Evaluation of the cost and medical resource use outcomes associated with nasal glucagon versus injectable glucagon for treatment of severe hypoglycemia in people with diabetes in Canada: a modeling analysis

Abstract

Objectives

Treatments for severe hypoglycemia aim to restore blood glucose through successful administration of rescue therapy, and choosing the most effective and cost-effective option will improve outcomes for patients and may reduce costs for healthcare payers. The present analysis aimed to compare costs and use of medical services with nasal glucagon and injectable glucagon in people with type 1 and 2 diabetes in Canada when used to treat severe hypoglycemic events when impaired consciousness precludes treatment with oral carbohydrates using an economic model, based on differences in the frequency of successful administration of the two interventions.

Methods

A decision tree model was prepared in Microsoft Excel to project outcomes with nasal glucagon and injectable glucagon. The model structure reflected real-world decision-making and treatment outcomes, based on Canada-specific sources. The model captured the use of glucagon, emergency medical services (EMS), emergency room, inpatient stay, and follow-up care. Costs were accounted for in 2019 Canadian dollars (CAD).

Results

Nasal glucagon was associated with reduced use of all medical services compared with injectable glucagon. EMS call outs were projected to be reduced by 45%, emergency room treatments by 52%, and inpatient stays by 13%. Use of nasal glucagon was associated with reduced direct, indirect, and combined costs of CAD 1,249, CAD 460, and CAD 1,709 per severe hypoglycemic event, respectively, due to avoided EMS call outs and hospital costs, resulting from a higher proportion of successful administrations.

Conclusions

When a patient with type 1 or type 2 diabetes is being treated for a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate, use of nasal glucagon was projected to be dominant versus injectable glucagon in Canada reducing costs and use of medical services.

Keywords:

• Diabetes mellitus
• glucagon
• hypoglycemia
• Canada
• costs

JEL CLASSIFICATION CODES:

• I10
• I1
• I
• I19

Introduction

Untreated severe hypoglycemia can result in cognitive impairment, loss of consciousness, seizure, coma, or death¹. Moreover, severe hypoglycemia is associated with increased risk of cardiovascular disease and mortality over the longer term, with 5-year mortality found to be 3.4 times higher in patients with diabetes who reported severe hypoglycemia compared with those who reported no/mild hypoglycemia^2,3. Analysis of 15 trials conducted as part of an insulin clinical trial program suggests that severe hypoglycemic events are most common in people with type 1 diabetes receiving basal-bolus therapy, followed by patients with type 2 diabetes using multiple daily insulin injections, and a small number of events occur in people with type 2 diabetes receiving basal-oral therapy⁴. Real-world studies have shown that severe hypoglycemia remains a challenge for people with diabetes in Canada. In a retrospective analysis of 552 people with diabetes (17% type 1 diabetes, 83% type 2 diabetes) treated with insulin and/or insulin secretagogues, severe hypoglycemia was reported by 41.8% of all respondents in the last year, at an average rate of 2.5 events per person-year⁵. Similar outcomes were observed in the Canadian cohort of the global Hypoglycemia Assessment Tool (HAT) study⁶. The study enrolled 498 patients with diabetes (37% type 1 diabetes, 63% type 2 diabetes) treated with insulin, and assessed hypoglycemic events retrospectively over 6 months and prospectively over 4 weeks. In the retrospective period, annualized event rates of severe hypoglycemia were 2.5 and 2.0 events per patient per year in those with type 1 and type 2 diabetes, respectively. Similarly, during the prospective period, 1.9 and 1.8 severe events per patient per year were observed in those with type 1 and type 2 diabetes, respectively.

The use of continuous glucose monitoring and flash glucose monitoring may provide opportunities for patients, particularly those receiving intensive insulin therapy, to reduce the frequency of severe hypoglycemia, but technology use does not eliminate the risk entirely^7–9. Severe hypoglycemia represents a barrier to achieving good glycemic control in patients with diabetes, as in an effort to avoid hypoglycemia, patients may target higher blood glucose levels and thereby increase their risk of related complications^10,11. Severe hypoglycemic events can result in significant medical resource use, such as emergency medical services (EMS) and treatment in the emergency room, and therefore an economic burden to healthcare payers^12,13. The majority of severe hypoglycemic events occur outside of a healthcare setting. A survey conducted in 184 people with diabetes and 140 caregivers found that 87.1 and 87.7% of severe hypoglycemic events occurred at home in people with type 1 and type 2 diabetes, respectively¹⁴.

Treatments for severe hypoglycemia aim to restore blood glucose to physiologic levels through the administration of a rescue therapy, namely oral carbohydrates, intravenous dextrose, or glucagon¹⁵. Diabetes Canada guidelines recommend that severe hypoglycemia is treated with 20 g of oral carbohydrate if the patient is able to swallow¹⁵. Patients who are unconscious, unable to swallow or tolerate oral carbohydrates, and have intravenous access are recommended to receive 10–25 g of glucose intravenously over 1–3 min, while administration of 1 mg of glucagon intramuscularly or subcutaneously is recommended in patients without intravenous access¹⁵. Glucagon treatments are particularly important for severe hypoglycemic events occurring outside of a healthcare setting, as patients will not have intravenous access. For administration by laypeople, injectable (intramuscular or subcutaneous) and intranasal glucagon are currently available in Canada for treatment of severe hypoglycemic events in insulin-treated people with diabetes when impaired consciousness precludes treatment with oral carbohydrates¹⁵.

Administration of injectable glucagon is a multi-step process that requires reconstitution of glucagon powder using a syringe and needle, followed by injection of the glucagon solution into the affected person¹⁶. The glucagon kit must be administered by a family member or acquaintance who may or may not have been trained to use the kit. Difficulties may arise during the administration of injectable glucagon, including handling difficulties (such as opening the package, sheath removal, mixing, bent needles), syringe separation and injection being aborted entirely, and injection of air or water only¹⁷. These difficulties may be enhanced during the stressful emergency situation of a severe hypoglycemic event, particularly if training on the use of injectable glucagon was not completed recently¹⁷.

Increasing the proportion of severe hypoglycemic events that are successfully treated may improve outcomes for people with diabetes, and may reduce the use of costly medical services, such as EMS. Nasal glucagon was developed to address this unmet need¹⁸. Nasal glucagon contains glucagon powder in a ready-to-use, single-use, one-step dispensing device that, upon pressing a plunger, releases a 3 mg dose into the patient’s nostril where it is passively absorbed in the nasal mucosa^19,20. The dispensed powder does not need to be inhaled and is therefore well-suited for administration in unconscious patients. The nasal glucagon device can be stored at temperatures up to 30 °C for 2 years²¹. In a clinical trial of adults with type 1 diabetes, nasal glucagon was demonstrated to be non-inferior to injectable glucagon in terms of the percentage of patients successfully treated, where success was defined as an increase in plasma glucose to ≥70 mg/dL or an increase of ≥20 mg/dL from the glucose nadir, within 30 min of receiving study glucagon, without receiving additional actions to increase the plasma glucose level²². As neither reconstitution nor injection is necessary for the use of nasal glucagon, caregivers, family, friends or passers-by are able to treat a severe hypoglycemic event with greater success than with injectable glucagon. A simulation study demonstrated that 94% of instructed caregivers and 93% of non-instructed acquaintances successfully administered a full dose of nasal glucagon, compared with only 13% of caregivers and 0% of non-instructed acquaintances using injectable glucagon²³. Another simulated study found that successful administration was increased with nasal glucagon compared with injectable glucagon in both trained users (90.3% versus 15.6%) and untrained users (90.9% versus 0%), with users and people with diabetes who observed the study reporting feeling safer with nasal glucagon over injectable glucagon²⁴. In a real-world study of nasal glucagon in adults with type 1 diabetes, 96% of moderate and severe hypoglycemic events resolved within 30 min, and all patients with severe hypoglycemic events awakened or returned to normal status within 15 min²⁵. In a real-world study of nasal glucagon in children with type 1 diabetes, all moderate hypoglycemic episodes resolved within 30 min of administration²⁶. In both real-world studies, no calls for EMS were made for any of the hypoglycemic events and ∼94% of caregivers reported overall satisfaction with the use of nasal glucagon^25,26.

The aim of the present analysis was to compare cost and use of medical services with nasal glucagon and injectable glucagon in Canada when used to treat severe hypoglycemic events when impaired consciousness precludes treatment with oral carbohydrates using an economic model, based on differences in the frequency of successful administration of the two interventions.

Methods

Evaluation of cost-effectiveness

A cost-effectiveness analysis was prepared, assessing the incremental cost per a series of medical services resource use avoided with nasal glucagon versus injectable glucagon. These were:

• Incremental cost per EMS call out avoided

• Incremental cost per emergency room treatment avoided

• Incremental cost per inpatient stay avoided

Incremental cost-effectiveness ratios (ICERs) can be used to express whether an intervention represents good value for money by combining cost and effectiveness outcomes. Where an intervention improves outcomes while reducing costs it is said to be dominant, and no ICER is calculated. Where applicable, ICERs in terms of the incremental cost per aspect of care avoided were calculated using the following formula: $$\mathit{Incremental\; cost\; per\; aspect\; of\; care\; avoided}=\frac{\mathit{total\; cost\; in\; the\; nasal\; glucagon\; arm}-\mathit{total\; cost\; in\; the\; injectable\; glucagon\; arm}}{(\mathrm{\%\; receiving\; the\; aspect\; of\; care\; with\; nasal\; glucagon}-\mathrm{\%\; receiving\; the\; aspect\; of\; care\; with\; injectable\; glucagon})}\mathrm{*}-1$$

Model structure

A decision tree model was developed in Microsoft Excel, reflecting the real-world decision-making and treatment outcomes that follow a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrates. The model structure was developed based on reviews of the published literature, treatment guidelines, and emergency care pathways. The model consists of 15 nodes to reflect the events and decisions that occur, and the model uses the same structure to evaluate outcomes for nasal glucagon and injectable glucagon (Figure 1).

Figure 1. Model structure. The model uses the same structure to evaluate outcomes for nasal glucagon and injectable glucagon. A decision node representing the choice of glucagon kit is not shown. Treatment success was defined as successful administration of treatment, considered to be delivery of a full dose of nasal glucagon or injectable glucagon.

This network results in a total of 16 possible outcomes, each associated with different costs depending on the events and decisions that took place. Severe hypoglycemic events are acute in nature and the modeling analysis, therefore, took a short-term perspective, capturing the acute treatment following an event and follow-up care. No discounting was applied as outcomes were not projected over time horizons >1 year. Outcomes were assessed from a societal perspective, capturing both direct and indirect costs.

Decision probabilities

Decision probabilities were taken from the published literature, with Canadian-specific sources used wherever possible (Table 1). The only difference between the treatment arms was at node b, the probability of successful administration of glucagon. Data on successful administration of treatment with nasal glucagon and injectable glucagon were taken from a simulation study²³. In the study, 16 instructed caregivers and 15 non-instructed acquaintances administered nasal glucagon and injectable glucagon to manikins, simulating unconscious people with diabetes during severe hypoglycemic episodes. With nasal glucagon, 15 caregivers (94%) and 14 acquaintances (93%) administered a full dose. One caregiver and one acquaintance did not successfully administer nasal glucagon because they did not fully depress the plunger on the device. With injectable glucagon, only two caregivers (13%) administered a full dose, and no acquaintances gave a full dose. All other decision probabilities were equivalent in the two arms and were taken from published sources (inputs applied in the base case are described below alongside alternative values applied in sensitivity analyses)^6,23–29. For each model input, the most recent value from a large-scale Canadian study was used where ever possible.

Table 1. Treatment and decision probabilities.

Treatment/decision	Nasal glucagon, mean (SE)	Intramuscular glucagon, mean (SE)	References
Initial treatment decision
Treatment attempted with glucagon kit following severe hypoglycemic event, %	100 (0)	100 (0)	Assumed
If treatment is attempted
Treatment is administered by care giver, %	68 (5)	68 (5)	^27^,^28
Treatment administered by care giver is successful, %	94 (6)	13 (8)	^23
Treatment is administered by acquaintance, %	32	32	^27^,^28
Treatment administered by acquaintance is successful, %	93 (6)	0 (0)	^23
If treatment is successful
If treatment is successful, further action taken, %	38 (2)		^14
Taken to emergency room by private car, %	15 (1)		^29
Followed by release, %	76 (2)		^27
Followed by inpatient stay, %	24		^27
Emergency services called, %	85		^29
Treat and release at scene, %	80 (4)		^23–25
Transported to emergency room, %	20		^23–25
Followed by release, %	76 (2)		^27
Followed by inpatient stay, %	24		^27
If treatment is unsuccessful/not attempted
Taken to emergency room by private car, %	15 (1)		^29
Followed by release, %	76 (2)		^27
Followed by inpatient stay, %	24		^27
Emergency services called, %	85		^29
Treat and release at scene, %	31 (2)		^27
Transported to emergency room, %	69		^27
Followed by release, %	76 (2)		^27
Followed by inpatient stay, %	24		^27
Follow up care
Appointment with healthcare professional, %	23 (2)		^6
Additional SMBG tests, n	3.1 (0.4)		^6

Abbreviations. SE. standard error; SMBG. self-monitoring of blood glucose.

Costs

The modeling analysis captured the direct costs of nasal glucagon, injectable glucagon, EMS call out, emergency room treatment, inpatient stay, follow-up with a healthcare professional, and self-monitoring of blood glucose testing (Table 2)^{28,30–32,35–37}.

Table 2. Resource use and costs associated with severe hypoglycemic events.

	Direct cost (CAD)	Reference	Indirect cost (CAD)	References
Nasal glucagon, mean (SE)	131.60 (0)	Data on file	0 (0)	Assumed
Intramuscular glucagon, mean (SE)	94.87 (0)	^30	0 (0)	Assumed
Emergency medical service treatment with release at the scene, mean (SE)	948.26 (47.41)	^31	0 (0)	Assumed
Emergency medical service treatment with transport to the emergency room, mean (SE)	948.26 (47.41)	^31	0 (0)	Assumed
Emergency room treatment, mean (SE)	287.00 (3.67)	^32	228.39 (1.24)	^32–34
Emergency room treatment followed by inpatient stay, mean (SE)	5,880.00 (546.99)	^32	2,897.22 (158.29)	^32^,^34
Appointment with healthcare professional, mean (SE)	98.13 (4.91)	^35	0 (0)	Assumed
SMBG test, mean (SE)	0.73 (0.04)	^36	0 (0)	Assumed

Abbreviations. CAD. 2019 Canadian dollars; SMBG. self-monitoring of blood glucose testing.

The modeling analysis also captured indirect costs. Despite the extensive review of the literature, the only robust sources that could be identified were those informing indirect costs relating to emergency room treatment and emergency room treatment followed by inpatient stay (Table 2). Therefore, a conservative approach was taken; indirect costs were applied for these two aspects of care only, and indirect costs for other aspects of care were not included (acquisition of glucagon, EMS treatment, an appointment with a healthcare professional, and acquisition of SMBG testing supplies). Total indirect costs for emergency room treatment and emergency room treatment followed by inpatient stay were comprised of the indirect cost identified by the OCCI and lost productivity³². Time off work for an emergency room visit was assumed to be 5.19 h, based on a retrospective cohort study conducted in Alberta for a 5-year period³³. Time off work for an emergency room treatment followed by inpatient stay was based on the mean length of stay identified in the OCCI for admissions related to hypoglycemia (5.3 days)³². Lost productivity was calculated by applying the mean hourly wage in Canada to these time off work estimates³⁴.

Sensitivity analyses

A series of sensitivity analyses were conducted, with alternative inputs applied at each node of the model (Table 3). Alternative inputs around treatment attempts, the success of treatment attempts, further action taken, and action following transport to the emergency room were considered in the sensitivity analyses and were sourced from published literature^{14,24,27,28,33,38,39}.

Table 3. Summary of sensitivity analyses.

Input in the base case	Sensitivity analysis	Sensitivity analysis reference
Base case	N/A	N/A
Node a: 100% attempt treatment in both arms	(1) 23.7% attempt treatment with nasal glucagon, 11.85% attempt treatment with injectable glucagon	^14
Node a: 100% attempt treatment in both arms	(2) 37.5% attempt treatment in both arms	^38
Node a: 100% attempt treatment in both arms	(3) 11.85% attempt treatment in both arms	^14
Node b: 68% of treatments administered by caregiver in both arms	(4) 100% of treatments administered by caregiver in both arms	Assumed
Node b: 32% of treatments administered by acquaintance in both arms	(5) 100% of treatments administered by an acquaintance in both arms	Assumed
Node b: 94% and 93% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 13% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful	(6) 90.3% and 90.9% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 15.6% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful	^24
Node b: 94% and 93% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 13% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful	7) Upper 95% confidence limit of successful treatment in both arms (100% and 100% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 29% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful)	Assumed
Node b: 94% and 93% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 13% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful	(8) Lower 95% confidence limit of successful treatment in both arms (82% and 80% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 0% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful)	Assumed
Node c: 38% take further action in both arms	(9) 100% take further action in both arms	Assumed
Node d: 85% call the EMS in both arms	(10) 100% call the EMS in both arms	Assumed
Node e and g: 76% treated in the emergency room and then released in both arms	(11) 100% treated in the emergency room and then released in both arms	Assumed
Node f: 20% transported the emergency room in both arms	(12) 69% transported to the emergency room in both arms	^27
Node h and i: 85% call the EMS	(13) 100% call the EMS	Assumed
Nodes j and n: 31% treated and released at the scene in both arms	(14) 37% treated and released at the scene in both arms	^33
Nodes j and n: 31% treated and released at the scene in both arms	(15) 25% treated and released at the scene in both arms	^39
Nodes i, k, m and o: 24% treatment in the emergency room followed by inpatient stay in both arms	(16) 27% treatment in the emergency room followed by inpatient stay in both arms	^28
Nodes i, k, m and o: 24% treatment in the emergency room followed by inpatient stay in both arms	(17) 23% treatment in the emergency room followed by inpatient stay in both arms	^33
After event care: 23% of patients attend an appointment	(18) 40% of patients attend an appointment with a healthcare professional in both arms	^28

Abbreviations. CAD. 2019 Canadian dollars; EMS. emergency medical services.

Results

Cost outcomes

Compared with the use of injectable glucagon, the use of nasal glucagon was associated with reduced direct, indirect, and combined costs of a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate (Table 4). Nasal glucagon was associated with a higher drug cost by CAD 36.73, due to the higher acquisition cost compared with injectable glucagon. However, this was completely offset by direct costs savings due to avoided EMS call outs and hospital costs (emergency room treatment and inpatient stays). Cost savings were driven by a reduced frequency of hospitalization, with equal costs accrued in both the nasal glucagon and injectable glucagon arms when hospitalization occurred. Follow-up costs were equal with nasal glucagon and injectable glucagon. Overall, nasal glucagon was associated with direct cost savings of CAD 1,249 per severe hypoglycemic event compared with injectable glucagon. Further indirect cost savings versus injectable glucagon of CAD 460 were identified due to avoided emergency room treatments and inpatient stays. When direct and indirect costs were combined, nasal glucagon was associated with overall cost savings of CAD 1,709 per event compared with injectable glucagon.

Table 4. Mean cost of a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate.

	Nasal glucagon (CAD)	Injectable glucagon (CAD)	Difference (CAD)
Direct costs
Glucagon	131.60	94.87	+36.73
EMS	336.66	758.99	−422.33
Hospital (emergency room and inpatient)	265.97	1,129.16	−863.20
Follow up	24.35	24.35	0.00
Total	758.58	2,007.37	−1,248.79
Indirect costs
Glucagon	0.00	0.00	0.00
EMS	0.00	0.00	0.00
Hospital (emergency room and inpatient)	141.69	601.53	−459.84
Follow up	0.00	0.00	0.00
Total	141.69	601.53	−459.84
Combined cost
Glucagon	131.60	94.87	+36.73
EMS	336.66	758.99	−422.33
Hospital (emergency room and inpatient)	407.65	1,730.69	−1,323.04
Follow up	24.35	24.35	0.00
Total	900.27	2,608.90	−1,708.64

Abbreviations. CAD. 2019 Canadian dollars; EMS. emergency medical service.

Effectiveness outcomes

The use of nasal glucagon was associated with reduced use of all medical services compared with the use of injectable glucagon (Figure 2). EMS call outs were reduced by 45% (36% versus 80%), emergency room treatments by 52% (16% versus 69%), and inpatient stays by 13% (4% versus 17%). Reduced use of medical services was driven by the higher rates of successful delivery of a full dose of nasal glucagon compared with injectable glucagon, supported by two usability studies, and therefore a reduced requirement for professional medical care.

Figure 2. Percentage of patients requiring medical services following a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate. Abbreviations. EMS. emergency medical service.

Cost-effectiveness outcomes

Nasal glucagon was associated with reduced direct and combined costs, reduced frequency of EMS call out, emergency room treatment, and inpatient stay versus injectable glucagon. Therefore nasal glucagon was considered dominant versus injectable glucagon, reducing the frequency of EMS call out, emergency room treatment and inpatient stay with cost savings, and ICERs were not calculated.

Sensitivity analyses

Variation in the decision probabilities applied showed that nasal glucagon was dominant over injectable glucagon for treatment of a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate was robust (Table 5). Nasal glucagon was associated with reduced costs versus injectable glucagon in all sensitivity analyses investigated. Furthermore, nasal glucagon was associated with reduced requirements for emergency room treatment and inpatient stays in all scenarios, and therefore was considered dominant in all sensitivity analyses.

Table 5. One-way and multi-way sensitivity analysis results.

Sensitivity analysis	Combined cost difference (CAD)	Difference in EMS call outs required (%)	Difference in emergency room treatments required (%)	Difference in inpatient stays required (%)	Combined cost per EMS call out avoided (CAD)	Combined cost per emergency room treatment avoided (CAD)	Combined cost per inpatient stay avoided (CAD)
Base case	−1,708.64	−45	−52	−13	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(1) 23.7% attempt treatment with nasal glucagon, 11.85% attempt treatment with injectable glucagon	−398.56	−11	−13	−3	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(2) 37.5% attempt treatment in both arms	−617.78	−17	−20	−5	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(3) 11.85% attempt treatment in both arms	−170.10	−6	−6	−2	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(4) 100% of treatments administered by caregiver in both arms	−1,630.22	−43	−50	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(5) 100% of treatments administered by an acquaintance in both arms	−1,877.17	−49	−58	−14	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(6) 90.3% and 90.9% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 15.6% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful	−1,606.43	−42	−49	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(7) Upper 95% confidence limit of successful treatment in both arms (100% and 100% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 29% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful)	−1,613.94	−42	−50	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(8) Lower 95% confidence limit of successful treatment in both arms (82% and 80% of nasal glucagon treatments administered by caregivers and acquaintances, respectively, are successful, 0% and 0% of injectable treatments administered by caregivers and acquaintances, respectively, are successful)	−1,196.57	−43	−50	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(9) 100% take further action in both arms	−547.41	0	−35	−9	Equal EMS use with reduced costs	Nasal glucagon dominant	Nasal glucagon dominant
(10) 100% call the EMS in both arms	−1,761.41	−40	−56	−14	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(11) 100% treated in the emergency room and then released in both arms	−1,390.01	−45	−52	−15	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(12) 69% transported to the emergency room in both arms	−1,371.17	−45	−39	−9	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(13) 100% call the EMS	−1,731.50	−58	−48	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(14) 37% treated and released at the scene in both arms	−1,592.65	−45	−48	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(15) 25% treated and released at the scene in both arms	−1,810.13	−45	−56	−14	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(16) 27% treatment in the emergency room followed by inpatient stay in both arms	−1,848.79	−45	−52	−14	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(17) 23% treatment in the emergency room followed by inpatient stay in both arms	−1,641.16	−45	−52	−12	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant
(18) 40% of patients attend an appointment with a healthcare professional in both arms	−1,708.64	−45	−52	−13	Nasal glucagon dominant	Nasal glucagon dominant	Nasal glucagon dominant

Abbreviations. CAD. 2019 Canadian dollars; EMS. emergency medical services.

Nasal glucagon was associated with equal EMS call out use in one scenario (analysis 9), when probabilities downstream from successful administration of a full dose of glucagon were varied (node c). When it was assumed that further action was taken in 100% of severe hypoglycemic events after successful administration of a full dose of glucagon, EMS use was equal, as the successful treatment of the severe hypoglycemic event did not result in a change in action from the caregiver. However, emergency room and inpatient stay use remained lower with nasal glucagon as it was assumed that the successful treatment of the severe hypoglycemic event would lead to reduced resource use. This assumption was based on real-world data showing that the majority of successfully treated severe hypoglycemic events resolve within 15 min, with median target EMS response times in urban, rural, and remote areas of 8, 20, and 40 min, respectively^25,26,40. Therefore severe hypoglycemic events are likely to have resolved either before the EMS arrive or before transportation to the emergency room has been initiated.

Discussion

Use of nasal glucagon for treatment of severe hypoglycemic events when impaired consciousness precludes treatment with oral carbohydrate was projected to result in reduced costs for healthcare payers and reduced use of medical services compared with the use of injectable glucagon. Cost savings and reduced use of medical services with nasal glucagon were driven by the higher proportion of successful, full-dose administrations and a corresponding reduction in professional medical care. Nasal glucagon was associated with overall cost savings of CAD 1,709 per severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrates. The modeling analysis found that the use of nasal glucagon to treat severe hypoglycemic events was associated with reduced use of all medical services compared with the use of injectable glucagon. Nasal glucagon was dominant versus injectable glucagon, reducing the use of medical services and resulting in cost savings.

A key strength of the analysis is that it follows the treatment pathways after a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrates. The model was designed to reflect real-world decision-making and treatment outcomes in the context of a severe hypoglycemic event, capturing the subsequent events, and using Canadian-specific data wherever possible. This results in a highly relevant analysis, reflecting clinical practice in the Canadian setting. The U.S. Food and Drug Administration has recently approved two new formulations of injectable glucagon that do not require reconstitution before use^41,42. These interventions have not yet been approved for use and are not currently under review by Health Canada. Therefore they were not included in the present analysis, as they are not a treatment option available to clinicians and patients. As new treatments for severe hypoglycemia become available in Canada, it will be important for updated cost-effectiveness analyses to capture all available treatment options. For such analyses to be accurate, additional head-to-head clinical data comparing ready-to-use injectable glucagon with nasal glucagon will be required.

The only difference between nasal glucagon and injectable glucagon in the base case analysis was the probability of successful administration of a full dose (node b), with all other parameters equal in both arms, including the probability of the resolution of a severe hypoglycemic event if the glucagon kit were administered successfully. This results in a fair analysis, with data sources to inform resource use following successful or unsuccessful treatment equal in both arms.

In the present analysis, cost-effectiveness was assessed only in patients experiencing a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate who used a glucagon kit. However, it is likely that only a small proportion of the dispensed glucagon kits are used, and the cost of these unused kits will have an impact on the overall cost-effectiveness of glucagon therapies. A previously published budget impact analysis for the US setting aimed to capture the impact of unused kits on the overall cost at a population level⁴³. This analysis found that nasal glucagon was associated with cost savings versus injectable glucagon even when a glucagon kit was not used in 75% of events meeting the criteria for treatment, despite a glucagon kit having been dispensed to the patient. The present analysis was focused on a Canadian healthcare payer perspective, rather than a budget impact analysis as used in the analysis for the US, but similar results could be expected in the Canadian setting. Appropriate prescription to patients at risk of severe hypoglycemic events, followed by appropriate education and training is key to maximizing the cost-effectiveness of nasal glucagon. Kits will also be unused if a device is considered difficult to use or intimidating by caregivers and acquaintances. In the Yale et al. study, 81% of caregivers and 100% of acquaintances were reported as recommending nasal glucagon over injectable glucagon²³. A real-world study has shown that in 100% of events, caregivers of adults with diabetes agreed that nasal glucagon would be less intimidating for caregivers than injectable glucagon, while an equivalent study found that in 96.9% of events, caregivers of children with diabetes agreed with this statement^25,26. Kits will also be unused if unavailable at the time of the severe hypoglycemic episode, such as severe hypoglycemia occurring outside the home or workplace. The easy portability of a nasal glucagon kit due to the small size and room temperature storage may result in greater availability of nasal glucagon versus injectable glucagon. Currently, it is not possible to accurately quantify the proportion of glucagon kits that are unused when a severe hypoglycemic event occurs, provide reasons why they were not used, and how these could change due to the introduction of nasal glucagon. As such, this represents an area for future research in order to conduct more accurate cost-effectiveness analyses.

Numerous studies have shown that hypoglycemic events have a substantial impact on quality of life^44–47. However, to date, no study has investigated the impact of the resolution of severe hypoglycemia on quality of life. It is likely that earlier resolution and avoidance of treatment by the EMS, in the emergency room, and hospitalization are associated with improved quality of life, but additional studies are required to quantify this. The modeling analysis did not capture mortality, as the costs accrued would depend on when this occurred in the model flow diagram (Figure 1), which is difficult to quantify. However, improved treatment of severe hypoglycemia with nasal glucagon may likely result in reduced mortality.

Conclusions

When a patient with type 1 or type 2 diabetes is being treated for a severe hypoglycemic event when impaired consciousness precludes treatment with oral carbohydrate, the use of nasal glucagon was projected to reduce costs for healthcare payers. The higher likelihood of successful administration with nasal glucagon was projected to reduce EMS call outs, emergency room treatments, and inpatient stays in the modeling analysis. Nasal glucagon was projected to be dominant versus injectable glucagon, reducing the use of medical services and resulting in cost savings.

Transparency

Declaration of funding

The study was funded by Eli Lilly and Company.

Declaration of financial/other relationships

JFY has received consulting fees for participation in advisory boards and giving lectures from Eli Lilly and Company Limited and Novo Nordisk Canada Inc. BO is an employee of Eli Lilly and Company Limited. BM is an employee of Eli Lilly and Company. DM is an employee of Eli Lilly Canada Inc. BH and WV are employees of Ossian Health Economics and Communications. Ossian received consulting fees from Eli Lilly and Company Limited to support the preparation of the analysis. GS was an employee of Eli Lilly Canada Inc. when the study was conducted. MJ was an employee of Eli Lilly Canada Inc. when the study was conducted. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

All authors designed the study. Data from previously published studies were identified by GS. The analysis was performed by BH. All authors interpreted the results. The manuscript was drafted by BH and revised critically for important intellectual content by all other authors. All authors have approved the final version of the manuscript, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Acknowledgements

No assistance in the preparation of this article is to be declared.