes Control and Complications Trial. The absence of a glycemic threshold for the development of long-term complications: the perspective of the Diabetes Control and Complications Trial. Diabetes. 1996;45:1289-1298.

28.

Klein

McBride

et al . Cardiovascular disease, mortality, and retinal microvascular characteristics in type 1 diabetes: Wisconsin Epidemiologic Study of Diabetic Retinopathy. Arch Intern Med. 2004;164:1917-1924.

29.

Selvin

Marinopoulos

Berkenblit

et al . Meta-analysis: glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Ann Intern Med. 2004;141:421-431.

30.

Sainfort

Tomar

et al . Development and application of a model to estimate the impact of type 1 diabetes on health-related quality of life. Diabetes Care. 1998;21:725-731.

31.

Eastman

Javitt

Herman

et al . Model of complications of NIDDM. II. Analysis of the health benefits and cost-effectiveness of treating NIDDM with the goal of normoglycemia. Diabetes Care. 1997;20:735-744.

32.

Eastman

Javitt

Herman

et al . Model of complications of NIDDM. I. Model construction and assumptions. Diabetes Care. 1997;20:725-734.

33.

Stevens

Kothari

Adler

Stratton

; United Kingdom Prospective Diabetes Study (UKPDS)Group. The UKPDS risk engine: a model for the risk of coronary heart disease in type II diabetes (UKPDS 56). Clin Sci (Lond). 2001;101:671-679.

34.

Hayes

Leal

Gray

Holman

Clarke

PM.

35.

NICE. Type 1 diabetes in adults: diagnosis and management, clinical guideline NG17, appendices H-U. 2015. Available at: https://www.nice.org.uk/guidance/ng17/evidence/appendices-hu-pdf-435400240. Accessed December 15, 2016.

36.

NICE. Type 1 diabetes in adults: diagnosis and management NICE guideline NG17. 2015. Available at: https://www.nice.org.uk/guidance/ng17/chapter/1-Recommendations#blood-glucose-management-2. Accessed December 15, 2016.

37.

NICE. Identifying and managing complications in adults with type 1 diabetes, eye disease. 2016 Available at: https://pathways.nice.org.uk/pathways/type-1-diabetes-in-adults#path=view%3A/pathways/type-1-diabetes-in-adults/identifying-and-managing-complications-in-adults-with-type-1-diabetes.xml&content=view-node%3Anodes-eye-disease. Accessed November 24, 2016.

38.

MIMS. Monthly Index of Medical Specialties, September 2016. 2016. Available at: http://www.mims.co.uk/. Accessed October 2016.

39.

Datapharm. Electronic medicines compendium (eMC). 2016. Available at: https://www.medicines.org.uk/emc/. Accessed October 2016.

40.

PSSRU. Unit Costs of Health and Social Care 2015. 2016. Available at: http://www.pssru.ac.uk/project-pages/unit-costs/2015/. Accessed November 25, 2016.

41.

Hammer

Lammert

Mejías

Kern

Frier

BM.

Costs of managing severe hypoglycaemia in three European countries. J Med Econ. 2009;12:281-290.

42.

NHS. NHS Reference costs 2015 to 2016. 2016. Available at: https://www.gov.uk/government/publications/nhs-reference-costs-2015-to-2016. Accessed December 15, 2016.

43.

Alva

Gray

Mihaylova

Leal

Holman

RR.

The impact of diabetes-related complications on healthcare costs: new results from the UKPDS (UKPDS 84). Diabet Med. 2015;32:459-466.

44.

Briggs

Uncertainty in cost-effectiveness models. Pharmacoeconomics. 2000;17:479-500.

45.

Schnell

Hummel

Weber

Economic and clinical aspects of diabetes regarding self-monitoring of blood glucose. Diabetes Technol Ther. 2008;10(s1):S72-S81.

46.

Menzin

Korn

Cohen

et al . Relationship between glycemic control and diabetes-related hospital costs in patients with type 1 or type 2 diabetes mellitus. J Manag Care Pharm. 2010;16:264-275.

47.

Menzin

Langley-Hawthorne

Friedman

Boulanger

Cavanaugh

Potential short-term economic benefits of improved glycemic control: a managed care perspective. Diabetes Care. 2001;24:51-55.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB