Combination of budesonide/formoterol more effective than. than fluticasone/salmeterol in preventing exacerbations

Transcription

1 Original Article doi: /joim Combination of budesonide/formoterol more effective than fluticasone/salmeterol in preventing exacerbations in chronic obstructive pulmonary disease: the PATHOS study K. Larsson 1, C. Janson 2, K. Lisspers 3, L. Jørgensen 4, G. Stratelis 4, G. Telg 4,B.St allberg 3 & G. Johansson 3 From the 1 Unit of Lung and Allergy Research, National Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Departments of 2 Medical Sciences Respiratory Medicine, 3 Public Health and Caring Science, Family Medicine and Preventive Medicine, Uppsala University, Uppsala, and 4 AstraZeneca Nordic, S odert alje, Sweden Abstract. Larsson K, Janson C, Lisspers K, Jørgensen L, Stratelis G, Telg G, St allberg B, Johansson G (Karolinska Institutet, Stockholm; Uppsala University, Uppsala; AstraZeneca Nordic, S odert alje, Sweden). Combination of budesonide/formoterol more effective than fluticasone/salmeterol in preventing exacerbations in chronic obstructive pulmonary disease: the PATHOS study. J Intern Med 2013; 273: Objectives. Combinations of inhaled corticosteroids (ICSs) and long-acting b 2 -agonists (LABAs) are recommended for patients with moderate and severe chronic obstructive pulmonary disease (COPD). However, it is not known whether different fixed combinations are equally effective. The aim of this study was to investigate exacerbation rates in primary care patients with COPD treated with budesonide/formoterol compared with fluticasone/salmeterol. Methods. Patients with physician-diagnosed COPD and a record of postdiagnosis treatment with a fixed combination of budesonide/formoterol or fluticasone/salmeterol were included. Data from primary care medical records were linked to those from Swedish national hospital, drug and cause of death registers. Pairwise (1 : 1) propensity score matching was carried out at the index date (first prescription) by prescribed fixed ICS/LABA combination. Exacerbations were defined as hospitalizations, emergency visits and collection of oral steroids or antibiotics for COPD. Yearly event rates were compared using Poisson regression. Results. Matching of 9893 patients (7155 budesonide/formoterol and 2738 fluticasone/salmeterol) yielded two cohorts of 2734 patients, comprising patient-years. The exacerbation rates were 0.80 and 1.09 per patient-year in the budesonide/ formoterol and fluticasone/salmeterol groups, respectively (difference of 26.6%; P < ); yearly rates for COPD-related hospitalizations were 0.15 and 0.21, respectively (difference of 29.1%; P < ). All other healthcare outcomes were also significantly reduced with budesonide/formoterol versus fluticasone/salmeterol. Conclusions. Long-term treatment with fixed combination budesonide/formoterol was associated with fewer healthcare utilization-defined exacerbations than fluticasone/salmeterol in patients with moderate and severe COPD. Keywords: budesonide/formoterol, exacerbations, fluticasone/salmeterol, inhaled corticosteroids, long-acting b 2 -agonists. Introduction Management of chronic obstructive pulmonary disease (COPD) involves short-term objectives, such as symptom relief, and long-term goals focusing on improving and maintaining health status and preventing exacerbations, comorbidities and mortality [1, 2]. Frequent exacerbations of COPD accelerate lung function decline [3], increase morbidity and mortality [4, 5] and account for a substantial proportion of the healthcare costs [6, 7]. It is well established that exacerbations occur at any stage of COPD, but are gradually more frequent and more severe as the disease progresses [3, 8 10]. Inhaled corticosteroids (ICSs) in fixed combination with long-acting b 2 -agonists (LABAs) reduce exacerbations [11 13] and are therefore recommended for prevention of COPD exacerbations [1]. Two products combining an ICS and an LABA are 584 ª 2013 The Association for the Publication of the Journal of Internal Medicine

2 available in Sweden, budesonide/formoterol (Symbicort â Turbuhaler â ) and fluticasone/salmeterol (Seretide â Diskus â ). Both of these combinations reduce the exacerbation rate in COPD and are more effective in this respect than the respective individual components alone [11, 12, 14, 15]. However, no randomized controlled trials have been conducted to compare the effects of different fixed combinations of ICSs and LABAs. Depending on the chosen outcome variables, controlled COPD trials are often designed using exclusion criteria that may limit their ability to reflect real-world effectiveness, in particular in patients with concomitant conditions such as heart disease, sarcopenia and osteoporosis. Controlled trials also generally include assessments and contact with healthcare professionals beyond usual clinical care, which may predispose patient selection to a more compliant phenotype and alter patient behaviour compared with the real-life clinical situation. In addition, predefined study periods are seldom well suited to accurately assessing the long-term effectiveness of chronic disease treatment due to high and differential withdrawal rates, particularly if placebo-treated control arms are included [16, 17]. Thus, because of limited possibilities for physicians to alter therapy during controlled trials focusing on COPD exacerbations, withdrawal rates frequently exceed 30% within the first year of the trial [11 15]. Comparative effectiveness data from observational databases of propensity score-matched cohorts provide an alternative means to balance study groups to minimize bias when randomization is not possible and can provide complementary data to overcome some of the limitations of controlled trials. However, information from observational studies of fixed ICS/LABA combination regimens delivered via dry powder inhalers in COPD is limited. In a recent observational, matched-cohort study, in which exacerbations requiring use of antibiotics were not reported, a lower rate of exacerbations requiring hospitalization and emergency treatment were observed in COPD patients treated with budesonide/formoterol than in those treated with fluticasone/salmeterol [18]. The aim of this propensity score-matched, population-based, retrospective, observational cohort study in 5468 patients with up to 11 years of follow-up in general practice in Sweden was to compare the effects of two different fixed ICS/LABA combinations on COPD exacerbations. Healthcare utilization associated with fixed ICS/LABA combinations of either budesonide/formoterol Turbuhaler â or fluticasone/salmeterol Diskus â in COPD was investigated. Methods Study design, protocol and data sources A population-based, retrospective, observational registry study was conducted by linking data from primary care medical records and from mandatory Swedish national registers. The linkage of data obtained from the national registers and primary care centres was performed by the Swedish National Board of Health and Welfare. The linked database was managed by the Department of Public Health and Caring Sciences, Uppsala University, Sweden. The Swedish primary care system is organized into primary healthcare centres (compulsory first point of contact). Each centre is responsible for a defined population (a patient list or a defined catchment area). For specialist consultation, patients are referred to secondary care. No formal stratification of primary healthcare centres was performed, but effort was made to ensure that the sample was representative of the Swedish population reflecting the COPD healthcare provision through a mixture of rural and urban areas, public and private providers and centre size. Seventy-six centres were included covering approximately 8% of Swedish primary care centres with a catchment area comprising 8% of the Swedish population. Data were extracted from medical records [e.g. date of birth, gender, diagnoses according to International Classification of Diseases, 10th revision, Clinical Modification (ICD-10-CM) codes, number of primary healthcare centre contacts, lung function assessments and collections of drug prescriptions] using an established software system (Pygargus Customized extraction Program, CXP; Pygargus AB, Stockholm, Sweden) [19, 20]. The personal identification number used to identify patients with respect to all contacts with healthcare professionals was replaced by a study identification number prior to further data processing. Data were also extracted retrospectively from mandatory Swedish national registries. Data regarding morbidity and mortality were collected from the National Patient Register, inpatient (admission and discharge dates, and main and secondary diagnoses) and outpatient hospital care (number of ª 2013 The Association for the Publication of the Journal of Internal Medicine 585

3 contacts and diagnoses according to ICD-10-CM codes), and the Cause of Death register [date and cause(s) of death]. Data regarding drug prescriptions from hospital and primary care were collected from the Swedish Prescribed Drug Register. General population data, such as number and gender distribution per each year of birth, were generated by Statistics Sweden (SCB, Stockholm, Sweden). The study protocol was reviewed and approved by the regional ethics committee in Uppsala, Sweden (Reference number 2010/040) and registered at ClinicalTrials.gov (clinical trial identifier NCT ). Study population The baseline population (primary data set) included women and men with physician-diagnosed COPD (ICD-10-CM code J44). Patients without a diagnosis of COPD, from medical records or national registers, and patients with a COPD diagnosis only present in the Cause of Death register were excluded. There were no predefined exclusion criteria. Matched study population Patients eligible for matching were receiving treatment with a fixed ICS/LABA combination (budesonide/formoterol Turbuhaler â or fluticasone/ salmeterol Diskus â ). The index date was defined as the date of the first fixed ICS/LABA combination prescription after COPD diagnosis. Patients were followed from 1 January 1999 to the end of the study on 31 December 2009, or end of treatment with the fixed ICS/LABA combination, emigration or death. Outcome variables Definition of exacerbations COPD exacerbations were defined as COPD-related hospitalizations (ICD-10-CM code J44 as primary diagnosis or J44.0/J44.1 as secondary diagnosis), emergency visits (ICD-10-CM code J44.0/J44.1 in outpatient hospital care) and collection of oral steroids [Anatomical Therapeutic (ATC) Classification System code H02AB] or antibiotics (ATC codes J01AA and J01CA). Exacerbations that occurred within 14 days were defined as a single event. Prescription events Dispensed prescriptions of inhaled drugs used in obstructive pulmonary diseases were defined as ICSs (ATC code R03BA), LABAs (ATC codes R03AC12 and R03AC13), short-acting b 2 -agonists (SABAs; ATC code R03AC) and fixed ICS/LABA combinations (ATC codes R03AK06 and R03AK07). All treatment periods on the same drug, from start to end (estimated based on prescribed pack size and prescribed number of daily inhalations), were summarized for treatment length and events. All events were assigned to the treatment that the patient received at the time of the event. Events during interruptions in treatment (no medication collection by the estimated end of the treatment period) were not included in analyses. If treatment was switched to the other fixed ICS/LABA combination, the date of prescription of the new drug was used as the start date. Medication utilization Drug usage after the index date was calculated per year, for all eligible patients in a given year. The end date of drug usage was calculated from the prescribed dose. Statistical analysis The yearly rate of healthcare utilization events (exacerbations or prescriptions) was compared using a Poisson regression analysis with events as the dependent variable and time on specific fixed-combination treatment as an offset variable. Propensity score matching was used to compensate for concerns related to nonrandom assignment of patients to treatments [21] to reduce potential confounding caused by unbalanced covariates [22, 23]. A five to one digit matching was performed without replacement and a nonhierarchal order of variables. Statistical analyses were performed using the matched data. Patients treated with either budesonide/formoterol or fluticasone/salmeterol were matched, and the propensity score calculated based on the following criteria 2 years preceding and/or at the index date: age; gender; available lung function measurements; number of outpatient visits for acute COPD; number of prescriptions for antibiotics, oral steroids, tiotropium, ipratropium, ICSs, SABAs, LABAs, angiotensin receptor blockers, b-blockers, statins, calcium antagonists and thiazides; diagnosis of diabetes, asthma, cancer, rheumatoid arthritis, heart failure, hypertension and stroke; and number of previous hospitalizations. The sum of outpatient visits, yearly rate of oral steroid and antibiotic prescriptions, and yearly rate of hospitalizations for acute COPD before the index date 586 ª 2013 The Association for the Publication of the Journal of Internal Medicine

4 was also included in the propensity score matching. A number of different sensitivity analyses were performed. The effect of propensity score matching was tested by adding the matching variables, both cumulative and one variable at a time, and by calculating the crude rates (without matching). A sensitivity analysis of the Poisson regression was conducted using only events that occurred up to 1 month after the first switch of fixed combination, and by adding the calendar year as a covariate. Data management and statistical analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, NC, USA). Results A total of patients who met the inclusion criteria of a recorded COPD diagnosis were identified within the study period. Of these patients, 9893 (53% female, mean age 67 years) had a record of fixed ICS/LABA combination therapy following COPD diagnosis and were eligible for matching: 7155 patients treated with budesonide/ formoterol and 2738 patients with fluticasone/ salmeterol at the index date, that is, the start of ICS/LABA treatment (Table 1). The scores for estimated probability of treatment between the budesonide/formoterol and fluticasone/salmeterol populations showed overlapping distributions, indicating well-balanced cohorts prior to propensity score matching (Fig. 1); however, there were some between-group differences (Table 1). Pairwise matching (1 : 1) resulted in two cohorts of 2734 patients each, with similar characteristics. All but four patients with fluticasone/salmeterol prescriptions at the index date could be matched into pairs with equivalent budesonide/formoterol patients. After matching, no important differences were observed with respect to demographic characteristics, baseline measures, comorbidities or COPDrelated variables (Table 1). The mean collected budesonide dose during the study period was lg day 1 for matched patients who were prescribed budesonide/formoterol and the mean fluticasone dose was lg day 1 for patients who were prescribed fluticasone/salmeterol. The study comprised patient-years. The follow-up time was years (mean standard deviation): years for budesonide/formoterol and years for fluticasone/salmeterol. Following matching, the postindex exacerbation rates were 0.80 and 1.09 per patient-year in the budesonide/formoterol and fluticasone/salmeterol treatment groups, respectively, corresponding to a 26.6% lower exacerbation rate [rate ratio 0.74, 95% confidence interval (CI) ; P < ] in the budesonide/formoterol group (Table 2). This corresponded to a number needed to treat with budesonide/formoterol versus fluticasone/salmeterol to prevent one exacerbation per patient-year equal to 3.4. Patients with less than one exacerbation per year more often used budesonide/formoterol than fluticasone/salmeterol, whereas those with one or more exacerbation per year were more likely to use fluticasone/salmeterol (Fig. 2). Further, there were 7% more fluticasone/salmeterol users among patients with at least two exacerbations per year (Fig. 2). There was some evidence from subpopulation analyses that the magnitude of the treatment difference increased with study duration, with the difference in exacerbation rates in favour of budesonide/formoterol in the first 1 3 years being less than during the full study period (Table 3). The significant reduction in favour of budesonide/formoterol also remained irrespective of age, gender, whether or not patients had an asthma diagnosis in the pre-index period, prior to exacerbation medication or prescription for a longacting bronchodilator. The difference in overall exacerbation rate between the two groups ranged from 18% to 31% (Table 3). Sensitivity analyses demonstrated a difference in the overall exacerbation rate between the two groups ranging from 25% to 31%. The crude rate difference (without propensity score matching) was 34% (P < ). In budesonide/formoterol-treated patients, the yearly rate of COPD-related hospitalizations was 0.15 compared with 0.21 in patients treated with fluticasone/salmeterol (P < ), corresponding to a difference of 29.1% between the two groups (rate ratio 0.71, 95% CI ; P < ; Table 2). Thus, the number needed to treat to prevent one COPD-related hospitalization per patient-year was 16.0 with budesonide/formoterol versus fluticasone/salmeterol. In total, 27.0% fewer days were spent in hospital due to exacerbations of COPD in the group treated with budesonide/formoterol (0.63 days year 1 ) compared with the fluticasone/salmeterol-treated group (0.95 days year 1 ; rate ratio 0.66, 95% CI ; P < ; Table 2). In addition, there were 21.0% fewer emergency visits in the ª 2013 The Association for the Publication of the Journal of Internal Medicine 587

5 Table 1 Baseline characteristics in the 2 years before a first ICS/LABA prescription after a physician-based COPD diagnosis (index date), by treatment. Unmatched and pairwise (1 : 1) propensity score-matched populations of budesonide/formoterol- and fluticasone/salmeterol-treated patients are shown Unmatched Matched Fluticasone/ salmeterol Budesonide/ formoterol Fluticasone/ salmeterol Budesonide/ formoterol Variable (n = 2738) (n = 7155) P-value (n = 2734) (n = 2734) P-value Age at diagnosis, mean (SD) years a 67.6 (10.4) 66.7 (10.8) (10.4) 67.6 (10.9) Female gender, n (%) b 1459 (53) 3815 (53) (53) 1446 (53) Exacerbation, n (%) b 2105 (77) 5584 (78) (77) 2106 (77) Hospitalization for COPD,% 24; (0.6) 21; (0.6) < ; (0.6) 23; (0.6) patients; mean (SD) yearly rate a Emergency department visits% 6.9; (0.08) 7.0; (0.2) ; (0.08) 7.7; (0.3) patients; mean (SD) yearly rate a Oral steroids% patients; 54; 1.99 (3.19) 55; 2.01 (3.28) ; 1.97 (3.10) 56; 1.98 (3.08) mean (SD); yearly rate a Antibiotics% patients; 61; 1.57 (1.94) 64; 1.59 (1.89) ; 1.57 (1.94) 61; 1.55 (1.71) mean (SD); yearly rate a SABA% patients; 60; 2.81 (4.00) 56; 3.24 (4.66) ; 2.81 (4.01) 59; 2.85 (4.16) mean (SD) yearly rate a LABA% patients; 27; 1.90 (4.39) 27; 2.61 (3.91) ; 1.90 (4.39) 28; 2.24 (3.53) mean (SD) yearly rate a LAMA% patients; 64; 2.73 (3.53) 63; 3.36 (4.35) < ; 2.73 (3.53) 62; 2.93 (3.91) mean (SD) yearly rate a Inhaled steroids% patients; 40; 2.33 (2.69) 43; 2.87 (3.24) < ; 2.32 (2.69) 43; 2.25 (2.49) mean (SD) yearly rate a Days from COPD 756 (965) 800 (993) (965) 786 (970) diagnosis to index date, mean (SD) a Osteoporosis, n (%) b 184 (7) 373 (5) (7) 202 (7) Diabetes, n (%) b 288 (11) 967 (14) < (11) 283 (10) Cancer, n (%) b 323 (12) 1023 (14) (12) 322 (12) Asthma, n (%) b 1053 (39) 2300 (32) < (39) 1069 (39) Depression, n (%) b 315 (12) 977 (14) (12) 323 (12) Anxiety, n (%) b 200 (7) 557 (8) (7) 205 (8) Smokers c, n (%) b 341 (48) 1337 (53) (48) 397 (49) Data are presented as number of patients and/or percentage and/or mean and SD. All prescriptions for COPD medications in the Swedish prescription database generally corresponds to 3 months drug supply (a value of 1.0 could represent more than one inhaler per month for up to 3 months). Group means are calculated including patients not in receipt of the listed medication before the index date. COPD, chronic obstructive pulmonary disease; ICS, inhaled corticosteroids; LABA, long-acting b2 agonist; LAMA, long-acting muscarinic antagonist; SABA, short-acting b2 agonist; SD, standard deviation. a T-test for unmatched and paired t-test for matched patients. b Chi-squared and McNemar tests. c Limited data available. 588 ª 2013 The Association for the Publication of the Journal of Internal Medicine

6 30 25 Patients (%) Budesonide/formoterol Patients (%) Fluticasone/salmeterol / Estimated probability Fig. 1 Estimated probability of treatment of the two unmatched populations of patients with chronic obstructive pulmonary disease treated with fluticasone/salmeterol (n = 2738) or budesonide/formoterol (n = 7155). A high probability indicates milder disease. Table 2 Yearly occurrence of events among pairwise (1 : 1) propensity score-matched populations of COPD patients treated with budesonide/formoterol versus fluticasone/salmeterol Fluticasone/ Budesonide/ Variable salmeterol (n = 2734) formoterol (n = 2734) Treatment contrast a Events, per patient-year Mean (95% CI) Mean (95% CI) Rate ratio (95% CI) P-value All exacerbations 1.09 ( ) 0.80 ( ) 0.74 ( ) < COPD hospitalizations 0.21 ( ) 0.15 ( ) 0.71 ( ) < COPD-related hospital stay, days 0.95 ( ) 0.63 ( ) 0.66 ( ) < Emergency visits ( ) ( ) 0.79 ( ) Oral steroid use 0.85 ( ) 0.63 ( ) 0.74 ( ) < Antibiotic use 0.54 ( ) 0.38 ( ) 0.70 ( ) < The adjusted yearly rate of healthcare utilization events was compared using Poisson regression analysis. See Methods for details of propensity score matching. CI, confidence intervals; COPD, chronic obstructive pulmonary disease. a With reference to fluticasone/salmeterol. budesonide/formoterol treatment group (0.027 events per patient-year) compared with those treated with fluticasone/salmeterol (0.034 events/patient-year; rate ratio 0.79, 95% CI ; P = ; Table 2). Budesonide/formoterol-treated patients had 26.0% fewer courses of oral steroids (0.63 vs events per year; rate ratio 0.74, 95% CI ; P < ) and 29.0% fewer antibiotic courses (0.38 vs events per year; rate ratio ª 2013 The Association for the Publication of the Journal of Internal Medicine 589

7 Patients (%) Budesonide/formoterol Fluticasone/salmeterol 0 >0 <1 1 <2 2 Overall exacerbation rate per patient per year Fig. 2 Number of exacerbations per patient per year for patients treated with budesonide/formoterol and fluticasone/salmeterol during the study period , 95% CI ; P < ) than patients treated with fluticasone/salmeterol (Table 2). Among patients who were treated with fluticasone/ salmeterol at the index date, 13.8% switched to budesonide/formoterol during the study period; among those who were treated with budesonide/ formoterol at the index date, 2.8% switched to fluticasone/salmeterol during the study. Patients who switched treatment had higher yearly exacerbation rates than those who did not switch (data not shown). The number of patients who required tiotropium in addition to the ICS/LABA combination was 16.0% lower for the budesonide/formoterol group (rate ratio 0.84, 95% CI ; P < vs. fluticasone/salmeterol). There were also fewer ipratropium and b-agonist prescription events in the budesonide/formoterol treatment group than in those treated with fluticasone/salmeterol Table 3 COPD exacerbation rates in subpopulations of pairwise (1 : 1) propensity score-matched patients treated with budesonide/formoterol versus fluticasone/salmeterol Yearly rate fluticasone/ Yearly rate budesonide/ Treatment contrast a Subpopulation Patients, n salmeterol formoterol Rate ratio (95% CI) Patients with asthma diagnosis ( ) Patients without asthma diagnosis ( ) Age 60 years ( ) Age >60 years ( ) 1 year after index date ( ) 2 years after index date ( ) 3 years after index date ( ) Females ( ) Males ( ) Patients with oral steroid/antibiotic use ( ) before index date Patients without oral steroid/antibiotic use before index date Patients with long-acting bronchodilator use before index date Patients without long-acting bronchodilator use before index date ( ) ( ) ( ) Patients with index date after 1 June 2003 b ( ) Long-acting bronchodilators include formoterol, salmeterol and tiotropium bromide. COPD, chronic obstructive pulmonary disease. All P-values < a With reference to fluticasone/salmeterol. b Index date after regulatory approval of both COPD indications. 590 ª 2013 The Association for the Publication of the Journal of Internal Medicine

8 Table 4 Yearly rates of collection of inhalation drugs in pairwise (1 : 1) propensity score-matched populations of COPD patients treated with budesonide/formoterol versus fluticasone/salmeterol Fluticasone/ Budesonide/ Drug type salmeterol (n = 2734) formoterol (n = 2734) Treatment contrast a Events, per patient-year Mean (95% CI) Mean (95% CI) Rate ratio (95% CI) P-value Tiotropium bromide 1.06 ( ) 0.89 ( ) 0.84 ( ) < Short-acting beta-agonists 1.22 ( ) 0.95 ( ) 0.78 ( ) < Ipratropium bromide 0.46 ( ) 0.39 ( ) 0.84 ( ) Formoterol ( ) 0.12 ( ) 1.52 ( ) < Salmeterol ( ) ( ) 0.37 ( ) < Data are presented as mean (95% CI) for all bronchodilator prescriptions, which in the Swedish prescription database generally corresponds to 3 months drug supply (i.e. a value of 1.0 could represent more than one inhaler per month for up to 3 months). Group means are calculated for up to 11 years of follow-up, including all years with no collection of the listed medication. The adjusted yearly rate of healthcare utilization events was compared using Poisson regression analysis. See Methods for details of propensity score matching. CI, confidence intervals; COPD, chronic obstructive pulmonary disease. a With reference to fluticasone/salmeterol. (Table 3). There was a higher rate of formoterol prescription in the budesonide/formoterol group (P < ) and of salmeterol prescription in the fluticasone/salmeterol group (P < ; Table 4). Discussion In the present observational, matched-cohort, register-linkage study of COPD patients with a followup period of up to 11 years, comprising more than patient-years, we showed that patients treated with budesonide/formoterol Turbuhaler â were significantly less likely to experience a COPDrelated exacerbation than those treated with fluticasone/salmeterol Diskus â. Furthermore, the numbers needed to treat with budesonide/formoterol versus fluticasone/salmeterol to avoid one COPD-related exacerbation and one hospital admission were less than 4 and 16 patients per year, respectively. In a previous population-based, retrospective, matched-cohort, 1-year study conducted in Canada by Blais and co-workers, there were 15% fewer COPD exacerbations requiring use of oral steroids with budesonide/formoterol compared with fluticasone/salmeterol; however, this difference failed to reach statistical significance [18]. In our larger study, we found that 18% fewer exacerbations occurred in the first year with budesonide/formoterol compared with fluticasone/salmeterol; this difference in exacerbation rates between the two ICS/LABA groups in the first year was statistically significant, and subsequently, the difference increased with study duration. In line with the current findings, healthcare utilization in the Canadian study with regard to acute visits to the emergency department, hospitalizations and additional tiotropium use due to COPD were all significantly lower in patients treated with budesonide/ formoterol than those treated with fluticasone/ salmeterol [18]. The Canadian study was based on 2262 treatment-years, whereas the current analysis was based on treatment-years. The long observation time, the high number of patient-years and the inclusion of treatment with antibiotics in the definition of exacerbations in the present study constitute a strong basis for our finding of statistically significant differences for all outcome variables in favour of the combination of budesonide/ formoterol compared with fluticasone/salmeterol. We demonstrated 29% fewer exacerbations, defined by the use of antibiotics, in the budesonide/formoterol group compared with the fluticasone/salmeterol group; however, antibiotic use was not assessed in the smaller Canadian trial. The results of both the present observational, matched-cohort study and that of Blais and coworkers [18] suggest differences in exacerbation outcomes, as well as a need for adjuvant tiotropium prescriptions, in the matched cohorts despite the fact that the most common prescription for fixed-combination therapy was not the same in the two countries (Sweden and Canada, respectively). In previous studies, for up to 1 year, both budesonide/formoterol and fluticasone/salmeterol decreased the rate of COPD exacerbations when ª 2013 The Association for the Publication of the Journal of Internal Medicine 591

9 compared with placebo, and the magnitude of reduction was similar for the two ICS/LABA combination products [11, 12, 14, 15]. However, to our knowledge, no direct, long-term comparisons of the efficacy of these regimens in COPD have been made in a randomized, double-blind, double-dummy trial setting. Budesonide/formoterol has a faster onset of action (bronchodilatation) than fluticasone/salmeterol [24], implying a more rapid onset of relief of acute symptoms [25, 26]. Whether this influences patient adherence to either regimen in the long term is unclear. In previous real-world COPD studies with a duration of 6 12 months [18, 27], adherence rates to the two fixed ICS/LABA combinations were similar, suggesting that the observed difference in effectiveness between budesonide/formoterol and fluticasone/salmeterol reported here and by Blais et al. is most likely to be related to the pharmacological properties of the combinations rather than differences in patient adherence. The 16 22% lower use of concomitant COPD medications in the budesonide/formoterol group and the lower rate of switching from budesonide/formoterol to fluticasone/salmeterol than vice versa indicate that patients and treating physicians generally perceive COPD control to be better during treatment with budesonide/formoterol than with fluticasone/salmeterol. Differences between ICSs with regard to pharmacokinetic and pharmacodynamic properties, such as oral bioavailability and clearance, volume of distribution and speed of airway uptake, may have implications for real-life clinical effectiveness [28 31]. In contrast to the sterile airways of normal lungs, bacterial colonization of the lower airways with a spectrum of pathogens is observed in a significant proportion of patients with stable COPD [32, 33]. This colonization may be an important pro-inflammatory airway stimulus influencing exacerbation frequency. Fluticasone is a more potent immunosuppressant as well as a more lipophilic glucocorticosteroid than is budesonide, which means that it is more likely to be retained in the mucosa and epithelial lining fluid for longer periods than budesonide; it has been proposed that these factors may contribute to prolonged suppression of local immunity and thereby facilitate the occurrence of a difference in exacerbations [28 31, 34]. This hypothesis may be supported by the finding of a 29% lower prevalence of exacerbations requiring antibiotic treatment in the budesonide/formoterol-treated cohort than in patients treated with fluticasone/salmeterol. A subgroup of patients in this real-life study had a concomitant diagnosis of asthma and COPD after the index date. Sensitivity analyses showed similar results in favour of budesonide/formoterol in this subgroup, confirming that our findings were not explained by inclusion of patients with this comorbidity. Crude population comparisons revealed results similar to those in the matched samples, indicating that the overall results were not affected by loss of information created by the budesonide/ formoterol-treated patients who were not included in the matching. As with all retrospective, observational registry studies, the present study may have limitations. Data retrieval is limited to the variables recorded in the databases. Although the included patients have been matched pairwise with respect to a number of variables, there may still be potential unknown confounding factors. The validity and completeness of the data regarding COPD diagnoses (e.g. lung function measurements and disease history) and the accuracy and severity of the physician diagnoses of COPD could not be fully verified by spirometry in all cases. Medication use is based on prescription claims, which do not fully reflect how patients actually use medications. Nevertheless, as the study is based on a large sample (almost 5500 individuals), it is not likely that potential confounding factors would differ between the cohorts. Nonpharmaceutical COPD therapy data (e.g. smoking cessation and exercise) were not complete and could not be included in the analyses. As the cohort mainly comprised patients of Scandinavian origin, the findings may not be generalizable to non-caucasians. Notwithstanding its limitations, this study also has several important strengths, not least the primary care setting used to initially identify the COPD patients and the lack of restrictions with regard to, for example, age, employment status, concomitant medication use, comorbidities or healthcare insurance. The nonbiased data extraction from electronic primary healthcare data linked with mandatory national healthcare registers with high coverage and quality, together with the opportunity to follow individual patients through their treatment journey by use of the personal identification number, provides robust and unique data. The generalizability of our findings to the treatment of COPD in general practice is therefore greater than in controlled trials. All patients were matched with respect to a large number of recognized COPD 592 ª 2013 The Association for the Publication of the Journal of Internal Medicine

10 parameters and other markers. Sensitivity analyses demonstrated the strength of the outcomes across COPD severity with or without concomitant asthma diagnoses. Finally, the duration of followup and number of patient-years were substantial for both medications studied, without the potential for increased and differential drop-out rates in one vs. the other therapy, which can often confound the results in longer-term studies. In summary, we found that long-term budesonide/ formoterol treatment was associated with fewer moderate and severe exacerbations than long-term treatment with fluticasone/salmeterol. The findings were robust irrespective of the exacerbation definition used and were not affected by several sensitivity analyses. Considering that, to our knowledge, no long-term controlled trials comparing the two ICS/LABA combinations in COPD have been conducted, and the fact that the numbers needed to treat to avoid one COPD exacerbation per year with one ICS/LABA regimen versus another were clinically meaningful, these real-world findings warrant further investigation. Conflict of interest statement AstraZeneca funded this study and was a member of the study steering committee that carried overall responsibility for the concept and design. Kjell Larsson has, during the last 5 years, on one or more occasion served on an advisory board and/or served as a speaker and/or participated in education arranged by AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline (GSK), Meda, Merck Sharp & Dohme (MSD), Nycomed, Novartis, Takeda and Pfizer. Kjell Larsson has also received unrestricted research grants from AstraZeneca, Boehringer Ingelheim, and GSK. Christer Janson has received honoraria for educational activities from AstraZeneca, GSK and MSD. Karin Lisspers has received speaking fees from AstraZeneca, Boehringer Ingelheim and MSD. Bj orn St allberg has received honoraria for educational activities from AstraZeneca, GSK and MSD. Christer Janson, Karin Lisspers, Bj orn St allberg, and Gunnar Johansson have served in, and/or received honoraria for, advisory board meetings arranged by AstraZeneca. Georgios Stratelis, Gunilla Telg and Leif J orgensen are fulltime employees of AstraZeneca Nordic. Acknowledgements This study was sponsored by AstraZeneca. References 1 Global Initiative for Chronic Obstructive Lung Disease. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease. wwwgoldcopdcom/guidelinesresources, Postma D, Anzueto A, Calverley P et al. A new perspective on optimal care for patients with COPD. Prim Care Respir J 2011; 20: Donaldson GC, Seemungal TA, Bhowmik A, Wedzicha JA. Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease. Thorax 2002; 57: Fabbri LM, Luppi F, Beghe B, Rabe KF. Update in chronic obstructive pulmonary disease Am J Respir Crit Care Med 2006; 173: Soler-Cataluna JJ, Martinez-Garcia MA. Roman Sanchez P, Salcedo E, Navarro M, Ochando R. Severe acute exacerbations and mortality in patients with chronic obstructive pulmonary disease. Thorax 2005; 60: Andersson F, Borg S, Jansson SA et al. The costs of exacerbations in chronic obstructive pulmonary disease (COPD). Respir Med 2002; 96: Wouters EF. Economic analysis of the Confronting COPD survey: an overview of results. Respir Med 2003; 97 (Suppl C): S Hurst JR, Vestbo J, Anzueto A et al. Susceptibility to exacerbation in chronic obstructive pulmonary disease. N Engl J Med 2010; 363: Jenkins CR, Celli B, Anderson JA et al. Seasonality and determinants of moderate and severe COPD exacerbations in the TORCH study. Eur Respir J 2012; 39: Jones PW, Willits LR, Burge PS, Calverley PM. Disease severity and the effect of fluticasone propionate on chronic obstructive pulmonary disease exacerbations. Eur Respir J 2003; 21: Calverley PM, Boonsawat W, Cseke Z, Zhong N, Peterson S, Olsson H. Maintenance therapy with budesonide and formoterol in chronic obstructive pulmonary disease. Eur Respir J 2003; 22: Szafranski W, Cukier A, Ramirez A et al. Efficacy and safety of budesonide/formoterol in the management of chronic obstructive pulmonary disease. Eur Respir J 2003; 21: Wedzicha JA, Calverley PM, Seemungal TA, Hagan G, Ansari Z, Stockley RA. The prevention of chronic obstructive pulmonary disease exacerbations by salmeterol/fluticasone propionate or tiotropium bromide. Am J Respir Crit Care Med 2008; 177: Calverley P, Pauwels R, Vestbo J et al. Combined salmeterol and fluticasone in the treatment of chronic obstructive pulmonary disease: a randomised controlled trial. Lancet 2003; 361: Calverley PM, Anderson JA, Celli B et al. Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 2007; 356: Calverley PM, Spencer S, Willits L, Burge PS, Jones PW. Withdrawal from treatment as an outcome in the ISOLDE study of COPD. Chest 2003; 124: Vestbo J, Anderson JA, Calverley PM et al. Bias due to withdrawal in long-term randomised trials in COPD: evidence from the TORCH study. Clin Respir J 2011; 5: ª 2013 The Association for the Publication of the Journal of Internal Medicine 593

11 18 Blais L, Forget A, 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: Kjeldsen SE, Stalhammar J, Hasvold P, Bodegard J, Olsson U, Russell D. Effects of losartan vs candesartan in reducing cardiovascular events in the primary treatment of hypertension. J Hum Hypertens 2010; 24: Martinell M, Stalhammar J, Hallqvist J. Automated data extraction a feasible way to construct patient registers of primary care utilization. Ups J Med Sci 2012; 117: Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70: Perkins SM, Tu W, Underhill MG, Zhou XH, Murray MD. The use of propensity scores in pharmacoepidemiologic research. Pharmacoepidemiol Drug Saf 2000; 9: Russell D, Stalhammar J, Bodegard J, Hasvold P, Thuresson M, Kjeldsen SE. Cardiovascular events in subgroups of patients during primary treatment of hypertension with candesartan or losartan. J Clin Hypertens (Greenwich) 2011; 13: Palmqvist M, Arvidsson P, Beckman O, Peterson S, Lotvall J. Onset of bronchodilation of budesonide/formoterol vs. salmeterol/fluticasone in single inhalers. Pulm Pharmacol Ther 2001; 14: Lindberg A, Szalai Z, Pullerits T, Radeczky E. Fast onset of effect of budesonide/formoterol versus salmeterol/fluticasone and salbutamol in patients with chronic obstructive pulmonary disease and reversible airway obstruction. Respirology 2007; 12: Partridge MR, Schuermann W, Beckman O, Persson T, Polanowski T. Effect on lung function and morning activities of budesonide/formoterol versus salmeterol/fluticasone in patients with COPD. Ther Adv Respir Dis 2009; 3: Roberts M, Mapel D, Petersen H, Blanchette C, Ramachandran S. Comparative effectiveness of budesonide/formoterol and fluticasone/salmeterol for COPD management. J Med Econ 2011; 14: Brattsand R, Miller-Larsson A. The role of intracellular esterification in budesonide once-daily dosing and airway selectivity. Clin Ther 2003; 25 (Suppl C): C Dalby C, Polanowski T, Larsson T, Borgstrom L, Edsbacker S, Harrison TW. The bioavailability and airway clearance of the steroid component of budesonide/formoterol and salmeterol/ fluticasone after inhaled administration in patients with COPD and healthy subjects: a randomized controlled trial. Respir Res 2009; 10: Thorsson L, Edsbacker S, Kallen A, Lofdahl CG. Pharmacokinetics and systemic activity of fluticasone via Diskus and pmdi, and of budesonide via Turbuhaler. Br J Clin Pharmacol 2001; 52: Volovitz B. Inhaled budesonide in the management of acute worsenings and exacerbations of asthma: a review of the evidence. Respir Med 2007; 101: Cabello H, Torres A, Celis R et al. Bacterial colonization of distal airways in healthy subjects and chronic lung disease: a bronchoscopic study. Eur Respir J 1997; 10: Monso E, Rosell A, Bonet G et al. Risk factors for lower airway bacterial colonization in chronic bronchitis. Eur Respir J 1999; 13: Ek A, Larsson K, Siljerud S, Palmberg L. Fluticasone and budesonide inhibit cytokine release in human lung epithelial cells and alveolar macrophages. Allergy 1999; 54: Correspondence: Prof Kjell Larsson, Unit of Lung and Allergy Research, National Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden. (fax: ; kjell.larsson@ki.se). 594 ª 2013 The Association for the Publication of the Journal of Internal Medicine