FormalPara Key Summary Points

Why carry out this study?

Chronic obstructive pulmonary disease (COPD) is a prevalent respiratory condition and is a leading cause of death worldwide

In patients previously treated with inhaled corticosteroid (ICS)/long-acting β2-agonist (LABA), there is a lack of real-world evidence on the occurrence of COPD exacerbations before and after initiation of single-inhaler triple therapy (SITT) with fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI)

These results support the Global Initiative for Chronic Obstructive Lung Disease 2023 recommendations that symptomatic patients with a history of frequent and/or severe exacerbations should be treated with a SITT

What was learned from the study?

In this real-world setting, we learned that patients who switched to FF/UMEC/VI after being treated with ICS/LABA had fewer COPD exacerbations and lower COPD-related healthcare resource utilization (HCRU) following the switch versus before

These observational study results build on previous clinical trial results demonstrating that SITT with FF/UMEC/VI can reduce exacerbations in patients with COPD stepping up from ICS/LABA dual therapy, while also reducing COPD-related HCRU

Introduction

Chronic obstructive pulmonary disease (COPD) is a prevalent respiratory condition estimated to affect 4.6–6.6% of American adults [1], and is a leading cause of death worldwide [2]. The economic and social burden of COPD is considerable, with annual total costs in the USA estimated to be $49 billion and rising [3]. Patients with COPD often have substantial comorbid diseases, placing additional burden on healthcare services and costs of care [3]. Treatments that improve the management of COPD and reduce the use of healthcare services can ultimately reduce healthcare expenditures and improve patient quality of life.

Previously, the combination of an inhaled corticosteroid (ICS) and a long-acting β2-agonist (LABA) was recommended for patients with moderate to very severe COPD and exacerbations [4], and ICS/LABA has become a common treatment option for patients with COPD who have exacerbations [5, 6]. However, following the publication of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2023 strategy document, the use of ICS/LABA is no longer encouraged in COPD, unless the patient is already being treated with stable ICS/LABA maintenance therapy [2].

The GOLD 2023 report states that if there is an indication for ICS, a triple therapy combination of ICS, LABA, and a long-acting muscarinic antagonist (LAMA) has been shown to be superior to ICS/LABA (improved lung function and patient-reported outcomes and reduced exacerbations) and is therefore the preferred treatment choice for patients with COPD at risk of exacerbations [2]. The single-inhaler triple therapy (SITT) Trelegy Ellipta (fluticasone furoate/umeclidinium/vilanterol [FF/UMEC/VI]) was approved by the US Food and Drug Administration (FDA) for once-daily maintenance treatment of patients with COPD in 2017 [7]. In the phase 3 randomized clinical trial FULFIL (Lung Function and Quality of Life Assessment in Chronic Obstructive Pulmonary Disease with Closed Triple Therapy), FF/UMEC/VI demonstrated statistically significant benefits—including reduced exacerbations—in patients with COPD compared with a twice-daily ICS/LABA combination of budesonide/formoterol (BUD/FORM) [8].

There is, however, limited real-world evidence on the occurrence of COPD exacerbations before and after initiation of FF/UMEC/VI in patients previously treated with ICS/LABA. Therefore, this study compared exacerbations and healthcare resource utilization (HCRU) in patients with COPD who used ICS/LABA—including a subgroup utilizing BUD/FORM—during the 12 months prior to initiation of FF/UMEC/VI. The outcomes were reported before and after initiation of FF/UMEC/VI.

Methods

Study Design and Objectives

This retrospective cohort study of patients enrolled in a commercial or Medicare Advantage with Part D (MAPD) health plan was performed using the US Optum administrative claims database for the study period of September 01, 2016, through March 31, 2020. Patients diagnosed with COPD who initiated FF/UMEC/VI treatment were included. The index date was the date of the first pharmacy claim for FF/UMEC/VI between September 01, 2017 (to coincide with FDA approval of FF/UMEC/VI), and March 31, 2019. The 12-month period before, including the index date, was used to assess pre-index (baseline) characteristics and study outcomes (Fig. 1). Study outcomes were also assessed during the 12 months following the index date (follow-up period). Outcomes were not measured beyond the 12-month follow-up period.

Fig. 1
figure 1

Study design. COPD chronic obstructive pulmonary disease, COVID-19 coronavirus disease 2019, FF/UMEC/VI fluticasone furoate/umeclidinium/vilanterol

Full size image

The primary objectives of this study were to compare the number of patients with COPD exacerbations, annualized exacerbation rate, and the all-cause and COPD-related HCRU, before and after initiation of FF/UMEC/VI among ICS/LABA patients. Baseline and follow-up comparisons were made among all ICS/LABA users (overall population) and among a subgroup who utilized BUD/FORM during baseline, to complement the FULFIL randomized controlled trial (RCT). The secondary objectives were to repeat the primary objectives among patients with ≥ 1 exacerbation during baseline.

This study complied with all applicable laws regarding subject privacy. As this was a retrospective analysis of existing data, no subject identifiers were included in these results and no direct subject contact or primary collection of individual human subject data took place. As per Title 45 of CFR, Part 46 (http://www.hhs.gov/ohrp/humansubjects/guidance/45cfr46.html#46.101), the data analysis of our study was exempt from institutional review board review. All study results are in tabular form and are presented as aggregate analyses that omit subject identification; therefore, informed consent and ethics committee or institutional review board approval were not required.

Eligibility Criteria

Patients included in the study were ≥ 35 years of age as of the year of the index date, had one or more pharmacy claims for FF/UMEC/VI between September 01, 2017, and March 31, 2019, and two or more medical claims with a diagnosis code for COPD in any position on separate dates during the baseline period. Patients also had to have 30 or more consecutive days of overlap in the days’ supply of ICS and LABA during the baseline period. Furthermore, patients must have been enrolled continuously in a commercial or MAPD health plan with medical and pharmacy coverage during both the 12-month baseline and follow-up periods.

Patients were excluded from the study if they had one or more medical claims with a diagnosis code for cystic fibrosis, interstitial lung disease, alpha-1 antitrypsin deficiency, or lung cancer, or two or more medical claims with a diagnosis code for asthma in any position on separate dates during the baseline period. Patients with one or more pharmacy claims for FF/UMEC/VI (SITT) or evidence of any other triple therapy (defined as 30 or more consecutive days of overlap in the supply of ICS, LABA, and LAMA [other SITT or multiple inhaler triple therapy]) during the baseline period were excluded from the study. Additionally, patients with evidence of one or more fill for biologic therapies for asthma (omalizumab, mepolizumab, reslizumab, benralizumab, tezepelumab, or dupilumab) during the baseline period were excluded from the study. Finally, patients with unknown information on their age, sex, geographic region, or insurance were not included.

Treatment Cohorts and Subgroups

The overall population included all patients with 30 or more consecutive days’ supply of any ICS and LABA dual therapy combination during the 12 months before starting FF/UMEC/VI. The BUD/FORM subgroup was a subset of the overall population and included patients with 30 or more consecutive days of supply of BUD/FORM during baseline. The baseline exacerbation subgroup included patients from the overall population with one or more exacerbations during the baseline period.

Study Outcomes

Outcome variables of COPD exacerbation (evidence of exacerbation in the 12-month period before and after starting FF/UMEC/VI treatment) and healthcare utilization (i.e., all-cause and COPD-related HCRU) were collected during baseline and follow-up periods. Among patients who had a COPD exacerbation during baseline, time to FF/UMEC/VI was calculated as the number of days from last COPD exacerbation (any, moderate, or severe) during baseline up to and including the index date. Time from FF/UMEC/VI treatment to the start of the first COPD exacerbation (any, moderate, or severe) was measured during the follow-up period. Moderate exacerbations were defined as those requiring a visit to the emergency room, physician’s office, or hospital as an outpatient with a primary diagnosis of COPD plus at least one claim for guideline recommended antibiotic or systemic corticosteroid within ± 5 days of visit. Severe exacerbations were classified as requiring hospitalization with a primary diagnosis of COPD. The definitions for moderate and severe exacerbations were based on previous validated literature [9].

Data Analysis

All study variables were analyzed descriptively. All analyses were based on an intent-to-treat approach, which does not take into account time-varying confounding factors and considers only the baseline characteristics of the patients [10]. For categorical variables, numbers and percentages were provided, and for continuous variables, means, medians, and standard deviations (SDs) were reported. Bivariate comparisons of COPD exacerbations, all-cause and COPD-related HCRU before and after starting FF/UMEC/VI treatment were performed. Mean exacerbations were compared using the McNemar test, and mean exacerbation episode lengths were compared using paired t tests. The length of an exacerbation episode was the service date of the first medical event (medical visit or pharmacy dispensing/administration) to the last observed exacerbation event date (or date of discharge if this was an inpatient event) plus 14 days. If two exacerbations occurred within 14 days of each other, they were combined into a single exacerbation.

Results

Baseline Demographic and Clinical Characteristics

A total of 1449 patients were included in the overall population. Patients had a mean age of 70.75 (SD 8.71) years, and more than half of the patients were female (54.18%), White (68.12%), and from the south (60.94%; Table 1). Most patients had MAPD insurance (88.13%). The mean (SD) Charlson comorbidity score (comorbidity burden) [12] at baseline was 2.33 (1.69), and the most common comorbidities were lower respiratory disease (other than COPD; 84.06%) and hypertension (80.47%). A total of 71.57% of patients had evidence of tobacco use. Baseline demographic and clinical characteristics for the BUD/FORM subgroup (N = 540) were similar to the overall population (Table 1).

Table 1 Baseline demographic and clinical characteristics for the overall population and the BUD/FORM subgroup
Full size table

COPD Exacerbations

Overall Population and BUD/FORM Subgroup

Significantly fewer patients experienced any exacerbation in the follow-up period versus baseline (53.49% vs 62.59%; p < 0.001 in the overall population and 55.00% vs 62.41%; p = 0.004 in the BUD/FORM subgroup; Table 2). The annualized rate of exacerbations was significantly lower during follow-up versus baseline for the overall population (1.11 vs 1.27; 12.60% decrease in patients with exacerbations; p < 0.001) and numerically lower for the BUD/FORM subgroup (1.21 vs 1.34; 9.70% decrease in patients with exacerbations; p = 0.066; Fig. 2).

Table 2 COPD exacerbations in the overall population and the BUD/FORM subgroup
Full size table
Fig. 2
figure 2

Count of COPD exacerbations in i the overall population and the BUD/FORM subgroup and in ii patients in the overall population and BUD/FORM subgroup who had ≥ 1 exacerbation during the baseline period. Baseline period includes claims on the index date. All patients are included in the comparisons of before and after the index date for overall comparisons. All patients with ≥ 1 baseline exacerbation are included in the comparisons of before and after the index date. Paired t test was used for continuous measures and McNemar test was used for binary measures. aExacerbations are classified according to the highest severity contributing event. Exacerbations within 14 days of each other are considered a single episode. Moderate exacerbations were defined as an emergency room, physician office, or hospital outpatient visit with a primary diagnosis code of COPD accompanied by an antibiotic or systemic corticosteroid dispensing within ± 5 days. Severe exacerbations were defined as a hospitalization with a primary diagnosis of COPD. BUD budesonide, COPD chronic obstructive pulmonary disease, FF/UMEC/VI fluticasone furoate/umeclidinium/vilanterol, FORM formoterol

Full size image

Significantly fewer patients experienced at least one moderate exacerbation in the follow-up period versus baseline (46.79% vs 54.87%; p < 0.001 in the overall population and 49.07% vs 56.11%; p = 0.009 in the BUD/FORM subgroup; Table 2). The annualized rate of moderate exacerbations was significantly lower during follow-up versus baseline for the overall population (0.88 vs 1.02; 13.73% decrease in patients with exacerbations; p < 0.001) and numerically lower for the BUD/FORM subgroup (0.96 vs 1.08; 11.11% decrease in patients with exacerbations; p = 0.072; Fig. 2).

Numerically fewer patients experienced at least one severe exacerbation in the follow-up period versus baseline for the overall population (17.32% vs 19.74%; p = 0.060) and the BUD/FORM subgroup (17.22% vs 20.19%; p = 0.161; Table 2). The annualized rate of severe exacerbations was similar during follow-up and baseline for the overall population (0.23 vs 0.25; p = 0.181) and the BUD/FORM subgroup (0.25 vs 0.26; p = 0.651; Fig. 2).

Figure 3 shows Kaplan–Meier curves for time from the start of the most recent COPD exacerbation during the baseline period and the time to first follow-up COPD exacerbation among patients initiating FF/UMEC/VI following ICS/LABA in the overall population and BUD/FORM subgroup. In the baseline period, 5.18% of patients in the overall population experienced an exacerbation < 1 month prior to index and 47.41% had an exacerbation ≤ 6 months prior to index. In the follow-up period, 6.83% of patients in the overall population experienced an exacerbation within 1 month of initiating treatment with FF/UMEC/VI and 36.58% had an exacerbation within 6 months of FF/UMEC/VI initiation.

Fig. 3
figure 3

Kaplan–Meier analysis for the overall population and BUD/FORM subgroup of i time to FF/UMEC/VI treatment from start of most recent COPD exacerbation (any severity) during the baseline period and ii time to the first COPD exacerbation during the follow-up period following start of FF/UMEC/VI treatment. BUD budesonide, COPD chronic obstructive pulmonary disease, FF/UMEC/VI fluticasone furoate/umeclidinium/vilanterol, FORM formoterol

Full size image

Patients with ≥ 1 Exacerbation During Baseline: Overall Population and the BUD/FORM Subgroup

Among 907 patients in the overall population who experienced any exacerbation during the baseline period, 563 (62.07%) experienced an exacerbation during follow-up (Table 3). Of the 907 patients, 337 were in the BUD/FORM subgroup; 218 of these (64.69%) experienced an exacerbation during follow-up. The annualized rate of exacerbations was significantly lower during follow-up versus baseline for both the overall population (1.40 vs 2.03; 31.03% decrease in patients with exacerbations; p < 0.001) and the BUD/FORM subgroup (1.54 vs 2.15; 28.37% decrease in patients with exacerbations; p < 0.001; Fig. 2). The total mean duration of exacerbation(s) was significantly shorter during follow-up compared with baseline for both the overall population and BUD/FORM subgroup (27.16 days vs 38.93 days; p < 0.001; and 30.07 days vs 41.01 days; p < 0.001, respectively; Table 3).

Table 3 COPD exacerbations in patients in the overall population and BUD/FORM subgroup who had ≥ 1 exacerbation during the baseline period
Full size table

Significantly fewer patients experienced at least one moderate exacerbation in the follow-up period versus baseline in the overall population (54.58% vs 87.65%; p < 0.001) and the BUD/FORM subgroup (57.57% vs 89.91%; p < 0.001; Table 3). The annualized rate of exacerbations was significantly lower during the follow-up period compared with baseline for both the overall population (1.11 vs 1.62; 31.48% decrease in patients with exacerbations; p < 0.001) and the BUD/FORM subgroup (1.22 vs 1.73; 29.48% decrease in patients with exacerbations; p < 0.001; Fig. 2). The total mean duration of moderate exacerbation(s) was significantly shorter during follow-up compared with baseline for the overall population and BUD/FORM subgroup (18.47 days vs 27.47 days; p < 0.001; and 20.37 days vs 29.16 days; p < 0.001, respectively; Table 3).

Significantly fewer patients experienced at least one severe exacerbation in the follow-up period versus baseline in the overall population (20.62% vs 31.53%; p < 0.001) and BUD/FORM subgroup (21.07% vs 32.34%; p < 0.001; Table 3). The annualized rate of exacerbations was significantly lower during follow-up versus baseline for both the overall population (0.29 vs 0.41; 29.27% decrease in patients with exacerbations; p < 0.001) and the BUD/FORM subgroup (0.32 vs 0.42; 23.81% decrease in patients with exacerbations; p = 0.021; Fig. 2). The total mean duration of severe exacerbation(s) was significantly shorter during follow-up compared with baseline in the overall population (8.70 days vs 11.46 days; p = 0.005) and numerically lower in the BUD/FORM subgroup (9.69 days vs 11.85 days; p = 0.172; Table 3).

HCRU

All-Cause HCRU

Overall Population and the BUD/FORM Subgroup

In the overall population, differences in all-cause HCRU counts were statistically significant during the follow-up compared to the baseline period for the number of mean all-cause outpatient visits (13.89 vs 11.98; p < 0.001) and mean all-cause emergency room visits (1.51 vs 1.31; p = 0.014) (Table S1). Similar results were seen in the BUD/FORM subgroup for mean all-cause outpatient visits (14.06 vs 12.64; p = 0.035) (Table S1). No numerical differences were observed in the proportion of patients with all-cause HCRU in the follow-up period compared with the baseline period for either the overall population or BUD/FORM subgroup (Fig. S1).

Patients with ≥ 1 Exacerbation During Baseline: Overall Population and the BUD/FORM Subgroup

Among patients with ≥ 1 baseline exacerbation, mean number of all-cause outpatient visits (15.50 vs 13.62; p = 0.002) during the follow-up period was significantly different compared with during baseline for the overall population (Table S2). Minimal numerical differences were observed for all-cause HCRU counts in the follow-up compared to the baseline period for the BUD/FORM subgroup (Table S2). Differences in the proportion of patients with all-cause outpatient visits (84.56% vs 87.87%; p = 0.010), all-cause emergency room visits (53.47% vs 60.86%; p < 0.001), and all-cause inpatient stays (33.08% vs 39.80%; p < 0.001) were statistically significant during the follow-up compared to the baseline period for the overall population (Fig. S1). Among the BUD/FORM subgroup, statistically significant differences in the proportion of patients with all-cause outpatient visits (83.38% vs 88.72%; p = 0.018) and all-cause emergency room visits (53.41% vs 62.91%; p = 0.005) were observed during follow-up versus baseline (Fig. S1).

COPD-Related HCRU

Overall Population and the BUD/FORM Subgroup

The difference in mean number of COPD-related office visits was statistically significant during the follow-up compared to the baseline period for both the overall population (2.81 vs 3.09; p < 0.001) and BUD/FORM subgroup (2.73 vs 3.09; p = 0.003; Table S1). The mean number of COPD-related outpatient visits was statistically different during the follow-up compared to the baseline period for the overall population (2.82 vs 2.29; p = 0.027; Table S1). A statistically significantly different proportion of patients had COPD-related office visits (78.40% vs 84.75%; p < 0.001) and emergency room visits (19.12% vs 22.22%; p = 0.018) during the follow-up compared to the baseline period for the overall population (Fig. 4). In the BUD/FORM subgroup, a statistically significant difference was observed in the proportion of patients who had COPD-related office visits (78.89% vs 84.63%; p = 0.003) during follow-up versus baseline (Fig. 4).

Fig. 4
figure 4

Proportion of patients with COPD-related HCRU in the i overall population and the BUD/FORM subgroup and in ii patients in the overall population and BUD/FORM subgroup who had ≥ 1 exacerbation during the baseline period. McNemar test was used for binary measures. p values for patients admitted to ICU and pharmacy use were not calculable; in pre–post studies, for a p value to be calculated the same set of patients needs to be in the denominator. BUD budesonide, COPD chronic obstructive pulmonary disease, ER emergency room, FORM formoterol, HCRU healthcare resource utilization, ICU intensive care unit

Full size image
Patients with ≥ 1 Exacerbation During Baseline: Overall Population and the BUD/FORM Subgroup

Among patients in the overall population who had ≥ 1 exacerbation during baseline, the mean number of COPD-related office visits (3.27 vs 3.75; p < 0.001), emergency room visits (0.54 vs 0.68; p = 0.011), inpatient stays (0.30 vs 0.37; p = 0.015), and pharmacy fills (13.77 vs 14.28; p = 0.038) was statistically different during the follow-up compared to the baseline period. For the BUD/FORM subgroup, only COPD-related office visits were statistically significant during follow-up versus baseline (3.23 vs 3.80; p < 0.001; Table S2). A statistically significant proportion of patients had COPD-related office visits (82.25% vs 89.97%; p < 0.001), outpatient visits (42.45% vs 48.18%; p = 0.004), emergency room visits (24.48% vs 33.52%; p < 0.001), and inpatient stays (19.51% vs 27.12%; p < 0.001) during the follow-up compared to the baseline period for the overall population (Fig. 4). Among the BUD/FORM subgroup, statistically significant differences were observed in the proportion of patients who had COPD-related outpatient visits (38.28% vs 49.55%; p < 0.001), emergency room visits (25.52% vs 36.20%; p < 0.001), and inpatient stays (20.18% vs 28.78%; p = 0.005; Fig. 4) during the follow-up compared to the baseline period.

Discussion

Using a longitudinal cohort design with claims-based data, our study compared COPD exacerbations and HCRU before and after initiation of FF/UMEC/VI SITT among a population of commercial healthcare insurance enrollees and MAPD beneficiaries diagnosed with COPD who had been treated with ≥ 30 consecutive days of ICS/LABA in the 12 months prior to FF/UMEC/VI initiation. The definitions of moderate and severe exacerbations used in this study were based on previous validated literature [9]. Overall, fewer exacerbations, of shorter duration, and lower COPD-related HCRU were observed during the 12-month follow-up period after initiation of FF/UMEC/VI compared with the 12-month ICS/LABA baseline period. Similar results were observed in a subgroup of patients receiving BUD/FORM during baseline, and among patients who experienced at least one exacerbation during baseline; the effect of FF/UMEC/VI initiation on reducing subsequent exacerbations was more pronounced among patients who experienced at least one exacerbation during the baseline period.

Real-world evidence on the occurrence of COPD exacerbations before and after initiation of SITT are lacking. A recent observational, noninterventional study in Germany demonstrated that patients with moderate-to-severe COPD with at least one exacerbation in the previous 12 months before switching to SITT (ICS/LAMA/LABA) from ICS/LABA (single or two inhalers) as part of routine clinical care experienced an improvement in health-related quality of life, lung function, and treatment adherence [13]. To our knowledge, there were no real-world studies evaluating the effectiveness of switching from dual therapy to SITT on reducing exacerbation frequency in US populations prior to the current study.

The results from the current effectiveness study using data from a routine clinical care setting are complementary to those seen in RCTs where SITT with FF/UMEC/VI was compared with dual therapy combinations. In the phase 3 randomized FULFIL trial in patients with advanced symptomatic COPD at risk of exacerbations, a reduction in the annualized rate of moderate/severe exacerbations at 24 weeks was observed in patients treated with FF/UMEC/VI versus BUD/FORM; a high proportion of patients had experienced at least one moderate/severe exacerbation in the prior year [8]. In our observational study, we also observed a reduction in the annualized rate of any (moderate/severe) exacerbation following the initiation of FF/UMEC/VI in the BUD/FORM subgroup; a high proportion of patients had experienced ≥ 1 exacerbation in the prior year and these patients experienced a reduction in any exacerbation as well. In FULFIL, a reduction in the annualized rate of severe exacerbations was observed at 24 weeks in patients treated with FF/UMEC/VI versus BUD/FORM with an exacerbation in the year prior to study entry [14]. In our study, we also observed a significant reduction in severe exacerbations in patients with ≥ 1 exacerbation during baseline following the initiation of FF/UMEC/VI in the BUD/FORM subgroup.

There are some differences between our real-world cohort study and the FULFIL RCT. Firstly, our study is a pre–post design, unlike the parallel control-based design used in RCTs. Secondly, patients enrolled in FULFIL without a prior exacerbation were required to have a forced expiratory volume in 1 s (FEV1) of ≤ 50%, which would indicate that the patient is at risk of a future exacerbation, in line with the GOLD 2017 guidelines in use at the time of the FULFIL study [15]; the population in our study were not selected by FEV1 value, but by using medical claims. However, as patients in our study had received a treatment escalation from ICS/LABA to FF/UMEC/VI, it could be presumed that these patients had experienced an increase in symptoms or exacerbations to warrant this escalation, based on clinical decision-making. Further to this, patients enrolled in FULFIL could have received any previous maintenance treatment, including triple therapy, which may indicate a more ill population than those included in our analysis. This may partly explain why the reduction in exacerbation rate observed in the BUD/FORM subgroup during follow-up was attenuated compared to that observed in FULFIL, since patients taking triple therapy during the baseline period would have been excluded from our study. Lastly, treatment adherence was assessed at each clinic visit for the population enrolled in the FULFIL study, with non-compliant patients discontinued, and only 90% of patients completing the study treatment. As an observational, real-world study, medication adherence was not required to be included in this study. Several studies have shown poor patient adherence to medication for COPD [16,17,18], and despite not monitoring adherence, which is assumed to be worse in real-world versus RCT settings, significant differences were still detected in the current analysis. Emulating a trial by conducting a study with a real-world evidence external control arm in parallel to the control/placebo group of the trial may be useful, and under such circumstances, the magnitude of the outcomes would be expected to be closer to those seen in the RCT. However, the results of our study may be a more accurate reflection of outcomes in a routine clinical care setting.

Reductions in moderate or severe exacerbations with step-up treatment from dual ICS/LABA therapy to SITT have also been reported in patients with symptomatic COPD treated with budesonide/glycopyrrolate/formoterol (BUD/GLY/FOR) in the phase 3 KRONOS trial, although the reduction with SITT was not statistically significant [19]. An additional phase 3 randomized trial that evaluated the efficacy and safety of triple therapy with BUD/GLY/FOR in patients with moderate to very severe COPD reported lower moderate/severe exacerbations in the BUD/GLY/FOR group in comparison to ICS/LABA [20].

While our study shows reduced COPD-related HCRU during the follow-up period compared with the baseline period, statistically higher all-cause HCRU utilization during the follow-up versus baseline was also reported. A possible explanation for this could be the progressive nature of the disease which can often worsen patients’ symptoms and exacerbations, which in turn increases the risk of other medical events, leading to increased all-cause HCRU and cost of care [21,22,23]. COPD is a multisystem disease and patients frequently have other comorbid conditions that require specific treatment, resulting in high resource use and healthcare costs [24]. Additionally, patients who switch to a new drug are also likely to be monitored more regularly, which could account for more office or outpatient visits during follow-up.

The results of this study should be viewed with the following limitations in mind. In an effort to exclude data during the coronavirus disease 2019 pandemic in the USA, the data included in this study were less contemporary and the study period ended on March 31, 2020. Additionally, this study population is likely to include patients who are moving to FF/UMEC/VI as a result of worsening of disease symptoms and are therefore likely to see a benefit; however, this population is thought to be representative of patients commonly seen in clinical practice who are appropriately escalated to FF/UMEC/VI on the basis of clinical decision-making, thus reducing the introduction of bias. Patients who died during the follow-up period (i.e., did not have 12 months' continuous insurance coverage) were excluded from the analysis, which may have introduced survivorship bias. Additionally, pharmacy claims were used as a surrogate for medication utilization. This could impact the results, as patients who had a pharmacy claim for FF/UMEC/VI may not have taken the medication as prescribed, or at all. Finally, all possible confounders in this study (especially clinical variables) will not have been captured by the claims data. Despite the limitations associated with real-world analyses, this effectiveness study captures treatment response from a real-world, heterogeneous population and reflects findings from the FULFIL clinical trial.

Conclusion

In a real-world setting, patients with COPD who initiated FF/UMEC/VI after receiving ICS/LABA treatment had fewer COPD exacerbations following initiation of FF/UMEC/VI versus before; this effect was more pronounced among patients with ≥ 1 baseline exacerbation. COPD-related HCRU was lower in the 12 months following FF/UMEC/VI initiation versus the 12 months prior to initiation. This retrospective, pre–post effectiveness study therefore adds to the evidence base of FF/UMEC/VI in patients with COPD in real-world settings and supports GOLD recommendations for SITT in symptomatic patients at risk of exacerbation. Findings from this study may guide treatment optimization in patients with COPD.