Prescription Patterns of New Use of Fixed-Dose Combination Inhalers in Patients with Chronic Obstructive Pulmonary Disease: Long-Acting β2 Agonists Plus Long-Acting Muscarinic Antagonists versus Long-Acting β2 Agonists Plus Inhaled Corticosteroids

Introduction

Chronic obstructive pulmonary disease (COPD) is a disease with substantial impacts on health globally, which is ranked as the third and the eighth leading cause of death worldwide and in Taiwan, respectively.^1,2 According to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guideline, long-acting bronchodilators are the mainstay pharmacological treatment of COPD. Combination therapies of long-acting β2 agonists plus long-acting muscarinic antagonists (LABA/LAMA) or long-acting β2 agonists plus inhaled corticosteroids (LABA/ICS) are recommended for patients with apparent dyspnea symptoms or frequent exacerbation history at initial diagnosis or with suboptimal response to monotherapy of long-acting bronchodilators during follow-up.³

Theoretically, when combination therapies are indicated, there is an advantage of using medications with different mechanisms in a single inhaler, a so-called fixed-dose combination (FDC) inhaler. The products of LABA/ICS FDC have been marketed in the world for several decades. A number of studies have also noticed that use of LABA/ICS combination therapy, mainly driven by LABA/ICS FDC, is very common in COPD patients,^4–6 although a potential increased risk of lower respiratory tract infections remains concerning.^7,8 In contrast, although LABA/LAMA combination treatment is also suggested by the GOLD guideline, the products of LABA/LAMA FDC have only been available in the market within the last 10 years (eg, 2013 in the European Union and the US; 2015 in Canada and Taiwan). It is expected that this may spur use of LABA/LAMA combination treatment in daily practice. Several studies have described temporal trends of LABA/LAMA combination treatment (including FDC and free combination forms);^9–12 however, few studies have specifically focused on the uptake of these new FDC products.¹² Furthermore, prescription patterns may vary across geographic areas. It is therefore necessary to analyze domestic data, examine the status of drug utilization in each country, and identify potential clinical issues, if any.

The products of LABA/LAMA FDC and LABA/ICS FDC were reimbursed in Taiwan on May 1, 2015 and March 1, 2002, respectively, which provides an opportunity to explore prescription patterns of new use of LABA/LAMA and LABA/ICS FDC in COPD patients in real-world settings. Specifically, the present study aimed to examine potential temporal trends, variations in healthcare provider characteristics, and differences in patient characteristics for both recommended alternative treatments.

Methods

Database

A compulsory, single-payer National Health Insurance program started in Taiwan in 1995; 99% of the 23 million residents were enrolled in the program by June 2021.¹³ The Taiwan National Health Insurance Research Database (NHIRD) contains complete information on enrollment and demographic records; diagnosis, procedure, and health services data from hospital admissions and outpatient visits; and pharmacy dispensing claims from hospital admissions and outpatient visits of each beneficiary.¹⁴ The dataset used in the present study consists of a random sample of 2 million individuals from the Taiwan NHIRD, with longitudinally linked data for beneficiaries from 2010 to 2018. All data are de-identified by scrambling the identification codes of patients and healthcare providers.¹⁴ The study protocol was approved by the National Yang-Ming Chiao Tung University Research Ethics Committee. Informed consent was waived given the retrospective nature of the study and the analysis of anonymous data.

Study Population and Study Drugs

We identified patients who initiated LABA/LAMA FDC or LABA/ICS FDC at outpatient visits between January 1, 2015 and December 31, 2018 using the World Health Organization’s Anatomical Therapeutic Chemical Classification System codes (see Appendix Table 1 for list of codes). Initiation of LABA/LAMA FDC or LABA/ICS FDC was defined as first LABA/LAMA FDC or LABA/ICS FDC dispensing during the study period with no prior dispensing for either study drug in the preceding 365 days; during this period, patients were required to have interactions with the healthcare systems, defined as having at least one hospital admission or outpatient visit in this window. The index date was the date of the first dispensing of either study drug. We further restricted the cohort patients to those aged between 40 and 100 years on the index date, with at least one inpatient or outpatient COPD diagnosis (see Appendix Table 2 for codes) within 365 days before the index date. This algorithm provided a sensitivity of 85.0% and a specificity of 78.4%.¹⁵

To specify if patients were LABA/LAMA FDC or LABA/ICS FDC initiators, we excluded patients who simultaneously initiated both LABA/LAMA FDC and LABA/ICS FDC or who simultaneously initiated LABA, LAMA, and ICS on the index date. We also excluded patients who initiated more than one type of LABA/LAMA FDC or LABA/ICS FDC on the index date.

Temporal Trends, Healthcare Provider Characteristics, and Patient Characteristics for Use of LABA/LAMA FDC or LABA/ICS FDC

To examine potential changes in the use of LABA/LAMA FDC or LABA/ICS FDC over time, we stratified LABA/LAMA FDC or LABA/ICS FDC initiators according to the years of index dates (2015, 2016, 2017, and 2018). To explore healthcare provider characteristics, we captured the accreditation level of the hospitals where the initiation prescriptions were from (medical centers, metropolitan hospitals, district hospitals, or primary care clinics) and the specialty of the physicians who prescribed the study drugs (chest, internal medicine, family medicine, and others). We also assessed patient characteristics of LABA/LAMA FDC or LABA/ICS FDC initiators, including demographics (age on the index date and sex), comorbidities (cardiovascular diseases, asthma, pneumonia, cancer, and psychiatric disorders), and the Charlson comorbidity score, which are widely used to assess general comorbidities),^16–18 as well as resource utilization at baseline.

We determined if patients had comorbidities based on any inpatient and outpatient diagnosis except asthma, which required at least 3 diagnosis records at baseline to prevent misclassification between asthma and other conditions. We measured resource utilization based on records of hospital admission or outpatient visits due to any reasons, cardiovascular diseases, or COPD. All the information was assessed within 365 days before the index date. See Appendix Table 3 for detailed definitions of comorbidities.

Statistical Analysis

We applied the Cochran–Armitage trend test to assess whether there was a significant, temporal trend for use of LABA/LAMA FDC and LABA/ICS FDC. We used the chi-square test to examine if there was a significant variation in healthcare provider characteristics, including the accreditation level of the hospitals and the specialty of the physicians, between treatment groups. A p-value of <0.05 was considered statistically significant. For patient characteristics, we computed standardized differences for each characteristic, with a value of >0.1 in absolute terms indicating a clinically meaningful difference between treatment groups.^19,20 We also conducted multivariate logistic regression, with a 95% CI not including the null suggesting an independent predictor of initiating LABA/LAMA FDC or LABA/ICS FDC. All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC).

Results

Eligible Patients

There were 12,455 eligible patients identified; the mean (SD) age of patients was 69 (12) years and 68% were male. Among LABA/LAMA FDC initiators (n=4019), vilanterol/umeclidinium (44%) was the most frequently used medication, followed by indacaterol/glycopyrronium (39%) and olodaterol/tiotropium (17%). Among LABA/ICS FDC initiators (n=8436), salmeterol/fluticasone was prescribed predominantly (36%), followed by formoterol/budesonide (28%), formoterol/beclomethasone (24%), vilanterol/fluticasone (11%), and formoterol/fluticasone (1%) (Figure 1).

Figure 1 Flow chart of the study population assembly. Abbreviations: COPD, chronic obstructive pulmonary disease; FDC, fixed-dose combination; ICS, inhaled corticosteroids; LABA, long-acting beta2-agonists; LAMA, long-acting muscarinic antagonists.

Temporal Trends Between LABA/LAMA FDC and LABA/ICS FDC Treatment Groups

As shown in Figure 2, the number of initiators rose only slightly between 2015 (n=2752) and 2018 (n=3229). During the entire study period, the number of initiators with LABA/LAMA FDC was lower than that with LABA/ICS FDC. However, there was an apparent increased use of LABA/LAMA FDC but an evident decreased use of LABA/ICS FDC over time. Specifically, among 2752 patients who initiated either FDC class in 2015, 12% (n=336) and 88% (n=2416) initiated LABA/LAMA FDC and LABA/ICS FDC, respectively. Nevertheless, in 2018, the proportion of each FDC class become closer, with 44% initiating LABA/LAMA FDC and 56% initiating LABA/ICS FDC. The Cochran–Armitage trend test suggested a significant temporal change in use of both FDC products (p-value <0.001).

Figure 2 Temporal trends for use of LABA/LAMA FDC and LABA/ICS FDC. Abbreviations: FDC, fixed-dose combination; ICS, inhaled corticosteroids; LABA, long-acting beta2-agonists; LAMA, long-acting muscarinic antagonists.

Healthcare Provider Characteristics Between LABA/LAMA FDC and LABA/ICS FDC Treatment Groups

Figure 3A shows number of initiators with each FDC class across the accreditation levels of the hospitals. Most patients received FDC products from regional hospitals (n=5161), followed by medical centers (n=3390), district hospitals (n=2434), and primary care clinics (n=517). There were 133 LABA/LAMA FDC initiators and 820 LABA/ICS FDC initiators with missing information on hospital accreditation levels. Use of LABA/LAMA FDC was less common than use of LABA/ICS FDC at each hospital accreditation level. However, the differences in proportions of both FDC initiators in the settings of non-primary care clinics tended to be smaller than those in primary care clinics. For example, among 3390 FDC initiators in medical centers, 37% (n=1265) used LABA/LAMA FDC and 63% (n=1265) used LABA/ICS FDC. Nevertheless, among 517 FDC initiators in primary care clinics, only 9% (n=44) received LABA/LAMA FDC and up to 91% (n=473) received LABA/ICS FDC. The chi-square test indicated a significant difference in the type of FDC products across hospital accreditation levels (p-value <0.001).

Figure 3 Healthcare provider characteristics for use of LABA/LAMA FDC and LABA/ICS FDC. (A) Hospital accreditation levelsa. (B) Physician specialties. Abbreviations: FDC, fixed-dose combination; ICS, inhaled corticosteroids; LABA, long-acting beta2-agonists; LAMA, long-acting muscarinic antagonists. Notes: aThere were 133 LABA/LAMA FDC initiators and 820 LABA/ICS FDC initiators with missing information on hospital accreditation levels.

Most patients received FDC prescriptions from physicians with the chest specialty (n=9737), followed by internal medicine (n=1096), other specialties (n=1052; otorhinolaryngology and cardiology mainly), and family medicine (n=570). Overall, use of LABA/LAMA FDC was less frequent than use of LABA/ICS FDC in different physician specialties. However, it was noticed that, among patients who received FDC products from chest physicians (n=9737), 37% (n=3648) had LABA/LAMA FDC and 63% (n=6089) had LABA/ICS FDC. While among patients who received FDC products from family medicine physicians (n=570), there were only 8% (n=43) with LABA/LAMA FDC and up to 92% (n=527) with LABA/ICS FDC. There was a significant difference in the type of FDC products across physician specialties (Figure 3B, chi-square p-value <0.001).

Patient Characteristics Between LABA/LAMA FDC and LABA/ICS FDC Treatment Groups

Table 1 shows patient characteristics of both FDC initiators. LABA/LAMA FDC initiators tended to be older (71 vs 68 years) and to be male (86% vs 59%); were more likely to have any cancer or lung cancer and to have higher Charlson comorbidity scores; and were more likely to have outpatient visits due to COPD, hospital admissions due to COPD, and hospital admissions due to any reason. In contrast, LABA/ICS FDC initiators were more likely to have asthma (ie, asthma–COPD overlap syndrome). In the multivariate logistic regression analysis, we noticed that old age, male sex, any cancer, lung cancer, anxiety, as well as outpatient visits or hospital admissions due to COPD were independent predictors of initiating LABA/LAMA FDC, with ORs >1 and 95% CIs not including the null. In contrast, asthma was a strong, independent predictor of initiating LABA/ICS FDC (0.36; 95% CI, 0.32–0.40).

Table 1 Demographics, Comorbidities, and Medication Use of LABA/LAMA FDC and LABA/ICS FDC Initiators

Discussion

This population-based, real-world study showed evident temporal trends, variations in healthcare provider, and differences in patient characteristics among COPD patients who initiated LABA/LAMA FDC or LABA/ICS FDC, both of which are recommended alternative choices for those with severe disease. There was an apparent increased use of LABA/LAMA FDC but an obvious decreased use of LABA/ICS FDC since the availability of the first LABA/LAMA FDC product in Taiwan in 2015. LABA/LAMA FDC tended to be prescribed in the setting of non-primary care clinics and by physicians with the chest specialty compared to LABA/ICS FDC. LABA/LAMA FDC initiators appeared to be more vulnerable than LABA/ICS FDC initiators.

Comparison with Existing Studies

Temporal Trends for LABA/LAMA FDC or LABA/ICS FDC Use

Several researchers have depicted temporal trends of COPD maintenance medications. Although the timeframes ranged from 1 to 6 years across studies conducted in South Korea, France, and Canada, these studies generally indicated that use of LABA/LAMA combination therapy increased over time.^9–12 Other maintenance medications, including LABA/ICS combination therapy, LABA/ICS/LAMA combination therapy, LABA monotherapy, and LAMA monotherapy, may follow a decreased or stable temporal trend during the same observational windows (see Appendix Table 4 for detailed information).^9–12 Most of these studies, however, did not specifically classify LABA/LAMA combination therapy into either FDC or free combination forms.^9–11 One Canadian population-based study noticed that there was a rapid uptake of LABA/LAMA FDC (up to 283 users per 100,000 populations from 2015 to 2016) among elderly COPD patients treated with maintenance medications.¹² The uptake of LABA/LAMA FDC, nevertheless, had no profound impacts on the reduction of LABA/ICS FDC use within the same window.¹²

The present study targeted patients who initiated either LABA/LAMA FDC or LABA/ICS FDC. This facilitates figuring out treatment selection when physicians would like to prescribe FDC maintenance medications indicated for similar disease severity. We observed that LABA/LAMA FDC became more dominant during a 4-year observation window. Cumulative literature pointed out that LABA/LAMA FDC may be more efficacious while having less adverse reactions such as lower respiratory tract infections than LABA/ICS FDC.^21–24 Our results, to some extent, reflect that FDC prescription patterns in daily practice may adhere to the recommendations of available evidence.

Healthcare Provider Characteristics for LABA/LAMA FDC or LABA/ICS FDC Use

Although some studies have indicated that COPD treatment selection may differ across healthcare providers, limited studies examined the scenario of determining LABA/LAMA FDC or LABA/ICS FDC therapy among healthcare providers with different backgrounds and in different environments.^25–27 In a US health survey study, Mannino et al found that chest physicians may have a higher preference for LABA/LAMA FDC compared to primary care physicians.²⁷

Extending Mannino et al’s findings, the present study suggested that physicians who were chest physicians or who worked in the setting of non-primary care clinics (eg, medical centers) were more likely to prescribe LABA/LAMA FDC than those who were non-chest physicians (eg, family medicine physicians) or who stayed in primary care clinics. It is possible that physicians with a chest specialty or working in the setting of non-primary care clinics are more familiar with the suggestions of available evidence and the role of LABA/LAMA FDC. In addition, LABA/LAMA FDC are generally more expensive than LABA/ICS FDC and therefore more likely to be available in medical centers, regional hospitals, or district hospitals than in primary care clinics. These may partially explain the variations in use of both FDC therapies across various settings. These findings also facilitate the development of strategies to reduce inappropriate use of LABA/ICS FDC, if concerned and needed, for the target audience.

Patient Characteristics for LABA/LAMA FDC or LABA/ICS FDC Use

Some comparative effectiveness or safety studies of LABA/LAMA and LABA/ICS combination therapies (either FDC or free combination forms) noticed that baseline patient characteristics differed between treatment groups.^23,24,28,29 For example, in one study using a US commercial claims database, Samp et al showed that patients who started LABA/LAMA combination therapy were older, more likely to be male, had more comorbidities, including cardiovascular diseases, any cancer, lung cancer, mental health disorder, and pneumonia, and had more frequent resource utilization due to COPD compared to patients who started LABA/ICS combination therapy.²⁹

Similar to Samp et al’s results, our study also found that LABA/LAMA FDC initiators were generally more vulnerable than LABA/ICS FDC initiators. In the multivariate logistic regression analysis, we further showed that old age, male sex, several comorbidities (including any cancer, lung cancer, and anxiety), and outpatient visits or hospital admissions due to COPD were independent predictors of using LABA/LAMA FDC rather than LABA/ICS FDC. Although both FDC products are listed as alternative treatments in the clinical guideline, it is possible that physicians still prefer to prescribe LABA/LAMA FDC, a new product, for sicker patients. Our results also deserveattention in controlling for patient characteristics (ie, confounders) when undergoing comparative effectiveness and safety research for both FDC therapies in real-world settings.

Of interest, in terms of cardiovascular comorbidities, it seemed that the differences between treatment groups were more obvious in Samp et al’s study than in ours. Specifically, in Samp et al’s study, proportions of hypertension, congestive heart failure, heart disease, and arrhythmia between LABA/LAMA combination therapy and LABA/ICS combination therapy were 39% versus 34%, 12% versus 6%, 20% versus 10%, and 20% versus 11%, respectively.²⁹ Nevertheless, in our study, corresponding proportions for individual diseases between LABA/LAMA FDC and LABA/ICS FDC groups were 55% versus 55%, 15% versus 14%, 26% versus 23%, and 14% versus 13%, respectively. Given the pharmacological effects of LABA on β₂ receptors and LAMA on M3 receptors, there may be potential concerns regarding cardiovascular safety when prescribing LABA and LAMA, either monotherapy or combination therapy, for patients with cardiovascular comorbidities. Our study, however, demonstrated that underlying cardiovascular diseases may not be a substantial issue when deciding between LABA/LAMA FDC and LABA/ICS FDC treatment in Taiwan.

Unlike Samp et al’s study in which patients with asthma were excluded,²⁹ the present study retained this sub-population in the cohort and observed that asthma was a strong, independent predictor of initiating LABA/ICS FDC. This result was in line with the GOLD guideline suggestions that LABA/ICS FDC is preferred rather than LABA/LAMA FDC when patients have concomitant diagnoses of asthma and COPD.³

Strengths and Limitations of the Present Study

The present study was conducted using a population-based Taiwanese database derived from a single-payer health insurance system. The comprehensive information on demographics, diagnosis and procedure records, medication use, and resource utilization enhances the representativeness and validity of our results for COPD patients in an Asian country. We demonstrated treatment selection of either LABA/LAMA or LABA/ICS FDC products, both of which are indicated for similar COPD disease severity. The 4-year temporal trends, variations in healthcare provider characteristics, and differences in patient characteristics are useful to examine prescription patterns in real-world settings and, to some extent, to assess if the patterns cohere with the suggestions of the GOLD guideline.

However, we need to recognize that prescription patterns may be different in a longer timeframe. In addition, information on COPD classification (eg, COPD GOLD A–D groups), lung function test measures (eg, forced expiratory volume within one second, FEV1), and laboratory data (eg, eosinophil counts) is not available in the current claims database. Therefore, continual follow-up with more complete information on clinical parameters would require evaluation of the appropriateness of use of both FDC products in COPD patients in future studies.

Conclusion

This population-based study demonstrated 4-year temporal trends, variations in healthcare provider characteristics, and differences in patient characteristics for new use of LABA/LAMA FDC and LABA/ICS FDC in Asian COPD patients.