Patient Characteristics

In total, 5421 patients were included: prompt initiators, n = 2057; delayed initiators, n = 3364 (Fig. 2). After weighting, baseline demographic and clinical characteristics were generally well balanced between the two cohorts (std. diff. < 10%; Table 1). Baseline demographic and clinical characteristics for the unweighted cohorts are listed in Supplementary Table S1. In each weighted cohort, mean age was 62.1 years and approximately 50% of patients were female. The type of index COPD exacerbation was similar between the two cohorts: 76.0% and 75.7% of prompt and delayed initiators, respectively, had a moderate index exacerbation. The most frequently prescribed respiratory medications during baseline were systemic corticosteroids (prompt cohort, 62.0%; delayed cohort, 62.3%) and short-acting β2-agonists (SABAs; prompt cohort, 59.6%; delayed cohort, 60.0%). The mean Quan-CCI score was 1.9 in each weighted cohort. The most common Elixhauser comorbidities were hypertension (64.4%), cardiac arrhythmias (23.9%), and diabetes (20.8%) (Supplementary Table S2).

Fig. 2

Patient attrition. aModerate COPD exacerbations: an outpatient or ED visit with a COPD exacerbation diagnosis code in the primary position and ≥ 1 dispensing/administration of a systemic corticosteroid or guideline-recommended antibiotic within 5 days before or after the visit. bSevere COPD exacerbations: hospitalization with a diagnosis code for COPD exacerbation in the primary position. cApproval date of FF/UMEC/VI in the USA. dContinuous eligibility: continuous health plan enrollment with medical and pharmacy coverage. eFF/UMEC/VI recognized using National Drug Codes: 0173-0887-10, 0173-0887-14, and 0173-0887-61. fInitiation of single-inhaler FF/UMEC/VI within 30 days following the index date. gInitiation of single-inhaler FF/UMEC/VI between 31 and 180 days following the index date. COPD chronic obstructive pulmonary disease, ED emergency department, FF/UMEC/VI fluticasone furoate, umeclidinium, and vilanterol

Table 1 Baseline demographics and clinical characteristics of the overall population and the PS-weighted patient cohortsaRespiratory Medication Use During Follow-up, Prior to Initiation of FF/UMEC/VI

The mean (SD) time from the index date to FF/UMEC/VI initiation was 15.7 (9.3) days in the prompt cohort and 97.6 (43.3) days in the delayed cohort (Supplementary Table S3). During follow-up, the most frequently used respiratory medications prior to FF/UMEC/VI initiation were systemic corticosteroids (prompt cohort, 40.3%; delayed cohort, 75.1%), SABAs (prompt cohort, 21.1%; delayed cohort, 53.4%), ICS/LABA dual therapy (prompt cohort, 6.5%; delayed cohort, 32.2%), and SABA/short-acting muscarinic antagonist dual therapy (prompt cohort, 7.9%; delayed cohort, 19.2%); 1.4% of patients in the prompt cohort and 10.2% of patients in the delayed cohort received multiple-inhaler triple therapy (MITT) prior to initiating FF/UMEC/VI (Supplementary Table S3).

Rate of COPD Exacerbations

During the follow-up period, prompt initiators of FF/UMEC/VI had significantly lower rates of overall, moderate, and severe COPD exacerbations versus delayed initiators; this was the case for up to 12 months of follow-up (Fig. 3), and for the entire follow-up period (mean 1.7 years) (Supplementary Fig. S1). For up to 12 months of follow-up, the rate of overall COPD exacerbations was 0.74 PPY in the prompt cohort versus 1.06 PPY in the delayed cohort (RR 0.70, 95% CI 0.64–0.77; P < 0.001), representing a 30% lower rate of overall exacerbations (Fig. 3).

Fig. 3

Rates of COPD exacerbations in the weighted prompt and delayed cohorts (analysis up to 12 months of follow-up). Prompt and delayed initiators were weighted using the inverse probability of treatment weighting approach based on the PS. Covariates used in the PS calculation included: age, sex, year and quarter of index date, US region, type of insurance plan (i.e., PPO, HMO, POS, or other), Quan-CCI score (categories of 0, 1, 2, and 3+), asthma diagnosis, type of COPD exacerbation on the index date, patients with multiple events at the index exacerbation, respiratory medication use, all-cause and COPD-related HCRU and medical costs (hospitalization, ED, and outpatient constituents), and comorbidities (with ≥ 5% prevalence in either cohort). aRRs were calculated from Poisson regression models with log-link. bCIs and P values were calculated using non-parametric bootstrap procedures with 999 replications. cModerate COPD exacerbations: an outpatient or ED visit with a COPD diagnosis code in the primary position and ≥ 1 dispensing/administration of a systemic corticosteroid or guideline-recommended antibiotic within 5 days before or after the visit. dSevere COPD exacerations: hospitalization with a diagnosis for COPD exacerbation in the primary position. CI confidence interval, COPD chronic obstructive pulmonary disease, ED emergency department, HMO health maintenance organization, HCRU healthcare resource utilization, POS point of service, PPO preferred provider organization, PPY per person-year, PS propensity score, Quan-CCI Quan-Charlson comorbidity index, RR rate ratio

For the entire follow-up period, the rate of overall COPD exacerbations was 0.67 PPY in the prompt cohort versus 0.90 PPY in the delayed cohort (RR 0.75, 95% CI 0.69–0.81; P < 0.001), representing a 25% lower rate of overall exacerbations (Supplementary Fig. S1). The rates of moderate and severe exacerbations were 26–30% and 22–30% lower, respectively, in the prompt versus delayed cohort.

Time to First COPD Exacerbation

At 12 months post-index, patients in the prompt versus delayed weighted cohort had significantly lower Kaplan–Meier rates of overall exacerbations (HR 0.68, 95% CI 0.62–0.73; P < 0.001), moderate exacerbations (HR 0.67, 95% CI 0.62–0.74; P < 0.001), and severe exacerbations (HR 0.71, 95% CI 0.61–0.83; P < 0.001) (Fig. 4). The median time to first COPD exacerbation was longer for patients in the prompt versus delayed weighted cohort for overall exacerbations (499 vs. 267 days) and moderate exacerbations (727 vs. 385 days) (Fig. 4).

Fig. 4

Time to first COPD exacerbation for a overall exacerbations, b moderate exacerbations, and c severe exacerbations (weighted analysis). Prompt and delayed initiators were weighted using the inverse probability of treatment weighting approach based on the PS. Covariates used in the PS calculation included: age, sex, year and quarter of index date, US region, type of insurance plan (i.e., PPO, HMO, POS, or other), Quan-CCI score (categories of 0, 1, 2, and 3+), asthma diagnosis, type of COPD exacerbation at the index date, patients with multiple events on the index exacerbation, respiratory medication use, all-cause and COPD-related HCRU and medical costs (hospitalization, ED, and outpatient constituents), and comorbidities (with ≥ 5% prevalence in either cohort). aHazard ratios, including CIs and P values, were calculated using Cox proportional hazards models. bThe number of patients observed at each timepoint. CI confidence interval, COPD chronic obstructive pulmonary disease, ED emergency department, HMO health maintenance organization, HCRU healthcare resource utilization, PPO preferred provider organization, POS point of service, PS propensity score, Quan-CCI Quan-Charlson comorbidity index

Association Between Time to FF/UMEC/VI Initiation and Rate of COPD Exacerbations

During the first 12 months post-index, a 1-day delay in FF/UMEC/VI initiation was associated with a 0.31% increase in the rate of overall, moderate, and severe exacerbations (HR 0.31, 95% CI 0.23–0.38), which translates to an approximately 2% increase in exacerbations for each week FF/UMEC/VI initiation is delayed (Table 2). Results for the overall observation period are shown in Supplementary Table S4.

Table 2 Rates of COPD-related exacerbations: continuous time to FF/UMEC/VI initiation (analysis up to 12 months of follow-up)

Kaplan–Meier estimates show similar results up to 12 months post-index, with a 1-day delay in FF/UMEC/VI initiation associated with a 0.32%, 0.32%, and 0.33% increase in the risk of experiencing an overall, moderate, and severe exacerbation, respectively (Table 3).

Table 3 Time to first COPD-related exacerbation: continuous time to FF/UMEC/VIAll-Cause and COPD-Related Medical Costs

Over a mean of 10.8 months post-index (up to 12 months’ follow-up analysis), total all-cause healthcare costs PPPY were trending lower but the difference did not reach statistical significance for patients in the prompt versus delayed weighted cohort ($34,666 vs. $37,233; cost difference $2567; P = 0.160) (Table 4). Conversely, total COPD-related healthcare costs PPPY were significantly lower for patients in the prompt versus delayed weighted cohort ($8483 vs. $10,587; cost difference $2104; P = 0.010) (Table 4). Cost differences were primarily driven by lower COPD-related medical costs ($4454 vs. $6449; cost difference $1996; P = 0.012) with hospitalization cost differences being the main component ($3473 vs. $5075; cost difference $1602; P = 0.038).

Table 4 Healthcare costs in the weighted prompt and delayed cohorts (analysis up to 12 months of follow-up)

For the overall follow-up period analysis (mean of 20.4 months), total all-cause healthcare costs PPPY were not significantly different for patients in the prompt versus delayed weighted cohort ($31,793 vs. $33,277; cost difference $1484; P = 0.350); however, pharmacy costs were significantly lower in the prompt versus delayed cohort ($9690 vs. $10,667; cost difference $977; P = 0.048) (Supplementary Table S5). Total COPD-related healthcare costs were not significantly different for patients in the prompt versus delayed weighted cohort ($8245 vs. $9225 PPPY; cost difference $981; P = 0.236); however, ED visit costs ($214 vs. $282 PPPY; cost difference $67; P = 0.030), outpatient visit costs ($621 vs. $804 PPPY; cost difference $183; P = 0.024), and pharmacy costs ($3887 vs. $4139 PPPY; cost difference $252; P = 0.036) were all significantly lower in the prompt versus delayed cohort (Supplementary Table S5).

During the first 12 months post-index, a 1-day delay in FF/UMEC/VI initiation was associated with a $14.1 increase in total COPD-related medical costs, which was largely driven by COPD-related hospitalization costs (Table 5). During the overall follow-up period, only increases in COPD-related ED and outpatient visit costs were statistically significantly lower for an additional 1-day delay in FF/UMEC/VI initiation (Supplementary Table S6).

Table 5 Healthcare costs: continuous time to FF/UMEC/VI (analysis up to 12 months of follow-up)Time to First Hospital Readmission

A total of 494 patients (24.0%) in the prompt cohort and 818 patients (24.3%) in the delayed cohort (weighted analysis) had a severe COPD exacerbation at index. At 90 days post-index, Kaplan–Meier rates of all-cause and COPD-related readmissions were statistically significantly lower in the prompt versus delayed cohort (Fig. 5). Specifically, patients in the prompt cohort had a 38% lower risk of all-cause readmission (HR 0.62, 95% CI 0.45–0.86; P = 0.004) and a 42% lower risk of COPD-related hospital readmission (HR 0.58, 95% CI 0.35–0.98; P = 0.042) versus patients in the delayed cohort (Fig. 5).

Fig. 5

Time to first a all-cause and b COPD-related hospital readmission (weighted analysis). Prompt and delayed initiators were weighted using the inverse probability of treatment weighting approach based on the PS. Covariates used in the PS calculation included: age, sex, year and quarter of index date, US region, type of insurance plan (i.e., PPO, HMO, POS, or other), Quan-CCI score (categories of 0, 1, 2, and 3+), asthma diagnosis, type of COPD exacerbation on the index date, patients with multiple events at the index exacerbation, respiratory medication use, all-cause and COPD-related HCRU and medical costs (hospitalization, ED, and outpatient constituents), and comorbidities (with ≥ 5% prevalence in either cohort). aHazard ratios, including CIs and P values, were calculated using Cox proportional hazards models. bThe number of patients observed at each timepoint. CI confidence interval, COPD chronic obstructive pulmonary disease, ED emergency department, HMO health maintenance organization, HCRU healthcare resource utilization, PPO preferred provider organization, POS point of service, PS propensity score, Quan-CCI Quan-Charlson comorbidity index

At 90 days post-index, among patients with a hospitalization on the index date, a 1-day delay in FF/UMEC/VI initiation was associated with a 0.33% higher risk of an all-cause hospital readmission (Table 6). There was a trend for a lower (0.29%) risk of a COPD-related hospital readmission for every 1-day delay in FF/UMEC/VI initiation, but the difference did not reach statistical significance (Table 6).

Table 6 Time to first hospital readmission: continuous time to FF/UMEC/VI