Mortality risk and causes of death in patients with non-cystic fibrosis bronchiectasis

This retrospective cohort study included participants aged ≥20 years who underwent chest computed tomography (CT) at the Seoul National University Hospital between January 1, 2005 and December 31, 2016. The diagnosis of bronchiectasis was determined based on the CT report, which was assessed by radiologists. Index date was the time when these patients underwent chest CT. We excluded those regarded to have previously undergone lung resection surgery (such as pneumonectomy, lobectomy, segmentectomy, wedge resection, metastasectomy, or video-assisted thoracoscopic surgery [VATS] lung biopsy) on the chest CT or those with significant non-respiratory (malignancy, liver cirrhosis, chronic kidney disease, heart failure, acquired immune deficiency syndrome, connective tissue disease, organ transplantation) or respiratory diseases (interstitial lung disease, pneumoconiosis, pleurisy, empyema, sarcoidosis, kyphosis, or tuberculous-destroyed lung) which may affect the prognosis. The identification of comorbidities or exclusion criteria was based on ICD-10 codes registered within two years before and after the index date. Those with traction bronchiectasis were also excluded. To ensure appropriate selection of individuals in the non-bronchiectasis control group, we applied the following additional exclusion criteria: 1) isolation of nontuberculous mycobacteria (NTM) from a respiratory specimen at least once; 2) a forced expiratory volume in one second (FEV₁)/forced vital capacity (FVC) ratio < 0.7 or FEV₁ < 80% predicted value; or 3) other abnormalities on CT such as pneumonia, active pulmonary tuberculosis, pneumothorax, pleural effusion, pericardial effusion, pulmonary embolism, embolization in coronary artery, aorta or bronchial artery, pleural calcification, or haemoptysis. In the bronchiectasis group, patients were divided into two categories according to the isolation of NTM, the isolation of Pseudomonas, smoking status, and airflow limitation defined as a FEV₁/FVC of less than 0.7. During the study period, patients with any evidence of NTM or Pseudomonas were classified into NTM or Pseudomonas subgroups. This study was approved by the Institutional Review Board of the Seoul National University Hospital (IRB No. H-1607-037-774) and conformed to the tenets of the Declaration of Helsinki. Informed consent was waived due to its retrospective nature.

Baseline demographic data, comorbidities, smoking history, sputum culture results, and lung function results were collected by using data based on electronic medical records (EMRs). Student’s t-test or analysis of variance was used for between-group comparisons of demographics involving continuous variables, and the chi-square test for those involving categorical variables. Mortality and cause of death before 2016 December 31 was determined based on the mortality data from the Korean National Statistical Office. Cox proportional hazard regression analysis was used to present adjusted hazard ratios (aHR) and their 95% confidence intervals (CI). In addition, the proportion of mortality risks attributable to risk factors (population attributable risk fraction, PAF) and their 95% CIs were estimated. P-values for multiplicative interactions, as well as relative excess risk due to interaction (RERI) and attributable proportion (AP) for additive interactions were evaluated, along with their 95% CIs. P-values less than 0.05 were considered statistically significant. Data analyses were conducted using Stata14.2 (Stata Corp, College station, Texas).

We initially identified 217,702 individuals who underwent a chest CT scan during the study period. Among those, 18,134 participants with bronchiectasis and 90,313 non-bronchiectatic control participants were included (Fig. 1). The baseline demographics of the subjects are shown in Table 1. The bronchiectasis group was older, had more females, and had a high proportion of comorbidities and poorer lung function.

In total, 1653 (9.12%) deaths in the bronchiectasis group and 714 (0.79%) deaths in the control group were observed during 5.9 ± 3.4 and 5.4 ± 3.5 years of mean follow up period, respectively. The all-cause mortality rate in the bronchiectasis group was 1608.8 per 100,000 person-years (95% CI, 1531.5–1690.0) which was higher than that of the control group (133.5 per 100,000 person-years; 95% CI, 124.1–143.8) (P < 0.001) (Fig. 2). Multivariable cox proportional hazard regression showed that bronchiectasis remained independently associated with a higher all-cause mortality risk (aHR, 1.22; 95% CI, 1.02–1.46) compared with the control group even after adjustment for age, sex, BMI, smoking history, presence of diabetes mellitus, hypertension, and dyslipidaemia, and baseline FEV₁(pred. %) (Table 2). In addition, bronchiectasis and smoking per se were accountable for 9.3% (population attributable fraction (PAF), 0.093; 95% CI, 0.012–0.167) and 9.2% (PAF, 0.092; 95% CI, 0.029–0.151) of all-cause mortality, respectively. Among participants with bronchiectasis, smoking was associated with 14.0% (PAF, 0.140; 95% CI 0.042–0.228) of all-cause mortality, while in the control group, the percentage attributed to smoking was 5.7% (PAF, 0.057; 95% CI, − 0.025-0.0133). We found multiplicative (P for interaction, 0.030) and additive interaction (RERI, 0.432; 95% CI, 0.097–0.769; AP, 0.274, 95% CI, 0.080–0.469) between the two factors, regarding the risk of all-cause mortality. Patients with both bronchiectasis and smoking history had the highest risk of all-cause mortality among four subgroups, categorized according to bronchiectasis and smoking history (aHRs on all-cause mortality (95% CI); 1.07 (0.86–1.34), 1.07 (0.87–1.32), 1.58 (1.25–1.99) for ever-smokers in the control group, never-smoker bronchiectasis patients and ever-smoker bronchiectasis patients, respectively) (Additional file 3).

The major causes of death in the bronchiectasis group were malignancy (31.2%), respiratory related (30.6%), neurological (9.0%), and cardiovascular death (7.1%). In the control group, malignancy (52.8%), cardiovascular (9.7%), neurological (9.2%), and suicide (9.1%) were major causes of death, in that order. Respiratory related deaths made up a small percentage (4.9%) of the total deaths in the control group (Table 3). Cox regression analysis showed bronchiectasis was significantly associated with increased respiratory death (aHR, 3.49; 95%; CI, 2.21–5.51) and lung cancer-related death (aHR, 3.36; 95% CI, 2.18–5.18) (Fig. 3). An additive interaction (RERI, 8.68, 95% CI, 1.631–15.736; AP, 0.586, 95% CI, 0.398–0.775) was observed between bronchiectasis and smoking, concerning the risk of lung cancer-related mortality. Bronchiectasis was not significantly related to other causes except for other cancer-related deaths compared with the control group.

Among 6957 bronchiectasis patients whose respiratory specimen was tested for acid-fast bacteria (AFB) at least once during the study period, NTM was isolated in 1740 (25%). Additionally, among 4000 bronchiectasis patients whose respiratory specimen was tested for bacteria at least once during the study period, Pseudomonas was isolated in 378 (9.5%). Among patients with available lung function data, 2493 (24.3%) exhibited airflow limitations and among patients with available smoking history, 2859 (36.3%) were ever-smokers. According to multivariate Cox regression analysis, which was performed to evaluate the factors related to increased mortality in patients with bronchiectasis, airflow limitation and a history of smoking were significantly associated with increased mortality (aHR, 1.39; 95% CI, 1.15–1.67; aHR, 1.39; 95% CI, 1.11–1.75) (Fig. 4, Additional file 1 and Additional file 2).