Bronchodilator reliever use and its association with the economic and humanistic burden of COPD: a propensity-matched study

Abstract

Background and aims: Short-acting bronchodilators are normally used as supplemental relief medication for breakthrough symptoms in COPD patients. The objective of this cross-sectional study was to assess if more frequent vs infrequent use of relief medication in maintenance-treated COPD patients, split by the severity dyspnea, was associated with an increase in the overall disease burden.

Methods: A population-based cross-sectional survey (Adelphi DSP) was conducted among patients with COPD in five European countries. Information was collected on demographic and clinical characteristics, reliever inhaler use, dyspnea (mMRC), health status (CAT, EQ-5D), sleep quality (JSEQ) and healthcare resource use including moderate–severe COPD exacerbations, physician visits, COPD medications and other COPD related resources. The humanistic and economic burden was compared between patients with infrequent reliever use (<1 occasion/week) and more frequent use (≥ 1 occasion/week). The association between increased reliever use and economic burden was also examined after matching patients based on propensity-scores balancing demographic and disease burden characteristics.

Results: Among the 1373 COPD patients prescribed a reliever inhaler, 29% reported using reliever medication ≥1 occasion/week. In the unmatched cohort, more frequent reliever use (n = 377) compared to infrequent use (n = 996) was linked to poorer health status (CAT: 25.7 vs 20.0; p < .0001; EQ-5D-3L: 0.63 vs 0.82; p < .0001) and poorer sleep quality (JSEQ: 8.6 vs 4.6 units; p < .0001). More frequent reliever use was also associated with higher annual rates of moderate/severe exacerbations (1.6 vs 1.0 events/year; p < .0001) and respiratory specialist visits (2.8 vs 2.2 events/year; p = .0001). In the propensity-score matched population, more frequent reliever use was also associated with significantly higher annual costs for COPD management (€5,034 vs €3,705, p = .0327) compared to patients with infrequent reliever use.

Conclusion: In moderate-to-severe COPD, more frequent reliever use is associated with increased exacerbation risk and increased management costs.

Keywords:

• COPD
• Reliever use
• Dyspnea
• Costs
• Resource use

Introduction

Chronic Obstructive Pulmonary Disease (COPD) is characterized by persistent, usually progressive, airflow limitation¹. Its progressive nature leads to chronic morbidity and mortality¹, and has become the fourth leading cause of death in the world². Despite effective COPD treatments, the under-diagnosis and under-reporting of chronic symptoms is a significant healthcare challenge that can lead to an avoidable added social and economic impact³.

COPD is characterized by persistent symptoms, with dyspnea being the most common. Patients with recognized moderate-to-severe dyspnea have been shown to have higher COPD related resource use and costs⁴. However, COPD patients may potentially under-report daily symptoms if they over use short-acting beta-2 agonist (SABA) bronchodilator to relieve dyspnea and other COPD symptoms⁵.

Supplemental reliever use has been shown to improve lung function and symptoms when used in addition to maintenance long-acting bronchodilators^6,7. As a result, they are often used in combination with a maintenance COPD treatment in stable moderate-to-severe COPD patients. Monitoring relief medication use is a key control measure in asthma management and increases in maintenance treatment are targeted to minimize over-reliance on reliever therapy⁸. Over-reliance on reliever in COPD has not received similar attention. Despite this, increasing reliever use has been shown to correlate with other COPD end-points such as a decrease in trough forced expiratory volume in 1 second (FEV₁)⁹ and more frequent exacerbations^10,11. Moreover, UK primary care records highlight that a shorter time to a long-term step up in therapy from mono bronchodilator to dual or triple maintenance therapy is associated with increased SABA prescription rates¹². Finally, a population-based case-control study in COPD has shown that SABA therapy in addition to maintenance treatment is associated with increased COPD mortality, albeit this risk is largely driven amongst the patient cohort using ≥400 mcg/day of salbutamol equivalents¹³. Whilst moderate-to-severe symptoms of dyspnea have been shown to predict current and future costs in COPD, the relationship between frequent reliever use and healthcare resource use is unknown.

The primary objective of this study was to examine if frequent self-reported reliever medication use among stable COPD patients was associated with increased humanistic and economic burden. Further, we estimated whether this impact was consistently observed among patients with or without clinically significant moderate-to-severe dyspnea.

Methods

Survey design and study population

The Respiratory Disease Specific Programme (DSP; Adelphi Real World, Macclesfield, UK) is a multinational survey which collects key health outcomes information from patients diagnosed with COPD and their physicians on an ongoing regular basis. The full methodology of the DSP programme¹⁴ and this current analysis¹⁵ has been described previously.

Briefly, 50 respiratory specialists and 50 primary care physicians from each of the five European countries including France, Germany, Italy, Spain, and the UK were randomly selected from public lists of healthcare professionals. These physicians were selected based on the quotas set by specialty type, geographical location, and number of COPD patients currently being treated by them to be representative of the national population of physicians. A combined dataset from three separate Respiratory DSP surveys (2010, 2011, and 2012) was used in the current analyses.

The recruited physicians completed a standardized record form on six consecutive consenting patients diagnosed with COPD visiting their practice and personally seen by the physician. Patients meeting the inclusion criteria of more than 40 years of age with history of smoking and diagnosis of airflow obstruction (COPD, emphysema and/or chronic bronchitis) were selected by physicians, irrespective of their reason for physician visit. Information was collected on patient demographics, diagnostic history and resource use in the past 12 months, current symptomatology and its impact on lifestyle, current and past treatments including reliever use, treatment adherence, comorbidities¹⁶, and inhaler preference. Of these, data on healthcare resource use were included in the present analysis.

The patient completed information on their COPD history, current symptomatology, and its impact on health status, current treatment, and attitude towards COPD and its treatment. They also provided information on dyspnea using modified medical research council (mMRC) scale¹⁷, COPD related health status using COPD assessment test (CAT)¹⁸, general health status using EQ-5D (3L)¹⁹, sleep quality based on the Jenkins sleep evaluation questionnaire (JSEQ)²⁰, and Morisky medication adherence scale (MMAS)²¹.

Included patients were also required to have received at least one short- or long-acting COPD-specific treatment for at least 12 months, to have a completed mMRC scale from patient completed questionnaire, and had to be currently prescribed reliever medication. The survey was conducted as a market research survey adhering to the ICC/ESOMAR International code on market and social research, and, therefore, ethical approval was not sought²². Prior to volunteering to complete a questionnaire, patients were asked to provide informed consent.

Study cohorts

Patients were divided into two cohorts based on their levels of dyspnea as no–mild (mMRC <2) and moderate-to-severe (mMRC ≥2) dyspnea. Each cohort was further stratified by how often they needed to use their reliever inhaler in the previous 4 weeks, as less frequent users (less than once a week) and more frequent users (once a week or more) (Table 1)²³.

Table 1. Patient disposition and demographic characteristics.^a

Patient variable, n (%)	All patients						Matched patients
	mMRC <2 (n = 678)			mMRC ≥2 (n = 695)			Reliever use		SMD (%)
	Reliever use			Reliever use		Total (n = 695)	Less than once a week (n = 241)^b	Once a week or more (n = 241)
	Less than once a week (n = 555)	Once a week or more (n = 123)	Total (n = 678)	Less than once a week (n = 441)	Once a week or more (n = 254)
Age, mean (SD)	62.1 (10.3)	62.1 (9.6)	62.1 (10.2)	68.3 (9.5)	68.0 (8.7)	68.2 (9.2)	65.6	65.5	−0.6
Gender, Male	401 (73.4)	85 (71.3)	486 (73.0)	318 (73.8)	176 (73.3)	494 (73.6)	67.6	66.8	−1.8
Time since diagnosis in months,  mean (SD)	63.5 (51.7)	82.1 (62.2)	66.7 (53.9)**	92.7 (79.6)	106.2 (75.9)	97.8 (78.2)*	92.0	92.5	0.8
Ex-smokers	332 (59.3)	76 (68.6)	408 (61.0)	301 (72.3)	172 (69.6)	473 (71.3)	65.6	65.1	−0.9
Currently on ICS treatment	320 (52.4)	80 (63.4)	400 (54.4)	324 (73.4)	193 (77.8)	517 (75.1)	72.2	71.4	−1.8
ICS alone	22 (9.4)	4 (4.9)	26 (8.5)	12 (2.6)	6 (4.4)	18 (3.3)	—	—	—
ICS combination^c	120 (37.0)	31 (35.5)	151 (36.7)	91 (25.8)	48 (26.6)	139 (26.1)	—	—	—
Triple therapy	178 (53.6)	45 (59.6)	223 (54.8)	221 (71.6)	139 (69.0)	360 (70.6)	—	—	—
Triple therapy^d	178 (28.1)	45 (37.8)	223 (29.8)	221 (52.5)	139 (53.7)	360 (53.0)	47.7	49	2.5
2+ exacerbations	145 (20.6)	35 (28.0)	180 (21.9)	198 (41.0)	167 (62.3)	365 (49.1)**	53.9	53.9	0
Number of comorbidities, mean (SD)	0.9 (1.2)	1.0 (1.0)	0.9 (1.1)	1.3 (1.3)	1.6 (1.6)	1.4 (1.4)	1.3	1.3	1.3
Serious CV comorbidity	120 (17.9)	30 (27.3)	150 (19.6)	189 (46.2)	139 (57.8)	328 (50.7)***	45.2	45.2	0

^a Results for all patients are based on weighted analyses and those for propensity matched cohort were based on unweighted data.

^b Consists of 241 patient profiles with 196 unique patients.

^c As a percentage of patients who currently receive ICS-containing therapy.

^d As a percentage of all patients (ideally this needs to remain for the linkage to the matching information); Triple therapy − long acting muscarinic antagonists (LAMA) + ICS + long acting beta agonist (LABA) or LAMA + ICS/LABA.

n, number of patients; SD, Standard deviation; SMD, Standardized mean difference.

* p = .0279;

** p = .0001;

*** p = .0288.

Study outcomes and their derivation

For each of these strata, the COPD-specific impact on health status was captured using CAT, impact on sleep was captured using JSEQ, and general health status was estimated using EQ-5D in the patient record forms. Utilities were then calculated using country-specific valuation sets of EQ-5D (3L). The resource use over the previous 12 months was estimated using data collected in the physician-completed patient form. It included primary and secondary care consultations, moderate and severe exacerbations, visits to the emergency department, hospital stays due to COPD, as well as COPD-related medications. Moderate exacerbations were defined as elevated symptoms requiring treatment with an antibiotic and/or oral corticosteroid and a severe exacerbation was defined as elevated symptoms requiring emergency room visit or hospitalization. Mean estimates for each resource use were derived for each country separately and were then converted to costs using appropriate unit cost for each event (Table 2).

Table 2. Unit costs of resource use in individual EU countries.

Costs (€)	France	Germany	Italy	Spain	UK
COPD treatments
SABA	5.8	3.0	3.0	2.2	4.1
SAMA	2.4	40.4	15.3	17.1	11.9
LABA	19.6	39.0	25.5	29.3	31.7
ICS/LABA	35.3	71.6	39.5	44.4	39.6
Xanthines	0.8	2.3	1.7	1.8	5.3
ICS	18.4	16.5	7.9	14.0	22.4
PDE4 Inhibitor	—	25.1	4.3	37.4	42.5
LAMA	12.9	50.2	20.2	25.5	26.3
Exacerbation treatments
OCS only exacerbation	33.7	47.0	30.7	62.7	90.3
Antibiotic only exacerbation	35.9	68.1	32.2	64.2	92.7
OCS + AB exacerbation	38.4	69.9	37.4	65.5	96.3
Consultations
PCP consultation	24.1	20.5	13.5	57.2	77.1
Pulmonologist consultation	40.6	40.2	22.9	169.1	275.0
Other COPD therapeutic options
Influenza vaccination	6.1	15.1	5.8	6.1	6.5
Pneumococcal vaccination	15.0	21.5	47.7	14.7	10.8
Lung transplant	18,000.0	49,137.0	72,572.0	117,948.2	3,150.7
Lung volume reduction surgery	63,000.0	5,701.5	2,270.0	1,314.7	3,150.7
Long-term oxygen therapy	3,868.8	2,604.6	2,847.7	2,000.2	1,701.7
Hospitalizations
Emergency department and hospitalization admission	8,413.5	7,174.2	3,344.0	6,564.1	1,766.4
Emergency department admission (no hospital stay)	148.5	190.4	241.0	264.9	132.5

OCS, Oral corticosteroids; AB, antibiotics.

The country-specific unit costs were obtained from literature wherever possible and were supplemented by a local health economic expert opinion if unavailable^23–36. If multiple costs were available, the lowest cost was used as a conservative estimate. The cost of a moderate exacerbation was estimated to include prednisolone (40 mg od) treatment for 5 days if treated with an oral corticosteroid and/or co-amoxiclav treatment (875 mg/125 mg bd) for 7 days if treated with an antibiotic. Country-specific unit costs for each class of COPD treatment were generated by averaging the unit costs of the top five (or less) selling products (in units) within that class for each country. The total medication cost for a therapy class was then estimated by multiplying the number of prescriptions in the last 12 months for that class (currently or previously received) by the unit cost estimated based on sales. Missing prescription data was imputed per country based on the average number of prescriptions per week for that class.

Data analysis

The frequency of reliever use was reported in each country and across Europe, by pooling patients to form a single cohort (EU5). In the descriptive analysis, patients were weighted to correct for design bias, ensuring that the patient sample was as close as possible to the COPD population in that country. Weighting was based on an inverse-probability of the patients’ 12-month consultation rate. For patients with less than 12 months of consultation history, this weight was imputed based on their available consultations and the length of time the patient has been managed by the physician.

Patients in each stratum were described using demographic and healthcare utilization descriptors, firstly by their frequency of reliever use and further, at the cohort level, by their level of dyspnea (mMRC <2 vs ≥2). Within each dyspnea-based cohort, the CAT score, JSEQ score, and EQ-5D utilities were calculated and compared between patients with less frequent reliever use (less than once a week) and more frequent reliever use (once a week or more). The resource use variables and costs were also compared across dyspnea and reliever use levels. Appropriate weighted regression models were performed to assess statistical differences, whereby the type of regression model used depended on the type or distribution of the outcome variable. All the comparisons were unadjusted at a significance level of 0.05.

Finally, the full sample was divided into two strata based on reliever use. Patients using reliever medications frequently were propensity matched with replacement, with patients using reliever medication less frequently to produce closely matched profiles^37–39. Covariates included in the matching were age, gender, smoking status, time since diagnosis, use of inhaled corticosteroid (ICS) containing and triple therapy, exacerbation history, number of concomitant conditions, and presence of severe cardiovascular (CV) comorbidity. Propensity-scores were generated using logistic regression models and each patient from an unweighted sample with less frequent reliever use was matched to a patient with more frequent reliever use in the previous 4 weeks. Standardized mean differences (SMDs) were used to assess covariate balance post-matching. A SMD between −10% and 10% (not inclusive) was assumed to provide adequate balance for each covariate³⁷. The total annual COPD management costs for the two groups were estimated and compared by calculating the Abadie-Imbens (AI) standard error (SE) and corresponding test statistic and p-value^40,41. The effect of unobserved confounders on the p-value was assessed using Rosenbaum sensitivity analyses⁴². Analyses were conducted using Stata v13.0 and R v3.0 or later versions.

Results

A total of 1,373 COPD patients (EU5 cohort) who had been prescribed a reliever medication across five European countries (France, Germany, Italy, Spain, and the UK) were included in this analysis. The mean age of this cohort was 65.2 years and 26.7% were females. The mean time since COPD diagnosis was 6.8 years and 33.8% were continuing smokers. Two-thirds of all patients were prescribed ICS containing treatment (64.9%), with 41.6% receiving triple therapy. The mean number of concomitant conditions for this cohort were 1.2, with a third (35.4%) suffering with major CV co-morbidity.

Frequency of reliever inhaler use and association with dyspnea

Among patients in EU5 cohort who were prescribed a reliever inhaler, 29% used it >1 occasion/week in the 4 weeks prior to the survey. In the individual countries, the proportion of patients who used their reliever inhaler at this level included 13% in Italy, 24% in France, 28% in Germany, 29% in Spain, and 53% in the UK. A significantly higher proportion of more frequent reliever users also experienced moderate-to-severe dyspnea (p < .0001). Patients with more frequent reliever use were diagnosed with COPD for a longer period (p < .0001), and a higher proportion of patients were on ICS containing treatments (p = .0025), including triple therapy (p = .0111). A higher proportion also had more than two exacerbations in the previous 12 months (p < .0001), and had more concomitant conditions (p = .0018) including CV disease (p < .0001). The demographic and clinical characteristics of the reliever and dyspnea based sub-strata are displayed in Table 1.

Health status by frequency of reliever inhaler use

The unadjusted results showed that patients with more frequent use of reliever inhaler had higher CAT (25.7 vs 20.0; p < .0001) and JSEQ (8.6 vs 4.6; p < .0001) scores and lower EQ-5D utility score (0.63 vs 0.82; p < .0001) compared to patients with less frequent use of reliever therapy. Within each of the reliever usage-based strata, patients with moderate–severe dyspnea had worse scores on CAT, JSEQ, and EQ-5D compared to patients with no-mild dyspnea (all p < .0001; Figure 1a and b).

Figure 1. Association of reliever use ± breathlessness on (a) sleep and COPD-related health status and (b) general health status at the time of survey*. *Values based on weighted data; RU, Reliever use; CAT, COPD assessment test; EQ-5D (3L), EuroQoL questionnaire (3 levels); p < .0001 for all unadjusted comparisons between RU-Less than once a week and RU-Once a week or more within each cohort.

Association of frequency of reliever inhaler use with healthcare resource utilization (HCRU)

Patients with more frequent reliever use experienced more moderate–severe exacerbations (mean annual rate: 1.6 vs 1.0; p < .0001) and severe exacerbations requiring accident and emergency department visit or hospital treatment (0.7 vs 0.3; p < .0001) in the previous year compared to patients with less frequent reliever use. These patients also experienced more emergency department visits leading to hospitalization (0.5 vs 0.2; p < .0001) and more emergency department visits not leading to a hospital admission (0.3 vs 0.1; p < .0001). In both the reliever usage based strata, patients with moderate–severe dyspnea had higher annual rates of moderate–severe exacerbations, severe exacerbations, and emergency department visits with/without hospital admissions compared to patients with no-mild dyspnea (all p < .01; Figure 2).

Figure 2. Resource use in prior 12 months by frequency of reliever inhaler use and dyspnea. Rates based on weighted data; p < 0.0001 for all unadjusted comparisons between RU-Less than once a week and RU-Once a week or more within each dyspnea cohort; p < .01 for all unadjusted comparisons between mMRC <2 and mMRC ≥2 within each reliever use strata.

Patients with more frequent reliever use experienced higher rate of specialist visits (2.8 vs 2.2; p = .0001) compared to less frequent reliever users, whilst the rate of primary care visits was similar (3.6 vs 3.3). A higher proportion of these patients also had undergone pulmonary rehabilitation (22.8% vs 10.8%; p < .0001) or received long-term oxygen therapy (13.4% vs 5.1%; p < .0001) in the previous 12 months. In the less frequent reliever use sub-strata, patients with moderate–severe dyspnea had higher rates of primary care and specialist visits and higher proportions had undergone long-term oxygen therapy and a pulmonary rehabilitation program. The resource use was comparable between dyspnea groups in the moderate–high reliever use sub-strata.

Impact on costs

Mean annual COPD management costs among more frequent reliever inhaler users were €5,761 (95% confidence interval [CI]: €4,452; €7,071). Exacerbation/hospitalization costs (€3,967 [€3,127; €4,806]) constituted a major share of these costs followed by other costs (€688 [€84; €1,292]), medication costs (€521 [€466; €575]), and consultation costs (€428 [€367; 488]). The total costs among less frequent reliever users were significantly lower, at €2,271 [€1,900; €2,642] (p < .0001). The corresponding cost categories also were lower, at €1,315 [€1,010; €1,621] for exacerbations/hospitalizations, €217 [€122; €313] for other costs, €408 [€376; €439] for medication costs, and €267 [€245; €290] for consultation costs (Figure 3). Within the dyspnea-based sub-strata, the total costs, medication costs, and exacerbation/hospitalization costs were consistently higher for patients with moderate–severe dyspnea compared to patients with no–mild dyspnea, irrespective of their levels of reliever use. Dyspneic patients also had higher consultation costs among less frequent reliever users, whilst the consultation costs were comparable among moderate–high reliever users.

Figure 3. Costs in prior 12 months by frequency of reliever inhaler use and dyspnea, based on weighted data.

In the EU5 cohort, propensity matching on the baseline demographics and clinical parameters except mMRC scores resulted in 241 patients with moderate–high reliever inhaler use in the previous 4 weeks and 241 matched profiles (196 unique patients) of patients with less frequent reliever inhaler use. All covariates were adequately balanced post-matching (Table 1). In this matched sample, the total annual COPD management cost was significantly higher among patients with more frequent reliever use compared to less frequent reliever use (€5,034 vs €3,705; p = .0327). The Rosenbaum sensitivity analysis produced a Γ of 1.3, indicating that the result was moderately insensitive to unobserved confounding factors. Although no statistical comparison was performed on individual cost components, patients with moderate–high reliever use also showed higher costs of exacerbations/hospitalizations (€3,471 vs €2,498), consultations (€431 vs €305), medications (€514 vs €488), and other costs (€618 vs €414) (Figure 4). Propensity matching on the sub-strata by levels of dyspnea was not performed due to insufficient sample size.

Figure 4. Costs by frequency of reliever inhaler use in the propensity matched patient cohort.

Discussion

Patients with COPD experience substantial variation in the level of symptoms, with seasonal variation that may also change from day-to-day or even within different periods of a single day^43,44. Symptoms are essential features of disease stability in COPD and may be the most visible factor guiding treatment decisions in a primary care setting. In randomized clinical trials patient reported outcomes usually capture this symptom information at a set point in time (e.g. by monthly questionnaire), capturing the longitudinal aspects of treatment responses on symptoms, but often failing to assess short-term fluctuations in symptoms or mild exacerbations, which frequently remain unrecorded. In asthma, monitoring the frequency of as needed reliever medication use in response to symptoms is an essential component of assessing disease control⁴⁵. However, this end-point has been largely under-reported in observational studies assessing patient well-being in COPD⁴⁶.

This European database study attempted to quantify the economic and social impact among patients with differing levels of reliever use. The results indicated that most COPD patients (71%) reported infrequent reliever medication use on a weekly basis. However, a higher frequency of reliever use above a modest threshold of once per week was associated with increased dyspnea. In addition, more frequent reliever use was also associated with a significant impact on sleep quality during the previous 4 weeks and reductions in health status at the time of the survey compared to infrequent reliever users. Furthermore, this detrimental health impact was pronounced among patients with clinically identifiable dyspnea. This observation suggests that patients with elevated dyspnea are also more frequent users of reliever medications, which may potentially mask some of their symptoms and, thus, controlling for both reliever use and dyspnea assessments may more accurately reflect ongoing disease activity and potential lack of stability in COPD.

The adverse impacts on patients’ health status of increased reliever use and increased dyspnea was also mirrored by higher exacerbation rates, healthcare resource use, and costs. This was particularly evident when looking at the number of hospitalized exacerbations and nights with hospital admissions. Historically, over-reliance on reliever medication based on collected prescription rates with or without under-use of maintenance inhaled corticosteroids has been well documented to increase the risk of hospital admissions due to asthma^47,48. More recently a post hoc analysis of data from a budesonide/formoterol vs formoterol exacerbation study has also shown elevated reliever use to be a reliable predictor of both short-term and long-term exacerbation risk in COPD⁴⁹. Our observational findings were consistent with these results. Our findings suggest that, similar to asthma, a more holistic approach considering symptoms, reliever use, and exacerbation history may more accurately assess future risk in COPD.

Our study has potential limitations. The data were collected using a cross-sectional survey. This limited our ability to estimate the temporal relationship between variables and assess any causal link with reliever medication use. The information on the use of reliever medication was collected as a categorical variable ranging from ‘not at all’ to ‘every day’ through the patient survey. As a result, the exact amount of reliever use in puffs/day could not be estimated and, therefore, we did not conduct any statistical analyses of a dose association between reliever medication use and resource use or costs. The information provided by patients is also subject to recall bias. The association between reliever use and HCRU and dyspnea and HCRU has limitation of variable exposure time such that the reliever use was recorded over the previous 4 weeks, dyspnea at the time of the survey, and HCRU over the previous 12 months. However, they can be reasonably assumed to be representative of the patient’s dyspnea over a 12-month period since the instrument is relatively insensitivity to change and it is uncommon for individuals to improve or deteriorate by an entire grade over relatively short periods⁵⁰. The sample collected was a convenience sample of consecutive eligible COPD patients who actively consulted their physician; therefore, it may not fully represent the broad population of COPD patients in the community in each country who may present less frequently to healthcare providers. This methodology, however, helps eliminate any selection bias from the physician. Physicians were selected to represent the national sample of healthcare professional. However, their inclusion was likely influenced by their willingness to take part, and practical considerations of geographical location. The results were further weighted to be generalizable to the COPD population of the specific country in which the data was collected. Due to an assumption of normality in calculating the 95% confidence intervals for right-skewed data, in some instances negative confidence intervals may be observed. We used a propensity matching method with replacement. Whilst matching with replacement does not use the full set of patients available for analysis by eliminating those patients in the control group that are not an ideal match for any patient in the treatment group, it produces matches of higher quality by increasing the set of possible matches³⁸.

Conclusion

Across multiple European countries, we found an association between more frequent use of reliever medication in COPD patients and increased disease burden and costs to the healthcare system. Interventions likely to reduce symptoms and inappropriate reliance on reliever use may offer significant benefit to patients and cost savings to healthcare systems.

Transparency

Declaration of funding

Funding for this study and article processing charges were provided by GlaxoSmithKline.

Declaration of financial/other relationships

YSP, HM, and IN are employees of GlaxoSmithKline and hold stocks at GlaxoSmithKline. MS, TH, and RW are employed by Adelphi and were paid consultants to GlaxoSmithKline. MV was a paid consultant to GlaxoSmithKline. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgements

The survey was conducted as a market research survey adhering to the ICC/ESOMAR International code on market and social research. The study was a retrospective analysis of secondary data and, therefore, ethical approval was not sought. Prior to volunteering to complete a questionnaire, patients were asked to provide informed consent. The Program instructions describe the purpose of the survey, why the respondent had been selected, and who might have access to the aggregated and anonymized dataset compiled.