Article Text
PDF
Abstract
Background The life trajectory of chronic obstructive pulmonary disease (COPD) remains unknown.
Patients and methods We collected data from two populations. In the first cohort, we recruited 375 patients with COPD from our hospital, and 1440 repeated assessments of quality of life (QoL) using the European Quality of Life-5 Dimensions questionnaire from 2006 to 2020. We analysed their dynamic changes using the kernel-smoothing method. The second cohort comprised 27 437 patients from the National Health Insurance (NHI) dataset with their first severe acute exacerbations (AEs) requiring hospitalisation from 2008 to 2017 were analysed for their long-term course of AEs. We employed a Cox hazard model to analyse the predictors for mortality or AEs.
Results Cohorts from our hospital and NHI were male predominant (93.6 and 83.5%, respectively). After the first severe AE, the course generally comprised three phases. The first was a 1-year period of elevated QoL, followed by a 2-year prolonged stable phase with a slowly declining QoL. After the second AE, the final phase was characterised by a rapid decline in QoL. For NHI cohort, 2712 died during the 11-year follow-up, the frequency of the first AE was approximately 5 per 10 000 per day. The median time from the first to the second AE was 3 years, which decreased to less than 6 and 3 months from 4th to 5th and 8th to 9th AE, respectively. The frequency of AE was increased 10-fold and 15-fold and risk of subsequent AE was increased 12-fold and 20-fold after the 6th and the 10th AE, relative to the first. Male gender, heart failure comorbidities were associated with the risk of subsequent AE and death.
Conclusions The life trajectory of COPD includes the accelerated frailty phase, as well as elevated health and prolonged stable phase after the first AE.
- COPD epidemiology
- COPD Exacerbations
- Emphysema
Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The life trajectory of chronic obstructive pulmonary disease (COPD) has traditionally been illustrated as a continuous accelerated decay in health status known as the accelerated frailty course. However, the life trajectory of COPD in the real world remains unknown.
WHAT THIS STUDY ADDS
COPD has three phases posthospitalisation: first, 1 year of elevated health and increased health-related quality of life (HRQoL); second, a 2-year stable phase with declining HRQoL; third, after the second hospitalisation for COPD, a 4-year rapid decline in HRQoL with severe acute exacerbations (SAEs) until death or severe disability. The median time from the first to the second SAE was 3 years, which decreased to less than 6 months from the fourth to the fifth SAE, and to 3 months from the eighth to the ninth SAE.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The life trajectory of COPD includes the accelerated frailty phase, as well as two other phases: elevated health and prolonged stable phase after the first SAE. To understand the life trajectory of COPD can help clinic physician to early intervention and support the evidence of life trajectory to share decision-making with patients and their family.
Introduction
Chronic obstructive pulmonary disease (COPD) is common and the leading cause of severe deterioration in health-related quality of life (HRQoL).1 2 The life trajectory of COPD has traditionally been illustrated as a continuous accelerated decay in health status known as the accelerated frailty phase.3 4 The deterioration of health status accompanied by frequent acute exacerbation (AE) and hospitalisation is thought to be an accelerated frailty phase for the natural course of patients with COPD. Patients with COPD could approach the end of life owing to disease exacerbation or comorbidities 2–5 years after entering this phase.3 Suissa et al observed that patients with COPD, after their first AE requiring hospitalisation, that is, severe AE (SAE), have 5 years of stable disease between their first and second AE, called the prolonged stable phase.5 However, the life trajectory of patients with COPD hospitalised for the first time, combined with the natural course of severe exacerbations and dynamic changes in HRQoL in reality, is unknown.
To understand the life trajectory of COPD and investigate factors influencing HRQoL a long-term and continuous study design to explore the dynamic change in HRQoL in COPD is necessary, but no attention has been given before. The European Quality of Life-5 Dimensions (EQ-5D) questionnaire is a generic health status questionnaire widely used in health and economic evaluation. The EQ-5D is also a valid and responsive measure of health status in COPD.1 To our knowledge, no study has used the EQ-5D to demonstrate the dynamic change in QoL in patients with COPD. This study aimed to explore the life trajectory of COPD, the natural course of subsequent severe exacerbations was analysed using the National Health Insurance (NHI) dataset, and dynamic changes in HRQoL evaluated by the EQ-5D after their first hospitalisation for AE or SAE by using the cohort from National Cheng Kung University Hospital (NCKUH). In addition, the factors influencing dynamic changes in HRQoL during long-term follow-up were evaluated.
Material and methods
Study design and population
NCKUH cohort for the dynamic change in HRQoL analysis
A total of 375 patients diagnosed with COPD in the outpatient clinic of pulmonary medicine at the NCKUH between January 2006 and December 2020 were recruited (online supplemental figure 1a). All patients received regular medical treatment for COPD for at least 3 months before recruitment. COPD cases and SAE were defined according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) diagnostic guidelines and criteria.6 7 Participants were classified according to the GOLD 2006 classification into four stages (mild to very severe) by pulmonary function test, which corresponds to the GOLD 2017 grades 1–4, based on the postbronchodilator forced expiratory volume (FEV1).6 7 In this study, we pooled all participants with an FEV1 of <50% to the ‘severe’ stage to obtain a sufficient sample for estimation.
Supplemental material
NHI cohort for long-term course of AE analysis
In Taiwan, more than 99% of residents are enrolled in the NHI system, which covers healthcare costs and all medical claims collected in the NHI dataset. We applied the criteria of the first SAE to select patients with COPD from the NHI reimbursement data. SAE was defined as an AE requiring hospitalisation according to the GOLD guidelines.6 All patients with an admission diagnosis of COPD (main diagnosis International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes: 491, 492, 493.2, 496; ICD-10: J40, J41, J42, J43 and J44) who were first hospitalised between January 2008 and December 2017 owing to AE were included (n=93 639).
Next, we excluded patients aged less than 40 years (n=1293) and >80 years (n=35 765) at the first hospitalisation for COPD. To ensure the validity of the SAE, we excluded patients who were admitted for less than 3 days (n=4284) and those whose sex-related data were missing (n=62) and included only those who returned to the clinic more than twice with the main diagnosis of COPD (including emergency room visits) after leaving the hospital. In total, 27 437 patients with COPD with a history of SAE were included and followed up until death or the end of the study (31 December 2018) (online supplemental figure 1b).
Unequal distribution of sex with male predominance in both NCKUH and NHI cohorts was expected according to our previous studies.8
Patient and public involvement
Patients were not involved in the development of study design or recruitment of participants.
Dynamic change in HRQoL
The HRQoL of patients with COPD was consistently monitored using the validated Taiwanese version of the three-level EQ-5D (EQ-5D-3L).9 The EQ-5D-3L questionnaire was administered by the COPD case manager at the outpatient department. The dynamic change in QoL was evaluated using EQ-5D utility and the kernel smoothing method, with a locally weighted scatterplot smoothing curve fitted for the utility score of the EQ-5D-3L questionnaire (R software V.4.2.1, Vienna, Austria.)
Statistical analyses
Patient characteristics for the two cohorts are presented as descriptive statistics, with mean±SD for continuous variables and sample size percentage for categorical variables. For statistical comparisons among groups, continuous variables were analysed using the t-test, and categorical variables were analysed using the χ2 test.
For assessing the natural course of subsequent severe exacerbations, we retrieved frequency and time of every SAE and date of death from the NHI cohort. The survival probability of patients was estimated using the Kaplan-Meier method. Cox proportional hazard models were applied to estimate the HR to assess how each severe exacerbation affected the subsequent exacerbation and mortality. The Cox model adjustment variables included age, sex, whether cohort entry was before or after 2012, and all comorbidities, as well as time-dependent variables, including the number of exacerbations and times. The estimation of the overall hazard function of successive exacerbations included two steps. First, the median time from each exacerbation to the next or death was estimated using the Kaplan-Meier method. Next, the life table approach was used to estimate the hazard rate of the next severe exacerbation for each day, forming the hazard function. The hazard functions of each successive severe exacerbation were then combined with the corresponding median interexacerbation times. A smoothing spline was applied to the hazard function to visualise the patterns. The results of the Cox models are presented as HRs and associated 95% CIs. Each significance test was two sided, with the significance level set at 0.05.
In the NCKUH cohort, HRQoL was evaluated according to EQ-5D utility score. To evaluate dynamic change of EQ-5D over time, linear regression using kernel smoothing method, with a locally weighted scatterplot smoothing curve fitted for the utility score of the EQ-5D-3L questionnaire was applied. When analysing the factors influencing the utility value of the QoL with EQ-5D, we employed a repeated-measures-mixed-effects model to account for random effects of the same subject measured at different times during the longitudinal follow-up. The fixed-effect variables included in the construction of the models were as follows: pulmonary function test, gender, the presence of previous SAE, the presence of comorbidities, educational level, age and time from previous SAE. All analyses were performed using R software (V.4.0.2; R Foundation for Statistical Computing, Vienna, Austria).
Results
NCKUH cohort
A total of 375 patients were enrolled, with 1440 repeated assessments of their EQ-5D scores and utility values from the NCKUH during the 15-year follow-up. The mean age was 62.3 years, and 93.6% were men; coronary artery disease (CAD) and heart failure were the most common comorbidities. Patients who had SAE had a higher ratio of heart failure (19.3%) and a higher ratio of having received triple therapy (30.2%) than those who had no SAE (online supplemental table 1). 26 patients with SAEs used triple therapy, all of them had checked their eosinophils before using triple therapy, 16 of them had eosinophil count of >300 cells/µL (16/26, 61.5%) and 22 (22/26, 84.6%) had eosinophils counts of ≥100 cells/µL. More patients with SAEs used triple therapy than those without SAEs (30.2% vs 10.5%), and similar results were found for SAEs patients with early decline of HRQoL than those without early declined (33.3% vs 13.1%) (online supplemental table 2).
NHI cohort
The NHI cohort included 27 437 patients hospitalised at first AE, of whom 12 712 (46.3%) died during the 11-year follow-up. At cohort entry from NHI, the patients had a mean age of 68.7 years, 83.5% were men, and CAD and heart failure were the major comorbidities, with 4863 (17.7%) and 3383 (12.3%) prevalence, respectively. Among them, 9711 (35.4%) had received at least one long-acting bronchodilator, whereas 2060 (7.5%) had received triple therapy after their discharge from the first COPD hospitalisation (table 1).
Demographic data for COPD with severe AE (required hospitalisation) from NHI cohort
Most patients in the NCKUH and NHI cohorts were male, having major comorbidities such as CAD and heart failure.
Dynamic change of HRQoL evaluated with the EQ-5D utility (NCKUH cohort)
After the first hospitalisation for COPD exacerbation, these patients had gradual increases in HRQoL for approximately 1 year, followed by a period of slow and minimal decrease for approximately 2 years. Patients then entered the accelerated frailty phase, which was accompanied by rapid HRQoL deterioration (figure 1). The same situation was observed in patients without AEs during the initial enrolment period, and they had distinctly better HRQoL than those with SAEs (figure 1). Most patients increased or keep constant of their QoL in the first year after SAE (figure 1). However, 19.5% (24/123) patients presented early decline of QoL in the first 2 years, the characterisations of these patients were having more severe symptoms of dyspnoea with higher mMRC dyspnoea scores (3 vs 2), higher ratio of using triple therapy (33.3% vs 13.1%) and lower ratio of them receiving pulmonary rehabilitation (79.2% vs 95.0%) as compared with those without early decline of QoL (online supplemental table 2).
Dynamic changes of utility in patients with COPD with and without severe acute exacerbations history. COPD, chronic obstructive pulmonary disease; QoL, quality of life; SAE, severe acute exacerbation.
Using a mixed effects model, an FEV1<50% of the predicted value and SAE history were independent factors associated with the dynamic change in HRQoL with declined utility after long-term follow-up (online supplemental table 3). Using kernel smoothing method, we also found that an accelerated decline in HRQoL with steepest descent occurred in patients combined with FEV1<50% of the predicted value and SAE history after a 3-year follow-up than the other two groups (no SAE and SAE). (online supplemental figure 2). In addition, comorbidities of chronic kidney disease (CKD) were factors associated with decreased utility. Moreover, having a college education was related to higher utility (online supplemental table 3).
Natural course of subsequent severe exacerbations (NHI cohort)
Online supplemental figure 3 shows the cumulative mortality over the 11-year follow-up period.
Figure 2A shows the hazard function for the time from the first SAE to the subsequent SAEs. The interval between exacerbations decreased from the first to the 12th SAE. The median time from the first to the second SAE was 3 years, which decreased to less than 6 months from the fourth to fifth SAE and 3 months from the eighth to ninth exacerbations (table 2). Figure 2B also reveals a similar and shorter time interval between exacerbation and death. The time from the first to the second exacerbation or death was approximately 2 years. For patients suffering more than one episode of exacerbation, the risk for subsequent exacerbations increased after each exacerbation. The baseline rate of SAE increased with every new AE, increasing from a rate of approximately 5 per 10 000 per day for the first AE to 50 per 10 000 per day for the sixth AE and 75 per 10 000 per day for the 10th AE (figure 2A).
Hazard function of successive hospitalised COPD exacerbations (per 10 000 per day) from the time of their first hospitalisation for exacerbation over the follow-up period, with the time between successive exacerbations estimated using: (A) the median interexacerbation times, conditional on survival with death as a competing risk and (B) the median interexacerbation times as the time to the next exacerbation or death, whichever occurs first. COPD, chronic obstructive pulmonary disease.
Crude and adjusted HR of a subsequent exacerbation and death associated with each successive exacerbation and mortality, as a function of the time since the first exacerbation, estimated by the Cox proportional hazards model with time-dependent covariate defining the exacerbation number and its timing
Table 2 shows that the risk of SAE gradually increased as the number of episodes increased. Compared with the first exacerbation, the risk of the second episode increased by 2.66-fold and increased by 24.26-fold after the twelfth occurrence. The risk of death was similar between groups. Compared with the first exacerbation, the risk of death in the 2nd and the 12th exacerbations increased 1.95-fold and 5.11-fold, respectively.
Online supplemental table 4 shows the impact of age, sex, calendar time and comorbidities on the risk of subsequent exacerbations and death. The risk of death increased with age, whereas the risk of exacerbation decreased. Men had a higher risk of exacerbation or death than women. In contrast, patients with their first exacerbation after 2012 had a lower risk of subsequent exacerbations or death than those with exacerbations before 2012. Patients with heart failure comorbidities have an increased risk of exacerbation and mortality. Heart failure was the only comorbidity associated with SAE and mortality. Both stroke and CKD were associated with an increased mortality risk but decreased AE risk. CAD was the only comorbidity associated with decreased mortality. Younger and male patients had more subsequent exacerbations (online supplemental table 5).
Life trajectory of COPD (NCKUH and NHI cohorts)
We integrated figures 1 and 2A to show the natural course of subsequent severe exacerbations after the first SAE from NHI cohort and the dynamic change in HRQoL for patients with COPD from the NCKUH cohort (online supplemental figure 4). Although, the patient sources for these two cohorts were different, we modified it to plot the life trajectory of patients with COPD (figure 3). There was an elevated HRQoL (increase in EQ-5D utility) for nearly 1 year. Subsequently, the patient entered the prolonged stable phase for approximately 2 years, wherein their condition was relatively stable with a lower risk of SAE requiring hospitalisation. However, QOL began to decline slowly at the same time. The stages mentioned above totaled 3 years until the second SAE, which can serve as the start of the accelerated frailty phase, characterised by frequent exacerbations and hospitalisations, accompanied by rapidly deteriorating HRQoL (figure 3).
The life trajectory of COPD. Figures 1 and 2 were combined to show the trend of HRQoL decrease in every acute exacerbation. COPD, chronic obstructive pulmonary disease; HRQoL, health-related quality of life.
Discussion
The life trajectory of COPD has traditionally been illustrated as a continuous accelerated decay in health status, accompanied by frequent hospital admissions and exacerbations over time. However, the accelerated frailty phase and two other phases, namely elevated health and prolonged stable phase, occur after the first hospitalisation exacerbation. In our study, patients with COPD presented with an increasing HRQoL in the elevated health phase for approximately 1 year. Subsequently, their condition was relatively stable with slowly decreasing HRQoL and without AE after standard medical therapy for approximately 2 years until the second SAE. We found SAE history and an FEV1<50% of the predicted value were predictors of dynamic change in HRQoL, with rapidly declining HRQoL after a 3-year follow-up. In addition, for patients with SAE, those with more dyspnoea symptoms, and those who did not receive pulmonary rehabilitation were associated with early decline of HRQoL in the first 2 years. Therefore, early pulmonary rehabilitation for patients with COPD before their second SAE; approximately 2 years after first SAE, may prevent entry into the accelerated frailty phase.
The second SAE can serve as the beginning of the accelerated frailty phase. At this time, patients experienced rapid deterioration in HRQoL and frequent severe exacerbations. The interval between sequential exacerbations decreased, and the risk of death increased gradually. According to our previous findings using the same study design and NHI cohort between January 2006 and December 2016, life expectancy was approximately 8 years after the first SAE.8 Therefore, patients who entered the accelerated frailty phase were expected to have approximately 4–5 years before their death caused by frequent exacerbations or comorbidities. After the fourth AE, the estimated interval between successive SAEs was approximately 6 months, which we consider a suitable time for introducing palliative care.
A prolonged stable phase between the first and second SAE for approximately 5 years was first found by Suissa et al5 and our results support their findings. In addition, we found that the stable phase was prolonged and a period of elevated health phase with an estimated 1-year duration between the first SAE and the prolonged stable phase. In total, the duration of the first two phases was approximately 3 years in this study. During the elevated health stable phase and prolonged stable phase, aggressive interventions such as early rehabilitation after the first SAE may provide the greatest benefit to the patient. A recent study found that pulmonary rehabilitation within 90 days after SAE benefited the mortality rate, but rehabilitation after 90 days did not.10 In addition, Ryrsø et al found that early pulmonary rehabilitation following COPD SAE reduced mortality and improved HRQoL.11 Therefore, by treating patients early during the phase of elevated health, we may prolong the period of increasing health status and avoid recurrent AE, and finally improve the disease trajectory.
In patients who experienced the first AE, the duration of improving health status after hospitalisation lasted only for 1 year. We also noticed a similar pattern but a more marked improvement in patients without a history of SAE. However, the improvement was sustained for approximately 1 year. This shows that the effect of current treatments for COPD, especially bronchodilators, cannot be sustained for a long time. The majority of clinical trials of current medication, including bronchodilators such as long-acting beta agonists (LABA), long-acting muscarinic receptor antagonists and inhaled corticosteroids (ICS) for COPD, had a follow-up period of no more than 1 year, and we had limited knowledge of their long-term efficacy.12–14 The short period of clinical improvement in our study indicates that we need to evaluate a longer treatment efficacy in future clinical trials. A larger proportion of patients with SAE had triple therapy in this study. The GOLD guideline recommends that triple therapy be considered as an initial treatment for patients with blood eosinophils ≥300 cells/µL and those who have had ≥1 severe exacerbation. Triple therapy is also recommended as a follow-up treatment in patients with blood eosinophils of ≥100 cells/µL who continue to exacerbate while receiving LABA+long-acting muscarinic antagonist (LAMA).6 In our study, more patients with SAEs used triple therapy than those without SAEs, 61.5% of them had eosinophils >300 cells/µL and 84.6% of them had eosinophils counts ≥100 cells/µL. Therefore, a higher ratio of using triple therapy for them may be due to having a ≥300 cells/µL eosinophil count and having had SAEs history or eosinophil count of ≥100 cells/µL and continue to exacerbate while receiving LABA+LAMA, with early and rapidly declined of their clinical condition of COPD.
Suissa et al found that older age was associated with a longer interval between exacerbations, which may be attributed to higher mortality in older patients.5 Our study also found that older age, stroke and CKD were associated with decreased SAE risk, which may result from competing risks owing to death, which could decrease the frequency of SAE.
Heart failure was the only comorbidity associated with increased AE and mortality in our study. This result was compatible with the study by Giezeman et al who demonstrated an increased risk of all-cause mortality and hospitalisation in patients with COPD with comorbid heart disease. The increased risk was unrelated to respiratory disease.15
CAD is a common and treatable comorbidity of COPD. In our study, patients with CAD had better survival than those without CAD. The result was not the same as that from previous studies showing that comorbid COPD and CVD increased the risk of mortality.14 16–18 One possible explanation was underdiagnosis. It has been shown that underdiagnosis of CAD is common even among patients whose heart failure is managed by cardiologists.19–21 In this study, the patients who had CAD diagnosis in the period of enrolment had a better survival rate may indicate the benefit of early diagnosis with early interventions for CAD. In addition, patients comorbid with CAD may have more symptoms such as exercise intolerance or dyspnoea, which makes physicians prescribe more medicines. Either treating CAD in patients with COPD or using ICS may result in lower mortality of patients with COPD. For example, treatment including ICS to reduce the mortality of COPD was shown in theTowards a Revolution in COPD Health (TORCH) and Efficacy and Safety of Triple Therapy in Obstructive Lung Disease (ETHOS) studies.22–24
Our study had some limitations. First, a small number of patients within only one medical centre were recruited, which may not represent the entire COPD population. Second, most of the population of both data sources were male, and only 6.4% of our patients were women in our hospital’s cohort, compared with 16.5% in the NHI cohort. This difference may be because patients visiting our medical centre differ from the general COPD population in Taiwan. According to statistical data published by the Health Promotion Administration Ministry of Health and Welfare, Taiwan, there were fewer women smokers than men smokers (3% compared with 25%, respectively).25 This difference was even more marked in south Taiwan, where our hospital is located. Moreover, owing to the high correlation between COPD and smoking, the fact that most smokers (95%) in Taiwan were men may account for the difference and the male predominant characteristics for both cohorts. However, the unequal distribution of sex means that our study results cannot be directly generalisable to other countries. Third, both smoking history and malnutrition were associated with the risk of subsequent AE and mortality.18 19 Therefore, the lack of data on smoking history and BMI in the Taiwan NHI cohort made potential impacts on the results in this study.
The strength of our study is that it showed dynamic changes in HRQoL after AE by following the EQ-5D utility. Previous studies on COPD QoL only showed that for approximately 1 year. In addition, by integrating quality data into the time course of SAEs, we could estimate the effect of exacerbation on quality-related health decline. This result may provide an opportunity to consider the starting point for terminal diseases and hospice care. Lastly, we showed a short period of quality improvement after treatment, which indicates the possibility of changing the disease trajectory through early treatment and improving patient outcomes.
In conclusion, the life trajectory of COPD includes the accelerated frailty phase and two additional phases: elevated health and prolonged stable phase after the first SAE. Early intervention, such as early pulmonary rehabilitation for patients with COPD before their second SAE, may prevent entry into the accelerated frailty phase.
Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.
Ethics statements
Patient consent for publication
Ethics approval
The Institutional Review Board of the National Cheng Kung University Hospital (NCKUH) approved this study (B-ER-98-289).
Acknowledgments
We are grateful to the Health and Welfare Data Science Center of the Ministry of Health and Welfare (NHIRD_MOHW) for providing all the data sets, facilities and linkage services that were required for this study. We are grateful to Chih-Hui Hsu for providing the excellent statistical consulting services from the Biostatistics Consulting Center, National Cheng Kung University Hospital. We are also grateful to the case manager I-Shan Lee from National Cheng Kung University Hospital and Pin-Rong Chen for image editing from National Cheng Kung University.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors C-ZC had full access to all of the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. C-ZC and J-DW were involved in the initial conception of the research questions. All authors were involved in the planning of the data analyses and interpretation of the data analyses. J-YH and C-HH analysed the data. S-HT wrote the manuscript. All authors were involved in reviewing, editing and final approval of the manuscript.
Funding The study was funded by grants from the National Health Research Institutes (NHRI-107A1-EMCO-02181810) and the Ministry of Science and Technology (MOST107-2627-M-006-007, MOST 109-2314-B-006-091 and MOST 110-2314-B-006-099) and the National Science and Technology Council (NSTC 112-2314-B-006-101-MY2) and grants from National Cheng Kung University Hospital (NCKUH-11203023 and NCKUH-11203033).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.