Cost-effectiveness of triple therapy with budesonide/glycopyrronium/formoterol fumarate dihydrate compared with dual therapy for the treatment of chronic obstructive pulmonary disease (COPD) in Canada

Abstract

RATIONALE: Chronic obstructive pulmonary disease (COPD) is associated with significant economic and societal burden in Canada. For patients whose symptoms remain uncontrolled with dual therapies including long-acting muscarinic antagonists (LAMA)/long-acting beta2-adrenergic agonists (LABA) or inhaled corticosteroids (ICS)/LABA, step-up to triple-therapy with ICS/LAMA/LABA is recommended. Budesonide/glycopyrronium/formoterol fumarate dihydrate (BGF) is an ICS/LAMA/LABA therapy with demonstrated efficacy compared to dual therapies in pivotal phase 3 trials (ETHOS and KRONOS).

OBJECTIVES

The objectives of this study were to conduct a cost-utility analysis to compare BGF with dual therapies (ICS/LABA and LAMA/LABA) for patients with COPD requiring escalation to triple therapy.

METHODS

A Markov cohort-based cost-utility analysis was conducted over a 30-year time horizon from a public payer perspective, to compare costs and outcomes in patients with COPD requiring escalation from dual therapy to triple therapy. The analysis was based on data from the ETHOS trial, Canadian administrative data, and published sources. Health states were characterized by forced expiratory volume (FEV1) severity and exacerbations (none, moderate, severe).

RESULTS

Over a 30-year time horizon in probabilistic analysis across several scenario analyses, BGF was associated with modest but consistent benefits to overall costs ($40,406 total for BGF, $40,685 for LAMA/LABAs and $43,261 for ICS/LABAs), as well as life years and quality-adjusted life year (QALYs) gained compared to dual therapies (8.7 total QALYs for BGF triple therapy, 8.5 for LAMA/LABAs and 8.4 for ICS/LABAs).

CONCLUSIONS

From a Canadian public healthcare perspective, BGF is associated with both cost-savings and improved health benefits compared with dual therapies for maintenance treatment of COPD patients at risk of exacerbations. These results may help to inform optimal decision-making.

RÉSUMÉ

JUSTIFICATION: La MPOC est associée à un fardeau économique et sociétal important au Canada. Pour les patients dont les symptômes ne sont pas maîtrisés par une bithérapie comprenant des antagonistes muscariniques à longue durée d'action (AMLA)/bêta₂-adrénergique à longue durée d'action (BALA) ou des corticostéroïdes en inhalation (CSI)/BALA, il est recommandé de passer à une trithérapie comprenant AMLA/BALA/CSI. Le budésonide/glycopyrronium/fumarate de formotérol dihydraté (BGF) constitue une option de traitement par AMLA/BALA/CSI dont l'efficacité a été démontrée par rapport aux bithérapies dans des essais pivots de phase 3 (ETHOS et KRONOS).

OBJECTIFS: Réaliser une analyse coût-utilité pour comparer le BGF aux bithérapies (BALA/CSI et AMLA/BALA) chez les patients atteints de MPOC nécessitant une transition vers une trithérapie.

MÉTHODES: Une analyse coût-utilité basée sur une cohorte de Markov a été réalisée sur un horizon de 30 ans du point de vue du payeur public, afin de comparer les coûts et les résultats chez les patients atteints de MPOC nécessitant une transition de la bithérapie à la trithérapie. Cette analyse s'est appuyée sur les données de l'essai ETHOS, des données administratives canadiennes et des sources publiées. Les différents états de santé ont été caractérisés en fonction de la gravité du VEMS et des exacerbations (aucune, modérée, grave).

RÉSULTATS: Sur un horizon de 30 ans, dans une analyse probabiliste de plusieurs scénarios, le BGF a été associé à des avantages modestes mais constants en ce qui concerne les coûts globaux (40 406 $ au total pour le BGF, 40 685 $ pour les AMLA/BALA et 43 261 $ pour les CSI/BALA), ainsi que les années de vie et les AVQ gagnées par rapport aux bithérapies (8,7 AVQ au total pour la trithérapie au BGF, 8,5 pour les AMLA/BALA et 8,4 pour les CSI/BALA).

CONCLUSIONS: Du point de vue des soins de santé publique au Canada, le BGF est associé à la fois à des économies de coûts et à une amélioration des avantages pour la santé par rapport aux bithérapies pour le traitement d'entretien des patients atteints de MPOC et présentant un risque d'exacerbations. Ces résultats peuvent contribuer à une prise de décision optimale.

Keywords:

• COPD
• cost-effectiveness
• Breztri
• triple therapy
• Canada

Introduction

Chronic obstructive pulmonary disease (COPD) is a chronic, progressive, debilitating respiratory disease characterized by persistent respiratory symptoms and airflow limitation.¹ Patients with COPD may experience exacerbations, which are associated with a decline in lung function and poor outcomes and results in the need for additional therapy.^1–4 Studies have shown that worsening airflow limitation, advancing disease severity and increasing exacerbation frequency may result in decreased health-related quality of life (HRQoL).⁵ A Canadian study reported that dyspnea, a cardinal symptom of COPD, is a major contributing factor to disability and anxiety associated with the disease.¹ Eventually, the impact of the disease becomes so advanced and debilitating that daily and social activities are negatively impacted.^1,6 COPD is also associated with considerable economic burden on the healthcare system costing an estimated $1.5 billion per year with more than 10% of the adult population in Canada affected by COPD, and 1 in 4 adults over the age of 35 developing COPD in their lifetime.^7,8 COPD exacerbations are responsible for more hospital admissions than any other disease, with an average stay of 7 days.⁸

Disease management of COPD is guided by numerous factors including the severity of symptoms, the risk of exacerbations, side effects and costs.⁴ For patients with mild to moderate symptoms, mono-inhaled therapies such as long-acting muscarinic antagonists (LAMA) or long-acting beta2-adrenergic agonists (LABA), or dual therapies such as LAMA/LABA or inhaled corticosteroids (ICS)/LABA are often incorporated into the treatment regimen. If patients on dual therapy remain uncontrolled, a step up to triple therapy with ICS/LAMA/LABA is recommended.⁴ International and Canadian guidelines recommend triple therapy for patients who are symptomatic and have poor health status despite the use of dual therapy, or for patients at high risk of exacerbations despite the use of dual therapy.^1,4 Despite this recommendation, many patients may not be receiving adequate treatment. The BALLISTIC study, an observational cohort study using administrative data from 2004 to 2020 in Ontario, showed 61.2% of patients with COPD received no pharmacotherapy or rescue therapies in the first year of follow-up, 13.7% received a LAMA or LABA only, 16.8% received dual therapy and 8.3% received triple therapy.⁹ Of patients who remained in the study in the last year of follow-up, 62.9% of patients received no pharmacotherapy or rescue therapies only, 8.8% received a LAMA or LABA only, 17.5% received dual therapy and 10.9% received triple therapy.

Budesonide/glycopyrronium/formoterol fumarate dihydrate (BGF) is an ICS/LAMA/LABA closed triple therapy indicated for long-term maintenance treatment to reduce exacerbations and treat airflow obstruction in patients with COPD who are not adequately treated by a dual therapy. BGF is the only closed triple therapy available as a pressurized metered dose inhaler and includes delivery technology consisting of phospholipid porous particles with a high fine particle fraction, which allows for efficient and uniform deposition of drug to large and small airways across a range of inspiratory rates.^10–12

The superior efficacy of BGF triple therapy versus dual therapy with ICS/LABA or LAMA/LABA, in terms of improved lung function, reduced rate of exacerbations, decreased symptom burden and improved HRQoL, was demonstrated in 2 pivotal phase 3 double-blind randomized controlled trials—ETHOS (NCT02465567) and KRONOS (NCT02497001) (study design descriptions in Appendix Table 1).^13–15 These trials enrolled patients with moderate to very severe COPD with a range of exacerbation history who were symptomatic while receiving at least 2 inhaled maintenance therapies.

Given the negative impact of COPD on patient well-being and HRQoL and the considerable cost of COPD to the Canadian economy, it is important to evaluate the cost-effectiveness of new interventions for COPD. This helps decision makers understand how the clinical benefit associated with a treatment could improve patient HRQoL, and whether these benefits justify the additional costs. Thus, the purpose of this study was to conduct a cost-utility analysis to compare BGF triple therapy with dual therapy (ICS/LABA and LAMA/LABA) for patients with moderate-to-very severe COPD.

Materials and methods

Model overview

A cost-utility analysis is a type of cost-effectiveness analysis that uses the quality-adjusted life year (QALY) to compared outcomes of different therapies.¹⁶ The QALY is a measure of the length of life adjusted to reflect the quality of life. These quality-of-life adjustments are made using utility values, which reflect the value or preference an individual places on symptoms or health states of a disease.^17,18 In this study, a cost-utility analysis was performed using a cohort-based Markov model to compare costs and outcomes (QALYs) of patients with COPD requiring escalation from dual therapy to triple therapy to manage symptoms and prevent exacerbations (Table 1).

Table 1. Decision problem overview.

Model component	Approach taken
Population	Patients with COPD requiring escalation to triple therapy to optimally manage symptoms and prevent exacerbations
Intervention	Budesonide/glycopyrrolate/formoterol fumarate (BREZTRI AEROSPHERE) administered as 2 oral inhalations twice daily
Comparators	ICS/LABA dual therapy Fluticasone furoate/vilanterol (Breo Ellipta) Budesonide/formoterol (Symbicort 200 Turbuhaler) Fluticasone propionate/salmeterol (Advair 125 or 250) Fluticasone propionate/salmeterol (Advair Diskus 100, 250 or 500) Fluticasone propionate/salmeterol (Wixela inhub 100, 250 or 500) Fluticasone propionate/salmeterol (PMS-fluticasone/salmeterol)	LAMA/LABA dual therapy Indacaterol/glycopyrronium (Ultibro Breezhaler) Umeclidinium bromide/vilanterol (Anoro Ellipta) Aclidinium bromide/formoterol fumarate (Duaklir Genuair) Tiotropium bromide/olodaterol (Inspiolto Respimat)
Outcomes	Total and incremental direct costs Quality-adjusted life years Life years Incremental costs per quality-adjusted life year
Perspective	Canadian public payer
Time horizon	30 years (lifetime analysis)
Discounting	Costs and outcomes are discounted at 1.5% annually

Abbreviations: COPD, chronic obstructive pulmonary disease; ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist.

This cost-utility analysis was based on data from the ETHOS trial, Canadian administrative data, and published data sources. Pair-wise comparisons were made between BGF triple therapy and dual therapies (LAMA/LABA and ICS/LABA) (Table 1). The perspective of this analysis was that of a Canadian public health care payer. A lifetime time horizon was used, which was assumed to be 30 years after consideration of the expected mean age at baseline of patients enrolled in the ETHOS trial (64.6 years) and the life expectancy of a moderate to severe COPD patient population.¹⁹ Each cycle in the model was 1 month. Costs and health outcomes were discounted at an annual rate of 1.5% as per the Canadian Agency for Drugs and Technologies in Health (CADTH) guidelines.²⁰

Model structure

The model structure was conceptualized to capture lung function, exacerbations and symptoms, in order to describe defining elements of the natural history of COPD and align to the structure and outcomes of the ETHOS trial. ETHOS was a 52-week trial comparing the efficacy and safety of BGF triple therapy versus dual therapy with ICS/LABA (budesonide/formoterol) or LAMA/LABA (glycopyrrolate/formoterol). The primary endpoint, the annual rate of moderate or severe COPD-related exacerbations, was significantly lower (24%) in patients treated with 320 μg BGF than in those treated with LAMA/LABA; similarly, the rate of exacerbations was significantly lower (13%) with 320 μg BGF than with ICS/LABA.¹³

A cohort-based semi-Markov model was selected as the preferred modeling approach. The model structure was based on a similar COPD model of BGF triple therapy in the UK.²¹ The model structure is characterized by health states that are differentiated by both forced expiratory volume (FEV1) severity and exacerbations (no events, moderate exacerbations and severe exacerbations) (Figure 1). This approach not only permits a decline in lung function, but also allows for the number of moderate and severe exacerbations experienced by patients to be captured. The symptom burden of COPD is expected to be captured indirectly in the resource use data and utility values ascribed to health states stratified by FEV1 and exacerbations. Other published cost-effectiveness models have used a similar modeling approach.^22,23

Figure 1. Cost-utility model structure. Abbreviations: FEV1, forced expiratory volume in 1 second. Note: Dashed black arrows indicate health state transitions between more severe FEV1 health states. Although not directly shown on this diagram, health state transitions between the moderate and very severe FEV1 health states are permitted.

The model comprised 10 health states including 9 FEV1-related health states and a death state (Figure 1). The FEV1-related health states were separated by 3 levels of severity, defined according to the 2023 Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines.¹ In the moderate (GOLD 2) FEV1 health state, patients had post-bronchodilator FEV1 ≥ 50% and <80% of predicted. In the severe (GOLD 3) FEV1 health state, patients had post-bronchodilator FEV1 ≥ 30% and <50% of predicted. In the very severe (GOLD 4) FEV1 health state, patients had post-bronchodilator FEV1 < 30% of predicted. Each moderate, severe, and very severe FEV1 health state was duplicated 3 times to capture patients who had experienced no exacerbation, moderate exacerbations and severe exacerbations.

A simulated cohort of patients entered the model with FEV1 transitions based on the ETHOS trial population at baseline (28.5% moderate FEV1, 60.6% severe FEV1 and 10.9% very severe FEV1) and received treatment with BGF triple therapy or dual therapy. At each monthly cycle, the simulated patients were permitted to transition to a more severe FEV1 health state without experiencing an exacerbation; transition to a moderate exacerbation health state; transition to a severe exacerbation health state; remain within the current FEV1 health state; or die (Figure 1). The probabilities of transitions to FEV1 and exacerbation health states were initially informed by the observed trial data, followed by updated risks reflecting a history of exacerbations over the modeled time horizon, as described in the following section.

Patients transitioning to an exacerbation health state experienced an immediate post-exacerbation period that patients occupied for 1 “tunnel state” cycle (1 month) on entry before transitioning to either the post-exacerbation or death states (Figure 1). Tunnel states were incorporated into the model since Markov models do not capture the history prior to entry of the health state and the time point at which entry occurs. This overcomes the memoryless feature of Markov models to a certain extent, allowing the analysis to capture exacerbation management costs as well as the utility decrement and increased risk of mortality associated with the exacerbation events experienced in an individual cycle. While the model included functionality and flexibility to assume longer-term post-exacerbation increased risk of disease progression or further exacerbations in a post-exacerbation state, the base case analysis did not make such assumptions, only a single post-exacerbation tunnel state cycle was included, to incorporate the immediate cost, HRQoL impact and risk of mortality post-exacerbation.

Patients were unable to transition back to the no exacerbation health states once they experienced an exacerbation (moderate or severe). Furthermore, since COPD is considered a progressive disease, patients were not able to experience an improvement in lung function and transition to a less severe FEV1 health state.

Model inputs

Patient population

The target population included patients with COPD, including chronic bronchitis and/or emphysema, who were not adequately treated by a combination of an ICS/LABA or a combination of a LAMA/LABA (as per clinical guidelines and BGF clinical trials, which enrolled patients with moderate-to-very severe COPD who were symptomatic while receiving at least 2 inhaled maintenance therapies and had a range of exacerbation history). Age and sex characteristics were based on the ETHOS clinical trial population.¹³

Treatment effects

FEV1 transitions were calculated based on the proportion of patients treated with BGF triple therapy in ETHOS who transitioned from moderate to severe FEV1 and severe to very severe FEV1 over 52 weeks. It was assumed that patients would not transition directly from moderate to very severe, such that there was always a period spent in the severe FEV1 state in the interim. Within ETHOS, 24.1% of patients treated with BGF triple therapy transitioned from moderate to severe FEV1, and 7% transitioned from severe to very severe FEV1 over 52 weeks, translating to monthly probabilities of 2.27% and 0.60%, respectively. In the base case of the model, it was assumed that these disease progression rates applied to all therapies, while in scenario analysis the potential for treatment effects on disease progression were explored. In this analysis, relative risks from ETHOS were applied for FEV1 transition rates for ICS/LABA and LAMA/LABA dual therapies. The risk ratios for FEV1 transition rates for BGF triple therapy versus LAMA/LABA were 1.13 for the moderate to severe transition and 0.85 for the severe to very severe transition. The risk ratios for FEV1 transition rates for BGF triple therapy versus ICS/LABA vs were 0.74 for the moderate to severe transition and 0.48 for the severe to very severe transition.

The rates of moderate and severe exacerbations were based on data from ETHOS for the 320 μg BGF patient population stratified by lung function (FEV1) severity at baseline (Table 2). Odds ratios from ETHOS were applied for exacerbation rates for patients treated with BGF triple therapy compared with dual therapies (Table 2).

Table 2. Model inputs associated with clinical events.

Model parameter	Base case	PSA distribution	Source
FEV1 health state transition probabilities (monthly) for triple therapy
Moderate to severe FEV1	0.0209		ETHOS trial
Severe to very severe FEV1	0.0060		ETHOS trial
Relative risks of transition probabilities for BGF triple therapy vs LAMA/LABA (used in sensitivity analysis)
Moderate to severe FEV1	1.13		ETHOS trial
Severe to very severe FEV1	0.85		ETHOS trial
Relative risks of transition probabilities for BGF triple therapy vs LAMA/LABA
Moderate to severe FEV1	0.74		ETHOS trial
Severe to very severe FEV1	0.48		ETHOS trial
Adjusted monthly exacerbation rates for BGF triple therapy
Moderate exacerbation, moderate FEV1	0.066	Normal	ETHOS trial
Moderate exacerbation, severe FEV1	0.077	Normal	ETHOS trial
Moderate exacerbation, very severe FEV1	0.088	Normal	ETHOS trial
Severe exacerbation, moderate FEV1	0.007	Normal	ETHOS trial
Severe exacerbation, severe FEV1	0.013	Normal	ETHOS trial
Severe exacerbation, very severe FEV1	0.017	Normal	ETHOS trial
Mean OR for exacerbation for BGF triple therapy vs LAMA/LABA
Moderate exacerbation, moderate FEV1	0.77	Lognormal	ETHOS trial
Moderate exacerbation, severe FEV1	0.89	Lognormal	ETHOS trial
Moderate exacerbation, very severe FEV1	0.73	Lognormal	ETHOS trial
Severe exacerbation, moderate FEV1	0.83	Lognormal	ETHOS trial
Severe exacerbation, severe FEV1	0.79	Lognormal	ETHOS trial
Severe exacerbation, very severe FEV1	0.75	Lognormal	ETHOS trial
Mean OR for exacerbation for BGF triple therapy vs ICS/LABA
Moderate exacerbation, moderate FEV1	0.93	Lognormal	ETHOS trial
Moderate exacerbation, severe FEV1	1.14	Lognormal	ETHOS trial
Moderate exacerbation, very severe FEV1	0.87	Lognormal	ETHOS trial
Severe exacerbation, moderate FEV1	0.83	Lognormal	ETHOS trial
Severe exacerbation, severe FEV1	0.79	Lognormal	ETHOS trial
Severe exacerbation, very severe FEV1	0.54	Lognormal	ETHOS trial
Mean OR for adverse events
BGF triple therapy vs. LAMA/LABA	0.95		ETHOS trial
BGF triple therapy vs. ICS/LABA	1.03		ETHOS trial
Relative risks of mortality for COPD population vs Canadian general population
Moderate FEV1	1.40	Normal	Shavelle et al,2009^24
Severe FEV1	2.60	Normal	Shavelle et al,2009^24
Very severe FEV1	2.60	Normal	Shavelle et al, 2009^24
Health state utility values (EQ-5D) for patients not experiencing exacerbations
Moderate FEV1	0.84	Beta	ETHOS trial
Severe FEV1	0.81	Beta	ETHOS trial
Very severe FEV1	0.78	Beta	ETHOS trial
Exacerbation utility decrements		Beta
Moderate exacerbation	0.010	Beta	Rutten-van Molken et al.^27
Severe exacerbation	0.042	Beta	Rutten-van Molken et al.^27

Abbreviations: BGF: budesonide/glycopyrrolate/formoterol fumarate; EQ-5D: EuroQoL 5-dimension; FEV1: forced expiratory volume in 1 second; ICS: inhaled corticosteroid; LABA: long-acting beta2-agonist; LAMA: long-acting muscarinic antagonist; OR: odds ratio.

Odds ratios from ETHOS were applied for adverse event rates for patients treated with BGF triple therapy compared with dual therapies (Table 2).

Discontinuation of therapy was based on the ETHOS trial for BGF triple and dual therapies. It was estimated that 1.88% of patients would discontinue BGF treatment per month, with a relative risk of 1.31 for LABA/LAMA relative to BGF and 1.16 for ICS/LABA relative to BGF.

Mortality

COPD-related mortality was modeled via lung function severity (i.e., for each FEV1-related health state) and severe exacerbation-related mortality. In each FEV1-related health state, the risk of COPD-related mortality was derived by applying a relative risk of COPD-related death versus the risk of death in the general population of an equivalent age. The general population all-cause mortality rates were obtained from Canadian life tables. In the base case, relative risks of COPD-related mortality compared with all-cause mortality in the Canadian general population were derived from the published literature and were assumed to be constant (Table 2).²⁴ Patients who experienced a severe exacerbation were at risk of severe exacerbation-related deaths. The severe exacerbation-related risk of death of 8.81% was based on an audit of COPD hospitalization data in Ontario.^25,26

Utilities

EuroQoL 5-dimension 5-level (EQ-5D-5L) health state utility data were collected in the ETHOS clinical trial and were stratified by lung function (FEV1) severity at baseline. The EQ-5D-5L data for the patient population not experiencing exacerbations were used (Table 2). Utility decrements were assigned to patients on treatment that experienced moderate or severe exacerbations (Table 2). The exacerbation-related utility decrements were obtained from published literature.²⁷ Decrements were not assigned to adverse events as they were assumed to be already captured in the underling EQ-5D analysis of the ETHOS clinical trial data.

Resource use and costs

Treatment costs included drug acquisition and dispensation costs, costs associated with rescue medication use and treatment-related adverse event costs. Drug costs were based on 30-day inhaler costs of $127.00 for BGF triple therapy plus $8.83 per dispensation. Canadian PharmaStat claims data were used to determine the distribution of LAMA/LABA and ICS/LABA dual therapies (Table 2; Appendix Table 2), which had resulting average monthly costs of $75.24 and $96.69, respectively.^28,29

Dispensation costs for dual therapies were $8.83 per 30 days. All drug costs were subject to an 8% markup and costs were converted from 30 days to monthly model cycles.

Rescue medication-related costs were included in the model as the average number of rescue medication inhalations required per month per treatment and by FEV1 state. Rescue medication in Canada is albuterol sulfate, and the number of inhalations required was derived from the ETHOS trial: 54.8 inhalations per month for moderate FEV1, 82.2 inhalations per month for severe FEV1, and 115.7 inhalations per month for very severe FEV1. The drug acquisition cost for rescue medication was $5.00 per a 100-inhalation package, sourced from the Ontario Drug Benefit (ODB) formulary.³⁰

Disease management costs included the costs of treatment of exacerbations and routine medical costs, such as the cost of routine physician visits, beyond the treatment of exacerbations were not included. Costs per severe exacerbation and routine medical costs were estimated based on resource utilization observed in the ETHOS trial, while costs per moderate exacerbation were based on assumption (oral corticosteroid utilization from 7-14 days and 1 general practitioner visit), with Canadian unit costs applied (Appendix Table 3). The estimated costs per exacerbation were $86.27 per moderate exacerbation and $7,776.83 per severe exacerbation (Table 2; Appendix Table 3). Routine medical costs per month were estimated to be $15.10 for moderate FEV1, $19.99 for severe FEV1, and $28.81 for very severe FEV1.

Treatment-related adverse event costs were obtained from the Ontario Schedule of Benefits (OSB) and the Ontario Case Costing Initiative (OCCI).^31,32 Mild adverse events were assigned a cost of $84.45, corresponding to the cost of a general assessment by a primary care physician, as stipulated in the OSB. Moderate adverse events (acute upper respiratory infection, acute nasopharyngitis, pneumonia, bronchitis, headache, COPD exacerbation) were costed at $282.00, which is equivalent to the weighted average of ambulatory care claims for COPD patients with these conditions. Severe adverse events, defined as those requiring inpatient stays for COPD-related conditions (acute upper respiratory infection, pneumonia, bronchitis, COPD exacerbation) cost $8,558.57 (Table 2; Appendix Table 4).

The model also accounted for subsequent therapy in patients who discontinued treatment. Once patients discontinued their initial therapy (either BGF or dual therapy), it was assumed that they continued to receive triple therapy, based on a lack of further step-up therapeutic options recommended in Canada. This assumption implied that discontinuation of dual therapy was primarily due to lack of efficacy; further assumptions were explored in scenario analysis. In the base case, it was assumed that efficacy post-discontinuation would be equivalent across treatment arms.

It was assumed that subsequent treatment would be equivalent to a weighted distribution of the cost of BGF triple therapy (31%), fluticasone furoate/umeclidinium/vilanterol (37%) and open triple therapy (32%). This distribution was based on a best-fit trending analysis of relevant dual and triple therapies using 5 years of historical public claims data from the IQVIA PharmaStat database.³³ The monthly cost of this was based on unit costs, markups and dispensation fees consistent with the overall model.

Analysis and model outputs

Probabilistic analysis was performed to account for statistical uncertainties of multiple key parameters and informed the base case probabilistic results. The probabilistic analysis simultaneously varied all parameters with uncertainty in the model, sampling various input parameters from the appropriate probability distributions. Probabilistic distributions used were: gamma distributions for health care resource use and costs, beta distributions for probabilities and utilities and lognormal distributions for relative treatment effects. The probabilistic analysis was run for 5,000 iterations in the base case. The model outputs included total and incremental costs, life years, QALYs and incremental cost-effectiveness ratios (ICERs), presented as incremental costs per QALY gained.

Sensitivity and scenario analyses

Scenario analyses were conducted to assess the impact of changes in model parameters on the results; the probabilistic analysis was repeated for each scenario considered. Treatment effects for disease progression were based on ETHOS trial results, in addition to the base case treatment effect of exacerbation risk. In another scenario, rather than apply mortality hazard ratios by health state, treatment-specific mortality rates were taken directly from the ETHOS trial (0.118% per month for BGF triple therapy, 0.223% for LABA/LAMA and 0.158% for ICS/LABA). Another scenario was conducted in which the treatment distribution of open dual therapies was obtained from the Ontario BALLISTIC observational study.^9,34 The base case analysis used Canadian PharmaStat data to determine the distribution of dual therapies.

A number of alternative discontinuation patterns were also explored in scenario analysis. In contrast to the base case in which all patients who discontinued were assumed to receive a triple therapy, the following scenarios were considered:

• All patients discontinue to dual therapy

• All patients discontinue to no therapy

• Patients on dual therapies step up to triple while patients on triple therapy step down to a dual therapy

• All patients switch medications but remain on the same treatment step: patients on triple therapy to another triple therapy, and patients on dual therapy to another dual therapy

Scenario analyses were also conducted excluding dispensing fees, drug markups or both markups and dispensing fees.

For all scenario analyses, all parameters in addition to those explicitly listed were set as equivalent to the base case settings.

In addition, one-way sensitivity analyses were applied across model parameters and represented as a tornado diagram.

Results

Base case results

Over the 30-year probabilistic analysis, BGF triple therapy was associated with lower overall costs compared with both ICS/LABAs and LABA/LAMAs, with modest differences vs. LABA/LAMA ($40,406 total for BGF triple therapy, $40,685 for LAMA/LABAs and $43,261 for ICS/LABAs) reflecting the higher disease management costs associated with dual therapies despite their lower treatment-related costs (Table 3). BGF triple therapy was associated with more life years and QALYs gained compared with dual therapies, with slightly higher values observed for LAMA/LABAs compared to ICS/LABAs (8.7 total QALYs for BGF triple therapy, 8.5 for LAMA/LABAs and 8.4 for ICS/LABAs) (Table 3). Incremental differences were relatively small, particularly relative to LAMA/LABAs, although consistently observed in probabilistic analysis (Table 3).

Table 3. Probabilistic cost-effectiveness results at 30 years.

	BGF triple therapy	LAMA/LABAs	ICS/LABAs	BGF vs. LAMA/LABA	BGF vs. ICS/LABA
Total costs per patient, CAD (Mean; 2.5th–97.5th percentiles^a)	40,406 (34,109–47,339)	40,685 (33,997–47,931)	43,261 (36,008–51,173)	−279 (−592–112)	−2,855 (−3,834- −1,899)
Treatment-related costs	8,122 (6767–9677)	4,899 (3720–6294)	6,068 (4678–7662)	3,223 (,3007–3,383)	2,054 (2,014 – 2,060)
Disease management costs	17,028 (12,740–22,138)	19,656 (14,958–25,088)	21,982 (16,495–28,303)	−2,627 (−2,218 – −2,949)	−4,954 (−3,755 – −6,165)
Subsequent therapy costs	15,256 (12,283–18,420)	16,131 (12,900–19,402)	15,211 (12,236–18,293)	−875 (−982 – −616)	44 (40–127)
Total QALYs per patient (Mean; 2.5th–97.5th percentiles^a)	8.7 (7.1–10.4)	8.5 (6.9–10.2)	8.4 (6.8–10.0)	0.2 (0.2–0.3)	0.4 (0.3–0.5)
Life years per patient (Mean; 2.5th–97.5th percentiles^a)	11.3 (9.7–12.9)	11.1 (9.6–12.6)	10.9 (9.4–12.4)	0.2 (0.2–0.3)	0.4 (0.3–0.6)
ICER ($/QALY gained)				Dominant	Dominant

Abbreviations: BGF, budesonide/glycopyrrolate/formoterol fumarate; CAD, Canadian dollars; FEV1, forced expiratory volume in 1 second; ICER, incremental cost-effectiveness ratio; ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist; QALY, quality-adjusted life-year. ^a Percentile ranges obtained from probabilistic sensitivity analyses.

Due to the overall lower costs ($2855 cost savings vs. ICS/LABA and $279 cost savings vs. LABA/LAMA) and higher QALYs (0.4 incremental QALYs vs. ICS/LABA and 0.2 vs. LABA/LAMA) associated with BGF triple therapy versus dual therapies, BGF triple therapy was the dominant treatment option relative to both dual therapies. In probabilistic sensitivity analysis, BGF triple therapy was dominant versus LABA/LAMA in 65% of iterations and versus ICS/LABA in 99% of iterations (Figures 2 and 3). At a willingness-to-pay threshold of $50,000/QALY, the probability that BGF triple therapy remained the most cost-effective option versus LAMA/LABAs or ICS/LABAs was greater than 99% (Figure 4).

Figure 2. Cost-effectiveness plane for budesonide/glycopyrrolate/formoterol fumarate triple therapy versus ICS/LABAs over 5000 iterations. Abbreviations: ICER, incremental cost-effectiveness ratio; ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; QALY, quality-adjusted life-year.

Figure 3. Cost-effectiveness plane for budesonide/glycopyrrolate/formoterol fumarate triple therapy vs LAMA/LABAs over 5000 iterations. Abbreviations: ICER, incremental cost-effectiveness ratio; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist; QALY, quality-adjusted life-year.

Figure 4. Cost-effectiveness acceptability curve for budesonide/glycopyrrolate/formoterol fumarate triple therapy versus dual therapies over 5000 iterations. Abbreviations: ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist.

Figure 5. One-way sensitivity analysis incremental cost results for budesonide/glycopyrrolate/formoterol fumarate triple therapy versus (a) ICS/LABA and (b) LABA/LAMA. (a) Abbreviations: BGF, budesonide/glycopyrronium/formoterol; FEV1, forced expiratory volume in 1 second; ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; RR, relative risk. (b) Abbreviations, BGF, budesonide/glycopyrronium/formoterol; FEV1, forced expiratory volume in 1 second; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist; RR, relative risk.

Scenario and sensitivity analyses

Results of the scenario analyses were generally consistent with that of the base case analysis, with BGF triple therapy remaining dominant versus both ICS/LABAs and LAMA/LABAs in the majority of scenarios (Table 4). Exceptions to this included incremental costs leading to small (<$5000) incremental cost-utility ratios (ICURs) vs. LAMA/LABA in scenarios in which mortality was taken from the ETHOS trial, and when all patients were assumed to discontinue to dual therapy or no therapy. The discontinuation scenario in which triple therapy patients were assumed to discontinue to another triple therapy, while dual therapy patients were assumed to discontinue to another dual therapy was associated with ICURs of $7494 versus ICS/LABA and $27,433 versus LABA/LAMA. This scenario by definition induced higher subsequent therapy costs to BGF discontinuers relative to dual therapy discontinuers, although ICURs versus both dual therapies remained within the standard $50,000 per QALY threshold.

Table 4. Scenario analyses results.

Scenario	Incremental cost ($): BGF triple therapy vs. comparator (Mean; 2.5th–97.5th percentiles)	Incremental QALYs: BGF triple therapy vs. comparator (Mean; 2.5th–97.5th percentiles)	ICER: BGF triple therapy vs. comparator
Treatment effects applied to disease progression
ICS/LABA	−4,872	0.64	Dominant
LAMA/LABA	−465	0.22	Dominant
Mortality taken from ETHOS trial (treatment-specific) rather than health state specific
ICS/LABA	−2,990	0.39	Dominant
LAMA/LABA	72	0.29	$245
BALLISTIC as source for treatment distributions
ICS/LABA	−2,902	0.37	Dominant
LAMA/LABA	−316	0.21	Dominant
Post-discontinuation treatment – all patients discontinue to dual therapy
ICS/LABA	−3,002	0.21	Dominant
LAMA/LABA	75	0.37	$327
Post-discontinuation treatment – all patients discontinue to no therapy
ICS/LABA	−3,002	0.21	Dominant
LAMA/LABA	75	0.37	$3008
Post-discontinuation treatment – dual therapies to triple therapy, triple therapy to dual therapy
ICS/LABA	−8,754	0.21	Dominant
LAMA/LABA	−6,027	0.37	Dominant
Post-discontinuation treatment – dual therapies to an alternative dual therapy, triple therapy to an alternative triple therapy
ICS/LABA	2,760	0.37	$7494
LAMA/LABA	5,838	0.21	$27,433
Dispensing fees excluded
ICS/LABA	−3,039	0.37	Dominant
LAMA/LABA	−274	0.21	Dominant
Drug markups excluded
ICS/LABA	−3,142	0.37	Dominant
LAMA/LABA	−434	0.21	Dominant
Dispensing fees and markups excluded
ICS/LABA	−3,189	0.37	Dominant
LAMA/LABA	−443	0.21	Dominant

Abbreviations: BGF, budesonide/glycopyrrolate/formoterol fumarate; ICS, inhaled corticosteroid; ICER, incremental cost-effectiveness ratio; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist; QALY, quality-adjusted life-year.

In one-way sensitivity analysis, incremental costs of BGF triple therapy remained negative (cost-savings) versus both ICS/LABAs and LAMA/LABAs. Incremental cost results were most sensitive to differential efficacy post-discontinuation (assumed to be equivalent in the base case) and treatment-specific exacerbation rates (Figure 5).

Discussion

This cost-utility analysis, based on the ETHOS phase 3 clinical trial and Canadian data on survival and treatment patterns in COPD patients, indicates that BGF triple therapy is dominant versus ICS/LABA dual therapy (cost savings of $4711 and 0.7 additional QALYs) and LAMA/LABA dual therapy (cost savings of $480 and 0.2 additional QALYs) in a population of individuals living with moderate to very severe COPD. Results were generally robust across scenario analyses. The greatest variation in results was associated with assumptions around post-discontinuation of therapy, particularly the scenarios in which differential post-discontinuation patterns are assumed for dual versus triple therapy patients. These scenarios reflect differential post-discontinuation behavior by treatment arm, such that the greater incremental costs are driven by subsequent treatment acquisition, rather than the initial intervention costs. However, across all scenarios, ICURs remain below standard thresholds for cost-effectiveness (e.g., $50,000 per QALY), with the highest estimated ICUR being $47,937 for BGF versus ICS/LABA in the scenario in which all patients step down (dual therapy patients to no therapy). In this scenario, a conservative assumption is made that the ICS/LABA patients discontinuing to no therapy would not experience a loss of efficacy, only reduced costs. Thus, the true difference would likely include a greater QALY reduction and a lower ICUR for BGF vs. ICS/LABA. In addition to post-discontinuation assumptions, cost savings were most reduced when markups and dispensing fees were excluded (resulting from less excess costs from dual therapies and potential for multiple inhalers vs. single-inhaler BGF triple therapy); however, the overall finding of economic dominance of BGF triple-therapy remained consistent.

This is the first cost-utility evaluation evaluating BGF triple therapy vs dual therapies in Canada. However, studies of other triple therapies in other countries have also concluded that triple therapies are cost-effective with improved medical benefits compared with dual therapies for maintenance treatment of COPD patients at risk of exacerbations. In an analysis conducted in the UK, BGF triple therapy was associated with additional QALYs versus dual therapies, at a higher cost, for ICERs per QALY of £9901 versus LABA/LAMA and £2164 versus ICS/LABA, both falling within typical UK willingness to pay thresholds for cost-effectiveness (e.g. £20,000).²¹ Additional cost-utility analyses have been published for another available closed triple therapy, fluticasone furoate/umeclidinium/vilanterol, which have shown that fluticasone furoate/umeclidinium/vilanterol was associated with gains in QALYs at an incremental cost compared to dual therapies, resulting in fluticasone furoate/umeclidinium/vilanterol being a cost-effective treatment option.^35–38 In a Canadian cost-effectiveness analysis, the ICER per QALY for fluticasone furoate/umeclidinium/vilanterol was $18,989 versus fluticasone furoate/vilanterol (ICS/LABA) and $13,776 versus vilanterol/umeclidinium (LABA/LAMA).³⁵ Two studies in the UK estimated ICERs per QALY for fluticasone furoate/umeclidinium/vilanterol at £4104 versus fluticasone furoate/vilanterol, £6418 versus vilanterol/umeclidinium and £1098 versus budesonide/formoterol (ICS/LABA).^36,37 A Spanish cost-effectiveness analysis showed fluticasone furoate/umeclidinium/vilanterol was associated an ICER per QALY of €642 versus budesonide/formoterol.³⁸ While differences in currencies precludes direct comparisons across jurisdictions, the consistent overall finding is that the ICERs are well below what is considered cost-effective at standard willingness-to-pay thresholds.

The current analysis further found economic dominance for BGF triple therapy versus dual therapies, relating to differences in acquisition costs across triple therapy options. A cost-utility comparison was not made between BGF triple therapy and other triple therapies (e.g., fluticasone furoate/umeclidinium/vilanterol single inhaler, or “open triple” combinations consisting of multiple dual and single therapy inhalers). A recent network meta-analysis (NMA) found comparable efficacy (in terms of exacerbation, lung function, symptom control and quality of life) across available triple therapy options³⁹ and as such, the analysis here focused on the ETHOS trial-based comparison of BGF triple therapy versus dual therapies. Based on available clinical data, the most appropriate comparison versus other triple therapies in Canada would be a cost-minimization analysis to assess differences in treatment acquisition costs under the assumption of comparable efficacy; the review by the Canadian Agency for Drugs and Technologies in Health (CADTH) recommended BGF reimbursement, based on cost-minimization analysis and the assumption of equal efficacy across triple therapies as supported by NMA.⁴⁰

The results reported from this model are relevant to a Canadian population as the ETHOS trial was multinational and included numerous Canadian sites across multiple provinces. Canada-specific parameters incorporated into the analysis included treatment and dispensing costs, treatment patterns for dual therapy and general population mortality rates upon which COPD rate ratios were applied.

Limitations of this study included the reliance on the ETHOS trial alone for efficacy; at the time of the analysis, real-world effectiveness data were not available. As such, the data used in this study may not be accurately representative of real-world COPD populations and the generalizability of study finding may be limited. Similarly, the cost of exacerbations were based on ETHOS resource utilization that may reflect a more intensive trial-based protocol than is typical in real-world settings; a cost analysis of severe exacerbations in the United Kingdom, which showed the cost of severe exacerbations based on healthcare resource utilization data from KRONOS and ETHOS was higher than estimates previously reported in the published literature.²¹ Variability was not reported for the relative risk of mortality by COPD severity, in the Shavelle et al. study used to inform this parameter, such that empirical uncertainty in this parameter could not be incorporated into probabilistic analysis; however, the sample size of this study was relatively large (>6000 patients included) and it is thus anticipated that the resulting standard error is likely to be relatively small.²⁴ A limitation to the overall model structure was the assumption that exacerbation risk was dependent only on treatment, and did not account for exacerbation history; that is, the presence of a commonly-exacerbating phenotype is not explicitly modeled. While it is anticipated that the modeled average values likely do reflect overall population values in aggregate, it may be the case that in a real-world population, some patients would exacerbate more frequently, and some less, regardless of lung function.

Conclusion

This analysis showed that, from the perspective of the Canadian public healthcare payer, BGF triple therapy is a cost-saving option leading to improved health benefits (“economically dominant”) compared with LAMA/LABA or ICS/LABA dual therapy for the maintenance treatment of patients with COPD who are at risk of exacerbations. These results may help to inform optimal decision-making treatment choices in Canada.

Author contributions

All authors contributed to the drafting and revising of the manuscript. CS and CA conceptualized the study. KJ and MF carried out the modeling and analysis. EP contributed clinical insight.

Disclosure statement

Karissa Johnston and Michael Friesen are employees of Broadstreet Health Economics & Outcomes Research and received payment from AstraZeneca Canada Inc. in the conduct of this study. Cal Shephard and Caitlin Azzalini are employed by and own stock or hold stock options in AstraZeneca Canada Inc. Erika Penz has received honoraria and research support from AstraZeneca Canada Inc.

Data availability statement

Data underlying the findings described in this manuscript may be obtained in accordance with AstraZeneca’s data-sharing policy described at https://astrazenecagrouptrials.pharmacm.com/ST/Submission/Disclosure.

Additional information

Funding

This study was funded by AstraZeneca Canada, Inc.

Appendix

Appendix Table 1. ETHOS and KRONOS design and characteristics.

Trial characteristics	KRONOS	ETHOS
Study design	Randomized, double-blind, parallel-group, multicenter	Randomized, double-blind, parallel-group, multicenter
Duration	24 weeks	52 weeks
N	1800	8000 (PFT sub-study 3000 out of 8000)
Key inclusion criteria
Age	40–80 years	40–80 years
Diagnosis	Clinical history of COPD defined by ATS/ERS or local guidelines	Clinical history of COPD defined by ATS/ERS or local guidelines
Therapy	≥2 inhaled maintenance therapies	≥2 inhaled maintenance therapies
Lung function	FEV1 <80% of PN	FEV1 <65% of PN
CAT score	≥10	≥10
Exacerbation history	Not required but captured	Post-bronchodilator FEV1 <50% of PN with ≥1 moderate or severe COPD exacerbation in the 12 months Post-bronchodilator FEV1 ≥50% of PN with ≥2 moderate exacerbations or ≥1 severe exacerbation in the 12 months
Key exclusion criteria	Hospitalization due to poorly controlled COPD in the 3 months prior to Visit 1 or during screening period Use of antibiotics or OCS in the 6 weeks prior to Visit 1 (unless stable use of OCS)	Hospitalization due to poorly controlled COPD in the 3 months prior to Visit 1 or during screening period Use of antibiotics or OCS in the 6 weeks prior to Visit 1 (unless stable use of OCS)
Treatment arms	BGF 320/14.4/9.6 µg BID BFF 320/9.6 µg BID (ICS/LABA) GFF 14.4/9.6 µg BID (LAMA/LABA) Symbicort^® Turbuhaler^® 400/12 µg BID (ICS/LABA)*	BGF 320/14.4/9.6 µg BID BGF 160/14.4/9.6 µg BID BFF 320/9.6 µg BID (ICS/LABA) GFF 14.4/9.6 µg BID (LAMA/LABA)
Primary endpoint	Trough FEV1 FEV1 AUC	Rate of moderate or severe COPD exacerbations
Secondary endpoints
Exacerbations	Rate	Rate Time to first
Lung function (PFT sub-study in ETHOS)	Trough FEV1 Peak FEV1 FEV1 AUC0-4 Rate of decline	Trough FEV1 Peak FEV1 FEV1 AUC0-4 Rate of decline
Symptoms	Time to onset TDI EXACT and ERS	Time to onset TDI EXACT and ERS
HRQoL	SGRQ EQ-5D-5L	SGRQ EQ-5D-5L
Health resource utilization	Work absence in days (patients, primary caregivers) Calls (GPs, HCPs) Visits (GPs, specialists, ERs) Hospitalizations (numbers of patients and days in ICU or CCU)	Work absence in days (patients, primary caregivers) Calls (GPs, HCPs) Visits (GPs, specialists, ERs) Hospitalizations (numbers of patients and days in ICU or CCU)
Discontinuation	Time to treatment failure (any, exacerbation, death) Time to death (respiratory and all-cause)	Time to treatment failure (any, exacerbation, death)
Rescue medication	Rescue medication use (Ventolin HFA)	Rescue medication use (Ventolin HFA)
Countries	Canada, China, EU, Japan, USA	Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, Czech Republic, France, Germany, Hungary, Italy, Japan, Mexico, Netherlands, New Zealand, Peru, Poland, Russia, Serbia, South Africa, South Korea, Spain, Sweden, Taiwan, UK, and USA

Abbreviations: ATS: American Thoracic Society; AUC0-4: area under the curve from 0 to 4 h; BFF: budesonide/formoterol fumarate; BGF: budesonide/glycopyrronium/formoterol fumarate; BID: twice daily; CAT: COPD assessment tool; CCU: coronary care unit; COPD: chronic obstructive pulmonary disease; EQ-5D-5L: EuroQoL 5-dimension 5-level; ER: emergency room; ERS: European Respiratory Society; EXACT: exacerbations of chronic pulmonary disease tool; FEV1: forced expiratory volume in 1 second; GFF: glycopyrronium/formoterol fumarate; GP: general practitioner; HFA: hydrofluoroalkanes; ICS: inhaled corticosteroid; ICU: intensive care unit; LABA: long-acting beta adrenoreceptor agonist; LAMA: long-acting muscarinic antagonist; OCS: oral corticosteroid; PFT: pulmonary function test; PN: predicted normal; SGRQ: St. George’s respiratory questionnaire; TDI: transition dyspnea index; UK: United Kingdom; USA: United States of America.

*Open-label product.

Appendix Table 2. Product distribution and acquisition costs for dual therapy treatments.

Therapy	Canadian claims data, % use^a	Ontario administrative data, % use^b	Acquisition cost per day
ICS/LABAs
Advair Diskus (Wixela or PMS) 250	14.52%	17.49%	$1.69
Advair Diskus (Wixela or PMS) 500	9.00%	5.34%	$2.40
Advair 125 MDI	4.69%	5.40%	$3.50
Advair 250 MDI	14.95%	12.42%	$4.97
Symbicort Turbuhaler 200	40.95%	45.07%	$3.01
Breo Ellipta 100	15.88%	14.27%	$2.89
Weighted average daily cost of ICS/LABAs			$3.06
LAMA/LABAs
Ultibro Breezhaler	23.82%	24.17%	$2.58
Anoro Ellipta	28.34%	34.32%	$2.85
Duaklir Genuair	4.79%	4.75%	$2.00
Inspiolto Respimat	43.05%	36.76%	$2.12
Weighted average daily cost of LAMA/LABAs			$2.43

Abbreviations: ICS: inhaled corticosteroid; LABA: long-acting beta2-agonist; LAMA: long-acting muscarinic antagonist MDI: Metered dose inhaler; PMS: Pharmascience Inc.

^aBased on data reported by Canadian PharmaStat national claims data (third quarter 2019 to second quarter 2020).

^bData from the Ontario BALLISTIC study was used in scenario analysis.

Appendix Table 3. Exacerbation costs based on ETHOS trial utilization applied to Canadian unit costs.

	Unit cost (CAD)	Moderate exacerbation utilization (assumption)	Severe exacerbation utilization (ETHOS trial)
Hospitalization- ICU days	$3592	N/A	0.36 days
Hospitalization- CCU days (assume equivalent cost to general ward)	$1135	N/A	0.07 days
Hospitalization- General ward days	$1135	N/A	5.61 days
Oral corticosteroids − 7 days	$1.21	50%	N/A
Oral corticosteroids − 14 days	$2.43	50%	N/A
Ambulance transport	$42.10	N/A	29%
General practitioner visit	$84.45	1 visit	N/A
Total		$86.27	$7776.83

Abbreviations: CAD: Canadian dollars; CCU: critical care unit; ICU: intensive care unit.

Appendix Table 4. Adverse event (AE) costs (Ontario Case Costing Initiative).

	Included codes	Cost
Mild AE		$84.45
Moderate AE	J069 (Ambulatory) – Acute upper respiratory infection, unspecified	$282.00
	J00 (Ambulatory) – Acute nasopharyngitis (common cold)
	J441 (Ambulatory) – Chronic obstructive pulmonary disease with acute exacerbation, unspecified
	J449 (Ambulatory) – Chronic obstructive pulmonary disease, unspecified
	R51 (Ambulatory) – Headache
	J40 (Ambulatory) – Bronchitis, not specified as acute or chronic
	J189 (Ambulatory) – Pneumonia, unspecified
Severe AE	J440 – COPD with acute lower respiratory infection	$8558.57
	J441 – COPD with acute exacerbation, unspecified
	J448 – Other specified COPD
	J449 – COPD, unspecified
	J189 – Pneumonia, unspecified
	J40 – Bronchitis, not specified as acute or chronic
	J069 – Acute upper respiratory infection, unspecified

Appendix Table 5. Disease management costs based on ETHOS trial utilization applied to Canadian unit costs.

	Unit cost (CAD)	Moderate FEV1	Severe FEV1	Very severe FEV1
ER visits	$76.90	0.017	0.022	0.029
Specialist visit	$169.65	0.048	0.067	0.105
GP visit	$84.45	0.039	0.043	0.058
Other health care provider visit	$68.24	0.012	0.014	0.015
Telephone calls physician	$40.45	0.037	0.059	0.069
Telephone calls other health care provider	$5.20	0.006	0.012	0.016
Total		$15.10	$19.99	$28.89

Abbreviations: CAD: Canadian dollars; ER: emergency room; GP: general practitioner.