Effectiveness of Inhaled Combined Corticosteroid/Long-Acting Bronchodilator Treatment in Reducing COPD Exacerbations and Short-Acting Bronchodilator Use
Douglas Mapel, MD, MPH, Melissa H. Roberts, MS, PhDc, Christopher M. Blanchette, PhD, Hans Petersen, MS, and Sulabha Ramachandran, PhD
From the Lovelace Clinic Foundation, Albuquerque, NM (Drs. Mapel and Roberts), IMS Consulting Group, Alexandria, VA (Dr. Blanchette), Lovelace Respiratory Research Institute, Albuquerque, NM (Mr. Petersen), and AstraZeneca Pharmaceuticals, Wilmington, DE (Dr. Ramachandran).
• Objective: To compare budesonide/formoterol (BFC) and fluticasone propionate/salmeterol (FSC) in their ability to reduce COPD exacerbations using a novel risk stratification approach.
• Design: Retrospective cohort study with propensity matching and stratification by short-acting beta2-agonist (SABA) use.
• Participants: 7704 COPD patients identified using 2 U.S. health claims databases (study period: 1/1/2004 to 4/30/2009).
• Measurements: Users of BFC and FSC were stratified by SABA fills in the 6 months prior to the start of therapy (index date). Exacerbation events before and after the index date were compared, and proportional hazard models used to examine exacerbation risk by baseline SABA use.
• Results: At baseline, 6% (6.2%, BFC; 5.8%, FSC) had a COPD-related hospitalization or emergency department (ED) visit. Post-index, this was reduced to 4.8% (P < 0.001) among all patients (4.9%, BFC, 4.7%, FSC). Change was greatest among high SABA users (9% pre-index to 4.9% post-index, P < 0.001). SABA use post-index was reduced in both groups, with slightly lower need for SABA refills seen among BFC users. SABA use was correlated with increased exacerbations during the baseline period and was predictive of exacerbations during the follow-up period.
• Conclusions: BFC and FSC had equivalent effectiveness in reduction in ED visits and hospitalizations for COPD exacerbations, but BFC users had lower subsequent need for SABA fills. Increased SABA use is a predictor of COPD exacerbation among COPD patients in the general population.
Chronic obstructive pulmonary disease (COPD) is a progressive disease of the lungs characterized by variable combinations of inflammation of the airways and emphysematous damage to the lung parenchyma, which result in airflow obstruction. Chronic respiratory symptoms associated with COPD, mainly dyspnea, wheezing, and cough, may acutely worsen during exacerbations . COPD exacerbations usually result in treatment changes, ranging from increased use of short-acting bronchodilators for dyspnea relief to hospitalization for hypoxemia and respiratory distress. In general, as the severity of COPD worsens, the frequency and severity of exacerbations increase, but there are individuals with relatively modest airflow obstruction who experience frequent exacerbations . In addition to the acute decrease in health status, exacerbations may result in permanent loss of lung function, faster disease progression, and increased risk of death [3–7].
Acute exacerbations of COPD generate high demand for health care services and account for the largest portion of health care expenditures for COPD [8–11]. Exacerbations are the most common reason for hospitalization of COPD patients, who have 2 or 3 times the annual risk for hospitalization of age- and gender-matched controls without COPD [9,12]. In a survey of over 600 hospitals across the United States, exacerbations resulting in an emergency department (ED) visit averaged $647 (2008 US$) per patient, $7242 for a 4-day hospital admission, $20,757 for a complex admission, and $44,909 for an intensive care unit admission . Hospital costs account for 45% of direct medical costs for COPD patients in the United States , which is similar to the estimate of 53% in the UK . Moderate exacerbations that prompt unscheduled ED and clinic visits but not hospitalization are also expensive, with mean direct cost estimates ranging $248 to $1011 (2007 US$) in international studies .
Main objectives of COPD treatment include the reduction of chronic symptoms and the prevention of acute exacerbations . Inhaled medications that have demonstrated effectiveness in reducing COPD exacerbations include long-acting beta2-agonists (LABAs), long-acting anticholinergics, and inhaled corticosteroids (ICS) . Combinations of these agents may result in additional reduction of exacerbation rates as compared with use of single agents used alone . Currently, 2 combinations of ICS/LABAs are approved for use in COPD in the United States: budesonide/formoterol fumarate dihydrate combination (BFC) and fluticasone propionate/salmeterol combination (FSC), although only FSC is specifically approved by the U.S. Food and Drug Administration for marketing as a prevention for COPD exacerbations.
Traditionally, the severity of COPD is graded by the degree of airflow obstruction, and patients with greater obstruction tend to experience more frequent and more severe exacerbations [2,4,7]. Short-acting beta2-agonist (SABA) medications are bronchodilators that work quickly with effects lasting 2 to 4 hours, and are prescribed for all patients with COPD on an as-needed basis for relief of symptoms [20–22]. Patients often use SABAs not only for relief but on a preventative basis, such as before undergoing activities that may require significant exertion. Thus, the degree to which SABA medications are used may be indicative of the level of COPD severity.
In a previous study we compared BFC and FSC on a variety of outcomes, including exacerbations and associated health care service use, adherence to prescribed treatment regimen, health care costs, the use of rescue inhalers, and pneumonia events. In that study, BFC and FSC demonstrated comparable real-world effectiveness as measured by the reduction in the number of severe exacerbations in the 6 months after initiation of treatment. However, patients who initiated therapy with BFC had a greater number of comorbidities and higher rates of COPD-related utilization, including SABA use, prior to the initial prescription, creating the possibility of residual confounding.
The objective of this study was to assess changes in hospitalizations and ED visits for COPD exacerbations among patients started on combination ICS/LABA medications with stratification for baseline SABA use. A secondary objective of this study was to examine SABA use before and after initiation of combination treatment and to see whether SABA use was an indicator for increased risk of exacerbation.
This retrospective analysis was based on data from the IMS PharMetrics Integrated Database (IMS, Norwalk, CT) and MarketScan (Thomson Reuters, Ann Arbor, MI), which gather de-identified health care claims data from more than 90 health insurance providers and 55 million patients from across the United States. Patients were required to be age 40 years or older and have a diagnosis of COPD (primary or secondary ICD-9-CM codes 491.xx [chronic bronchitis], 492.xx [emphysema], or 496 [chronic airway obstruction]) associated with an ED visit, hospitalization, or outpatient visit at least once during the study period (1 July 2006 to 30 June 2010). Patients were also required to have a new prescription fill for either BFC or FSC, which we define as the index date, and have continuous insurance coverage for 6 months before and at least 3 months after the index date. Patients who had a respiratory tract cancer (ICD-9-CM codes 161, 161.x, 162, 163, 231, or 231.x) or who were oral corticosteroid–dependent (defined as those with a medication possession ratio for oral corticosteroids ≥ 0.50 in the 6 months before the index date) were excluded.
To reduce selection biases that may arise from the nonrandom nature of an observational study, we used a propensity-matching method to match BFC patients to FSC patients using clinical characteristics in the baseline period (6 months before the index date) [23,24]. BFC patients were propensity-matched to FSC patients based on sex, age, geographic region, year of treatment initiation, comorbid conditions, months of follow-up time, and pre-index utilization for any COPD-related hospitalization, any pneumonia-related hospitalization (associated with a pneumonia ICD-9 code), any COPD-related ED visit, any oral corticosteroid fill, any antibiotic fill, and ipratropium use (0, 1, or > 1 fill). Comorbidities were identified based on Elixhauser classifications—asthma, obstructive sleep apnea, heart disease (rheumatic heart disease, hypertensive heart disease, ischemic heart disease, and cardiovascular heart disease) . The Elixhauser measure has the advantage of being able to provide information on relationships between individual comorbidities and study outcomes, as opposed to an index comorbidity score like the Charlson measure . The existence of other serious lung conditions during the baseline period (bronchiectasis, cystic fibrosis, coal worker pneumoconiosis, asbestosis, pneumoconiosis due to silica, inorganic dust, inhalation of other dust, or unspecified respiratory conditions due to chemical fumes and vapors or other unspecified external agents, post-inflammatory pulmonary fibrosis, other alveolar and parietoalveolar pneumonopathy, lung involvement in conditions classified elsewhere, other respiratory diseases, lung cancer, tuberculosis, extrinsic allergic alveolitis, lipid pneumonia, detergent asthma, or other diseases of the lung ) was also included. Patients were not matched on SABA use.
Analyses and Statistics
Patients were characterized according to SABA use as evidenced through prescription claim information in the baseline period. The 3 stratifications were no SABA (no prescription claim), low SABA (1 prescription claim), and high SABA (> 2 prescription claims), which were stratified based on the overall distribution of SABA prescription fills. The main outcome measures of interest were severe exacerbation events, defined as hospitalizations or ED visits with a primary respiratory diagnosis (COPD, pneumonia, or respiratory distress). Patients in the main analyses were required to have at least 6 months of follow-up information after initiation of therapy.
To examine whether SABA use in the baseline period predicted risk of exacerbation in the follow-up period, we developed a Cox proportional hazard model using data from the entire cohort of COPD patients who met all inclusion criteria, were propensity-matched 1:1 BFC to FSC, and had a minimum of 3 months follow-up (n = 11,830). All statistical tests were 2-sided with a level of significance of 0.05 and were conducted using SAS version 9.1.3 for Windows (SAS Institute, Cary, NC).
Of the patients who were initiated on BFC during the study period, 99% were matched to patients initiating with FSC (total n = 7704). The demographic and clinical information for these groups are compared in Table 1. Mean age was 63 years, 45% of each treatment group was male, and subjects were predominantly from the Northeast/East (31%), Midwest/North Central (31%), and South (29%) regions. The most common comorbidities were hypertension (41%), asthma (33%), other lung conditions (20%), heart disease (19%), and diabetes (17%). Mean rates of COPD-related and pneumonia-related medical services utilization at baseline were similar. A low proportion of patients (13%) were using any form of inhaled ipratropium. At baseline, there was no difference between groups in the percentage with no SABA use (Table 1), but a lower percentage of BFC patients had 1 SABA fill (14.7% vs. 16.6%, P = 0.02) and a higher percentage 2 or more SABA prescription fills (21.7% vs. 19.3%, P = 0.01).
Following initiation of maintenance treatment, the percentage of patients having at least 1 COPD-related hospitalization did not significantly differ between the 2 treatment groups (BFC 3.6%, FSC 3.1%, P = 0.23) (Table 2). In both the BFC and FSC groups, there was a significant decrease in hospitalizations in the 6 months after initiation of therapy as compared with baseline (BFC = 4.6% prior, 3.6% post, P = 0.02; FSC = 4.4% prior, 3.1% post, P = 0.002), and also for the combined measure of hospitalization or ED visit (BFC = 6.2% prior, 4.9% post, P = 0.008; FSC = 5.8% prior, 4.7% post, P = 0.023). There was no difference between treatment groups in the percentages with at least 1 COPD-related ED visit (BFC = 1.5%, FSC = 1.9%, P = 0.21) or with at least 1 hospitalization or ED visit (BFC = 4.9%, FSC = 4.7%, P = 0.67).
When study subjects were stratified by their SABA use during the baseline period, the most substantial decrease in exacerbations resulting in ED visits or hospitalization was found among those with 2 or more SABA fills at baseline (Table 3). Among these patients, hospitalizations decreased from 6.1% to 3.7% (P < 0.001); ED visits decreased from 3.0% to 1.5% (P = 0.002), and the combination measure of hospitalization or ED visits decreased from 9.0% to 4.9% (P < 0.001).
Table 3 also demonstrates that SABA use in the baseline period was correlated with increased ED and hospital visits during the baseline period. Among patients with no SABA use at baseline, 3.8% had 1 or more baseline hospitalizations compared with 5.1% of patients with low SABA use (P = 0.06) and 6.1% of patients with high SABA use (P < 0.001). Similarly, among patients with no SABA use, 1.2% had at least 1 baseline ED visit, compared with 1.9% with low SABA (P = 0.06) and 3.0% with high SABA use (P < 0.001). There is also an association between baseline SABA use and exacerbations during the follow-up period, but the relationship is not linear since the risk for ED visits and hospitalizations was lower after initiation of combined therapy among those with 2 or more baseline SABA fills as compared to those with just 1 SABA fill.
The Cox proportional hazard analysis also confirmed that baseline SABA use was a significant predictor of exacerbation events even after initiation of combined ICS/LABA therapy (Figure). The demographic and clinical characteristics for this larger group of matched patients (n = 11,830) were very similar to those of the smaller cohort (data not shown). The low SABA use patients had an increased risk of hospitalization (hazard ratio [HR] 1.44 for low SABA vs. no SABA [95% CI 1.03–1.99]), ED visit (HR = 1.80, 95% CI 1.16–2.78), or the combined risk of hospitalization or ED visit (HR = 1.58, 95% CI 1.21–2.06) compared with no SABA use patients. The high SABA use patients had an increased risk of hospitalization (HR = 1.55 for high SABA vs. no SABA [95% CI 1.16–2.07]) or the combined risk of hospitalization or ED visit (HR 1.38, 95% CI 1.08–1.76), but the increased risk for ED visit did not reach statistical significance for this relatively rare event.
SABA use during the follow-up period was stratified according to SABA use during the 6-month baseline period and is found in Table 4. In general, persons who did not use SABAs in the baseline period tended not to use them during the follow-up (approximately 70%), and persons who used ≥ 2 SABA inhalers at baseline tended to continue to use ≥ 2 SABA inhalers during the follow-up (approximately 60%). However, there were substantial numbers who were able to reduce their SABA use after initiating combined ICS/LABA therapy. Among those with 1 SABA fill in the baseline period, over 50% had no SABA fills in the follow-up period, and among those with > 2 SABA fills in the baseline period approximately 40% had fewer SABA fills. Overall, a higher proportion of BFC patients had no SABA fills after initiation of therapy (59.3% vs. 56.2%, P < 0.01).
We found that for the primary endpoint for this study, exacerbations resulting in ED visits or hospitalizations, BFC and FSC had equivalent efficacy, reducing the risk for exacerbations by approximately 20% overall. However, patients treated with BFC had fewer SABA prescription fills in the 6 months after initiation of therapy, suggesting that they may have had better symptom control. This difference was observed even though the BFC cohort had a higher percentage of patients with 2 or more SABA fills in the 6 months prior to initiation of therapy. There were no substantial differences between BFC and FSC in any of the other outcome measures. Patients started on either controller therapy had substantial reductions in COPD hospitalizations and ED visits, with the greatest benefit seen among those who had high SABA use at baseline. The significance of this analysis is that it confirms the “real-world” effectiveness of ICS/LABA therapy among COPD patients treated in the general population.
We found that SABA use at baseline increased proportionally with ED visits and hospitalizations for exacerbations, which is to be expected among patients with poor COPD control. We also found that SABA use at baseline was associated with increased ED visits and hospitalizations in the follow-up period. The Cox proportional hazards model demonstrated that this increased risk for exacerbations among the SABA users was sustained beyond the initial ICS/LABA treatment date. This suggests that SABA use is a useful indicator of patients who have more severe COPD and is a valid predictor of their future exacerbation risk. This finding is useful for large retrospective or cross-sectional studies of COPD patient populations that do not have spirometry data available to describe the severity of airflow obstruction. It is also important for models of COPD progression or exacerbation risk, which can be used to describe the cost effectiveness of maintenance therapies .
These results are consistent with an earlier analysis that did not include adjustment for SABA use . That study was limited to the LifeLink claims database, was of a smaller population (n = 6770 vs. 11,830 in our cohort) with overall similar age (62 years vs. 63 years) and with a greater percentage with asthma (36% vs. 32%). The previous study cohort had a slightly higher incidence of hospitalizations in the 6 months prior to initiation of ICS/LABA combined therapy (> 6% vs. < 5%); however, the earlier study cohort was also less likely to have a SABA fill in the baseline period (32% vs. 36% in our study). Although the prior study was also propensity matched, there were imbalances in SABA utilization and comorbidity that prompted this additional investigation. This augmented analysis confirms that the comparative effectiveness between BFC and FSC is robust to different study populations and different degrees of COPD severity as indicated by hospitalization rates and SABA use.
We are aware of only one other study that has compared the effectiveness of use of BFC with FSC among COPD patients in the general population. This was a study of Canadian COPD patients who used BFC as a dry powder inhaler and FSC as either a metered-dose inhaler or a dry powder inhaler . The authors’ conclusion was that BFC patients were significantly less likely to have COPD-related ED visits and hospitalizations as compared with FSC patients. There were some important differences in the study design between the Canadian study and the current analysis. The baseline and analysis periods were longer in the Canadian study, observing patients for 12 months before and after initiation of an ICS/LABA. Patients in the Canadian study were also more likely to use SABA or ipratropium bromide in the baseline period, suggesting that their cohort had more severe COPD. There were also significant demographic differences: the Canadian COPD population was older (78% were aged 65 or older) and 52% of the sample were male. The hospitalization rate over 12 months was 8.6% of BFC patients and 12.4% of FSC patients in that cohort. We can only speculate that with longer follow-up we may have been able to see long-term differences in effectiveness between our BFC and FSC cohorts.
Retrospective COPD studies are often limited by the difficulties of adjusting for differences in severity because spirometry data are rarely included in administrative databases. In absence of being able to use the traditional mild, moderate, severe, and very severe stratifications based on percentage of predicted FEV1, investigators are limited to utilization factors that are likely to be markers of unstable disease, such as hospitalizations, ED visits, outpatient visits, and oxygen use. One of the important findings of this study is that baseline SABA use is correlated with current and future hospitalizations, confirming that it is both a marker of patients with unstable disease and a predictor of future severe exacerbations.
There are several limitations to this study that should be considered. Propensity-matching can adjust for many confounding variables, but there is still the potential for residual confounding from variables that are not measured or included in the match. Selection biases based on factors that are not included in the match could affect the results, especially if the biases substantially affected the clinicians’ choice of one treatment versus the other. There may also be misclassification errors in the diagnosis of COPD or the comorbidities, although the prevalence of asthma and other nonpulmonary conditions was very similar after the propensity match. The pre-period and follow-up time frame for this study were each relatively short (6 months); a longer follow-up analysis period may have allowed examination of effects related to patients’ gaining proficiency in tailoring treatment use to more effectively control symptoms and exacerbations. No retrospective study design can replace a randomized clinical trial in the ability to demonstrate the efficacy of the study treatments. However, this analysis does help to confirm the real-world effectiveness, especially in terms of reduction in exacerbation events.
In conclusion, we find that BFC and FSC have equal efficacy in terms of reduction in COPD exacerbation rates among patients enrolled in large managed health care systems. BFC had a slight advantage in terms of reduction in SABA use in the 6 months after the initiation of therapy. We also found that increased SABA use is associated with increased COPD-related hospitalizations and ED visits at baseline, and is predictive of increased future exacerbation events. Exacerbation reduction after initiation of ICS/LABA therapy was similar to that seen in clinical trials, confirming that these treatments have real-world effectiveness.
Corresponding author: Douglas Mapel, MD, MPH, Lovelace Clinic Foundation/LCF Research, 2309 Renard Pl. SE, Ste. 103, Albuquerque, NM 87106, email@example.com.
Funding/support: This study was funded by a research grant from AstraZeneca.
Financial disclosures: Dr. Mapel has served as a consultant to and received research grants from AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, and Pfizer Pharmaceuticals. Dr. Blanchette has served as a consultant to GlaxoSmithKline, AstraZeneca, and Sunovion. Dr. Ramachandran is an employee of AstraZeneca.
Author contributions: conception and design, DM, MHS, CMB; analysis and interpretation of data, DM, MHS, CMB, SR; drafting of article, DM, MHS; critical revision of the article, DM, MHS, CMB, HP, SR; provision of study materials or patients, SR; statistical expertise, DM, MHS; obtaining of funding, CMB; administrative or technical support, DM; collection and assembly of data, HP.
1. Rodriguez-Roisin R. Toward a consensus definition for COPD exacerbations. Chest 2000;117(5 Suppl 2):398S–401S.
2. Hurst JR, Vestbo J, Anzueto A, et al. Susceptibility to exacerbation in chronic obstructive pulmonary disease. N Engl J Med 2010;363:1128–38.
3. Seemungal TA, Donaldson GC, Bhowmik A, et al. Time course and recovery of exacerbations in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2000;161:1608–13.
4. Donaldson GC, Seemungal TAR, Bhowmik A, Wedzicha JA. Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease. Thorax 2002;57:847–52.
5. Seemungal TAR, Donaldson GC, Paul EA, et al. Effect of exacerbation on quality of life in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998;157:1418–22.
6. Miravitlles M, Ferrer M, Pont A, et al. Effect of exacerbations on quality of life in patients with chronic obstructive pulmonary disease: a 2-year follow up study. Thorax 2004;59:387–95.
7. Soler-Cataluna JJ, Martinez-Garcia MA, Roman Sanchez P, et al. Severe acute exacerbations and mortality in patients with chronic obstructive pulmonary disease. Thorax 2005;60:925–31.
8. Hilleman DE, Dewan N, Malesker M, Friedman M. Pharmacoeconomic evaluation of COPD. Chest 2000;118:1278–85.
9. Mapel DW, Hurley JS, Frost FJ, et al. Health care utilization in chronic obstructive pulmonary disease. A case-control study in a health maintenance organization. Arch Intern Med 2000;160:2653–8.
10. Strassels SA, Smith DH, Sullivan SD, Mahajan PS. The costs of treating COPD in the United States. Chest 2001;119:344–52.
11. Halpern MT, Stanford RH, Borker R. The burden of COPD in the USA: results from the Confronting COPD survey. Resp Med 2003;97:S81–S9.
12. Mapel DW, Robinson SB, Dastani HB, et al. The direct medical costs of undiagnosed chronic obstructive pulmonary disease. Value Health 2008;11:628–36.
13. Dalal AA, Shah M, D’Souza AO, Rane P. Costs of COPD exacerbations in the emergency department and inpatient setting. Resp Med 2011;105:454–60.
14. National Heart, Lung, and Blood Institute. Morbidity & mortality: 2009 chart book on cardiovascular, lung, and blood diseases. Accessed 5 Jan 2012 at www.nhlbi.nih.gov/resources/docs/2009_ChartBook.pdf.
15. Halpin DM. Health economics of chronic obstructive pulmonary disease. Proc Am Thorac Soc 2006;3:227–33.
16. Toy EL, Gallagher KF, Stanley EL, et al. The economic impact of exacerbations of chronic obstructive pulmonary disease and exacerbation definition: a review. COPD 2010;7:214–28.
17. Global strategy for the diagnosis, management and prevention of COPD. Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2011. Accessed 5 Jan 2102 at www.goldcopd.org.
18. Baker WL, Baker EL, Coleman CI. Pharmacologic treatments for chronic obstructive pulmonary disease: a mixed-treatment comparison meta-analysis. Pharmacotherapy 2009;29:891–905.
19. Mapel DW, Hurley JS, Dalal AA, Blanchette CM. The role of combination inhaled corticosteroid/long-acting beta-agonist therapy in COPD management. Primary Care Respir J 2010;19:93–103.
20. Celli BR, MacNee W, Agusti A, et al. Standards for the diagnosis and treatment of patients with COPD: a summary of the ATS/ERS position paper. Eur Respir J 2004;23:932–46.
21. Rabe KF, Hurd S, Anzueto A, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease—GOLD executive summary. Am J Respir Crit Care Med 2007;176:532–55.
22. Dzierba AL, Jelic S. Chronic obstructive pulmonary disease in the elderly: an update on pharmacological management. Drugs Aging 2009;26:447–56.
23. D’Agostino RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81.
24. Parsons LS. Reducing bias in a propensity score matched-pair sample using greedy matching techniques. Proceedings of the 26th Annual SAS User Group International Conference). 2001 April 22-25; Paper 214-26.
25. Elixhauser A, Steiner C, Harris DR, Coffey RN. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
26. Baldwin LM, Klabunde CN, Green P, et al. In search of the perfect comorbidity measure for use with administrative claims data—does it exist? Med Care 2006;44:745–53.
27. Mapel DW, Robinson SB, Lydick E. A comparison of health-care costs in patients with chronic obstructive pulmonary disease using lightweight portable oxygen systems versus traditional compressed-oxygen systems. Respir Care 2008;53:1169–75.
28. Mapel DW, Schum M, Lydick E, Marton JP. A new method for examining the cost savings of reducing COPD exacerbations. Pharmacoeconomics 2010;28:733–49.
29. Roberts M, Mapel D, Petersen H, et al. Comparative effectiveness of budesonide/formoterol and fluticasone/salmeterol for COPD management. J Med Econ 2011;14:769–76.
30. Blais L, Forget AF, Ramachandran S. Relative effectiveness of budesonide/formoterol and fluticasone propionate/salmeterol in a 1-year, population-based, matched cohort study of patients with chronic obstructive pulmonary disease (COPD): effect on COPD-related exacerbations, emergency department visits and hospitalizations, medication utilization, and treatment adherence. Clin Ther 2010;32:1320–8.