Peer-Reviewed

Roflumilast: Who Is Using It and How It Affects Health Care Resource Utilization and Costs


Rahul Jain
HealthCore Inc., Wilmington, DE
Qian Cai
HealthCore Inc., Wilmington, DE
Shawn X. Sun
Forest Research Institute, an affiliate of Actavis, Jersey City, NJ
Hiangkiat Tan
HealthCore Inc., Wilmington, DE

Abstract

Objective: Compare baseline characteristics, health care resource utilization (HCRU), and associated costs of COPD patients treated with add-on roflumilast with those of other combination medications.

Design: Retrospective cohort study.

Methodology: Patients aged ≥40 years with a diagnosis of chronic obstructive pulmonary disease (COPD) between March 1, 2011, and Nov. 30, 2012, were identified from the HealthCore Integrated Research Database and classified as roflumilast or nonroflumilast combination-therapy cohorts. Baseline characteristics were compared for all patients. HCRU and costs were compared between matched (M) roflumilast and nonroflumilast cohorts, using propensity score as a partial balancing score and then complementing the score with exact matching on specifically important variables. Generalized linear model and Poisson regression were used to estimate the adjusted differences in total costs and hospitalization rates, respectively, between the 2 matched cohorts.

Results: A total of 695 roflumilast and 30,542 nonroflumilast combination therapy users were identified. At baseline, the roflumilast cohort had more complex COPD and a higher number of severe and moderate COPD exacerbations relative to the nonroflumilast cohort. After matching, the roflumilast (M) and nonroflumilast (M) cohorts (n=328 in each) had similar mean age, gender distribution, and follow-up time. The roflumilast (M) cohort had significantly higher pharmacy-related, per-patient, per-month (PPPM) costs (P<.001) and similar total cost (P=.90). After adjusting for confounding variables, no difference was observed between the 2 cohorts in total costs (P=.86) and number of hospitalizations (P=.65).

Conclusion: Findings suggest that patients in the roflumilast cohort, relative to the nonroflumilast cohort, were more severely ill in the real-world setting. Despite higher pharmacy costs, the total cost for the roflumilast cohort was statistically similar to the nonroflumilast cohort. Future studies with longer follow-up are needed to evaluate the long-term economic impact of roflumilast use.

Key words: COPD, roflumilast, patient characteristics, matching, health care resource utilization, and health care costs

Introduction

Chronic obstructive pulmonary disease (COPD) exacerbations are defined as episodes of worsening in respiratory symptoms that necessitate a change in regular medication (Pauwels 2004). Frequent exacerbations may lead to reduced lung function; exacerbations are strongly associated with reduced quality of life and high rate of hospitalization (Donaldson 2006, Solem 2013). Exacerbations are a substantial contributor to the economic burden of COPD; a recent retrospective study estimated the cost of a severe exacerbation to be as high as $18,000 per severe exacerbation (Abudagga 2013).

Chronic obstructive pulmonary disease

Prevention and adequate treatment of COPD exacerbations is a cornerstone in COPD management. The Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommended that long-acting formulations of β2-agonist (LABA) and long-acting muscarinic antagonists (LAMA), alone or with an inhaled corticosteroid (ICS), be used for patients with high risk of exacerbation. Additionally, combined use of LABA and LAMA may be considered if symptoms do not improve, and ICS, phosphodiesterase-4 (PDE4) inhibitors, or theophylline may also be added for severely ill patients (GOLD 2014).

Roflumilast was the first PDE4 inhibitor drug approved by the US Food and Drug Administration (FDA) as a treatment to reduce the risk of COPD exacerbations. Approved in February 2011 and sold under the brand name Daliresp, roflumilast has been incorporated into the GOLD guidelines for severe COPD associated with chronic bronchitis and a history of exacerbations (GOLD 2014, Daliresp 2011). Evidence from large, international, multicenter, randomized placebo-controlled trials of 6–12 months’ duration shows that roflumilast decreased COPD exacerbations in patients with severe chronic bronchitis (Calverley 2009, Fabbri 2009).

But COPD is a highly heterogeneous disease. Posthoc analysis of pooled clinical trial data has shown that roflumilast is effective in reducing exacerbations for specific subpopulations (Rennard 2008, Rennard 2011). These results, in conjunction with the GOLD recommendations, emphasize the need to understand the characteristics of patients who are treated with roflumilast in the real world. In addition, in spite of the efficacy shown in randomized, placebo-controlled trials, little is known about the impact of roflumilast use on health care resource utilization and costs relative to other COPD treatments in a real-world setting.

The objectives of this study were to describe characteristics of COPD patients using roflumilast and compare their characteristics, health care utilization, and associated costs to those of patients on other combination medications among COPD patients with a history of exacerbations.

Methods

Study Design and Data Source

For this retrospective observational study, medical, pharmacy, and eligibility claims data were extracted from the HealthCore Integrated Research Database (HIRD), which consists of claims from 14 geographically dispersed US health plans representing 30 million lives at the time this study was conducted. HIRD is one of the largest datasets of a commercially insured US population. Overall, it is comparable with US Census data (American Community Survey) by age and gender, with the HIRD population being slightly younger because all members are commercially insured.

All personally identifiable data used in this observational study were de-identified and accessed with protocols compliant with the Health Insurance Portability and Accountability Act (HIPAA). Patient confidentiality was preserved and the anonymity of all patient data was safeguarded throughout the study. This study was exempt from institutional review board (IRB) approval, as it was compliant with HIPAA and with federal guidance on Public Welfare and the Protection of Human Subjects. Specifically per the Code of Federal Regulations (45 CFR 46 §46.101(b)(4)), IRB review was not needed for a study of this nature because “subjects cannot be identified, directly or through identifiers linked to the subjects . . . .”

Study Patients

Medical, pharmacy, and eligibility claims data for the designated study period, March 1, 2010, to Nov. 30, 2012, were extracted and analyzed. Study patients were identified from March 1, 2011 (after roflumilast was approved in the US) to Nov. 30, 2012, which was regarded as the patient-identification period.

Patients were identified for the study if they had at least 1 medical claim for COPD (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM] codes 491.xx, 492.xx, 496.xx) during the study period and had at least 1 pharmacy claim for a COPD medication class of interest, including roflumilast, LAMA, LABA, ICS, theophylline, or LABA-ICS, during the patient-identification period. Additional requirements were applied while defining the roflumilast and nonroflumilast cohorts, as elaborated below.

Cohort Definition

Roflumilast is indicated for severe COPD and is recommended as an add-on to other combination therapy. Therefore, to make cohorts comparable, the nonroflumilast cohort was identified as patients with no pharmacy claims for roflumilast and with at least 2 pharmacy claims of different COPD medication classes occurring within a 90-day window during the patient-identification period. If the patient received combination therapy on the same day, the index date was defined as that fill date. If the first medication class (drug A) was filled, followed by a different medication class (drug B and/or C) within 90 days, then the index date was defined as the first fill date of the last occurring medication class (drug B or C).

Patients were assigned to the roflumilast cohort if they had at least 1 pharmacy claim for roflumilast during patient-identification period. If a patient had a pharmacy claim for any other COPD medication on or 90 days before the first fill date of roflumilast, then the first fill date of roflumilast was assigned as the index date. On the other hand, if a patient had a pharmacy claim for any other COPD medication within 90 days after the first fill date of roflumilast, the index date was defined as the fill date of the last occurring medication class.

Inclusion/Exclusion Criteria

Patients included in the study were aged ≥40 years on the index date, had at least 12 months of continuous health insurance enrollment prior to the index date (preindex period) and at least 3 months of continuous enrollment after the index date (follow-up period), were commercially insured or had Medicare Advantage.

Patients were also required to have at least 1 treatment-based exacerbation (Seemungal 2009) during the preindex period. With claims data, exacerbation was defined as: (1) an inpatient or emergency room (ER) admission with a primary diagnosis of COPD acute exacerbation (ICD-9 codes 491.21, 491.22, 493.22) or emphysema (ICD-9 code 492.8x) or a primary diagnosis of respiratory failure combined with a secondary diagnosis of COPD acute exacerbation or emphysema (Lindenauer 2010, Rothberg 2010); (2) a physician office visit with qualifying diagnosis for pulmonary conditions of interest (eg, pneumonia or pulmonary insufficiency) (Mapel 2011); (3) a pharmacy claim for qualifying drug therapy, (ie, systemic steroids or antibiotics commonly used for respiratory infection such as β2-lactamase inhibitors, second- or third-generation cephalosporins, macrolides, and doxycycline) (Mapel 2011).

Patients were excluded from the study if they had a medical claim for cystic fibrosis (ICD-9-CM code 277.0x), respiratory tract cancer (ICD-9-CM codes 160.xx–164.xx, or 231.xx), active tuberculosis (ICD-9-CM codes 010.xx–018.9x), interstitial lung disease diagnosis (ICD-9-CM codes 516.3x), or α1-antitrypsin deficiency (ICD-9-CM codes 273.4x) during the entire study period.

Matching

Patients receiving newly approved medication, such as roflumilast, may be sicker, more likely to have failed existing therapies, or have more generous health plan benefits than patients in the nonroflumilast group (Schneeweiss 2011). Use of the propensity score facilitates the construction of matched sets with similar distributions of covariates, without requiring close or exact matches on all the individual variables (Stuart 2010). However, there are reasons to believe some confounders play an important role in the assignment of patients to the treatment and comparison groups and outcomes of interest (Caliendo 2008, Brookhart 2006).

Roflumilast is recommended for patients with severe exacerbation and, therefore, is likely to be initially prescribed to patients who are severely ill, have already failed on existing therapies, or have generous health plan benefits. As time passes and roflumilast's market share increases, it is expected that it will be prescribed to patients with less severe exacerbations as its market share increases. If patients were matched solely on the basis of a propensity score that included the number of severe exacerbations and time since the FDA's approval of roflumilast, it is possible that a patient with a greater number of exacerbations closer to the FDA approval date would have been matched with a patient with a lesser number of exacerbations further from the FDA approval date—thereby defeating the purpose of the matching exercise. Therefore, based on methods proposed by Lechner (2002), propensity score (generated on the basis of demographic characteristics, preindex Deyo-Charlson comorbidity index [DCI], preindex COPD exacerbation, preindex COPD severity proxy, and preindex all-cause total costs) was used as a partial balancing score and then complemented with exact matching on the preindex number of inpatient or ER exacerbations (categorized as 0, 1, 2, 3, and ≥4 exacerbations), number of COPD maintenance medication classes in combination therapy (classified as 2 vs 3–4 combinations), time of the index date since approval of roflumilast (expressed as sequential quarterly indicator), and 75 categories of the estimated propensity score (Lechner 2002).

Outcome Measures
Patient Characteristics

Patient characteristics were examined during the 12 months prior to the index date. Demographic characteristics included age, gender, type of health insurance plan (PPO, HMO, or other insurance plan type), and geographic region of patients’ residence (as determined by the US Census classification). Clinical characteristics included comorbidity burden, as measured by DCI, and specific comorbid conditions of interest (eg, asthma, acute bronchitis, bronchitis) during the preindex period. DCI consists of 19 diagnoses identified by ICD-9-CM codes with a scoring weight from 1 to 6 identified for each diagnosis. The final summary DCI score consists of a sum of weighted values for existing comorbidities, ranging from 0 to 33, with higher scores indicating a greater comorbidity burden (Deyo 1992).

Severity of COPD during the preindex period was evaluated using measures to approximate COPD complexity (Wu 2006), including number of hospitalization days due to acute exacerbation of chronic bronchitis (AECB), episodes for AECB, spirometry test, pulmonologist visit, and use of COPD medications (anticholinergics, oral corticosteroids, ICS, short-acting β2-agonists [SABA], or LABA) (Wu 2006). The preindex number of COPD exacerbations also was evaluated. Furthermore, patients who had an ER- or inpatient-associated exacerbation were considered as having severe exacerbation, whereas those with a qualifying drug therapy were considered as having moderate exacerbations.

Healthcare Resource Utilization and Costs

Outcomes were measured within the entire follow-up period on a per-patient, per-month (PPPM) basis in any health care setting (including inpatient admissions, ER visits, and outpatient visits). Postindex all-cause health care resource utilization was defined as the use of any health care service associated with any condition identified from medical claims. Postindex all-cause health care costs were calculated using health plan allowed amounts, which are the full fees contractually allowed by health plans (inclusive of the amount paid by the health plan and the patient), and were considered as the direct health care cost associated with any condition identified from medical and pharmacy claims.

Statistical analysis

For unmatched cohorts, the unadjusted differences between 2 cohorts were assessed using χ2 test for categorical variables, Wilcoxon-Mann-Whitney test for discrete/nonnormal continuous variables, and 2 sample t-test for continuous variables. For unadjusted differences between matched cohorts, the McNemar test for categorical variables, the Wilcoxon signed-rank test for discrete/non-normal continuous variables, and paired t test for continuous variables were used. Adjusted differences in number of hospitalizations and total costs between matched cohorts were estimated using a Poisson regression and generalized linear model with log link and gamma variance function, respectively, controlling for statistically and/or clinically important demographic and clinical characteristics, COPD-related resource utilization and costs. Differences were considered statistically significant at P<.05. All data analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC) and STATA version 12 (StataCorp, College Station, TX).

Results

Unmatched Cohorts
Demographic and Clinical Characteristics

Of the 116,247 patients with at least 1 medical claim for COPD and 1 pharmacy claim for a medication class of interest, 31,237 (84.6%) patients met all study criteria and were used in the analyses, of whom 695 patients were classified in roflumilast cohort and 30,542 patients were in nonroflumilast cohort (Figure 1).

Figure 1 Patient identification flow chart

ICS=inhaled corticiosteroids, LABA=long-acting b2-agonists, LAMA=long-acting muscarinic antagonist, (M)=matched.

More than half of the study patients were female, with mean age of about 63 years for both cohorts (Table 1, below). Patients in the roflumilast cohort had higher mean DCI score (2.3 vs 2.1; P<.01) than the nonroflumilast cohort, indicating an overall relatively higher aseline comorbidity burden for patients with roflumilast. In addition, among the comorbid conditions examined in this study, the roflumilast cohort showed greater proportions of acute bronchitis and bronchitis, cardiovascular disease, diabetes, anxiety, and depression diagnoses than the nonroflumilast cohort (Table 1).

Table 1  Patient characteristics of roflumilast vs nonroflumilast cohorts
  Roflumilast (n=695) Nonroflumilast (n=30,542) P value
Age, (mean ± SD) 62.6 ±9.3 63.2 ±11.4 .12
Female (n, %) 382 55.0% 17,503 57.3% .22 
Health plan type (n, %)
HMO 177 25.5% 9165 30.0% .02
PPO 484 69.6% 19,755 64.7%  
Others 34 4.9% 1622 5.3%  
Geographic region (n, %)
Northeast 90 12.9% 6904 22.6% <.001
Midwest 262 37.7% 10,814 35.4%  
South 281 40.4% 9050 29.6%  
West 62 8.9% 3774 12.4%  
Pre-index Deyo-Charlson Comorbidity Index (mean ± SD) 2.3 ±1.9 2.1 ±1.8 .003
Pre-index comorbidities of interest (n, %)
Asthma 298 42.9% 13,768 45.1% .25
Acute bronchitis and bronchitis 269 38.7% 10,654 34.9% .04
Myocardial infarction 69 9.9% 1761 5.8% <.001
Congestive heart failure 133 19.1% 4257 13.9% <.001
Ischemic heart disease 211 30.4% 7233 23.7% <.001
Hypertension 474 68.2% 19,711 64.5% .05
Diabetes mellitus 204 29.4% 7435 24.3% .002
Gastroesophageal reflux disease 192 27.6% 7009 22.9% .004
Anxiety 183 26.3% 4900 16.0% <.001
Depression 146 21.0% 4608 15.1% <.001
Pre-index COPD severity proxy (mean ± SD)
Number of hospitalization days due to AECB or CB 2.3 ±6.4 0.7 ±3.7 <.001
Number of episodes for AECB 1.1 ±1.4 0.4 ±0.9 <.001
Number of spirometry test 1.0 ±1.6 0.7 ±1.3 <.001
Number of pulmonologist visit 1.5 ±2.5 0.7 ±1.6 <.001
Number of anticholinergics 3.4 ±4.1 1.3 ±2.9 <.001
Number of oral corticosteroids 3.2 ±3.7 1.5 ±2.3 <.001
Number of ICS use 0.5 ±2.0 0.3 ±1.3 <.001
Number of SABA use 4.2 ±5.0 2.2 ±3.4 <.001
Number of LABA use 0.4 ±1.8 0.2 ±1.1 <.001
Pre-index COPD exacerbation (mean ± SD)
Inpatient exacerbation (severe exacerbation) 0.4 ±0.8 0.1 ±0.4 <.001
Emergency room exacerbation (severe exacerbation) 0.1 ±0.6 0.04 ±0.2 <.001
Exacerbation by qualifying diagnosis (moderate exacerbation) 2.0 ±2.3 1.2 ±1.8 <.001
Exacerbation by qualifying drug therapy (moderate exacerbation) 2.3 ±2.6 1.6 ±1.9 <.001
AECB=acute exacerbation of chronic bronchitis, CB=chronic bronchitis, HMO=health maintenance organization, ICS=inhaled corticosteroids, LABA=long-acting β2-agonists, LAMA=long-acting muscarinic antagonist, PPO=preferred provider organization, SABA=short-acting β2-agonists, SD=standard deviation.
Note: Two-sample Student t test was used to compare differences in continuous variables and χ2 test was used to compare differences in categorical variables between the 2 study cohorts.

Patients in the roflumilast cohort showed a significantly higher number of episodes of AECB and hospitalizations due to AECB or chronic bronchitis, had more visits to a pulmonologist, underwent a greater number of spirometry tests, and used more oral corticosteroids than the nonroflumilast cohort (P<.05 for all above measures), indicating that patients with roflumilast were in more severe condition than those without roflumilast at baseline (Table 1). The roflumilast cohort also experienced a greater number of severe (P<.05) and moderate (P<.05) exacerbations than the nonroflumilast cohort during the baseline period (Table 1).

Among the roflumilast cohort, 18.7% (n=130) of patients had roflumilast as monotherapy. Nearly 42% (n=289) patients were on a combination of 4 medication classes (ie, roflumilast plus 3 other classes), followed by 21.4% (n=149) with 3 classes, 13.8% (n=96) with 2 classes, and <5% (n=31) having roflumilast with 4 other classes as the index medication. In contrast, among the nonroflumilast cohort, 89.2% of patients were on a combination of 2 medication classes, 10.5% with 3 classes, and 0.3% with 4 medication classes as the index medication (Table 2).

Table 2  of combination medications occurring within a 90-day window prior to matching
Category Roflumilast group Nonroflumilast group
N % N %
  695 100.0% 30,542 100.0%
2 96 13.8% 27,239 89.2%
3 149 21.4% 3219 10.5%
4 289 41.6% 84 0.3%
5 31 4.5% 0 0.0%
Roflumilast monotherapy 130 18.7% 0 0.0% 
Matched cohorts:
Demographic and Clinical Characteristics

To make these 2 cohorts more comparable for outcome evaluation, patients in the roflumilast cohort were matched (M) with patents in the nonroflumilast group, as described in the methods section. As a result, 328 patients in the roflumilast cohort were matched with patients in the nonroflumilast cohort, with a match rate of 61.4%.

After matching on preindex number of severe exacerbations, number of medication classes, 6 quarters since the approval month of roflumilast, and 75th deciles of estimated propensity score, the roflumilast (M) and nonroflumilast (M) cohorts consisted of similar proportions of males, with equivalent mean age and follow-up time (Table 3). In general, comorbidity burden and COPD severity were similar between the roflumilast (M) cohort and the nonroflumilast (M) cohort, although a higher percentage of patients with asthma diagnosis (41.2% vs 32.3%; P=.01) and higher number of SABA prescription fills (3.4 vs 2.7; P=.03) were still observed for roflumilast cohort during the 12 months prior to the index date (Table 3).

Table 3 Patient characteristics of roflumilast vs nonroflumilast exact matched (M) cohorts
  Roflumilast (M) (n=328) Nonroflumilast (M) (n=328) P value
Female (n, %) 182 55.5% 169 51.5% .32 
Age (mean ± SD) 63.5 ± 10.2 63.0 ± 9.8 .56
Health plan type (n, %)
HMO 104 31.7% 92 28.0% .48
PPO 212 64.6% 219 66.8%  
Others 12 3.7% 17 5.2%  
Geographic region (n, %)
Northeast 44 13.4% 41 12.5% .58
Midwest 130 39.6% 119 36.3%  
South 121 36.9% 138 42.1%  
West 33 10.1% 30 9.1%  
Pre-index Deyo-Charlson comorbidity index (mean ± SD) 2.4 ± 2.0 2.4 ± 1.9 .97
Pre-index period comorbidities of interest (n, %)
Asthma 135 41.2% 106 32.3% .01
Acute bronchitis and bronchitis 124 37.8% 134 40.9% .43
Myocardial infarction 33 10.1% 30 3.0% .69
Congestive heart failure 66 20.1% 61 18.6% .61
Ischemic heart disease 103 31.4% 94 28.7% .43
Hypertension 225 68.6% 222 67.7% .79
Diabetes mellitus 90 27.4% 85 25.9% .65
Gastroesophageal reflux disease 82 25.0% 98 29.9% .17
Anxiety 77 23.5% 57 17.4% .06
Depression 52 15.9% 50 15.2% .82
Pre-index COPD severity proxy (mean ± SD)
Number of hospitalization days due to AECB or CB 1.4 ± 5.1 1.2 ± 3.7 .38
Number of episodes for AECB 0.9 ± 1.2 1.0 ± 1.4 .90
Number of spirometry tests 0.9 ± 1.4 0.8 ± 1.7 .38
Number of pulmonologist visits 1.1 ± 1.9 1.2 ± 2.1 .40
Number of anticholinergics prescription fills 2.5 ± 3.2 2.3 ± 3.5 .31
Number of oral corticosteroids prescription fills 2.6 ± 3.2 2.2 ± 3.2 .05
Number of ICS prescription fills 0.4 ± 1.4 0.3 ± 1.1 .26
Number of SABA prescription fills 3.4 ± 4.2 2.7 ± 4.1 .03
Number of LABA prescription fills 0.2 ± 1.2 0.2 ± 1.0 .31
AECB=acute exacerbation of chronic bronchitis, CB=chronic bronchitis, HMO=health maintenance organization, ICS=inhaled corticosteroids, LABA=long-acting β2-agonists, LAMA=long-acting muscarinic antagonist, PPO=preferred provider organization, SABA=short-acting β2-agonists, SD=standard deviation.
Note: Paired t test was used to compare differences in continuous variables and McNemar test was used to compare differences in categorical variables between the 2 study cohorts.
Health Care Resource Utilization and Costs

During the postindex period, utilization of all-cause inpatient admissions, ER visits, and outpatient visits was not significantly different between the roflumilast (M) cohort and the nonroflumilast (M) cohort (Table 4). Compared to the nonroflumilast (M) cohort, the roflumilast (M) cohort had significantly higher unadjusted PPPM all-cause pharmacy costs ($702 vs $608; P<.001) but did not show statistically significant difference in medical and total (pharmacy + medical) costs (Table 4).

Table 4 Unadjusted health care resource utilization and costs of roflumilast vs nonroflumilast exact matched (M) cohorts
  Roflumilast (M) (n=328) Nonroflumilast (M) (n=328)  
PPPM all-cause health care resource utilization
  Mean SD Mean SD P value
Hospitalizations 0.07 ±0.15 0.09 ±0.18 .11
Emergency room visits 0.05 ±0.1 0.05 ± 0.1 .64
Outpatient visits 2.74 ±2.12 3.03 ±2.73 .13
PPPM all-cause health care costs
  Mean SD Mean SD P value
Inpatient admissions $1252 ±$3696 $1316 ±$3525 .77
Emergency room visits $76 ±$280 $62 ±$207 .66
Outpatient visits $651 ±$870 $815 ±$1231 .12
Medical $1980 ±$3994 $2193 ±$4,42 .41
Pharmacy $702 ±$464 $608 ±$646 <.001
Total (Medical + pharmacy) $2682 ± $4061 $2802 ± $4229 .90
PPPM=per-patient, per-month, SD=standard deviation.

The adjusted numbers of hospitalizations were not significantly different between the 2 cohorts (relative risk, 0.86 [95% CI, 0.46–1.62]; P=.65) after controlling for patient characteristics, including demographics, health plan characteristics, baseline comorbidities, Medicare Advantage enrollment, proxies for the severity of COPD complexity, and resource utilization, including preindex COPD-related costs. The adjusted all-cause health care costs showed a similar pattern. After controlling for the same patient characteristics, the roflumilast (M) cohort exhibited lower adjusted total health care costs on a PPPM basis (ratio, 0.9858 [95% CI, 0.84–1.16]compared with the nonroflumilast (M) cohort, although this comparison did not reach statistical significance (P=.86).

Discussion

There is limited real-world evidence of the demographic and clinical characteristics, as well as resource utilizations and associated costs, among COPD patients who were on roflumilast vs other maintenance combination medications. Results from this study indicate that, at baseline, most patients in the roflumilast cohort use roflumilast along with other maintenance medications as combination therapy. The roflumilast cohort showed a larger proportion of patients with >3 drug classes in their combination therapy, a greater comorbidity burden, more severe COPD conditions, and higher exacerbation history compared with the nonroflumilast cohort.

These results are not unexpected. Roflumilast is a relatively new drug indicated for the treatment of severe COPD, and previous research has suggested that newer drugs are more likely to be prescribed to patients who have failed to respond to other treatments, tend to be sicker, or both (Schneeweiss 2011). In addition, the GOLD guidelines recommend adding roflumilast to a combination therapy for severely ill patients (GOLD 2014). Therefore, it is expected that patients using roflumilast would be more severely ill. These results imply that when evaluating the impact of the use of roflumilast, careful attention should be given to verify that the cohorts being evaluated are comparable. If the cohorts are not well matched, the comparison may not be meaningful. This study used propensity score as a partial balancing score. We exactly matched patients on number of severe exacerbations and medication classes, quarters since approval month of roflumilast, and 75th categories of estimated propensity score. This approach produced a roflumilast cohort that was very similar to the nonroflumilast cohort in terms of demographic and clinical characteristics. Therefore, results based on the matched cohorts were robust. Because of large differences between the 2 groups, the match rate was 61.4%.

As expected, this study found that patients in the roflumilast (M) group had higher PPPM pharmacy costs than those in nonroflumilast (M) cohort. However, despite higher pharmacy costs, the roflumilast (M) cohort had similar total costs relative to the nonroflumilast (M) cohort. Additionally, this study found that the roflumilast (M) cohort had lower health care resource utilization although the comparison did not reach statistical significance. The adjusted results that control for the baseline demographic and clinical characteristics convey the same message.

Previous studies have shown that COPD exacerbations increase health care resource use (Blasi 2014). Treatment strategies that lead to the reduction of COPD exacerbations, therefore, have the potential to improve quality of life and mitigate use of limited health care resources. Roflumilast has been shown to decrease the occurrence of exacerbations in randomized, placebo-controlled trials (Calverley 2009, Fabbri 2009). This study provides evidence, using the real-world data, that affirms the findings from those trials. The nonsignificant results could be due to a limited follow-up period (average of 9 months) and a limited sample. Even with these limitations, this study found that total costs of care in the romiflumast (M) cohort were no higher than in the nonromiflumast (M) group, despite use of a costlier brand-name drug in the treatment group. In fact, most of the numerical estimates of health care utilization and costs are lower, except for pharmacy costs. The higher pharmacy cost is to be expected for a brand-name drug, but it is offset by lower health care utilization and medical costs. The net results are comparable total costs of care, with lower numerical estimates for roflumilast users.

Most physicians regard balancing costs with effectiveness as an appropriate component of clinical decision making (Ginsburg 2000). This study suggests that total health care costs are less of a concern for roflumilast use. For physicians and policy makers, a critical point when considering roflumilast as an add-on therapy may be the effectiveness of roflumilast for the appropriate severe chronic bronchitis patients with history of exacerbation and inadequate control. In the roflumilast population in this study, total costs of care and health care utilization were comparable between users and nonusers, with the lower point estimates in majority of the outcomes for roflumilast users. A longer-term evaluation with larger sample size is needed to warrant or manifest the current findings in this study.

Limitations

This study presented and compared the patient characteristics of a roflumilast cohort and a nonroflumilast cohort and estimated the impact of the use of roflumilast on healthcare resource utilization and costs. As such, confidence intervals were used to quantify the uncertainty pertaining to these estimated effects; therefore, it is possible that sample size may have influenced the estimated uncertainty. Additionally, the results are applicable only for a relatively short duration follow-up. In this study, we assigned index medications based on prescription use as observed through the pharmacy claims. For patients to be assigned to a combination therapy, a 90-day window before and after the first observed pharmacy claim was allowed. This identification strategy may have misclassified a “switching” from one drug treatment to another as being a combination therapy use. However, it is unlikely that a patient may switch from drug A to drug B and then switch back to drug A (or similar combinations of >2 drug classes); we believe the risk of this type of misclassification is very low.

Similar to other retrospective database studies, this study is subject to limitations, including coding errors of omission and commission, incomplete claims, unreliable clinical coding, and unobservable factors that may also have influenced the outcomes. Only direct costs that were billed and reimbursed by insurance claims were evaluated. Other costs outside of administrative data, such as over-the-counter drug cost, were not considered. Indirect costs were also ignored. Finally, study results were derived from commercial health plans in the US; therefore, results may not be generalized to people with other forms of insurance coverage. Nonetheless, the data used in this study were derived from a large, geographically diverse population, which enhances the generalizability of the results.

Conclusion

Findings from this study offer data to health plans regarding the characteristics of COPD patients on roflumilast in the real-world setting. This study found that, in general, patients in the roflumilast cohort were sicker and had more severe COPD. In most cases, roflumilast was prescribed as an add-on therapy to at least 2 other COPD maintenance medications. Therefore, this result implies that, in retrospective studies, attention needs to be given to defining a comparable group when evaluating the impact of roflumilast on health or economic outcomes of COPD patients.

Although the statistical evidence was not strong enough to show significance, the study found that despite higher pharmacy costs in the roflumilast group, total health care costs were lower than in the nonroflumilast group, and health care resource utilization was lower in the roflumilast cohort as well. Future research with a longer follow-up period and larger sample size is needed to evaluate health care resource utilization and costs with roflumilast treatment in long-term management of severe COPD patients.

References

Abudagga A, Sun SX, Tan H, Solem CT. Exacerbations among chronic bronchitis patients treated with maintenance medications from a US managed care population: an administrative claims data analysis. Int J Chron Obstruct Pulmon Dis. 2013;8:175–185.

Blasi F, Cesana G, Conti S, et al. The clinical and economic impact of exacerbations of chronic obstructive pulmonary disease: a cohort of hospitalized patients. PLoS One. 2014;9(6):e101228.

Brookhart MA, Schneeweiss S, Rothman KJ, et al. Variable selection for propensity score models. Am J Epidemiol. 2006;163(12):1149–1156.

Caliendo M, Kopeinig S. Some practical guidance for the implementation of propensity score matching. J Econ Surveys. 2008;22(1):31–72.

Calverley PM, Rabe KF, Goehring UM, et al. Roflumilast in symptomatic chronic obstructive pulmonary disease: two randomised clinical trials. Lancet. 2009;374(9691):685–694.

Daliresp [package insert]. St. Louis, MO: Forest Pharmaceuticals Inc; February 2011.

Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–619.

Donaldson GC, Wedzicha JA. COPD exacerbations .1: Epidemiology. Thorax. 2006;61(2):164–168.

Fabbri LM, Calverley PM, Izquierdo-Alonso JL, et al. Roflumilast in moderate-to-severe chronic obstructive pulmonary disease treated with longacting bronchodilators: two randomised clinical trials. Lancet. 2009;374(9691):695–703.

Ginsburg ME, Kravitz RL, Sandberg WA. A survey of physician attitudes and practices concerning cost-effectiveness in patient care. West J Med. 2000;173(6):390–394.

GOLD (Global Strategy for the Diagnosis, Management and Prevention of COPD, Global Initiative for Chronic Obstructive Lung Disease). http://goldcopd.org/uploads/users/files/GOLD_Report2014_Feb07.pdf. 2014.

Lechner M. Some practical issues in the evaluation of heterogeneous labour market programmes by matching methods. J Royal Stat Soc: Series A (Statistics in Society). 2002;165(1):59–82.

Lindenauer PK, Pekow PS, Lahti MC, et al. Association of corticosteroid dose and route of administration with risk of treatment failure in acute exacerbation of chronic obstructive pulmonary disease. JAMA. 2010;303(23):2359–2367.

Mapel DW, Dutro MP, Marton JP, et al. Identifying and characterizing COPD patients in US managed care. A retrospective, cross-sectional analysis of administrative claims data. BMC Health Serv Res. 2011;11:43.

Pauwels R, Calverley P, Buist AS, et al. COPD exacerbations: the importance of a standard definition. Respir Med. 2004;98(2):99–107.

Rennard SI, Calverley PM, Goehring UM, et al. Reduction of exacerbations by the PDE4 inhibitor roflumilast—the importance of defining different subsets of patients with COPD. Respir Res. 2011;12:18.

Rennard SI, Vestbo J. The many “small COPDs”: COPD should be an orphan disease. Chest. 2008;134(3):623–627.

Rothberg MB, Pekow PS, Lahti M, et al. Antibiotic therapy and treatment failure in patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. JAMA. 2010;303(20):2035–2042.

Schneeweiss S, Gagne JJ, Glynn RJ, et al. Assessing the comparative effectiveness of newly marketed medications: methodological challenges and implications for drug development. Clin Pharmacol Ther. 2011;90(6):777–790.

Seemungal TA, Hurst JR, Wedzicha JA. Exacerbation rate, health status and mortality in COPD—a review of potential interventions. Int J Chron Obstruct Pulmon Dis. 2009;4:203–223.

Solem CT, Sun SX, Sudharshan L, et al. Exacerbation-related impairment of quality of life and work productivity in severe and very severe chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2013;8:641–652.

Stuart EA. Matching methods for causal inference: A review and a look forward. Stat Sci. 2010;25(1):1–21.

Wu EQ, Birnbaum HG, Cifaldi M, et al. Development of a COPD severity score. Curr Med Res Opin. 2006;22(9):1679–1687.

* Appendix is published online at managedcaremag.com/archives/2015/8/roflumilast

Key points

  • Exacerbations of chronic obstructive pulmonary disease (COPD) are strongly associated with reduced quality of life and high rates of hospitalization and are a substantial contributor to the economic burden of the disease.
  • Roflumilast was the first PDE4 inhibitor drug approved by the FDA as treatment to reduce the risk of COPD exacerbations.
  • Little is known about the effect that roflumilast use has had on health care utilization and costs relative to other COPD treatments since its approval by the FDA in 2011.
  • This retrospective study of commercial claims data showed that 41.6% of patients prescribed roflumilast had been prescribed 3 other COPD drugs, a finding that is consistent with guidelines that position the medication as an add-on to combination therapy for COPD.
  • Pharmacy costs were higher, on average, for patients prescribed roflumilast than those who were not ($702 vs $608 per patient per month).
  • Medical costs and overall costs (the sum of medical and pharmacy costs) were lower, on average, for patients prescribed roflumilast than those who were not, although this comparison did not reach statistical significance.
  • A longer study with a larger group of patients may answer the question whether roflumilast lowers overall COPD costs through reductions in hospitalizations, emergency department visits, outpatient visits, or some combination of these interventions.

Disclosure: Jain, Cai, and Tan are employed by HealthCore, a consultancy whose activities for this project were funded by Forest Research Institute. Sun is an employee of Actavis, the sponsor of the study and the maker of roflumilast, and owns Actavis stock and stock options.

Corresponding author: Rahul Jain, PhD
Research Manager
HealthCore Inc.
800 Delaware Avenue, 5F
Wilmington, DE 19801
Email: rjain@healthcore.com

UP NEXT

Legislation and regulation
By Richard Mark Kirkner
Legislation and regulation
By Richard Mark Kirkner
Legislation and regulation
By Richard Mark Kirkner