Cost-effectiveness of insulin degludec compared with insulin glargine in a basal-bolus regimen in patients with type 1 diabetes mellitus in the UK

Abstract

Objective:

The aim of this study was to evaluate the cost-effectiveness of insulin degludec (IDeg) vs insulin glargine (IGlar) as part of a basal-bolus treatment regimen in adults with T1DM, using a short-term economic model.

Methods:

Data from two phase III clinical studies were used to populate a simple and transparent short-term model. The costs and effects of treatment with IDeg vs IGlar were calculated over a 12-month period. The analysis was conducted from the perspective of the UK National Health Service. Sensitivity analyses were conducted to assess the degree of uncertainty surrounding the results. The main outcome measure, the incremental cost-effectiveness ratio (ICER), was the cost per quality-adjusted life-year (QALY).

Results:

IDeg is a cost-effective treatment option vs IGlar in patients with T1DM on a basal-bolus regimen. The base case ICER was estimated at £16,895/QALY, which is below commonly accepted thresholds for cost-effectiveness in the UK. Sensitivity analyses demonstrated that the ICER was stable to variations in the majority of input parameters. The parameters that exerted the most influence on the ICER were hypoglycemia event rates, daily insulin dose, and disutility associated with non-severe nocturnal hypoglycemic events. However, even under extreme assumptions in the majority of analyses the ICERs remained below the commonly accepted threshold of £20,000–£30,000 per QALY gained.

Conclusions:

This short-term modeling approach accommodates the treat-to-target trial design required by regulatory bodies, and focuses on the impact of important aspects of insulin therapy such as hypoglycemia and dosing. For patients with T1DM who are treated with a basal-bolus insulin regimen, IDeg is a cost-effective treatment option compared with IGlar. IDeg may be particularly cost-effective for sub-groups of patients, such as those suffering from recurrent nocturnal hypoglycemia and those with impaired awareness of hypoglycemia.

Keywords::

• Cost-effectiveness
• Insulin analogs
• Insulin therapy
• Type 1 diabetes
• Nocturnal hypoglycemia
• Pharmaco-economics

Introduction

Diabetes mellitus is a chronic metabolic disorder that persists over a patient’s lifetime and is associated with considerable morbidity and mortality1^,2. The disease is characterized by hyperglycemia (elevated levels of blood glucose) which, when prolonged, causes long-term complications including retinopathy, nephropathy, neuropathy, and cardiovascular disease1. These complications have a considerable impact on health, a negative impact on patient health-related quality-of-life (HRQoL)3^,4, and represent a significant proportion of the economic burden of diabetes. In the UK in 2010/11, the cost burden of diabetes was ∼10% of total NHS resource expenditure, with a direct cost of ∼£9.8 billion. The direct cost of diabetes in the UK is projected to increase to £13.9 billion by 2035/365. The clinical goal in the treatment of diabetes is to achieve good glycemic control with minimal hypoglycemia or other adverse effects of treatment. Tight control of blood glucose levels with intensive diabetes therapy prevents or delays microvascular complications and reduces cardiovascular and all-cause mortality6^,7, thus improving HRQoL3.

The International Diabetes Federation (IDF) estimated that, in 2013, ∼382 million people worldwide (8.3% of the global population) in the 20–79-year-age group had diabetes8. Type 1 diabetes mellitus (T1DM) accounts for ∼10% of all cases of diabetes and usually develops in children/adolescents2. T1DM is caused by the destruction of pancreatic β-cells, leading to absolute insulin deficiency. For people with T1DM, insulin is the only pharmacological treatment, and is highly efficacious when dosed appropriately. The current recommended therapies include basal-bolus regimens (injections of short-acting or rapid-acting insulin analog before meals, together with one or more separate daily injections of intermediate-acting insulin or long-acting insulin analog), and twice daily regimens (injections of short-acting or rapid-acting insulin analog mixed with intermediate-acting insulin)9.

Despite clear guidelines, many patients still experience sub-optimal glycemic control. In the UK only 6.5% of patients with T1DM are achieving a HbA_1c target below 6.5%2. A common side-effect of insulin therapy is hypoglycemia, which occurs when the plasma glucose level becomes too low. Hypoglycemia can occur suddenly and with varying severity; symptoms include pounding heart, trembling, hunger, sweating, difficulty concentrating, and confusion. Severe events can include seizure, coma, and even death10–12. Hypoglycemia has a major impact on patients’ lives in terms of physical, mental, and social functioning, and is a substantial cost burden to the community through increased treatment costs and reduced productivity5^,13^,14. Importantly, the negative consequences and unpleasant symptoms associated with hypoglycemia often lead to significant fear or anxiety of future events14. Nocturnal hypoglycemia is particularly detrimental to quality-of-life, as the usual warning symptoms may not be felt and recovery time is prolonged compared with daytime events13.

Current basal insulin analogs such as insulin glargine (IGlar) more closely mimic physiological insulin and have an improved time action profile compared with intermediate acting insulin (neutral protamine Hagedorn (NPH) insulin), resulting in reduced potential for hypoglycemia15. Thus, long-acting insulin analogs are recommended as a treatment option by the National Institute for Health and Care Excellence (NICE) for patients with T1DM9. However, there remains an unmet need for further refinements in insulin pharmacology, with the goal of achieving good glycemic control with minimal risk of hypoglycemia.

Insulin degludec (IDeg) is a basal insulin analog with an ultra-long duration of action and a distinct, slow absorption mechanism which results in a flat and stable action profile. Upon subcutaneous injection IDeg forms multi-hexamers, resulting in a soluble depot from which it is slowly and continuously absorbed into the circulation. This contributes to lower blood glucose variations compared with IGlar16^,17.

Two phase III, open label, treat-to-target trials investigated the efficacy and safety of IDeg vs IGlar, both administered once daily (OD) as part of a basal-bolus regimen in adults with T1DM18^,19. In both trials IDeg effectively improved glycemic control; IDeg was non-inferior to IGlar as measured by change from baseline in HbA_1c, as expected in treat-to-target studies. Both trials also demonstrated significantly lower insulin doses (daily basal, daily bolus, and daily total) in the IDeg treatment arm compared with the IGlar group18^,19. Data from the two trials were pooled in a meta-analysis which demonstrated that similar glycemic control was achieved with IDeg with a significantly lower rate of nocturnal non-severe hypoglycemia (estimated IDeg vs IGlar rate ratio 0.83 [95% CI = 0.69;0.99])20.

The aim of the current study was to evaluate the cost-effectiveness of IDeg vs IGlar in patients with T1DM. Historically, cost-effectiveness analyses of diabetes interventions have been performed by estimating long-term clinical consequences as a function of differences in glycemic control. However, as treat-to-target trials enforce a similar level of glycemic control across interventions, resulting in non-inferior and non-significant differences, long-term modeling based on HbA_1c differences is not necessary when evaluating these studies. We, therefore, used a short-term model which focuses on the impact of other important aspects of insulin therapy such as hypoglycemia and dosing, as described previously for patients with type 2 diabetes mellitus (T2DM) on a basal oral therapy regimen21.

Patients and methods

Model overview

A cost-effectiveness analysis comparing IDeg OD with IGlar OD in patients with T1DM was performed. Cost-effectiveness analyses combine the incremental cost of an intervention with the benefit it produces in terms of quality and quantity of life. This cost-effectiveness analysis is typically termed a cost-utility analysis as the benefit is measured in quality-adjusted life-years (QALYs). The main outcome measure, the incremental cost-effectiveness ratio (ICER), is the cost per QALY22. The ICER allows comparison of the cost-effectiveness of treatments across disease areas and is the preferred outcome measure of health technology assessment bodies such as NICE23. An ICER threshold of £20,000–£30,000 per QALY gained is generally considered to represent acceptable value for money in the UK24. IGlar was selected as the comparator in the two confirmatory IDeg trials18^,19, as it is the most widely used basal insulin analog globally, and has an established efficacy and safety profile25.

The cost-effectiveness of IDeg was analyzed over a 1-year time horizon, with the outcomes therefore representing the average annual cost-effectiveness in a steady state. This short-term approach was designed to accommodate the treat-to-target efficacy trials required by regulatory bodies26. As the time horizon for the analysis was 1-year, no discounting was applied. The analysis was conducted from the perspective of the UK National Health Service.

A simple and transparent short-term model was developed in Microsoft Excel 2010 (Figure 1). Costs include treatment costs and the costs associated with hypoglycemic events, and QALYs are calculated by applying a disutility per hypoglycemic event.

Figure 1. Model schematic. The model calculates treatment costs (including insulin, needles, and costs associated with self-monitored blood glucose [SMBG] testing) and the costs associated with hypoglycemic events (the resource utilization associated with a hypoglycemic event multiplied by the hypoglycemic event rate) for both IDeg and IGlar. QALYs are calculated by applying a disutility per hypoglycemic event to the baseline health utility for IDeg and IGlar.

Patients

The analysis was based on 958 patients with T1DM who participated in the two phase III clinical trials comparing IDeg once daily (OD) + IAsp three times daily (TID) (n = 629) with IGlar OD + IAsp TID (n = 329)18^,19. Eligible subjects were adults with clinically diagnosed T1DM who had been treated with any basal-bolus regimen for at least 1 year, HbA_1c ≤10% and BMI ≤35 kg/m². Subjects with recurrent severe hypoglycemia (>1 severe event during the last 12 months), hypoglycemic unawareness, or significant concomitant illness were excluded from the trials.

Data used in the model

Clinical data

Insulin use

Daily use of basal and bolus insulin for the IDeg and IGlar treatment groups was captured from the clinical trial data. The IGlar dose and IDeg/IGlar dose ratio were derived from a meta-analysis of insulin dose20 and were used to calculate the corresponding IDeg dose (Table 1), in order to allow for adjustment of covariate factors such as trial, treatment anti-diabetic therapy at screening, age, sex, region, and baseline insulin dose.

Table 1. Insulin use.

Insulin use	Basal insulin	Bolus insulin
IGlar group† (units/day)	33.10	35.00
Dose ratio† (IDeg/IGlar)	0.87*	0.88*
IDeg group‡ (units/day)	28.80	30.80

*p < 0.05.

†Derived from a meta-analysis of insulin dose20;

‡Calculated from IGlar dose and dose ratio.

Needle use

Based on the instructions for needle use in the clinical trials and the Forum for Injection Technique (FIT) recommendations for needle use in the UK27, it was assumed that patients administer one basal injection and three bolus injections per day, and use a new needle for each injection. It was assumed that the NovoFine needle was used in the IDeg group for both basal and bolus injection, whereas patients in the IGlar group were assumed to use NovoFine needles for bolus injections (insulin aspart) and ClickFine needles for basal injections.

Self-monitored blood glucose (SMBG) testing

For basal injections, the titration schedule recommended for use with IGlar28 was used (seven SMBG tests per week), and for bolus injections it was assumed that one SMBG test was conducted with every main meal (21 tests per week). Thus, patients were assumed to test their blood glucose (for the purposes of monitoring and titrating their insulin dose) 4 times per day (or 28 times per week). Patients may monitor their blood sugar for reasons other than the monitoring and titration of their insulin (e.g., before exercise or driving). However, this impact was assumed similar across treatments and the additional tests were not included in the economic evaluation.

The long half-life of IDeg, with its corresponding flat and stable profile in steady state and low blood glucose variability, means that patients can titrate, predict, and monitor their blood glucose more efficiently29. IDeg, therefore, has the potential to be monitored and titrated with only two SMBG tests per week (rather than seven). However, due to expected inertia in changing established testing patterns, a reduction in SMBG testing was only explored in the sensitivity analyses.

Hypoglycemia event rates

The baseline values for severe and non-severe hypoglycemic event rates were derived from the clinical trial data. The clinical trial event rates were taken as the base-case event rates for the IGlar group. Event rates for the IDeg group were calculated using the relative rate ratio of hypoglycemia (IDeg vs IGlar) derived from the meta-analysis of hypoglycemia20. In this meta-analysis, rates of hypoglycemia were analyzed in mutually exclusive groups: severe events, non-severe events occurring during the day (diurnal), and non-severe events occurring during the night (nocturnal), to avoid the potential double counting of events (Table 2). Hypoglycemia event rates from other published sources30^,31 were explored in the sensitivity analyses.

Table 2. Hypoglycemic event rates.

Calculation of hypoglycemic event rates	Non-severe hypoglycemia		Severe hypoglycemia
Baseline hypoglycemia rate (IGlar)	53.85		0.27
Daytime/nocturnal split‡	Daytime	Nocturnal	–
	86.64%	13.36%
Total events per patient per year for IGlar	46.66	7.19	0.27
IDeg/IGlar hypoglycemic event rate ratio§	NS (1.00)	0.83*	NS (1.00)
Calculated IDeg hypoglycemic event rate	46.66	5.97	0.27

NS, non-significant.

*p < 0.05.

‡Proportion of daytime/nocturnal events for IGlar taken from the clinical trials.

§Derived from meta-analysis of hypoglycemia20, relative rates were adjusted for trial, type of diabetes, treatment, anti-diabetic therapy at screening, sex, and region as fixed factors, and age as a continuous covariate. In the case of non-significant results a relative rate of one was used in the calculation.

Cost data

Cost of insulin, needles, and SMBG tests

The costs of insulin, needles, SMBG test strips, and lancets were based on prices published in the Monthly Index of Medical Specialities (MIMS) November 2013 (Table 3).

Table 3. Unit costs for insulin, needles, and SMBG tests.

	Product	Price per pack†	Units per pack	Price per unit
Insulin	Degludec (in FlexTouch® pen)	£72.00	1500	£0.0480
	Glargine (in Solostar® pen)	£41.50	1500	£0.0277
Needles‡	NovoFine® 8 mm 30G	£9.24	100	£0.0924
	ClickFine® 8 mm	£9.11	100	£0.0911
SMBG tests§	Test strip (OneTouch® Ultra®)	£15.00	50	£0.30
	Lancet (OneTouch® Ultra®)	£3.49	100	£0.0349
	Unit Cost, SMBG test			£0.3349

SMBG, self-measured blood glucose.

†Based on MIMS November 2013.

‡It was assumed that the NovoFine® needle was used with IDeg, the ClickFine® needle was used with IGlar, and that all patients used a new needle for every injection as recommended.

§The unit cost of one SMBG test was calculated as the price of one test strip plus one lancet.

Cost of hypoglycemic events

The direct cost of a single hypoglycemic event was comprised of the cost of treating the hypoglycemic event itself, plus the cost of additional SMBG tests in the week following the event.

The proportion of patients contacting a hospital/healthcare professional (HCP) was based on data derived from patient hypoglycemia safety questionnaires, completed during the clinical trials. For severe events, data were obtained from serious adverse event case reports. Both questionnaires and case reports were completed by the investigator or study nurse to ensure validation of the patient-reported data. It was assumed that all patients who were hospitalized used the ambulance service to get to hospital.

The number of additional SMBG tests in the week following a non-severe event was also based on the hypoglycemia safety questionnaire, and was an average of 1.46 tests per non-severe event. As no data on the testing pattern following a severe event were available, it was conservatively assumed that a similar number of additional tests were used following a severe event as a non-severe event. The number of additional SMBG tests in the week following the event is likely to be conservative; an observational study reported that patients in the UK used an average of 6.2 extra SMBG tests in the week following a non-severe event32.

The average costs of a severe and non-severe event were estimated at £219.87 and £0.85, respectively (Table 4). These costs are lower than those reported for patients with T2DM on a basal oral therapy regimen21 due to different resource utilization by the two patient groups. In the clinical trials, fewer patients with T1DM used the ambulance and hospital for severe events, contacted a HCP following a non-severe event, and conducted additional SMBG tests in the week following a hypoglycemic event. This lower utilization is likely due to a long learning curve; patients with T1DM have been using insulin for longer and they (and their friends and family) may have developed experience in dealing with hypoglycemia.

Table 4. Total cost of an average severe/non-severe hypoglycemic event, T1DM.

		Utilization per hypoglycemic event
		Severe		Non-severe
	Unit cost	Rate	Cost	Rate	cost
Ambulance	£256.98†	0.14	£35.98	0.0	£0.00
Hospital	£1310‡	0.14	£183.40	0.0	£0.00
GP/diabetes clinic visit	£36§	0	£0.00	0.01	£0.36
SMBG	£0.3349¶	1.46	£0.49	1.46	£0.49
Total			£219.87		£0.85

GP, general practitioner; SMBG, self-measured blood glucose.

†ISD Scotland, R910 – Scottish Ambulance Service, expenditure and statistics33.

‡HRG charge: ‘Tariff for Admitted Patient Care & Outpatient Procedures’: Diabetes with Hypoglycaemic Disorders ≤69 years34.

§Unit Costs of Health & Social Care 201135 using the fee for ‘Per surgery consultation lasting 11.7 minutes’ in the table ‘10.8b General practitioner—unit costs’.

¶MIMS November 2013.

Utility data

QALYs were calculated by applying a disutility (reduction in quality-of-life) per hypoglycemic event. To estimate the incremental impact of IDeg over 1 year, the disutility per hypoglycemic event was multiplied by the annual number of events observed in the treatment group (see Table 2 for event rates). This was done for severe and for non-severe events separately.

The disutility incurred per hypoglycemic event was obtained from a recent time trade-off (TTO) study by Evans et al.36. This large-scale international study involved more than 8000 respondents in the UK, US, Canada, Germany, and Sweden. The TTO approach is an established health economic methodology that requires subjects to consider the relative amounts of time (for example, number of life-years) that they would be prepared to sacrifice to avoid a certain poorer health state. Descriptions of hypoglycemia health states (well-controlled diabetes and diabetes combined with hypoglycemia of differing event types and frequencies) were derived from a survey of 247 UK patients with diabetes37. Based on individual responses, the average utility value for each health state was calculated, and a disutility per hypoglycemic event derived by dividing the difference between the average utility and the baseline diabetes state utility by the number of annual events, to ensure that the resulting value reflected the effect of hypoglycemia alone. The TTO approach directly elicits preferences for acute health states that are difficult to evaluate with EQ-5D, and this study provides a quantitative disutility value for the three mutually exclusive hypoglycemic event classes (non-severe daytime, non-severe nocturnal, and severe). Evans et al.36 reported a disutility of 0.0565 for a severe event and disutilities of 0.0041 and 0.0067 for non-severe daytime and non-severe nocturnal events, respectively.

There was no difference between IDeg and IGlar in daytime non-severe and severe hypoglycemia rates (and therefore no difference in utility). The annual rate of nocturnal non-severe events was 5.97 events/year for IDeg and 7.19 events/year for IGlar (a difference of 1.22 events/year). This results in a utility increment for IDeg-IGlar of 0.0082 (−1.22 × −0.0067).

Sensitivity analysis

One-way sensitivity analysis

One-way sensitivity analyses were performed to assess the impact of varying key assumptions on the results, and to identify key drivers of the outcomes. The parameters that were varied are shown in Table 7 and summarized below.

Hypoglycemia event rates

The base case rates of hypoglycemia were taken from the clinical trial data. In the sensitivity analyses, alternative event rates from published sources (UK Hypoglycaemia Study group38, Donnelly et al.30, and Östenson et al.31) were investigated. In addition, the hypoglycemia rates observed in the maintenance period of the clinical trials (week 16 to end of trial) were used. The maintenance period represents the time during which a stable dose of basal insulin was expected to be obtained for the majority of subjects, and was defined by regulatory authorities prior to unblinding of clinical trial data. The maintenance period is important as this reflects the benefits most patients would experience over the duration of a chronic disease.

Disutility for hypoglycemia

The base case disutility incurred per hypoglycemic event was obtained from a recent large-scale TTO study36. Alternative disutilities, reported by Currie et al.39, were explored in the sensitivity analysis.

Hypoglycemia costs

An alternative published source32 for the cost of non-severe hypoglycemia was investigated. As there is no difference between treatments in the rate of severe hypoglycemia, the costs associated with severe hypoglycemia were not explored in sensitivity analyses.

Utility for flexible dosing

While it is recommended to inject basal insulin at the same time every day, the stable action profile of IDeg, combined with the low day-to-day variation in glucose-lowering effect at steady state, allows subjects to advance or delay the daily administration of IDeg if needed, without compromising short-term glycemic control and risk of hypoglycemia18^,40. Two studies have estimated the utility associated with a flexible dosing regimen. Boye et al.41 estimated the utility benefit of the option of flexible dosing time with IDeg to be 0.006 (p < 0.05), and a recent, large TTO study demonstrated that time-flexible basal injections (as part of a basal-bolus regimen) are associated with a 0.013 higher utility vs a fixed time of injection42. These utility values were explored in the sensitivity analysis to accommodate the possibility of dosing flexibly when needed.

SMBG

The base case analysis assumes that an equal number of SMBG tests are performed per week by patients in the two treatment groups. However, after overcoming the initial inertia in changing established testing habits, a reduction in SMBG testing may be observed with IDeg. The sensitivity analysis examined the effect of a reduction in the number of weekly SMBG tests with IDeg compared with IGlar.

Twice daily dosing of IGlar

For a proportion of patients with T1DM, basal insulin needs to be taken twice daily (BID) to ensure optimal glycemic control43–45. These individuals will, therefore, administer a total of five insulin injections per day (two basal and three bolus). The half-life of 25.4 h with IDeg (vs 12.5 h for IGlar17) may allow these patients to be successfully treated with just one basal injection daily, thus removing the burden of additional injections. The sensitivity analysis explored the effect of using twice as many needles for basal injections in the IGlar group, with either all or a proportion of patients assumed to use IGlar twice-daily. This affects cost only, efficacy and insulin dose are assumed to be the same.

Extension study data

An extension to one of the clinical trials has been conducted that extends the treatment period up to 104 weeks46. Data on insulin dose and hypoglycemia rates from the extension study were used to explore the impact on the results.

Insulin dose

Finally, the effect of varying the dose of both IDeg and IGlar was explored. In the base case analysis the IDeg/IGlar dose ratios were based on the findings from the clinical trials, sensitivity analyses investigated assumptions of equal doses (basal, bolus, and both) and doses from an alternative published source47 (applying both clinical trial dose ratios and equal doses).

Probabilistic sensitivity analysis

To capture the uncertainty of the results caused by statistical uncertainty with respect to the stochastic parameter inputs, a probabilistic sensitivity analysis (PSA) was conducted. PSA allows the model parameters to be varied simultaneously within a plausible range, and allows an estimate of the certainty that the intervention in question is cost-effective at different thresholds of cost-effectiveness. The standard errors around the parameters were used and a log-normal distribution around the rate-ratios and normal distributions around the continuous variables were assumed (see Table 5). The PSA was run with 10,000 iterations. In the primary analysis the standard error was applied to differences that were statistically significant. That is, if there was no statistical significance proven, then the rate ratio was set to 1 (assumed equivalent) and the SE was set to 0 (so as not to introduce random uncertainty).

Table 5. PSA input variables.

Hypoglycemia rate-ratios	Ratio	SE of log ratio	p-value	Distribution
Daytime	1.00	0	0.0716	Lognormal
Nocturnal	0.83	0.1	0.0495	Lognormal
Severe	1.00	0	0.6565	Lognormal
Insulin dose	Mean	SE		Distribution
Basal	33.10	1.110		Normal
Bolus	35.00	1.676		Normal
Insulin dose ratios	Ratio	SE of log ratio	p-value	Distribution
Basal	0.87	0.02	0.0001	Lognormal
Bolus	0.88	0.03	0.0001	Lognormal

SE, standard error.

Results

The results of the base case analysis are shown in Table 6. IDeg was found to be a cost-effective treatment option vs IGlar in adults with T1DM. The ICER was estimated at £16,895/QALY (Table 6), which is below commonly accepted thresholds for cost-effectiveness in the UK. The total cost per patient in the IDeg group was £138 higher than in the IGlar group, primarily due to the increased cost of insulin.

Table 6. Results of the base case analysis.

	IDeg (£/year)	IGlar (£/year)	Incremental cost (£/year)
Pharmacy costs
Insulin	734.36	595.27	139.09
Needles	135.00	134.52	0.47
SMBG tests	489.29	489.29	0.00
Hypoglycemic events
Non-severe daytime events	42.87	42.87	0.00
Non-severe nocturnal events	5.49	6.61	–1.12
Severe events	843.10	843.10	0.00
Total costs	2250.11	2111.7	138.44
Incremental QALYs (IDeg–IGlar)	0.0082
ICER (cost/QALY)	£16,894.70

QALY, quality-adjusted life-year.

One-way sensitivity analysis

The results of the one-way sensitivity analyses are shown in Table 7. In the majority of analyses the ICERs remained below the commonly accepted threshold of £20,000–£30,000 per QALY gained.

Table 7. Results of the one-way sensitivity analyses.

		ΔCost (£)	ΔQALY	ICER (£/QALY)
Base case results		138.44	0.0082	16,895
Hypoglycemia rates Base case: Non-severe = 53.85/year Severe = 0.27/year; % nocturnal = 13.36%; IDeg/IGlar nocturnal event rate ratio = 0.83	Östenson31: non-severe: 91, severe: 0.7	136.44	0.0228	5,983
	Donnelly 200530: non-severe: 41.74, severe: 1.15	138.69	0.0064	21,836
	UKHSG (T1DM using insulin >15 years): non-severe: 29, severe: 3.2	138.96	0.0044	31,489
	Maintenance period: non-severe: 45.16, severe: 0.27; % nocturnal: 13.62%; IDeg/IGlar nocturnal event rate ratio: 0.75	138.15	0.0103	13,410
	Hypoglycemia disutilities Base case: Non-severe daytime = 0.0041, non-severe nocturnal = 0.0067; severe = 0.0565	Currie 200639; Non-severe 0.0035	138.44	0.0043	32,341
Hypoglycemia costs Base case: Non-severe = £0.85	Published non-severe cost £11.30 (calculated from Brod 201132)	125.75	0.0082	15,345
Flexible dosing utility Base case: not included	Evans 201342: Flexible dosing utility = 0.013	138.44	0.0212	6,532
	Boye 201141: Flexible dosing utility = 0.006	138.44	0.0142	9,753
SMBG rates Base case: equal	–2 SMBG tests/week in the IDeg arm	103.49	0.0082	12,630
	–3 SMBG tests/week in the IDeg arm	86.02	0.0082	10,497
Injection frequency Base case: IGlar 1/day, IDeg 1/day	IGlar: 2/day	105.17	0.0082	12,834
	Weighted (25.44% IGlar BD)*	129.98	0.0082	15,862
2 year extension study	Whole treatment period – IDeg dose ratio 0.91; hypoglycemia rates: non-severe: 46.41, severe: 0.27; % nocturnal: 13.66%; IDeg/IGlar nocturnal event rate ratio: 0.74	173.30	0.0110	15,692
	Maintenance period – IDeg dose ratio 0.91; hypoglycemia rates: non-severe: 40.55; severe: 0.27; % nocturnal: 13.93%; IDeg/IGlar nocturnal event rate ratio: 0.69	173.20	0.0117	14,763
Insulin dose Base case: Basal (IGlar): 33.1 U/day Bolus (IAsp): 35 U/day Ratio (IDeg/IGlar) = 0.87 (bolus ratio = 0.88)	Equal basal dose (1:1 ratio)	213.88	0.0082	26,101
	Equal bolus dose (1:1 ratio)	169.74	0.0082	20,714
	Equal doses (1:1 ratio for both basal and bolus)	245.18	0.0082	29,920
	Heller 2009 doses47+ ratios from the clinical trials (basal ratio 0.87, bolus ratio 0.88)	123.12	0.0082	15,025
	Heller 2009 doses47+ equal(1:1 ratio for both basal and bolus)	209.53	0.0082	25,570

BD, twice daily; SMBG, self-measured blood glucose; UKHSG, UK Hypoglycemia Study Group. *BD weighting calculated from trial 3583 study report.

Hypoglycemic event rates had an important effect on the results. Drawing on higher event rates from the study of Östenson et al.31, IDeg was highly cost-effective vs IGlar, with an estimated ICER of £5983/QALY. Conversely using the lower rates of hypoglycemia reported by the UKHSG, the ICER increased to £31,489/QALY, on the border of cost-effectiveness thresholds. Increasing the cost of non-severe hypoglycemia from £0.85/event to £11.30/event had little impact on the ICER; however, using a lower disutility per non-severe hypoglycemic event increased the ICER to £32,341/QALY.

Insulin dose also influenced the results. In a meta-analysis of insulin doses used in the clinical trials, a 13% reduction in mean daily basal insulin dose was observed with IDeg vs IGlar20, giving a dose ratio of 0.87 for use in the base case analysis. When equal mean doses were assumed, the ICERs were borderline cost-effective (£20,714–£29,920/QALY). Using daily insulin doses from an alternative published source47 had little impact on the ICER.

IDeg was found to be highly cost-effective vs IGlar when published utility values associated with flexible dosing were applied. The ICERs were £9753/QALY and £6532/QALY when utility benefits of 0.00641 and 0.01342, respectively, were applied. A reduction in the rate of SMBG testing also influenced the ICERs. A reduction of two SMBG tests per week with IDeg compared with IGlar produced an ICER of £12,630/QALY and a reduction of three SMBG tests per week further reduced the ICER to £10,497/QALY.

When the effect of injecting IGlar twice daily (vs once-daily IDeg) was explored, the ICERs were slightly reduced compared with the results of the base case analysis. Finally, using data from the extension study had little impact on the ICERs, they were reduced slightly compared with the base case analysis.

Probabilistic sensitivity analyses

At the £20,000/QALY and £30,000/QALY willingness-to-pay thresholds (thresholds considered to represent acceptable value for money in the UK), the probability that IDeg was cost-effective relative to IGlar was 55.98% and 67.89%, respectively (Figure 2).

Figure 2. Cost-effectiveness acceptability curve—IDeg vs IGlar.

Discussion

There have been many advances and improvements in insulin therapy since its implementation almost a century ago, however there are still notable limitations associated with existing treatments. Insulin degludec is the first new treatment for patients with T1DM in the UK for almost 10 years.

Decision-making based on clinical and economic evidence is essential in order to optimize resource use and service delivery for patients with T1DM. The results of our short-term cost-utility model show that, for adults with T1DM using a basal-bolus regimen, IDeg is a cost-effective treatment option compared with IGlar. The ICER was estimated at £16,985/QALY (Table 6), which is below commonly accepted thresholds for cost-effectiveness in the UK24.

Compared with IGlar, IDeg is associated with an incremental treatment cost of £138 per patient, primarily due to the increased cost of insulin. However, treatment with IDeg is associated with an improvement in quality-of-life. The two confirmatory trials18^,19 showed that IDeg reduces the risk of nocturnal hypoglycemia compared with IGlar, at a similar level of glycemic control. The lower rates of nocturnal confirmed hypoglycemia were confirmed by a meta-analysis of pooled trial data20, which demonstrated a 17% lower rate of nocturnal non-severe hypoglycemia (p < 0.05) with IDeg vs IGlar.

Hypoglycemia and fear of hypoglycemia may cause patients to reduce or omit an insulin dose, leading to inadequate glycemic control and increased risk of long-term complications48. Thus, consideration of hypoglycemia is important when evaluating the efficacy and safety of novel insulin therapies. Rates of nocturnal hypoglycemia are particularly relevant in a basal-bolus setting as, with these regimens, night time episodes are more reflective of the action of the basal insulin component than the bolus component. The lower rate of nocturnal confirmed hypoglycemia with IDeg vs IGlar may be a function of the pharmacokinetic profile and lower day-to-day variability in glucose lowering action of IDeg.

Hypoglycemia exerts significant indirect societal costs that are not considered in our model. A recent study of the economic burden of nocturnal hypoglycemia identified major consequences for patients, including lost work hours, reduced work productivity, and additional medical costs due to falls or injuries incurred during the event49. Our evaluation is, therefore, likely to be conservative and may under-estimate the economic value of IDeg from a societal perspective.

Sensitivity analysis showed that hypoglycemic event rates had a notable impact on the ICER, a consequence of the reported difference in nocturnal hypoglycemia between IDeg and IGlar. Using the higher event rate from the study of Östenson et al.31, IDeg was highly cost-effective vs IGlar, with an estimated ICER of £5983/QALY. The study by Östenson et al.31 investigated the rates of self-reported hypoglycemia in 3827 subjects with T1DM and T2DM in real-world clinical practice, across seven European countries. These rates may represent a better estimation of real-life event rates than data from clinical trials. Hypoglycemic event rates from clinical trials suffer from both bias in the selection of patients and the setting in which the treatment occurs. First, the clinical trials excluded patients with a history of severe hypoglycemic episodes (more than one severe hypoglycemic event in the previous 12 months) and anyone considered to have impaired awareness of hypoglycemia (IAH), which is known to be a strong marker for future hypoglycemic episodes30^,50–52. Second, in the clinical trials only hypoglycemic events confirmed by a SMBG reading <3.1 mmol/L were included (to ensure high data validity), which may under-estimate the number of events where patients in real-life would experience the negative consequences of hypoglycemia, potentially causing them to seek healthcare professional (HCP) contact and use additional SMBG tests.

Another parameter that had a noticeable influence on the ICER was the disutility associated with a non-severe nocturnal hypoglycemic event. Using the lower disutility value reported by Currie et al.39 increased the ICER to £32,341/QALY. However, the TTO study by Evans et al.36 used in the base case analysis was large-scale, involving over 8000 respondents, and used an established, robust health economic methodology. It also reported a quantitative disutility value for mutually exclusive hypoglycemic event classes, that is: severe, non-severe daytime, and non-severe nocturnal hypoglycemia, and thus provides a more accurate estimate.

Daily insulin dose also influenced the ICERs, indicating that this is another key driver of the results. In the clinical trials daily insulin dose was significantly lower with IDeg vs IGlar: Heller et al.18 observed a mean basal insulin dose reduction of 14% with IDeg at end-of-trial, as well as bolus and total insulin dose reductions of 10% and 11%, respectively (all statistically significant); and Mathieu et al.19 reported that mean basal, bolus, and total insulin doses were 4%, 18%, and 11% lower, respectively, in the IDeg treatment arm. However, in sensitivity analyses, even when insulin dose (basal, bolus, and both) was assumed to be equal, the ICERs were still below the £30,000/QALY cost-effectiveness threshold.

IDeg is highly cost-effective compared with IGlar when the advantage of flexible dosing or the cost saving associated with reduced SMBG testing is applied. These potential benefits were not included in the base case analysis as it could be argued that they will not be realized by patients with T1DM who need to administer bolus insulin injections three times daily and perform regular SMBG testing. However, over time, the low variability and ultra-long duration of action of IDeg may lead to patients becoming increasingly confident in their glycemic control, and therefore able to reduce the frequency of SMBG testing.

As previously reported for patients with T2DM on a basal oral therapy regimen21, specific groups of patients with T1DM may particularly benefit from treatment with IDeg. In these sub-populations IDeg is likely to provide even better value for money. First, those who suffer from recurrent nocturnal hypoglycemia are highly likely to benefit, given the reduced risk of these events with IDeg compared with IGlar53. Nocturnal hypoglycemia is of particular importance for patients as it often goes undetected and can have serious consequences, including unconsciousness and death51^,54. Several international studies exploring the impact and burden of nocturnal hypoglycemia found wide-ranging negative consequences for patients13^,55^,56. These include sleep disruption, reduced next-day functioning, poor emotional state (feelings of anxiety, sadness, and helplessness), reduced social interaction, and absence from work.

Those with IAH are also likely to benefit from the reduced risk of hypoglycemia with IDeg treatment. IAH, defined as a reduced ability or failure to recognize hypoglycemia at the physiological plasma glucose concentration at which warning symptoms normally occur52, affects ∼20% of adults with T1DM57 and can adversely affect quality-of-life. Severely affected patients are susceptible to sudden onset of confusion, unusual behaviour, or loss of consciousness without warning and, in extreme cases, may need a relative or carer with them at all times52.

IDeg is likely to be a favorable treatment option for those who currently require twice-daily (BID) dosing of their basal insulin. Studies have shown that between 24–41% of patients with T1DM split their daily basal dose into two separate injections in order to achieve a balance between glycemic control and the risk of hypoglycemia43–45. The ultra-long action profile of IDeg may enable these patients to be successfully treated with just one daily basal dose, thus reducing daily needle use and the burden of additional injections. The sensitivity analysis explores the impact of one additional needle per day for patients in the IGlar group. This is conservative as there are several other assumptions that could have been included for BID dosing; higher costs associated with glucose monitoring to adjust basal insulin before each injection, and decreased utility due to more frequent injections and SMBG42. There is also evidence that T1DM patients using BID dosing have a higher average basal insulin dose than those dosing once-daily43^,58.

Finally, the potential for flexible dosing with IDeg may benefit those who find it difficult to adhere to a strict dosing schedule (such as shift workers or frequent travelers), or who rely on a carer to administer their insulin.

As with all model-based evaluations, our analysis was limited by the quality of input parameters. However, meta-analyses using clinical trial data were used in our model to increase the sample size and power of the parameter estimates derived from the individual trials. The model only used parameter estimates for which a statistically significant difference between the treatment arms was documented, and assumed that all other differences were due to random variation.

Our assessment of the cost-effectiveness of IDeg vs IGlar is limited by several factors. First, the data used to inform the model are derived from a highly selected clinical trial population which is not totally representative of the general population. Second, the cost estimates used for hypoglycemic events are derived from a variety of sources and, as such, may or may not reflect the true health economic impact of hypoglycemia. Third, the paucity of real-world data prevents a clear understanding of how the observed hypoglycemia benefit for IDeg will translate into clinical practice. Finally, our model is based on comparable glycemic control between IDeg and IGlar, and a lower insulin dose requirement with IDeg. In clinical practice, however, the time action profile of IDeg and a lower potential risk of hypoglycemia may translate into more effective insulin dose titration and differences in glucose control. Further economic evaluation based on more extensive clinical experience would help to address these considerations.

Conclusion

In summary, our short term cost-utility model demonstrates that IDeg provides a cost-effective alternative to IGlar for adults with T1DM on a basal-bolus insulin regimen. IDeg would be a particularly cost-effective treatment option for sub-groups of patients, such as those suffering from recurrent nocturnal hypoglycemia and those with IAH.

Transparency

Declaration of funding

This study was funded by Novo Nordisk. MW, JG, BC, and TC are employees of Novo Nordisk. All authors contributed to the design, conduct/data collection, analysis, and interpretation of results, and to the writing and final approval of this manuscript.

Declaration of financial/other relationships

ME has received honoraria and research awards from Novo Nordisk, Sanofi Aventis, MSD, and Novartis. MW, JG, BC, and TC are employees of Novo Nordisk. JME Peer Reviewers on this manuscript have no relevant financial relationships to disclose.

Acknowledgments

The authors acknowledge writing and editorial support from Abacus International (sponsored by Novo Nordisk).