Pulmonary exacerbations in insured patients with bronchiectasis over 2 years

Background Patients with bronchiectasis experience persistent symptoms and frequent pulmonary exacerbations; this study investigated the frequency of exacerbations and all-cause hospitalisation. Methods This longitudinal, retrospective, claims database study (IBM® MarketScan®) identified patients aged ≥18 years from 1 July 2015 through 30 September 2018. Exacerbations were identified by bronchiectasis inpatient claim or a healthcare interaction, followed by antibiotic prescription within 7 days. Patients with ≥36 months of continuous health plan enrolment (12 months preceding the first bronchiectasis claim, i.e., baseline period and ≥24 months of follow-up) were included. Patients with cystic fibrosis at baseline were excluded. A multivariable logistic regression model identified baseline factors associated with having ≥2 exacerbations over the 2-year follow-up period. Results In total, 14 798 patients with bronchiectasis were identified; 64.5% were female, 82.7% were aged ≥55 years and 42.7% had ≥2 exacerbations at baseline. Having ≥2 exacerbations after 2 years was positively associated with chronic macrolide use, long-acting β2 agonist use, gastro-oesophageal reflux disease, heart failure and Pseudomonas aeruginosa. Frequent exacerbations (≥2) at baseline were significantly associated with greater likelihood of experiencing ≥2 exacerbations during the first and second year's follow-up (unadjusted odds ratios 3.35 (95% CI 3.1–3.6) and 2.96 (95% CI 2.8–3.2), respectively). The proportion of patients experiencing ≥1 all-cause hospitalisation cumulatively increased from 41.0% in the first year of follow-up to 51.1% over 2 years' follow-up. Conclusion Frequent exacerbations in patients with bronchiectasis may increase the likelihood of future exacerbations over 2 years of follow-up, with increased hospitalisation rates over time.


Introduction
The chronic respiratory condition bronchiectasis is a heterogeneous disease characterised by permanent dilation of the bronchi, persistent cough and recurrent respiratory infections [1][2][3]. Estimates of the prevalence of bronchiectasis vary but are reportedly increasing globally [4][5][6]. In 2013, prevalence in the USA was estimated to be 213 cases per 100 000 persons [5]. Most patients with bronchiectasis endure daily symptoms of cough and sputum production, and some experience episodes of worsening of these symptoms, typically called pulmonary exacerbations [1]. frequent (e.g., two to three or more) exacerbations each year [3]. However, patients continue to experience frequent pulmonary exacerbations, despite antibiotic treatment.
Pulmonary exacerbations are associated with a greater number of emergency department (ED) visits and increased rates of hospitalisation, leading to increased hospitalisation costs and considerable economic burden [1,3,[9][10][11][12]. A systematic literature review indicated that bronchiectasis has a significant economic burden driven by hospitalisation costs, particularly for patients with frequent pulmonary exacerbations [11]. However, these studies may have underestimated the frequency of pulmonary exacerbations and disease progression, as exacerbations were identified using International Classification of Diseases, Ninth Revision (ICD-9) codes only and evaluated pulmonary exacerbations over a 1-year follow-up period [9][10][11].
Therapeutic interventions that can prevent patients from experiencing exacerbations over time are of potential clinical importance for improvement in quality of life and the preservation of lung function, in addition to reducing the associated economic burden. An understanding of the factors associated with pulmonary exacerbations is also needed to identify patients who may be best suited to preventative therapies.
This longitudinal, retrospective, claims database study used an algorithm to define exacerbations based on inpatient or outpatient claims and antibiotics prescription. We evaluated pulmonary exacerbation frequency and all-cause hospitalisation frequency in patients with bronchiectasis over 1 and 2 years of follow-up, as well as baseline factors associated with having ⩾2 exacerbations.

Data collection
Data were derived from the IBM® MarketScan® Commercial Claims and Encounters and Medicare Supplemental Core databases (hereinafter, the "MarketScan database"), a large healthcare claims repository that captures medical and drug data from employers and health plans for over 24.5 million individuals [13]. Data from patients with Medicare are captured while they are still employed and using a commercial plan [13]. The Medicare data encompass employees, spouses and dependants who are covered by employer-sponsored private health insurance in the USA [13].

Study population
Patients with bronchiectasis ⩾18 years of age were identified in the MarketScan database from 1 July 2015 through 30 September 2018 (figure 1). Patients must have had ⩾36 months of continuous health plan enrolment, including 12 months preceding the first bronchiectasis claim (i.e., baseline period) and ⩾24 months of follow-up. Patients met the criteria of having bronchiectasis if they had one or more of the following: ⩾2 outpatient claims for bronchiectasis occurring ⩾30 days apart, high-resolution computed tomography scan followed by two subsequent claims for bronchiectasis (one claim occurring ⩽60 days after the computed tomography scan) or ⩾1 hospitalisation with a bronchiectasis diagnosis. Patients with a diagnosis of cystic fibrosis during the 1-year baseline were excluded.

Data analysis
Patient characteristics, comorbidities of interest and healthcare resource use were summarised descriptively. Demographics were measured on the index date starting from 1 July 2015. Other baseline characteristics were measured during the 1-year baseline period. Comorbidities were required to be on ⩾2 separate outpatient claims, >30 days apart, or a diagnosis on one inpatient hospitalisation claim during the 1-year baseline period. Chronic macrolide/antibiotic use was defined as having at least 90 total treated days. Repeat antibiotic use was defined as using the same antibiotic for 2 consecutive months. Patients met the criteria for having a pulmonary exacerbation if they had an inpatient claim with a bronchiectasis diagnosis, or a healthcare office visit with a bronchiectasis diagnosis followed by a prescription for macrolide, amoxicillin/clavulanic acid or fluoroquinolone antibiotics within 7 days. Pulmonary exacerbation events were not included as bronchiectasis-related events if the primary claim was otitis media, acute rhinosinusitis or COPD. The number of pulmonary exacerbations and all-cause hospitalisation frequency are reported categorically as a proportion of patients experiencing 0, 1, 2 or ⩾3 events over the first and second year of follow-up. First year of follow-up was the period between the index date and 12 months after. The second year was the period from the end of the first year through 12 months after. All-cause ED visits are reported as mean and median occurrences during the 1-year baseline period. For the purpose of this analysis, frequent exacerbations were defined as patients experiencing ⩾2 exacerbations per year, to establish a threshold for patients who experience multiple exacerbations compared with those who do not.

Statistical analyses
To determine the p-value of continuous variables, t-tests were used. The Pearson chi-squared test was used for binary and categorical variables, apart from lung transplant comorbidity, which used the Fisher exact test due to the small, expected sample size.
Unadjusted odds ratios (ORs) with 95% confidence intervals (CIs) were derived for the cross-tabulation of number of exacerbations (⩾2; yes versus no) during the study observation periods (baseline, follow-up year 1 and follow-up year 2). The OR was derived from the ratio of the probability of having ⩾2 exacerbations during the baseline period or year 1 of follow-up and the probability of exacerbations in the second year of A multivariable logistic regression model was fit to assess if selected baseline factors were significantly associated with having ⩾2 exacerbations in the 2-year period following the bronchiectasis index date.
Covariates were selected through a parsimonious analysis assessing statistical differences in baseline demographics and clinical characteristics, comparing those who had ⩾2 exacerbations versus those who had <2 exacerbations in the 2-year follow-up period. Input was then collected from expert clinicians to provide recommendations on defining covariates. For example, medications used to manage respiratory conditions were used as a proxy for respiratory comorbidities such as asthma and COPD, due to the availability of this information in the databases. Using drug claims as proxy minimised the impact of errant ICD-9/10 coding given these data are firstly for administrative purposes. Variance inflation factors (VIFs) were calculated among the final set of model covariates to assess multicollinearity. The specific covariates used in the final model included the following variables at baseline: age at index, sex, number of ED visits, exacerbations, nontuberculous mycobacterial (NTM) disease, chronic macrolide use, long-acting β2 agonist (LABA) use, inhaled steroid use, cough, dyspnoea, GERD, heart failure, haemoptysis and the presence of Pseudomonas aeruginosa. Outcome modifiers for the multivariable analysis were also identified as an effort to avoid bias.

Study population and patient characteristics at baseline
A total of 14 798 patients with bronchiectasis and 2 years of follow-up were identified (figure 1). Most patients were female (64.5%), ⩾55 years of age (82.8%) and eligible for Medicare supplemental insurance (54.0%) (table 1). Demographic characteristics were mostly well balanced, and no significant differences were noted in age or payer when patients were stratified by number of exacerbations over the entire 2-year follow-up period (table 1).
During the 1-year baseline period, 28.3% of the overall population had ⩾1 all-cause hospitalisation and 30.9% had ⩾1 all-cause ED visit. Number of hospitalisations, ED visits and use of interventions such as antibiotics and LABAs at baseline were significantly higher in patients with ⩾2 exacerbations compared with those with <2 exacerbations ( p<0.0001; table 3). Nearly half of the patients in the ⩾2 exacerbations group had a claim for a macrolide or LABA prescription, and 7.6% had multiple macrolide claims for ⩾3 consecutive months. Patients with ⩾2 exacerbations a year were significantly more likely to receive repeat antibiotic treatment (2 consecutive months; 18.3%) compared with patients with <2 exacerbations (5.2%, p<0.0001) (table 3).

Longitudinal trends in exacerbations and hospitalisations
The mean±SD number of pulmonary exacerbations in the overall population was 2.5±3.5 in year 1 and 3.9 ±5.4 in year 2. The overall proportion of patients experiencing ⩾1 exacerbation cumulatively increased from 67.4% in the first year of follow-up to 76.6% over the entire 2 years. Similarly, the overall proportion of patients experiencing ⩾3 exacerbations cumulatively increased from 32.4% in the first year of follow-up to 46.2% over the entire 2 years ( p<0.0001; figure 2). The overall proportion of patients who experienced no exacerbations during the baseline period through year 2 of follow-up was 14.7% (n=2178).
At baseline, 28.3% of the overall bronchiectasis population had ⩾1 all-cause hospitalisation. The overall proportion of patients experiencing ⩾1 hospitalisation cumulatively increased from 41.0% in the first year of follow-up to 51.1% over the 2-year follow-up period. In contrast, the overall proportion of patients experiencing ⩾3 all-cause hospitalisations cumulatively increased from 10.1% in the first year of follow-up to 17.8% over the 2-year follow-up period ( p<0.0001; figure 3).
Factors associated with ⩾2 exacerbations during follow-up Unadjusted ORs were calculated to understand the association between number of exacerbations at baseline and in the subsequent 2 years of follow-up. Patients with ⩾2 pulmonary exacerbations in the baseline period had 3.4 times greater odds (95% CI 3.1-3.6) of experiencing ⩾2 exacerbations during the first year of follow-up, and 3.0 times greater odds (95% CI 2.8-3.2) of ⩾2 exacerbations during the second year (table 4). Furthermore, patients with ⩾2 pulmonary exacerbations in the first year of follow-up had 3.6 times greater odds (95% CI 3.4-3.9) of experiencing ⩾2 exacerbations during the second year of follow-up (table 4).
The multivariable logistic regression model assessed baseline factors that were associated with having ⩾2 exacerbations in the 2 years of follow-up. Covariates, listed in table 5, were selected based on known impacts on bronchiectasis prognosis or number of exacerbations and statistical differences in baseline data between the two number-of-exacerbations groups. The multivariable model had a c-statistic value of 0.71, and a Pr(>Chi) value of <0.0001, demonstrating a strong goodness of fit. The following baseline variables were positively associated with having a greater probability of ⩾2 exacerbations in the follow-up period: number of baseline exacerbations, number of ED visits (association not significant), chronic macrolide use, LABA use, or having GERD, heart failure or P. aeruginosa (table 5). When adjusted, the odds of ⩾2 exacerbations during the follow-up increased by 3% for each additional ED visit at baseline, 31% for each additional exacerbation at baseline and 2.6 times for each patient receiving chronic macrolides at baseline (table 5). Minimum and maximum VIF values were 1.01 and 1.21, respectively, indicating very low correlation among selected independent variables (table 5).

Discussion
The results from this study demonstrate that in patients with bronchiectasis, frequent exacerbations (⩾2) at baseline were associated with more exacerbations, ED visits, all-cause hospitalisations, and both LABA and macrolide use after 2 years of follow-up. Additionally, higher rates of pulmonary exacerbations at baseline were associated with greater occurrence in the future (1-2 years of follow-up). The overall proportion of patients experiencing ⩾1 pulmonary exacerbation or hospitalisation increased from year 1 to year 2, with a minority of the overall population experiencing no exacerbations during the study period. These findings add to the evidence currently in the literature, due to the additional use of the IBM® MarketScan® commercial claims database, and an algorithm with unique definitions of bronchiectasis and pulmonary exacerbations.
Our findings are consistent with previous studies in both bronchiectasis and COPD that have similarly shown that patients with a greater number of exacerbations at baseline have a greater risk of future exacerbations and hospitalisations [11,12,14]. It is worth noting that the algorithm used in this study to identify exacerbations is consistent with studies on economic burden of exacerbations in COPD and cystic fibrosis [13,15]. For example, CHALMERS et al. (2018) [12] demonstrated that frequent exacerbations were the strongest predictor of future exacerbations in patients with bronchiectasis; incidence rate ratios for future exacerbations increased from 1.73 for patients with one baseline annual exacerbation, to 3.14 for patients with two baseline annual exacerbations. In a systematic literature review of resource use and costs associated with bronchiectasis, mean annual baseline hospitalisation rates of 0.3-1.3 were reported, which is consistent with the mean 0.58 baseline annual hospitalisations in the present study [11]. Furthermore, significant increases in ED visits and physician office visits in patients with frequent exacerbations were found [11]. A higher Bronchiectasis Severity Index score is also indicative of a greater number of exacerbations and is associated with a greater risk of mortality [8].
Factors that are associated with patients with bronchiectasis having ⩾2 pulmonary exacerbations include having a greater number of ED visits, having a greater number of exacerbations at baseline, chronic macrolide use, LABA use, having GERD and the presence of P. aeruginosa. Furthermore, most of the target population experienced ⩾2 exacerbations over the 3 years of study, with very few experiencing none. Therefore, it is likely that at least one of these characteristics will impact patients with bronchiectasis. Future studies may consider these key patient characteristics when deciding on clinical trial inclusion criteria, since patients with frequent exacerbations may benefit the most from treatment aimed to reduce exacerbations.
As macrolides are the recommended treatment for patients with frequent exacerbations, this may account for the finding that higher macrolide use at baseline is associated with increased number of exacerbations [3]. These results are not necessarily indicative of a causal relationship, since macrolides are used to treat exacerbations and all claims for macrolides were used in the analyses and were not separated by the indication.
P. aeruginosa commonly colonise the airways of individuals with bronchiectasis [16]. The finding that the presence of P. aeruginosa at baseline is related to increased exacerbations is consistent with data collected from a prospective study by CHAWLA et al. (2015) [16] which showed that patients with both bronchiectasis and P. aeruginosa colonies had a higher number of exacerbations. Additional research is required to identify the mechanisms of P. aeruginosa that may account for this association. In contrast, NTM was not associated with increased exacerbations; this may be because patients with NTM at baseline are already receiving active therapies to treat infections, which overlap with the treatment of both acute and frequent exacerbations. However, it is important to note that this study neither directly investigated whether patients with NTM were being treated nor accounted for potential patients with NTM who were not listed in the claims databases.
Similarly, causal relationships cannot be inferred between incidence of exacerbations, and the other variables of the multivariate model: having a greater number of both ED visits and exacerbations at baseline, having GERD and LABA use. Incidentally, however, previous research has also associated such variables with having a higher number of exacerbations [11,12,[17][18][19][20][21]. Of particular interest, comorbidities with asthma and COPD, which are proxied by LABA use, have the most established Data are presented as n (%), mean±SD or median (interquartile range). IQR: interquartile range; ED: emergency department; LABA: long-acting β2 agonist. # : repeat is defined as the same antibiotic for 2 consecutive months; ¶ : chronic was defined as having at least 90 total treated days in the time period.
relationship with increased exacerbations in the literature [18,19,21]. However, it should be noted that in contrast to our findings, an earlier prospective observational study in a Spanish cohort by MENÉNDEZ et al.
(2017) [20] showed that the number of previous exacerbations was not significantly associated with future hospitalisations. This emphasises the importance of further research into the associations of these variables with having ⩾2 pulmonary exacerbations to better understand the factors that lead to frequent exacerbations, which could improve our understanding of ways to prevent them.
A potential limitation of this study is that Medicare data may not be generalisable for all patients ⩾65 years of age because the majority of data are only captured for those who are still employed. Additionally, inherent to claims data research, the clinical accuracy of the coding could not be assessed, and symptoms related to exacerbations are not captured. In addition, patients with health maintenance organisation or full or partial capitated point-of-service insurance coverage were excluded from this study because financial information for this population was incomplete. Furthermore, small-and medium-sized insurance firms may be underrepresented in the sample, as most data come from larger employers. No information was available for disease severity and intervention use outside of hospitals. Data not captured in claims, such as smoking status, disease severity and the presence of respiratory comorbidities, are a potential source of confounding known to claims data. Moreover, findings were based on retrospective claims data, which   Data are presented as n, n (%) or unadjusted OR (95% CI). OR: odds ratio; CI: confidence interval. have inherent limitations. Lastly, causal relationships could not be derived from the associations found using the multivariate model developed in this study.
Strengths of these data and method employed include that COPD events and pulmonary exacerbations were separated, ensuring that only pulmonary exacerbations were included. Additionally, unlike previous studies using claims databases [5,11,12], the present study did not use ICD-9/10 codes alone to identify individuals with bronchiectasis and pulmonary exacerbations. Instead, an algorithm with unique definitions of bronchiectasis and pulmonary exacerbations was developed that may better reflect the true frequency of exacerbations [11]. Furthermore, previous studies have only used the Medicare database to capture outpatient claims, whereas this study also uses the IBM® MarketScan® claims database to capture additional data from a commercially insured population that otherwise have not been characterised [13,[22][23][24]. This study was also consistent with how studies on the economic burden of exacerbations have been operationalised in COPD and cystic fibrosis [15,25]. Moreover, the latest data available in the USA were utilised to perform a detailed characterisation of people with bronchiectasis. Lastly, potential confounding variables were accounted for by using proxies to categorise respiratory comorbidities, such as utilising prescription paid claims for LABA to categorise COPD and asthma rather than relying on ICD-9/10 codes alone.

Conclusions
Overall, our findings improve our understanding of characteristics that may increase the number of exacerbations in patients with bronchiectasis, provide insights that could support the design of future clinical trials and highlight the need to address the prevention of exacerbations as part of bronchiectasis management. The longitudinal trends of increasing pulmonary exacerbations and hospitalisation rates observed over the 2-year follow-up period suggest that frequent exacerbations in patients with bronchiectasis may increase the likelihood of future exacerbations. In future work it would be interesting to assess the overlap between factors associated with ⩾2 exacerbations at year 1 and year 2, to determine if additional factors may influence progression of the disease.