Increased hospitalizations and economic burden in COPD with bronchiectasis: a nationwide representative study

With the increasing use of computed tomography, bronchiectasis has become a common finding in patients with chronic obstructive pulmonary disease (COPD). However, the clinical aspects and medical utilization of COPD with bronchiectasis (BE) remain unclear. We aimed to investigate the BE effect on prognosis and medical utilization in patients with COPD. Among 263,747 COPD patients, we excluded patients lacking chest X-ray, CT, or pulmonary function test codes and classified 2583 GOLD-C/D patients matched according to age, sex, and medical aid as having COPD-BE (447 [17.3%]) and COPD without BE (2136 [82.7%]). Patients with COPD-BE showed a higher rate of acute exacerbation requiring antibiotics than those without BE. Moreover, multivariable analysis showed that BE co-existence was a crucial factor for moderate-to-severe exacerbation (incidence rate ratio [IRR] 1.071; 95% CI 1.012–1.134; p = 0.019). Patients with COPD-BE had a significantly higher rate of exacerbations requiring antibiotics, as well as treatment cost and duration (meant as number of days using hospitalization plus outpatient appointment), than those with COPD without BE (52.64 ± 65.29 vs. 40.19 ± 50.02 days, p < 0.001; 5984.08 ± 8316.96 vs. 4453.40 ± 7291.03 USD, p < 0.001). Compared with patients with COPD without BE, patients with COPD-BE experienced more exacerbations requiring antibiotics, more hospitalizations, and a higher medical cost.

Chronic obstructive pulmonary disease (COPD) is characterized by airflow limitation and is comprised of a complex disease group with varying effects and pathophysiologies. In patients with COPD, the bronchiectatic feature of airway wall thickening and dilatation is positively correlated with ageing and airflow obstruction severity 1,2 . With the increased use of computed tomography (CT), bronchiectasis (BE) has become a common finding among patients with COPD 3,4 . Moreover, a recent study using national patient sample data reported that 19.3% of Korean patients with BE presented with COPD as a comorbidity 5 .
It has been found that the prevalence of BE with COPD (COPD-BE) ranges from 25.6 to 69% 4,6-11 , with patients with COPD-BE presenting more exacerbations and worse lung function than those without BE. Further, a study on the Canadian Cohort of Obstructive Lung Disease (CanCOLD) reported that the BE prevalence in patients with mild-to-moderate and severe COPD was 14.1-22.2% and 35.1%, respectively 12 . In the CanCOLD cohort, COPD-BE was not associated with an increased risk of acute exacerbation; however, it was associated with dyspnoea and other respiratory symptoms. Other small-scale studies on COPD-BE have reported a wide prevalence range and varying characteristics with respect to exacerbation episodes and respiratory symptoms 4,12,13 . There is a need for a nationwide large-scale study to elucidate the clinical aspects, prognosis, and medical utilization in patients with COPD-BE.
Therefore, this nation-wide representative study aimed to investigate BE-associated outcomes and medical utilization in patients with COPD. www.nature.com/scientificreports/

Results
Baseline characteristics of study population. Among 263,747 patients with COPD, we excluded 243,247 participants who did not undergo chest X-ray/CT examination or a PFT and 16,754 participants without ≥ 2 exacerbations or ≥ 1 hospitalization within previous 1 year. Consequently, we enrolled the remaining 3746 GOLD C/D patients. Among them, 447 patients with COPD-BE underwent 1:5 PS matching according to age, sex, and medical aid while 2136 patients with COPD without BE were matched (Fig. 1). After PS matching, we analysed 2583 patients with GOLD C/D COPD. There was no significant between-group difference in age, sex, and medical aid. Exacerbation related clinical outcomes.   www.nature.com/scientificreports/ gender (male), medical aid, visits to a tertiary hospital, and medications were significant factors (Table 5). Multivariable analysis of medical utilization revealed that co-existing BE was a significant factor in increasing medical use days (β = 7.304, p = 0.006) ( Table 6).

Discussion
This study investigated the effect of comorbid BE on clinical outcomes and medical utilization in patients with COPD. Our study yielded two main findings. First, compared with the COPD without BE group, the COPD-BE group showed has an increased risk of moderate-to-severe exacerbation and severe exacerbation requiring ICU admission. Second, the cost of medical utilization was significantly higher in patients with COPD with BE compared with patients without BE. An emerging approach for COPD treatment is the determination of specific phenotypes. BE features in radiologic imaging is among the most prominent phenotypes and is frequently identified using the diagnostic assessment of chest CT scans. Previous studies have reported a COPD-BE prevalence widely ranging from 25.6 to 69% 4 ; moreover, COPD-BE has revealed frequent exacerbations and poor prognoses 4,7,14 . However, most of these were small-scale studies with others not investigating the significance of COPD exacerbation with BE 7 . To our knowledge, this is the first study to analyse the relationship between COPD and BE using large-scale national claim data. The COPD-BE group showed an increased rate of moderate-to-severe exacerbation, especially severe exacerbation requiring hospitalizations and ICU admission. Moreover, BE coexistence was a significant risk factor for moderate-to-severe exacerbations.
Acute COPD exacerbation is generally caused by bacterial infections 15 . Ni et al. 4 reported that the structural abnormalities of BE in patients with COPD can result in chronic colonization with potentially pathogenic micro-organisms, which causes recurrent infection, frequent exacerbations, and poorer prognosis. This can be prevented by the administration of effective antibiotics as additional treatment in patients with COPD-BE 16,17 . In Table 1. Baseline characteristics of study populations. Data are expressed as mean ± standard deviation or N (%). Propensity score matching was performed in this study population, considering age, sex, and medical aid covering patients with low socioeconomic status. BE bronchiectasis, Chest CT chest computed tomography, PFT pulmonary function test. *Patients with low socioeconomic status were covered by medical aid in Korea. www.nature.com/scientificreports/  www.nature.com/scientificreports/ our study, moderate and severe exacerbation events resulted in a significant increase in the rate of oral antibiotic prescription, which was consistent with previous findings 11 . The present study investigated access to BE-related medical costs and clinical outcomes. A matched-cohort study on US health-care claim data analysed health care utilization in patients with COPD-BE 18 and found that COPD-BE treatment involved greater medical resources (USD 15,685 [14,678]) than only treating COPD (USD 6262 [5655-6868]). These expenses were related to acute care management or the prescription of antibiotics, steroids, or bronchodilators at outpatient visits. In our study, we monitored medications, including ICS, bronchodilators, and oral medications, with the COPD-BE group showing a higher prescription than the COPD without BE group. Additionally, the percentage of prescribed systemic steroids or antibiotics was high in patients with hospitalizations, including ER visits and hospitalizations requiring ICU care. BE contributed to higher medical costs, particularly related to hospitalization. Additionally, factors associated with increased medical costs were tertiary hospital visits, as well as the prescription of ICS, bronchodilators, and mucolytic medications.
In our study, there was a significant occurrence of anaemia and GERD in the COPD-BE group. A previous study reported that the GERD prevalence was higher in the BE group, especially when accompanied with nontuberculous mycobacterial (NTM) disease 19 . Although the causal relationship remains unclear, one explainable hypothesis is that recurrent acid reflux irritates the airway, which causes NTM lung infection that leads to BE. Furthermore, in patients with COPD-BE, anaemia might result from systemic inflammation or erythropoiesis 20,21 . The anaemia prevalence is much higher in patients with COPD-BE due to the synergic effect of systemic inflammation.
This study has several limitations. First, COPD is diagnosed as having a postbronchodilator FEV 1 /FVC < 0.7 in the spirometry test 15 . However, this was a national cohort study using claim data and we could not determine the COPD diagnosis through lung function results. To make up for this limitation, the study population was defined using the operational definitions (ICD-10 code for COPD plus COPD drug prescription history), which were validated from previous studies using Korean National Health Insurance data [22][23][24][25][26][27][28][29][30][31][32][33][34] . Moreover, the inclusion www.nature.com/scientificreports/ of patients who underwent PFT and radiological examinations, including chest X-ray/CT, allowed more precise definition of the COPD and BE diagnoses. Previous studies have indicated that BE was more common among elderly patients with COPD with lower FEV 1 4,35 . Since we could not determine each individual's FEV 1 , we could not adjust the values. In case the COPD-BE group had lower FEV 1 values than the COPD without BE group, we could not determine the causal effect of BE coexistence on increased exacerbation since the effect could have resulted from lower FEV1 values rather than BE coexistence in the COPD-BE group. Nonetheless, we attempted to adjust disease severity by including only GOLD C/D patients with COPD with exacerbations confirmed by ≥ 2 prescriptions of systemic steroids or antibiotics over one year. In the additional sensitivity analysis, even after including all patients with COPD, the COPD-BE group showed higher exacerbations and was an independent risk factor in multivariate analysis (data not shown). Second, potentially pathogenic micro-organisms can cause recurrent pulmonary infection 4,11 . Our study used Korean claim data; therefore, we could not determine the colonized pathogen in the COPD-BE group. Third, regression analysis revealed that any ICS or mucolytic use was associated with exacerbations. However, given that this was an observational study with cross-sectional analysis of 1-year insurance database rather than a cohort study with long-term follow-up, the ICS or mucolytics could have been prescribed more in patients with frequent exacerbations regardless of BE coexistence, which represents as a reverse causality issue. Traditionally, clinicians prescribe ICS and mucolytics for patients with a high exacerbation risk and bronchitis, respectively.
In conclusion, this study showed that patients with COPD-BE experienced a higher frequency of moderateto-severe exacerbations than those with COPD only. Specifically, the COPD-BE group showed a higher rate of exacerbations requiring hospitalization, including ER visits and ICU care, which led to a high hospitalization rate with subsequently increased medical costs. To increase the credibility of the recruited group, we identified GOLD C/D patients who underwent the pulmonary function test (PFT). Moreover, we performed propensity-score (PS) matching between the patients with COPD with and without BE according to age, sex, and medical aid covering patients with low socioeconomic status. BE was defined as at least one claim under the J47 ICD-10 code (except E84 [cystic fibrosis]) with the inclusion of only patients who underwent chest X-ray or chest CT examination to ensure BE diagnosis accuracy. The patients with COPD were allocated to the COPD-BE group and COPD without BE group. Additionally, comorbidities were defined using the following ICD-10 diagnosis codes: ischemic heart disease [angina pectoris (I20.x), myocardial infarction (I21.x, I22.x, or I25. Clinical outcomes and medical utilization measurement. In this study, the primary outcome was moderate-to-severe exacerbation. Moderate exacerbation was defined as an outpatient clinic visit with ICD-10 code for COPD (J43.x-J44.x, except J430) and prescription of systemic steroids and/or antibiotics. Severe exacerbation was defined as exacerbation requiring hospitalization or emergency room (ER) visits and prescription of systemic steroids and/or antibiotics. Additionally, severe exacerbation with a worsening condition resulting in admission to the Intensive care unit (ICU) was considered as severe exacerbation with ICU care.

Study design and study population.
We analyzed medical utilization and cost based on the 2012 HIRA database with the exclusion of medical utilization or costs not associated with COPD. We excluded patients with > 1 reimbursement per year for cancer (ICD-10 code: C00.x-C97.x), renal failure (N17.x-N19.x), and/or cerebrovascular disease (I60.x-I69.x) given the difficulty in distinguishing the significant expenses reimbursed for treatment of these diseases from those of COPD-related medical services. Regarding medication utilization and costs, we only analyzed COPDrelated medications (LAMA, LABA, LABA + LAMA, ICS, ICS + LABA, SAMA, SABA, theophylline, LTRA, SABA + SAMA, PDE4-I, systemic steroids, systemic beta-agonist, mucolytics, and antitussive agent).
Regarding outpatient services, we only analyzed visits with a principal or secondary (within the fourth position) COPD diagnosis (J43.x-J44.x, except J430). For inpatient services, the analysis was limited to admission when the principal or secondary (within the fourth position) diagnosis was COPD (J43.x-J44.x, except J430) or a COPD-related disease (pneumonia: J12.x-J17.x; pulmonary thromboembolism: I26, I26.0, and I26.9; dyspnea: R06.0; or acute respiratory distress syndrome: J80). We collected data regarding age, sex, co-morbidity, hospitalizations, ER visits, ICU admissions, medical costs, and medication use. Further, we separately analysed the number of days that a particular outpatient and inpatient service was used ("used days"). All costs were presented in US dollars (USD) with an exchange rate of 1 USD for 1100 Korean won (exchange rate based on April 2018). This study was approved by the Institutional Review Board of Seoul National University Bundang Hospital