Predictors of Nebulized Arformoterol Treatment: A Retrospective Analysis of Medicare Beneficiaries with Chronic Obstructive Pulmonary Disease*

Abstract

This study examined sociodemographic and clinical characteristics, treatment patterns, and health resource utilization among Medicare beneficiaries with chronic obstructive pulmonary disease (COPD) to identify predictors of nebulized arformoterol treatment. Using Medicare administrative data from 2010 to 2014, beneficiaries with ≥2 COPD outpatient visits ≥30 d apart or ≥1 COPD-related hospitalization(s) (ICD-9-CM 491.xx, 492.xx, and 496) were identified. Inclusion criteria required ≥1 COPD medication claim(s) and continuous enrollment in Parts A, B, and D. Four cohorts were identified: (a) 11,887 arformoterol users, (b) a subsample of arformoterol users (n = 1,778) who were hospitalized and discharged 30 d before initiating arformoterol, (c) 450,178 controls who had not received arformoterol, and (d) a subsample of controls (n = 21,910) who had hospitalizations. Logistic regression analysis was used to evaluate predictors of arformoterol treatment. The majority of beneficiaries were older than 70 years of age, female, Caucasian, and 47% were dual-eligible. The strongest predictors of arformoterol treatment were oxygen therapy, systemic corticosteroid or methylxanthine use, an exacerbation, a COPD-related hospitalization, and receiving care from a pulmonologist (all p < .001). Dual-eligibility, being a racial/ethnic minority, and having severe psychiatric comorbidity or immunodeficiency lowered the odds of receiving nebulized arformoterol (all p < .001). Among beneficiaries with recent hospitalizations, exacerbations and COPD-related admissions increased the odds of receiving arformoterol (p < .001). Nebulized arformoterol treatment was more likely to be initiated in sicker patients with COPD. Ensuring access to nebulized maintenance therapy is important and particularly warranted for COPD populations with greater medical needs.

Keywords:

• COPD
• Medicare
• nebulized arformoterol treatment
• long-acting beta2 agonist
• exacerbation
• hospitalization

Introduction

Chronic obstructive pulmonary disease (COPD) affects an estimated 16 million people in the United States (US) (1, 2). The condition accounts for more than $32.1 billion in annual medical costs, an economic burden that is projected to increase to $49 billion by 2020 (3). The use of pharmacotherapy in COPD aims to provide symptom relief and avoid costly exacerbations that frequently lead to hospitalizations to (4–7). To this end, therapeutic strategies for COPD recommend the use of maintenance therapy with long-acting bronchodilators (LABDs), (8, 9). However, past studies have shown that patients with COPD are often undertreated with LABDs despite exacerbations and repeated hospitalizations (7, 10–13).

The largest segment of the US population affected by COPD is adults over 65 years of age (14). Among this older adult population, the Medicare program serves as the primary payer for health coverage through its various components (14). Medicare Part A (hospital insurance) covers inpatients hospital stays, care in a skilled nursing facility, hospice care, and some home health care. Medicare Part B (medical insurance) covers two types of services: medically necessary services and preventive care including doctors’ office visits, outpatient care, mental health care (inpatient, outpatient, and partial hospitalization), ambulance service, clinical research, medical supplies, and durable medical equipment (DME) (e.g., nebulizer devices and nebulizer medications). Medicare Part D covers prescription drugs. In addition, Medicare Advantage (also known as Medicare Part C) is an “all in one” alternative to Medicare Parts A, B, and D that bundles all three components as inclusive coverage.

Approximately one in nine Medicare beneficiaries is diagnosed with COPD, of whom 17% are dual-eligible individuals (i.e., receive Medicare and Medicaid benefits) (12). Dual-eligible individuals are enrolled in Medicare Part A and/or Part B and receive full Medicaid benefits and/or assistance with Medicare premiums or cost sharing through various Medicare Savings Programs (MSPs) (15). Since Medicaid is managed by individual states rather than by the US government, dual-eligible benefits vary by state. Through federal statutes, the US government defines income and resource standards for full Medicaid coverage and MSPs, but states have discretion to effectively raise those limits above the federal floor. Some states offer Medicaid through managed care plans, some provide fee-for-service coverage, while others provide comprehensive plans.

Studies have shown that 40% to 71% of Medicare beneficiaries receive no maintenance treatment for COPD (13, 16–19). Although a recent longitudinal study on Medicare populations reported that use of maintenance medications has been improving (20), in general, it is recognized that Medicare beneficiaries with COPD remain undertreated (16, 20, 21).

COPD inhalation treatments are administered using several types of devices including pressurized metered-dose inhalers, dry powder inhalers, soft-mist inhalers, and nebulizers (22). Factors that limit the successful administration of pharmacotherapy in patients with COPD include complex inhaler regimens that require hand-breath coordination steps and mismatching inhaler devices to patients’ needs (23–26). Inhalation technique errors with handheld devices are common (27–29), and can result in incomplete medication dose delivery to the lungs (30). For some patients, COPD management may be improved with nebulizer use since it requires little to no hand-breath coordination or breath-holding (30, 31).

There is limited research on the factors associated with prescribing a nebulized therapy for COPD maintenance treatment. Moreover, no study has focused on predictors of nebulized treatment in the Medicare population. The purpose of our study was to examine the sociodemographic and clinical characteristics, treatment patterns, and health resource utilization (HRU) trends among Medicare beneficiaries with COPD to identify predictors of nebulized therapy using Brovana® (arformoterol tartrate) Inhalation Solution, one of two commercially available nebulized long-acting beta₂ agonists (LABA) in the US.

Methods

Data sources

This study used a 100% sample of Medicare administrative data from the Chronic Condition Warehouse at the Centers for Medicare and Medicaid Services (CMS). Demographic data were obtained from Master Beneficiary Summary Files. Information on hospitalizations was extracted from MedPar files, outpatient services from carrier and outpatient files, and home healthcare utilization from home health files. Information on medication utilization was obtained from Part D and DME files. Our study was approved by the Human Research Protections Program at the University of California San Diego. In addition, our study was subject to a data use agreement with CMS.

Sample selection

As illustrated in Figure 1, our primary study population was 10,371,035 Medicare beneficiaries who had a COPD-related claim (i.e., International Classification of Diseases, 9th Revision, Clinical Modification codes 490.xx, 492.xx, 494.xx, 496.xx) between 2010 and 2014 and a COPD diagnosis before 1st January 2013. Our study sample was further refined to 3,642,497 beneficiaries who met the inclusion criteria of having continuous coverage under Medicare Parts A, B, and D. Among these beneficiaries, 106,442 used nebulized LABA therapy during 2010 and 2014 and 36,061 of these were new users during 2012 and 2103; 3,536,055 beneficiaries did not use nebulized LABA therapy during this time period.

Figure 1. Flow chart depicting sample selection. a: Full coverage and COPD diagnosis <1 year prior to first nebulized LABA claim. Excluded deaths before first nebulized LABA claims. COPD: chronic obstructive pulmonary disease; LABA: long-acting beta2 agonist; SNF: skilled nursing facility; ESRD: end stage renal disease.

To support the analyses described below, a start date was randomly assigned to control group beneficiaries consistent with the distribution of initiation dates for nebulized LABA therapy in the study group. Both samples of new users and controls were limited to beneficiaries with: a diagnosis of COPD for at least 1 year; at least two office claims with COPD 30 d apart or one hospitalization for COPD in the prior year; without an admission to a skilled nursing facility; using COPD medication; without End Stage Renal Disease, who survived for at least 364 d following initiation of nebulized LABA therapy or their assigned start date. We further limited the sample of users of nebulized LABA therapy to those using only arformoterol.

As a final restriction, we limited both samples to beneficiaries who were hospitalized and discharged within 30 d prior to initiation of nebulized LABA therapy or their assigned start date. After applying additional exclusion and inclusion criteria, four cohorts were identified: (a) 11,887 nebulized arformoterol users, (b) a subsample of nebulized arformoterol users (n = 1,778) who had been discharged within 30 d before initiating arformoterol, (c) 450,178 controls who had received no nebulized arformoterol treatment, and (d) a subsample of controls (n = 21,910) who had been discharged within 30 d prior to the assigned start date.

Measures

The following predictor variables were examined for the year prior to initiating nebulized LABA or the assigned start date: (a) sociodemographic characteristics including age, sex, race/ethnicity, and US region in which care was delivered; (b) dual-eligibility status; (c) comorbidity as measured by the Chronic Illness and Disability Payment System (CDPS) diagnostic classification (32) and the Charlson Comorbidity Index (33); (d) COPD medications based on drug refill rates for short-acting bronchodilators (SABDs) which included short-acting muscarinic antagonists (SAMAs) and short-acting beta2 agonists (SABAs), LABAs, long-acting muscarinic antagonists (LAMAs), inhaled corticosteroids (ICS), systemic corticosteroids (CS), methylxanthines, phosphodiesterase (PDE)-4 inhibitors, non-specific PDE inhibitors, mucolytics, and antibiotics; (e) oxygen therapy; (f) exacerbation frequency, defined by an emergency department (ED) visit or a hospitalization for acute exacerbation of COPD (ICD-9 491.21 or 491.22) or a physician office visit followed by prescription for a systemic CS and/or a respiratory antibiotic; and (g) HRU including DME use, hospitalizations (all cause and COPD-related), intensive care unit stays, length of stay (LOS) during hospitalizations; pulmonary specialist visits (inpatient and outpatient), ED visits, and home health visits.

Statistical analysis

Descriptive statistics including means, frequencies, and proportions were used to compare sociodemographic characteristics, COPD treatment patterns, exacerbation frequency, and HRU between nebulized arformoterol users and controls. Logistic regression analysis was used to evaluate predictors of nebulized arformoterol treatment. Odds ratios (OR), 95% confidence intervals (CI), and p values were computed. p < .05 denoted statistical significance. All analyses were performed using SAS Enterprise Guide 7.1.

Results

Sociodemographic characteristics

All sociodemographic characteristics were comparable between nebulized arformoterol users and controls, except for a higher proportion of controls being dual-eligible compared to those treated with arfomoterol (47.3% vs. 36.5%, respectively) (Table 1). Across both groups, the majority of beneficiaries were >70-years-old, female, and Non-Hispanic White. The largest proportion of beneficiaries received care in the Southern or Midwest regions of the US.

Table 1. Sociodemographic characteristics of nebulized arformoterol users and controls.

Characteristic	All Medicare beneficiaries		Medicare beneficiaries with recent hospitalizations
	Arformoterol users	Controls	Arformoterol users	Controls
	(N=11,887)	(N=450,178)	(N=1,778)	(N=21,910)
	N (%)	N (%)	N (%)	N (%)
Age (years)
≤64	2,268 (19.1)	106,308 (23.6)	325 (18.3)	6,061 (27.6)
65–69	1,809 (15.2)	62,483 (13.9)	300 (16.9)	3,144 (14.4)
70–74	2,833 (23.8)	94,858 (21.1)	405 (22.8)	4,222 (19.3)
75–79	2,242 (18.9)	78,210 (17.4)	326 (18.3)	3,621 (16.5)
≥80	2,735 (23.0)	108,319 (24.0)	422 (23.7)	4,862 (22.2)
Sex
Male	4,738 (39.9)	180,242 (40.0)	685 (38.5)	8,451 (38.6)
Female	7,149 (60.1)	269,936 (60.0)	1,093 (61.5)	13,459 (61.4)
Race
Non-Hispanic White	10,804 (90.9)	384,890 (85.5)	1,602 (90.1)	18,035 (82.3)
Hispanic	336 (2.8)	18,993 (4.2)	52 (2.9)	1,142 (5.2)
African-American	573 (4.8)	34,076 (7.6)	91 (5.1)	2,198 (10.0)
Other	174 (1.5)	12,219 (2.7)	30 (1.9)	535 (2.5)
Dually-eligible
Yes	4,338 (36.5)	212,845 (47.3)	701 (39.4)	11,805 (53.9)
No	7,549 (63.5)	237,333 (52.7)	1,077 (60.6)	10,105 (46.1)
Region of care
Midwest	3,505 (29.5)	116,034 (25.8)	643 (36.2)	5,831 (26.6)
Northeast	1,514 (12.7)	81,715 (18.2)	183 (10.3)	3,765 (17.2)
South	5,394 (45.4)	192,437 (42.8)	792 (44.5)	9,780 (44.6)
West	1,474 (12.4)	59,992 (13.3)	160 (9.0)	2,534 (11.6)

COPD treatment patterns

The top 10 most common COPD treatments received by nebulized arformoterol users and controls are shown in Figure 2. Combination therapy with multiple medications was highly prevalent across all beneficiaries, irrespective of hospitalization history. A significantly higher proportion of nebulized arformoterol users (n = 11,887) compared to controls (n = 450,178) were receiving systemic CS (72.2% vs. 55.2%, respectively; p < .05), antibiotics (65.9% vs. 55.4%, respectively; p < .05), nebulized SABDs (61.2% vs. 36.7%, respectively; p < .05), oxygen therapy (58.2% vs. 34.8%, respectively; p < .05), and nebulized anticholinergics (AC) (31.3% vs. 17.0%, respectively; p < .05).

Figure 2. Top 10 most common COPD treatments among nebulized arformoterol users and controls in the 12 months prior to initiation of nebulized LABA (for the study group) or the assigned start date (for controls). COPD: chronic obstructive pulmonary disease; CS: corticosteroids; SABD: short-acting bronchodilator; ICS: inhaled corticosteroids; LABA: long-acting beta2 agonist; SABA: short-acting beta2 agonists; LAMA: long-acting muscarinic antagonist; AC: anticholinergics; *p < .05.

Exacerbations and HRU

Unadjusted results revealed that, compared to controls, a significantly higher proportion of nebulized arformoterol users had at least one exacerbation (63.5% vs. 76.4%, respectively; p < .05), an all-cause (35.2% vs. 46.2%, respectively; p < .05) or a COPD-related (12.3% vs. 21.1%, respectively; p < .05) hospitalization, and received inpatient (12.3% vs. 20.6%, respectively; p < .05) or outpatient care (42.3% vs. 57.6%, respectively; p < .05) from a pulmonologist within the previous 12 months (Figure 3a).

Figure 3. a: Exacerbations and health resource utilization patterns in the 12 months prior to initiation of nebulized LABA (for the study group) or the assigned start date (for controls). ICU: intensive care unit; COPD: chronic obstructive pulmonary disease; ED: emergency department; * p < .05. b: Exacerbations and health resource utilization patterns in the 12 months prior to initiation of nebulized LABA (for the study group) or the assigned start date (for controls) among Medicare beneficiaries with recent hospitalizations. ED: emergency department; ICU: intensive care unit; COPD: chronic obstructive pulmonary disease; * p < .05.

Among beneficiaries with recent hospitalizations (Figure 3b), nearly all nebulized arformoterol users and controls had at least one exacerbation within the previous 12 months (97.6% vs. 90.7%, respectively; p < .05). In addition, prior COPD-related hospitalization was significantly higher among nebulized arformoterol users than controls (57.1% vs. 38.9%, respectively; p < .05).

The average number of exacerbations, outpatient visits, all-cause and COPD-related hospitalizations, and ED visits within the previous 12-months were similar between nebulized arformoterol users and controls (Figure 4a). Charlson comorbidity scores were also similar between the two groups. Beneficiaries who had hospitalizations had frequent exacerbations and their HRU was high (Figure 4b). Although not statistically different from one another, within the previous 12 months, beneficiaries treated with nebulized arformoterol and controls had an average LOS that exceeded 11 d for all-cause admissions (11.45 ± 11.8 vs. 13.01 ± 16.5, respectively) and 6 d for COPD-related admissions (6.15 ± 6.1 vs. 5.96 ± 5.4, respectively). Moreover, these beneficiaries had a high number of comorbidity (3.86 ± 2.5 for nebulized arformoterol group vs. 4.38 ± 2.7 for controls).

Figure 4. a. Average number of exacerbations, health resources used, and comorbidity in the 12 months prior to initiation of nebulized LABA (for the study group) or the assigned start date (for controls). COPD: chronic obstructive pulmonary disease; ED: emergency department. b. Average number of exacerbations, health resources used, and co-morbidities in the 12 months prior to initiation of nebulized LABA (for the study group) or the assigned start date (for controls) among Medicare beneficiaries with recent hospitalizations. COPD: chronic obstructive pulmonary disease; LOS: length of stay; ED: emergency department.

Predictors of nebulized arformoterol treatment

After adjusting for potential confounders such as age and sex, several factors including sociodemographic and clinical characteristics, COPD treatments, and HRU patterns were found to be associated with the initiation of nebulized arformoterol therapy (Figure 5). The strongest predictors of nebulized arformoterol treatment were being on oxygen therapy (OR: 2.01, 95% CI: 1.93, 2.09; p < .001); using a systemic CS (OR: 1.50, 95% CI: 1.43, 1.57; p < .001) or methylxanthine (OR: 1.37, 95% CI: 1.28, 1.47; p < .001); having an exacerbation (OR: 1.33, 95% CI: 1.26, 1.41; p < .001) or a COPD-related hospitalization (OR: 1.31, 95% CI: 1.24, 1.39; p < .001); and receiving outpatient care from a pulmonologist (OR: 1.40, 95% CI: 1.34, 1.46; p < .001) or a respiratory therapist (OR: 1.23, 95% CI: 1.11, 1.36; p < .001).

Figure 5. Predictors of nebulized arformoterol treatment among Medicare beneficiaries with COPD. COPD: chronic obstructive pulmonary disease; OR: odds ratio; CI: confidence interval; US: United States; AIDS: acquired immune deficiency syndrome; GI: gastrointestinal; CNS: central nervous system; CS: corticosteroids; LAMA: long-acting muscarinic antagonist; SABD: short-acting bronchodilator; HH: home healthcare; ED: emergency department; OP: outpatient; IP: inpatient; * p < .05; ** p < .01; *** p < .001.

By contrast, the odds of being treated with nebulized arformoterol were significantly lower for Asians/Pacific Islanders (OR: 0.44, 95% CI: 0.34, 0.57); p < .001), African-Americans (OR: 0.71, 95% CI: 0.65, 0.78; p < .001), or and Hispanics (OR: 0.81, 95% CI: 0.72, 0.91; p < .01); the dual-eligible (OR: 0.73, 95% CI: 0.70, 0.77; p < .001); and beneficiaries with a severe psychiatric comorbidity (OR: 0.65, 95% CI: 0.56, 0.76; p < .001) or acquired immune deficiency syndrome (AIDS) (OR: 0.69, 95% CI: 0.52, 0.94; p < .01).

Among beneficiaries with recent hospitalizations (Figure 6), the strongest predictors of receiving nebulized arformoterol were having an exacerbation (OR: 2.62, 95% CI: 1.88, 3.63; p < .001) or a recent COPD-related hospitalization (OR: 1.83, 95% CI: 1.55, 2.14; p < .001); and receiving inpatient care from a pulmonologist (OR: 1.78, 95% CI: 1.58, 2.01; p < .001). Other significant predictors followed a similar trend to those observed in the larger cohort of beneficiaries (Figure 5), with the exception of three additional positive predictors: having a metabolic disease (OR: 1.41, 95% CI: 1.14, 1.75; p < .001), a central nervous system (CNS disease) (OR: 1.40, 95% CI: 1.10, 1.78; p < .01), or eye disease (OR: 1.55, 95% CI: 1.16, 2.08; p < .01); and one negative predictor: receiving care in the northeast region of the US (OR: 0.66, 95% CI: 0.56, 0.79; p < .001).

Figure 6. Predictors of nebulized arformoterol treatment among Medicare beneficiaries with recent hospitalizations. COPD: chronic obstructive pulmonary disease; OR: odds ratio; CI: confidence interval; US: United States; AIDS: acquired immune deficiency syndrome; GI: gastrointestinal; CNS: central nervous system; CS: corticosteroids; LAMA: long-acting muscarinic antagonist; SABD: short-acting bronchodilator; HH: home healthcare; ED: emergency department; OP: outpatient; IP: inpatient; *p<.05; **p<.01; ***p<.001.

Discussion

In this case–control study between Medicare beneficiaries who initiated nebulized arformoterol and control beneficiaries who did not, the distribution of nearly all of the sociodemographic characteristics were as anticipated except for the higher proportion of women than men in both cohorts. Since reports have shown that in the Medicare population the prevalence of COPD is higher in men than women (34), our results initially appeared to be somewhat counterintuitive. However, further examination of our data revealed that 56% of beneficiaries 65 years and older were female. Since most Medicare beneficiaries with COPD are older than 70 years of age, the higher percentage of women than men in our two cohorts can be explained by the age distributions that ensued as a result of applying our various inclusion and exclusion criteria. Moreover, our findings with regards to gender distribution are consistent with previous studies that have applied similar sampling criteria to ours to explore factors associated with maintenance treatment patterns among Medicare beneficiaries with COPD (20).

Our study revealed that the most important factors associated with the initiation of nebulized arformoterol were oxygen therapy, use of systemic CS or methylxanthine, and a previous exacerbation or COPD-related hospitalization. In contrast, beneficiaries who were dual-eligible, had severe psychiatric comorbidity or acquired immune deficiency syndrome (AIDS) were less likely to be treated with nebulized arformoterol. These results reveal several insights that can inform clinical practice.

From a clinical management perspective, we found several important predictors that collectively suggest sicker beneficiaries were more likely to be treated with nebulized arformoterol. The odds of receiving nebulized arformoterol treatment were increased if a beneficiary was receiving oxygen therapy, had an exacerbation, or a COPD-related hospitalization. While similar studies have not been conducted on Medicare populations and limited to nebulized therapy, there is some evidence from an international study that concurs with our findings that having a previous exacerbation is a positive predictor of initiating LABA treatment with handheld devices among patients with COPD (35). Among Medicare beneficiaries, the use of any maintenance medication substantially reduces the risk of hospitalizations (19). Unfortunately, past research conducted on large commercial and Medicare databases has found that hospitalizations do not necessarily impact LABD initiation, and that most patients with COPD continue to be undertreated even after a hospitalization episode (36). Our findings suggest a more positive outlook in that clinicians seem to be incorporating knowledge from empirical studies (37–39) into their patient treatment plans. Further support for our interpretation comes from the other positive predictors we found to be associated with initiating nebulized arformoterol therapy, namely receiving care from a pulmonologist or respiratory therapist. This latter finding is supported by past research showing that non-pulmonary clinicians tend to have more limited awareness and knowledge of current COPD treatment guidelines, a factor that has been posited to contribute to the under treatment of COPD (21, 40, 41).

We also found that certain pharmacotherapies were significant predictors of nebulized arformoterol treatment including systemic CS, methylxanthines, antibiotics, and handheld LAMAs and SABDs (all p < .001). Our study did not evaluate whether nebulized arformoterol was used in combination or as a replacement therapy. However, past studies have shown that patients with COPD who receive nebulized therapies generally have more advanced disease compared to those who receive other forms of pharmacotherapy (30, 42). Thus, it is reasonable to assume that nebulized arformoterol was likely added as a combination therapy rather than as a monotherapy (11, 43).

Our results further revealed that certain predictors were associated with lower odds of receiving nebulized arformoterol. Beneficiaries with African-American, Hispanic, and Asian/Pacific Islander race/ethnicity as well as the dual-eligible were less likely than non-Hispanic Whites and non-dual-eligibles to receive nebulized maintenance therapy. These results are concerning in light of past research that has demonstrated many negative health and economic outcomes associated with disparities in access to treatment among these groups of beneficiaries (44–49). For example, dual-eligible status is associated with a higher rate of COPD exacerbations and hospitalizations, a greater number of co-morbidities, and higher HRU (48–50). Although recent studies in Medicare populations have found that dual-eligible beneficiaries were more likely to receive maintenance medications compared to non-dual eligible beneficiaries, the rates of access to nebulized therapies have remained low (20). Past research has shown that treatment with nebulized arformoterol can substantially improve the health status of COPD patients through better disease management, leading to fewer exacerbations and lower COPD-related costs (primarily related to hospitalizations) (37–39, 51). Thus, increasing access to nebulized maintenance therapy is likely to benefit dual-eligible beneficiaries, many (>40%) of whom despite being much younger than 65 years old, have substantial unmet medical needs (8, 52, 53). Future research employing well-controlled study designs that quantify this potential benefit would be insightful.

Another important finding of the current study was that certain co-morbidities including severe psychiatric disorders, AIDS, diabetes, musculoskeletal disease and cardiovascular conditions significantly lower the odds of receiving nebulized arformoterol treatment. Although not specific to Medicare populations, past studies have shown cardiovascular diseases, diabetes, musculoskeletal disease, and psychiatric disorders to be negatively associated with initiation of LABA treatment, a finding that resonates with our results (32, 54, 55). Among these predictors, the most disconcerting result was that beneficiaries with severe psychiatric illness were less likely to be prescribed nebulized maintenance therapy. In Medicare populations, patients with concurrent COPD and psychiatric conditions experience more exacerbations, have higher HRU, and subsequent healthcare costs (55, 56). In addition, poor adherence to inhaled COPD maintenance medications is common among beneficiaries with psychiatric illness (13, 16, 57). Since nebulizers require more passive participation from users to administer medications than handheld inhalers, increasing reliance on nebulized maintenance therapy may improve adherence to prescribed COPD treatment regimens (13, 27, 40).

There are several limitations to our study. As with any retrospective claims database study, there is the potential for coding inaccuracy, missing data, and other reliability and validity indicators related to data quality. Specifically, the diagnosis of COPD is not based on spirometry but rather on the appropriate Medicare coding; therefore, the diagnosis was likely based on clinical interpretation in many of the patients included in our study. Our analysis was limited to fee-for-service Medicare beneficiaries with Parts A, B, and D. Thus, the findings may not be generalizable to managed Medicare or commercially insured populations. Information was not available on factors that may have influenced treatment management such as patient’s cognitive impairments, limited hand-breath coordination abilities, poor dexterity, inhalation device preferences, or inadequate adherence (29, 58–62). We had no direct measures of COPD severity, such as spirometry values and dyspnea scores. However, our inclusion criteria required a COPD diagnosis of ≥1 year and past research has provided ample evidence that by the time a COPD diagnosis is given, most patients have progressed beyond mild disease (1, 2, 63–65). In addition, medication use was determined based on prescription fills. Thus, we could not account for medications that may have been prescribed but never filled or filled but not used as prescribed. Reliance on ICD-9-CM codes to identify beneficiaries with psychiatric disorders may not have accurately captured the true prevalence of these conditions (66–68). Lastly, the role of formulary placement may have influenced the decision to initiate nebulized arformoterol therapy, an unobservable heterogeneous treatment effect that was not considered in the current analysis (69).

Conclusions

Nebulized arformoterol was more likely to be initiated in sicker patients with COPD. Given the results of this study, increasing access to nebulized maintenance therapy is warranted for select populations with COPD including racial/ethnic minorities, the dual-eligible, and those with certain co-morbidities, such as psychiatric disorders. Future studies are needed to explore the optimal time to initiate nebulized maintenance therapy, and the potential differential impact of early versus late initiation on patient outcomes.

Disclosure statement

T.P.G. and Z.X. are employees of the University of California San Diego and received institutional research funding from Advance Health Solutions to conduct this study. B.R.C. received consultation remuneration as a member of the Medical Advisory Board at Advance Health Solutions. He has also been a consultant for Glaxo Smith Kline, Boehringer-Ingelheim, Astra Zeneca, Novartis, and Pulmonix. S.C. and M.N. are employed by Advance Health Solutions, LLC. C.D. is an employee of Sunovion Pharmaceuticals Inc. This project was funded by Sunovion Pharmaceuticals Inc.

Additional information

Funding

This study was funded by Sunovion Pharmaceuticals Inc.