Therapy persistence and adherence in patients with chronic obstructive pulmonary disease: multiple versus single long-acting maintenance inhalers

Abstract

Objective:

To compare persistence and adherence among patients with chronic obstructive pulmonary disease (COPD) treated with either multiple- or single- long-acting maintenance inhalers.

Methods:

Patients with ≥2 COPD medical claims and ≥2 prescriptions for a long-acting inhaler within 1 year were classified as single- or multiple-inhaler users based on their treatment regimen (MarketScan database; 2004–2008) and matched on demographics and statistically significant baseline characteristics. Persistence, analyzed via time to treatment discontinuation, and treatment adherence, measured by proportion of days covered (PDC), were compared between the two groups over a 12-month period. Sensitivity analyses were conducted in patients with poorly and well-controlled disease.

Results:

A total of 23,494 patients were grouped into 11,747 matched pairs. After adjusting for confounding factors, multiple-inhaler users had a significantly higher discontinuation rate [Hazard ratio = 1.40, p < 0.0001] compared with single-inhaler users. Multiple-inhaler users were less likely to be adherent than single-inhaler users with an average PDC of 0.51 (SD = 0.272) vs. 0.55 (SD = 0.279), respectively (p < 0.0001). These results were consistent for the poorly- and well-controlled disease groups.

Conclusions:

Multiple long-acting inhaler users demonstrated lower treatment persistence and adherence rates than single long-acting inhaler users. Limitations of the study are related to the retrospective, observational design and use of claims data.

Key words::

• Adherence
• Combination therapies
• COPD
• Persistence

Introduction

Chronic obstructive pulmonary disease (COPD) is a progressive, debilitating respiratory disease caused primarily by long-term smoking and is characterized by persistent airflow limitation, inflammation, and pathological changes in the lungs1,2. COPD is primarily diagnosed in patients over 40 years of age and is responsible for significant morbidity and mortality3. It is currently the third leading cause of death in the United States and worldwide1,4, and it is expected that the COPD-related mortality rate will continue to rise5.

Treatment guidelines put forth by the Global Initiative for Chronic Obstructive Lung Disease (GOLD) recommend regular use of long-acting inhaled bronchodilators, including long-acting beta₂ agonists (LABAs) and long-acting muscarinic antagonists (LAMAs), for patients with moderate-to-very-severe COPD who remain symptomatic in spite of using short-acting bronchodilators1. The addition of inhaled corticosteroids (ICS) is also recommended for patients with repeated exacerbations.

In recent years, clinicians seeking to improve patient outcomes have begun prescribing combination therapies for patients with COPD. These complex regimens involve combining two or more long-acting inhaled medications in order to capitalize on the strengths of each therapeutic class6,7. While some medications are available in combination inhalers, many are not; thus a number of patients with COPD use multiple inhaler devices on a daily basis. However, while multiple inhaled medications may provide patients with more comprehensive treatment, there may also be some disadvantages due to treatment complexity. For instance, inhaled therapies come in a number of different devices, each one unique with respect to device loading and medication administration. Patients using multiple inhaled therapies may thus become confused by the different techniques required to properly use each individual device. Previous work supports this idea8, showing that the use of multiple inhalers, particularly inhalers with different methods of activation and inhalation techniques, may result in the greater likelihood of improper utilization (e.g., inadequate coordination of inspiration and actuation and inability to achieve a high enough respiratory flow rate)9 compared to the use of a single inhaler or the use of a single type of inhaler (e.g., the use of dry-powder inhalers only).

Proper use of inhaled medications is critical as it directly influences the amount of medication that reaches patients’ lungs. Therefore, improper use may result in a sub-therapeutic dose being received, which is likely to compromise treatment effectiveness, which may ultimately reduce patients’ long-term treatment adherence and persistence10--12. Work by Chrischilles and colleagues (2002) is supportive of the link between treatment simplicity and increased adherence13, showing that the use of a single inhaler combining ipratropium and albuterol significantly increased treatment persistence compared with separate inhaled therapies among patients with COPD. Similarly, in a patient survey by George and colleagues (2005)14, the degree to which patients indicated that they were confused about their medication was a significant independent predictor of non-adherence in patients with COPD.

Sustained adherence to long-term therapies is a key component in effective management of chronic diseases15. Long-acting inhaled medications must be used regularly and correctly to achieve therapeutic goals such as improved daily functioning, symptom reduction, exacerbation prevention, and decreased healthcare resource utilization16,17. A recent study by Vestbo and colleagues (2009) demonstrated that a higher adherence rate was strongly associated with a significantly lower risk of mortality and hospital admission due to COPD exacerbation18. However, in spite of its benefits, adherence to inhaled medications among patients with COPD has been shown to be very low, with some studies providing estimates as low as 34–60%13,16,19. It has also be shown that poor adherence in COPD patients seems to be influenced more by treatment complexity, patient beliefs and coping behavior, than by demographical factors or disease severity20.

While studies have suggested that increased treatment complexity can result in low COPD treatment adherence, to date there have been no studies comparing the relative impact of multiple-inhaler use over single-inhaler use on treatment adherence and persistence for patients with COPD21. This retrospective observational study compares treatment persistence and adherence between patients with COPD who used multiple long-acting inhalers versus those who used a single long-acting inhaler in a real-world setting.

Patients and methods

Data source

Healthcare claims data were extracted, based on data availability, for the period between January 2004 and December 2008 from a database combining the Thomson MarketScan Commercial Database (TMCD) and the MarketScan Medicare Supplemental and Coordination of Benefits (COB) Database for patients in the United States. The TMCD includes fully integrated patient-level data, including inpatient, outpatient, pharmacy, and laboratory services and enrollment data for approximately 25 million individuals covered annually by self-insured employers and private health insurance plans. The MarketScan Medicare Supplemental COB database focuses on patients who receive supplemental Medicare benefits through employer-sponsored health plans; it contains data on the employer-paid proportion, Medicare-paid benefits (i.e., COB amount), and patients’ out-of-pocket expenses for their medical and pharmacy services. Data are de-identified and comply with the patient confidentiality requirements of the Health Insurance Portability and Accountability Act (HIPAA).

Cohort selection

Multiple-inhaler users and single-inhaler users were identified based on their prescription drug claims among all the patients who received at least two diagnoses (either primary or secondary) for COPD [International classification of diseases, clinical modification (ICD-9-CM): 490.xx–492.xx or 494.xx–496.xx] during their period of continuous eligibility. Multiple-inhaler users were patients with COPD who had at least two occasions of filling two or more long-acting bronchodilator inhalers concomitantly or at least one long-acting bronchodilator inhaler and a separate ICS inhaler concomitantly for at least 7 consecutive days. Patients who had no record of filling more than one long-acting bronchodilator inhaler concomitantly or a long-acting bronchodilator and a separate ICS inhaler concomitantly at any time within a period of at least 18 months were defined as single-inhaler users. The long-acting bronchodilator inhalers considered were LAMAs, LABAs, and LABA/ICS combination inhalers. As LABA/ICS combination inhalers are contained in one device, these were considered as single inhalers.

It is possible patients on multiple inhalers were more severe than patients on a single inhaler at the time of initiation. In order to reduce any potential imbalance of disease severity between the two cohorts, the index date for each patient was randomly selected among all potential index dates, where each potential index date had an equal probability of being selected22. All long-acting inhaler prescription dates were considered as potential index dates for single-inhaler users, and all start dates for episodes of multiple long-acting inhaler use were considered as potential index dates for multiple-inhaler users. Potential index dates had to be preceded by at least 6 months of health plan continuous eligibility and followed by at least 12 months of continuous eligibility. Multiple and single-inhaler users had to be at least 40 years of age23 on that date and have received at least two prescriptions for a long-acting inhaler within 12 months following the index date.

Once the index date was selected, the baseline period was defined as the 6 months prior to the index date, and the study period was defined as the 12 months after the index date. The inhaler(s) being prescribed or used on the index date was defined as the index inhaler(s).

Since patients who use multiple long-acting bronchodilator inhalers, with or without ICS, are likely to have more severe COPD compared with single-inhaler users, an exact matching was performed to balance disease severity at baseline by selecting a cohort of single-inhaler users and a cohort of multiple-inhaler users. Single- and multiple-inhaler users were matched in a 1:1 ratio based on age, gender, index year, baseline Deyo–Charlson Comorbidity Index (CCI) score24, and different proxies of disease severity, including COPD subtype based on ICD-9 code classification (bronchitis, chronic bronchitis, emphysema, bronchiectasis, extrinsic allergic alveolitis, or chronic airway obstruction), baseline COPD pharmacotherapeutic classes used (LAMAs, LABAs, oral or inhaled corticosteroids, methylxanthines, or antibiotics), number of COPD prescriptions during the baseline period, and number of COPD exacerbations during the baseline period. A COPD exacerbation was defined as one of the following occurrences observed in the claims database: (1) a medical claim with a diagnosis code of 491.21 or 494.1x, (2) an inpatient or an emergency department visit with a diagnosis for COPD, or (3) a prescription for an antibiotic or an oral systemic corticosteroid within 7 days after an outpatient visit with a diagnosis of COPD. By matching and controlling for a comprehensive list of severity proxies during the baseline period, the potential bias associated with the cohort selection based on different treatment regimens is reduced. Patients were not matched on race or ethnicity, as this information was not contained in the dataset.

Since patients’ clinical status of COPD during the post-index study period may have an impact on treatment adherence and persistence additional analyses were performed which aimed to remove the potential confounding effect of treatment outcomes (whether patient disease was well- or poorly-controlled). A well-controlled patient was defined as one who had no COPD exacerbation during the 1-year post-index period, and a poorly-controlled patient was defined as one who had at least one COPD exacerbation event during the 1-year post-index period. After well-controlled and poorly-controlled patients were identified, patients were re-matched into two separate cohorts; well-controlled single-inhaler users were matched to well-controlled multiple-inhaler users and poorly-controlled single-inhaler users were matched to poorly-controlled multiple-inhaler users.

Study outcomes and statistical analysis

Patients’ baseline characteristics, including demographics (age and gender), index year, COPD treatments, CCI score, comorbidities frequently associated with COPD1,3,25, and number of COPD exacerbations, were summarized and compared between single-inhaler and multiple-inhaler users by applying the McNemar test for categorical variables and Wilcoxon signed rank test for continuous variables.

Treatment persistence was analyzed by comparing the discontinuation rates between the single- and multiple-inhaler cohorts over the 12-month study period. Discontinuation of treatment was defined as an interruption in prescription drug claims that lasted for 30 continuous days or longer for either the index inhaler (single-inhaler users) or one of the index inhalers (multiple-inhaler users). The rates of treatment discontinuation in both cohorts were estimated using Kaplan–Meier survival analyses and compared using log-rank tests. Cox proportional hazard models with random effects for the matched pairs estimated the time-adjusted risks of discontinuation controlling for age, gender, CCI score, COPD treatments, and comorbidities frequently associated with COPD. Treatment persistence rates being defined as one minus treatment discontinuation rates, higher discontinuation rates can be interpreted as lower persistence rates.

One-year adherence to inhaler treatment was estimated using the proportion of days covered (PDC). For the single-inhaler cohort, PDC was calculated as the number of days of supply covered by the index inhaler in the 1-year study period divided by 365 days. For the multiple-inhaler cohort, PDC was calculated as the smallest number of days of supply covered by any of the index inhalers divided by 365 days. Patients who had a PDC of 0.8 or higher were considered as adherent. Adherence rates were compared between the single- and multiple-inhaler cohorts using linear and logistic regression models. Univariate and multivariate linear regression models with random effects for the matched pairs were used to estimate the incremental effect on adherence of being a multiple-inhaler user versus being a single-inhaler user. Univariate and multivariate logistic regression models with random effects for the matched pairs were used to compare the probability of being adherent between the two study cohorts. The multivariate regression models controlled for baseline characteristics, including age, gender, CCI score, COPD treatments, and COPD comorbidities with a prevalence of at least 5% in either group.

Analyses listed above were also conducted among the well-controlled and poorly-controlled rematched cohorts. Statistical significance was set at a two-sided alpha-level of 0.05 or less. All statistical analyses were performed using SAS 9.2 (SAS Institute, Inc., Cary, NC).

Results

Study population

A total of 11,747 pairs of single- and multiple-inhaler users (i.e., 23,494 patients) with COPD who met the inclusion and exclusion criteria were matched by age, gender, index year, COPD subtype, baseline CCI score, pharmacotherapeutic classes of COPD treatment used during the baseline period, number of COPD prescriptions filled during the baseline period, and number of COPD exacerbations during the baseline period (Figure 1). As shown in Table 1, the 11,747 matched pairs of single- and multiple-inhaler users had a mean age (standard deviation [SD]) of 66.0 (SD 9.8) years, and over half were women. The most common drug therapies used during the baseline period were antibiotics, LABAs, LAMAs, and inhaled or oral corticosteroids. Although a majority of study patients had 0 to 5 prescriptions during the baseline period, a substantial proportion of patients had more complex COPD polypharmacy regimens. In addition, poorly-controlled patients had relatively more COPD-related prescriptions (10 or more COPD-related prescriptions) than well-controlled patients (9.6 vs. 2.7%, respectively).

Figure 1.  Sample selection flowchart. *Patients with COPD who had at least two occasions of filling two or more long-acting bronchodilator inhalers concomitantly or at least one long-acting bronchodilator inhaler and a separate ICS inhaler concomitantly for at least 7 consecutive days. ^†Patients who had no record of filling more than one long-acting bronchodilator inhaler concomitantly or a long-acting bronchodilator and a separate ICS inhaler concomitantly at any time within a period of at least 18 months.

Table 1.  Matched patient baseline characteristics.

Matching criteria	All patients	Well-controlled patients	Poorly-controlled patients
Matched patient pairs, n*	11,747	5371	3352
Demographics
Age, mean (SD)	66.0 (9.8)	65.5 (9.7)	66.1 (9.7)
Female, n* (%)	6232 (53.1)	2740 (51.0)	1896 (56.6)
Index year, n* (%)
2004	1301 (11.1)	484 (9.0)	462 (13.8)
2005	2211 (18.8)	1049 (19.5)	566 (16.9)
2006	3014 (25.7)	1405 (26.2)	802 (23.9)
2007	5221 (44.4)	2433 (45.3)	1522 (45.4)
COPD treatments
Drug use by treatment class, n* (%)
Anticholinergics	4921 (41.9)	2284 (42.5)	1193 (35.6)
Beta antagonists	4951 (42.1)	1939 (36.1)	1453 (43.3)
Oral and inhaled corticosteroids	4904 (41.7)	1774 (33.0)	1603 (47.8)
Methylxanthines	91 (0.8)	22 (0.4)	39 (1.2)
Antibiotics	5034 (42.9)	1872 (34.9)	1629 (48.6)
COPD prescriptions, n* (%)
0–5	7425 (63.2)	3633 (67.6)	1872 (55.8)
5–10	1941 (16.5)	667 (12.4)	592 (17.7)
>10	709 (6.0)	143 (2.7)	322 (9.6)
COPD exacerbations, mean (SD)	0.2 (0.6)	0.1 (0.3)	0.3 (1.0)
Charlson Comorbidity Index, mean (SD)	1.0 (0.9)	0.9 (0.8)	1.0 (0.9)

*Number of patients in each group (single-inhaler users and multiple-inhaler users).

The prevalence of comorbidities of the single- and multiple-inhaler cohorts during the baseline period is presented in Table 2. A significantly higher proportion of single-inhaler users had upper respiratory tract infections, musculoskeletal disorders, metabolic syndrome, diabetes, arthritis, and reflux esophagitis during the baseline period. However, a higher proportion of patients in the multiple-inhaler cohort had asthma during the baseline period.

Table 2.  Prevalence of comorbidities in the single- and multiple-inhaler users during the baseline period, n (%).

Comorbidities, n (%)	Single-inhaler users n = 11,747	Multiple-inhaler users n = 11,747	p-value
Anemia	565 (4.8)	515 (4.4)	0.1200
Arthritis	2,237 (19.0)	1,929 (16.4)	<0.0001
Asthma	1,971 (16.8)	2,215 (18.9)	<0.0001
Cachexia	1 (0.0)	0 (0.0)	–
Carcinoma in situ of respiratory system	3 (0.0)	10 (0.1)	0.0522
Cataracts and visual disturbances	1,025 (8.7)	1,043 (8.9)	0.6785
Chronic renal failure	103 (0.9)	103 (0.9)	–
Congestive heart failure	730 (6.2)	738 (6.3)	0.8306
Depression or anxiety	612 (5.2)	550 (4.7)	0.0616
Diabetes	1,238 (10.5)	1,132 (9.6)	0.0219
Glaucoma	700 (6.0)	716 (6.1)	0.6575
Heart disease (ischemic heart disease, coronary heart disease, angina)	1,769 (15.1)	1,684 (14.3)	0.1172
Hypoxemia with cognitive impairment	177 (1.5)	185 (1.6)	0.6741
Lower respiratory tract infection	630 (5.4)	607 (5.2)	0.5033
Lung cancer	130 (1.1)	140 (1.2)	0.5428
Malnutrition and weight loss	105 (0.9)	88 (0.7)	0.2162
Metabolic syndrome	2,733 (23.3)	2,437 (20.7)	<0.0001
Musculoskeletal disorders	4,549 (38.7)	4,168 (35.5)	<0.0001
Nutritional deficiencies	92 (0.8)	72 (0.6)	0.1183
Peptic ulcer disease	36 (0.3)	33 (0.3)	0.7180
Reflux esophagitis	143 (1.2)	110 (0.9)	0.0365
Sleep disorder	773 (6.6)	794 (6.8)	0.5817
Upper respiratory tract infection	1,718 (14.6)	1,516 (12.9)	0.0001

For the sensitivity analyses, 5371 matched pairs of well-controlled patients with COPD and 3352 matched pairs of poorly-controlled patients were further analyzed separately. The demographics and baseline characteristics of well-controlled and poorly-controlled patients are summarized in Table 1.

Treatment persistence

Table 3 and Figure 2 show that 9235 (78.6%) patients in the single-inhaler cohort and 10,190 (86.7%) patients in the multiple-inhaler cohort discontinued the index inhaler during the 1-year study period. More specifically, after 1 month and 3 months, 235 (2.0%) and 5403 (46.0%) of patients in the single-inhaler cohort and 587 (5.1%) and 6695 (57.0%) of patients in the multiple-inhaler cohort discontinued the index inhaler, respectively. The time-adjusted risk of treatment discontinuation was significantly higher for multiple-inhaler users (hazard ratio [HR] = 1.40, 95% confidence intervals [CI]: 1.35–1.46) after adjusting for patients’ baseline characteristics and comorbidities. This confirms that treatment persistence was significantly higher in the single-inhaler cohort than in the multiple-inhaler cohort.

Figure 2.  Treatment persistence (discontinuation rates) of single- and multiple-inhaler cohorts among (a) all, (b) well-controlled, and (c) poorly-controlled patients.

Table 3.  Comparisons of treatment discontinuation rates between single- and multiple-inhaler users during the 12-month study period.

	Discontinuation rates, n (%)		Unadjusted hazard ratio* (95% CI)	p-value	Adjusted hazard ratio* (95% CI)	p-value
	Multiple-inhaler users	Single-inhalers users
All patients	10,190 (86.7%)	9,235 (78.6%)	1.40 (1.35–1.45)	<0.0001	1.40 (1.35–1.46)	<0.0001
Well-controlled	4,624 (86.1%)	4,119 (76.7%)	1.44 (1.36–1.52)	<0.0001	1.44 (1.36–1.52)	<0.0001
Poorly-controlled	2,963 (88.4%)	2,751 (82.1%)	1.40 (1.30–1.50)	<0.0001	1.40 (1.30–1.50)	<0.0001

*HR > 1 indicates that multiple-inhaler users had a greater discontinuation rate compared to single-inhaler users.

Lower discontinuation rates among single-inhaler users compared to multiple-inhaler users were also observed in both well-controlled and poorly-controlled patients during the study period (Figure 2, Table 3). In the well-controlled group, after adjustment for confounding factors, multiple-inhaler users had 44% higher discontinuation rates compared to single-inhaler users (HR = 1.44, 95% CI: 1.36–1.52), with 4624 (86.1%) multiple-inhaler users and 4119 (76.7%) single-inhaler users discontinuing treatment after 1 year, respectively. Among patients who were poorly-controlled during the study period, multiple-inhaler users had a 40% higher discontinuation rate compared to single-inhaler users (HR = 1.40, 95% CI: 1.30–1.50) with 2963 (88.4%) multiple-inhaler users and 2751 (82.1%) single-inhaler users discontinuing treatment after 1 year, respectively.

Adherence

The average PDC was 0.55 (SD 0.279) for the single-inhaler users and 0.51 (SD 0.272) for the multiple-inhaler users (p < 0.0001). As shown in Table 4, after controlling for patients’ characteristics and baseline comorbidities, the PDC of the multiple-inhaler cohort was significantly lower than that of the single-inhaler cohort by 8.6%. Similarly, multiple-inhaler users were 34% less likely to be adherent (PDC ≥ 0.8) compared to single-inhaler users (odds ratio [OR] = 0.66, 95% CI: 0.62–0.70). Figure 3 shows that 4568 (38.9%) of single-inhaler users had a PDC of 0.8 or higher and thus were considered adherent, while only 3793 (32.3%) of multiple-inhaler users met this criterion.

Figure 3.  Adherence of single- and multiple-inhaler users among (a) all, (b) well-controlled, and (c) poorly controlled patients.

Table 4.  Adherence (PDC) comparison between single- and multiple-inhaler users during the 12-month study period.

	Incremental PDC (%)†	p-value	Odds ratio‡ (95% CI)	p-value
Univariate
All patients*	−8.4%	<0.0001	0.68 (0.64–0.72)	<0.0001
Well-controlled patients	−9.8%	<0.0001	0.63 (0.58–0.69)	<0.0001
Poorly-controlled patients	−5.4%	0.0038	0.74 (0.66–0.83)	<0.0001
Multivariate
All patients	−8.6%	<0.0001	0.66 (0.62–0.70)	<0.0001
Well-controlled patients	−10.1%	<0.0001	0.61 (0.56–0.67)	<0.0001
Poorly-controlled patients	−5.7%	0.0022	0.73 (0.65–0.82)	<0.0001

*The average PDC was 0.55 (SD 0.279) for the single-inhaler users and 0.51 (SD 0.272) for the multiple-inhaler users.

†For the linear regression, PDC < 0 indicates that single-inhaler users had higher PDCs compared to multiple inhaler users.

‡OR > 1 indicates that multiple-inhaler users had a higher probability of being compliant than single-inhaler users.

Consistent with the overall comparison, well-controlled single-inhaler users showed a significantly higher PDC and a higher likelihood of being adherent compared to well-controlled multiple-inhaler users (adjusted OR = 0.61; 95% CI: 0.56–0.67) (Table 4). Similarly, poorly-controlled single-inhaler users had a significantly higher PDC (by 5.7%) and a higher likelihood of being adherent (adjusted OR = 0.73, 95% CI: 0.65–0.82) (Table 4).

Discussion

The results of this large observational study show that the use of multiple long-acting inhalers for the treatment of COPD is associated with an increased risk of treatment discontinuation and decreased adherence to treatment compared to the use of a single long-acting inhaler within a 1-year period. These differences remained statistically significant after controlling for patients’ baseline characteristics and proxies of COPD severity. Results from the sensitivity analyses conducted on well-controlled and poorly-controlled patients further suggest that treatment complexity has a negative impact on treatment adherence and persistence, irrespective of patient outcomes. These findings are consistent with previous evidence on the impact of multiple inhaler use on treatment persistence and patient adherence with prescribed treatment in chronic diseases and COPD specifically5–7. However, to our knowledge, the current study is the first to examine the difference in treatment persistence and adherence between users of a range of multiple long-acting inhalers and single long-acting inhalers.

Given that patients with COPD have been shown to exhibit poorer physical functioning26, more comorbidities27, and worse cognitive performance28 when compared to matched controls, it is not surprising that the use of multiple inhalers may be difficult for these patients. It has been estimated that 34–60% of patients with COPD do not use metered-dose or dry-powder inhalers correctly14. Improper use is likely compounded by the addition of a second inhaler, as patients using multiple types of inhaled therapies must learn and follow a separate set of instructions in order to properly use each device. This is supported with work by van der Palen and colleagues (1999) who showed that adult asthma patients using both a dry-powder inhaler and metered-dose inhaler were less likely to perform correct techniques than patients who used only one type of inhaler8. Results from the persistence analysis are consistent with an array of different studies where it has been shown that treatment discontinuation is high amongst patients with COPD and asthma. For asthma, it has been found that treatment discontinuation with single- and fixed-combination ICS can be as high as 85% and 90%, respectively29. For COPD, a recent 1-year study of inhaled beta-agonist, anticholinergic and glucocorticoid medications found that patients taking anticholinergics had lower discontinuation rates compared to the other two medications (47 vs. 93, and 70% discontinuation rates, respectively; p < 0.0001)30.

It is well-established that correct use of prescribed treatments, including proper adherence and persistence, is critical to achieving optimal clinical outcomes and maximizing quality of life14,16,31. Other studies have also shown that a more complex regimen often leads to incorrect device use7, which in turn results in lower treatment effectiveness8,9. A decrease in treatment effectiveness may lead patients to see a decreased value in continuing their prescribed therapy and consequently increased the risk of discontinuation8.

Reducing the complexity of treatment regimens and confusion that patients face in properly using different types of inhalers has the potential to reduce discontinuation and improve adherence. It is also important to consider that in addition to their maintenance therapy device(s), most if not all patients also have a rescue inhaler device for acute episodes. Treatment complexity may be reduced in a number of ways including expanded use of combination inhalers instead of separate monotherapy inhalers, technology improvements to create simpler inhaler devices, and development and use of oral COPD medications once available.

Although this was a retrospective, observational study, one key strength of the study design was the number of precautions undertaken to minimize the potential influence of confounders on persistence and adherence. First, multiple-inhaler and single-inhaler cohorts were exactly matched on age, gender, index year, COPD subtype, baseline general health status (measured by the CCI index score), baseline COPD drug classes used, number of baseline COPD prescriptions, and the number of COPD exacerbations during the baseline period. The combination of these variables served as indirect indicators of patients’ disease severity. Second, regression analyses further controlled for patients’ baseline characteristics, most notably their comorbidities. Third, separate analyses in well-controlled and poorly-controlled cohorts accounted for the patients’ clinical status during the study period. Given these precautions, the results of this study suggest that the complexity of inhaler regimen is highly associated with patients’ adherence and persistence to prescribed therapy.

This study is subject to a number of limitations due to the observational design and limited availability of information in the healthcare claims database. First, COPD severity was not available for patients in this database; therefore, proxies for COPD severity were identified and controlled for in order to reduce any bias. Second, no information was available on patients’ impressions and use of the various devices, or on patient–physician interactions. Therefore, it cannot be known for certain whether the low persistence and adherence seen in this study were due to difficulty with properly using the inhaler devices or whether they were due to other unobserved variables, such as clinical intent or patients’ perception of treatment efficacy. Third, information was not available on the regimens that patients were prescribed, only on which prescriptions were filled. Therefore, it is unknown whether patients were advised by physicians to take both medications concomitantly during the entire 1-year study period. It is possible that some patients were prescribed a single core maintenance therapy and were told to use the other sporadically (e.g., seasonally). In this case, patients who may have been compliant with their therapy were categorized as non-adherent as they had not used both products for the entire 1-year period. However, in order to alleviate this limitation, sensitivity analyses using a range of definitions of PDC for multiple-inhaler users were conducted and results were consistent with the overall findings (results not presented).

Misclassification, omission, and lack of drug-related adverse events in the data may have introduced inaccuracies in the study analysis. Since there is no true measure of COPD exacerbation severity in healthcare claims, a patient’s baseline number of exacerbations and exacerbations during the study were most likely underestimated as only exacerbation events requiring medical intervention could be considered. However, these limitations are expected to equally affect patients in the single- and multiple-inhaler cohorts and are unlikely to bias the study results in a particular direction.

This study has demonstrated that simplification of a COPD treatment regimen may have a beneficial effect on patients’ treatment persistence and adherence. Additional data regarding the impact of treatment complexity on patients’ clinical outcomes and healthcare resource utilization will be helpful for healthcare providers and patients in making therapeutic decisions to optimally manage COPD, minimize symptoms and exacerbations, and maximize quality of life. Investigation into new technologies that can improve the efficiency of drug delivery and simplify drug regimens for patients with COPD is needed.

Transparency

Declaration of funding

This research was sponsored by the Forest Research Institute.

Declaration of financial/other relationships

M.M. and S,B. are current employees of the Forest Research Institute, J.S. worked for the Forest Research Institute while performing this analysis, while all other co-authors are employees of Analysis Group, Inc., which has received consultancy fees from the Forest Research Institute.

Acknowledgments

Prescott Medical Communications Group (Chicago, IL, USA) provided editorial assistance funded by Forest Research Institute.