CKD indicates chronic kidney disease; ESA, erythropoiesis-stimulating agent.aIncludes angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, α-blockers, β-blockers, calcium channel blockers, and diuretics.
A, Long-term dialysis. B, Composite outcome of long-term dialysis or death. We used the Nelson-Aalen method12 to calculate the cumulative hazards of both outcomes in users and nonusers of angiotensin-converting enzyme inhibitors (ACEIs) and/or angiotensin II receptor blockers (ARBs). A multivariate analysis was adjusted for all variables listed in Table 1. P < .001, users vs nonusers.
Outcomes include long-term dialysis and the composite outcome of long-term dialysis or death associated with use of angiotensin-converting enzyme inhibitors (ACEIs) and/or angiotensin II receptor blockers (ARBs). Each factor was adjusted for all other factors listed in Table 1. CCB indicates calcium channel blocker; CHD, coronary heart disease.
eTable 1. Oral antihypertensives analyzed in the study
eTable 2. Incidence rate ratios of long-term dialysis and dialysis or death between ACEI/ARB users and nonusers
eTable 3. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients, 2000-2004
eTable 4. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients, 2005-2009
eTable 5. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients with exposure to antihypertensives within 30 days and observation period beginning 30 days after the index date
eTable 6. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients with exposure to antihypertensives within 60 days and observation period beginning 60 days after the index date
eTable 7. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients with exposure to antihypertensives within 120 days and observation period beginning 120 days after the index date
eTable 8. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients with propensity score–matched analysis
eTable 9. Risks of long-term dialysis and/or death in predialysis stage 5 CKD patients with receiving ESA treatment persistently
eTable 10. Risk of long-term dialysis (including dialysis initiation without recovery within 90 days) and/or death in predialysis stage 5 CKD patients using different types of ACEIs/ARBs compared with ACEI/ARB nonusers
eTable 11. Hyperkalemia-associated hospitalization during follow-up
Hsu T, Liu J, Hung S, Kuo K, Chang Y, Chen Y, Hsu C, Tarng D. Renoprotective Effect of Renin-Angiotensin-Aldosterone System Blockade in Patients With Predialysis Advanced Chronic Kidney Disease, Hypertension, and Anemia. JAMA Intern Med. 2014;174(3):347-354. doi:10.1001/jamainternmed.2013.12700
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
The benefit of using a renin-angiotensin-aldosterone system blocker such as an angiotensin-converting enzyme inhibitor (ACEI) or an angiotensin II receptor blocker (ARB) for patients with advanced chronic kidney disease (CKD) remains undetermined.
To assess the effectiveness and safety of ACEI/ARB use for advanced predialysis CKD in patients with hypertension and anemia.
Prospective cohort study.
From January 1, 2000, through June 30, 2009, we selected 28 497 hypertensive adult patients with CKD. Serum creatinine levels were greater than 6 mg/dL, hematocrit levels were less than 28%, and patients were treated with erythropoiesis-stimulating agents.
Users (n = 14 117) and nonusers (n = 14 380) of ACEIs/ARBs.
Main Outcomes and Measures
We used Cox proportional hazards regression models to estimate hazard ratios (HRs) for commencement of long-term dialysis and all-cause mortality for ACRI/ARB users vs nonusers.
In a median follow-up of 7 months, 20 152 patients (70.7%) required long-term dialysis and 5696 (20.0%) died before progression to end-stage renal disease requiring dialysis. Use of ACEIs/ARBs was associated with a lower risk for long-term dialysis (HR, 0.94 [95% CI, 0.91-0.97]) and the composite outcome of long-term dialysis or death (0.94 [0.92-0.97]). The renal benefit of ACEI/ARB use was consistent across most patient subgroups, as was that of ACEI or ARB monotherapy. Compared with nonusers, the ACEI/ARB users had a higher hyperkalemia-associated hospitalization rate, but the risk of predialysis mortality caused by hyperkalemia was not significantly increased (HR, 1.03 [95% CI, 0.92-1.16]; P = .30).
Conclusions and Relevance
Patients with stable hypertension and advanced CKD who receive therapy with ACEIs/ARBs exhibit an association with lower risk for long-term dialysis or death by 6%. This benefit does not increase the risk of all-cause mortality.
Inhibition of the renin-angiotensin-aldosterone system with an angiotensin-converting enzyme inhibitor (ACEI) or an angiotensin II receptor blocker (ARB) delays the progression of chronic kidney disease (CKD) in patients with and without diabetes mellitus, particularly those with mild to moderate renal insufficiency as reflected by a serum creatinine level of 1.5 to 3.0 mg/dL (to convert to micromoles per liter, multiply by 88.4).1- 4 Two randomized clinical trials in China5,6 and a post hoc analysis of the Ramipril Efficacy in Nephropathy trial7 further demonstrated that the use of ACEIs or ARBs is beneficial in patients with stage 4 CKD, defined by a glomerular filtration rate (GFR) of 15 to 29 mL/min/1.73 m2 and a serum creatinine level of approximately 3.0 to 5.0 mg/dL, who do not have diabetes mellitus. However, data focusing on the use of ACEIs and/or ARBs in patients with serum creatinine levels of greater than 5.0 mg/dL or a GFR of less than 15 mL/min/1.73 m2 are limited. Because most prospective, large-scale clinical trials of ACEI/ARB use excluded individuals with stage 5 CKD who were not undergoing dialysis (predialysis CKD), whether ACEI/ARB therapy should be used continuously in patients with advanced CKD remains unclear.
The prevalence and incidence of end-stage renal disease (ESRD) are extremely high in Taiwan compared with other countries.8 The timing for initiation of dialysis in Taiwan is late (characteristically starting at an estimated GFR <5 mL/min/1.73 m2).9 According to National Health Insurance (NHI) reimbursement regulations, patients with CKD in Taiwan who have a serum creatinine level of greater than 6 mg/dL (approximately equivalent to GFR <15 mL/min/1.73 m2) and a hematocrit level of less than 28% could receive treatment with an erythropoiesis-stimulating agent (ESA) to maintain a hematocrit level not exceeding 36%. Therefore, treatment with ESAs provides us a unique opportunity to perform a risk-benefit analysis of ACEIs/ARBs in patients with advanced CKD. To bridge the evidence gap in the transition from stages 1 to 4 CKD to ESRD, we assessed the association between the use of ACEIs/ARBs and the risk of long-term dialysis and/or death in a large nationwide cohort of patients with predialysis stage 5 CKD who had hypertension and anemia and were treated with ESAs.
The present study used data from the NHI Research Database, which contains health care utilization data for more than 95% of the hospitals in Taiwan and 99% of the entire 23 million persons enrolled in the NHI program.10 Deidentified information kept in the NHI Research Database included date of birth, sex, residency area, diagnostic codes, drug prescriptions, and medical procedures. We used codes from the International Classification of Diseases, Ninth Revision (ICD-9) to define diseases. The study was approved by the institutional review board at Taipei Veterans General Hospital, and the informed consent was waived due to the personal information having been deidentified in the NHI Research Database.
Quiz Ref IDThe study was designed as a population-based longitudinal cohort study. We selected individuals who had a primary diagnosis of CKD (ICD-9 codes 016.0, 042, 095.4, 189, 223, 236.9, 250.4, 271.4, 274.1, 403-404, 440.1, 442.1, 446.21, 447.3, 572.4, 580-589, 590-591, 593, 642.1, 646.2, 753, and 984) and received ESA treatment, indicating that their serum creatinine levels were greater than 6 mg/dL and hematocrit levels were less than 28%, from January 1, 2000, through June 30, 2009. Based on an internal report of the Taiwan Department of Health, the rate of ESA use was 85% in 2012 for the patients with advanced stage 5 CKD who had not yet commenced dialysis. In addition, the median hematocrit level at the initiation of dialysis was 24.2% (interquartile range, 20.6%-27.5%) in Taiwan.9 Therefore, the selected cohort in this study was the most representative of patients with predialysis stage 5 CKD in Taiwan. We excluded patients younger than 20 years or older than 100 years and those who had received dialysis or a renal transplant before the ESA prescription. Because we used prescription information within 90 days after the first ESA treatment to ascertain ACEI/ARB use, the 91st day after the ESA prescription was defined as the index date. Patients who died, who commenced renal replacement, or who had not been prescribed any antihypertensives from the first ESA treatment to the index date were also excluded. Ultimately, we selected 28 497 patients for the current study (Figure 1). Comorbidities, including diabetes mellitus, coronary artery disease, stroke, and cancer, were defined as diseases diagnosed within 3 years before the index date. The Charlson comorbidity index was used to quantify patient comorbidity profiles.11
The antihypertensives analyzed in this study included diuretics, α-blockers, β-blockers, calcium channel blockers, ACEIs, and ARBs. A detailed drug classification is found in the Supplement (eTable 1). Patients who had taken any ACEI/ARB within 90 days after the first ESA prescription were defined as ACEI/ARB users; the remaining patients were defined as ACEI/ARB nonusers. All analyses were conducted on an intention-to-treat basis according to the patients’ initial ACEI/ARB assignment regardless of subsequent changes to their antihypertensive regimen.
The observation period started from the index date to death, to the start date for long-term dialysis, or to December 31, 2009, whichever occurred first. The onset of the renal outcome was defined as the commencement date of long-term dialysis for at least 90 days; the onset of the composite outcome of long-term dialysis or death was the start date of long-term dialysis or death, whichever came first. We identified the first event of hospitalization involving a diagnosis of hyperkalemia (ICD-9 code 276.7) during the observation period as a hyperkalemia-associated hospitalization.
Baseline characteristics were compared by the 2-sided t test and χ2 test. In the multivariate Cox proportional hazards regression models, the effects of ACEIs/ARBs were further adjusted for age, sex, Charlson comorbidity index, diabetes mellitus, coronary artery disease, stroke, cancer, frequency of visits to nephrologists within 3 years before the index date (0, 1-6, or >6 visits), geographic location (northern, middle, southern, or eastern/other islands, according to NHI registration location), and types of non-ACEI/non-ARB antihypertensives used. Study entry was defined as the index date. For the long-term dialysis end point, observations were censored at the end of the study or the date of death, whichever occurred first. For the composite outcome, observations were censored at the end of the study. Results were expressed as hazard ratios (HRs) compared with ACEI/ARB nonusers. The proportional hazards assumption, the constant HR over time, was evaluated by comparing estimated log-log survival curves for all time-independent covariates. All assessed log-log survival plots graphically showed 2 parallel lines, indicating no violation of the assumption. Adjusted HRs for long-term dialysis and the composite outcome associated with ACEI/ARB use were further analyzed among subgroups based on participants’ characteristics (see below). The cumulative hazards of long-term dialysis and the composite outcome over time were compared between the ACEI/ARB users and nonusers using the Nelson-Aalen method12 to adjust covariates adopted in the Cox proportional hazards regression models. All P values were 2-sided, and the significance level was set at .05. Analyses were performed using commercially available software (SAS, version 9.2 [SAS Institute Inc], and Stata SE, version 11.0 [StataCorp]).
To assess the reliability of our findings, we conducted an additional series of analyses. First, we conducted separate analyses for patients with index dates from 2000 to 2004 and from 2005 to 2009 to look for any evidence of a cohort effect. Second, we conducted a series of analyses defining ACEI/ARB use at intervals of 30, 60, and 120 days after the first ESA prescription to minimize misclassification bias. Third, we restricted the analysis to patients receiving ESA therapy at 2 or more consecutive ambulatory care visits to exclude acute exacerbation of chronic renal failure with transient creatinine levels of greater than 6 mg/dL. Finally, we conducted a propensity score–based matching technique to control for residual confounding factors. For each ACEI/ARB user, we identified 1 nonuser from our selected cohort who was frequency-matched with propensity scores calculated from all baseline covariates (Table 1). A nearest-neighbor algorithm was applied to construct matched pairs, assuming that the proportion of 0.95 to 1.00 is perfect.13
We enrolled 28 497 patients with predialysis advanced CKD and comorbid hypertension and anemia in the present study (Figure 1). Quiz Ref IDAmong this population, 14 117 patients (49.5%) had at least 1 prescription for an ACEI/ARB within 90 days after the first ESA prescription. Among the patients with ACEI/ARB prescriptions, 3810 (27.0%) had been treated with ACEIs, 8203 (58.1%) with ARBs, and 2104 (14.9%) with both administered concurrently. The mean age of the ACEI/ARB users was 64.7 years, of whom 53.4% were women and 57.7% had diabetes mellitus (Table 1). Compared with ACEI/ARB users, nonusers were older, had fewer comorbidities, were less likely to visit nephrologists in the preceding 3 years, and had greater use of other antihypertensives, including α-blockers and dihydropyridine calcium channel blockers. The prescription of ACEIs/ARBs after the ESA treatment was generally in accordance with their use in the period before the ESA prescription. More than 40% of ACEI/ARB users and nonusers were from northern Taiwan.
A total follow-up summation was 27 678 person-years during the study period; 20 152 patients (70.7%) required long-term dialysis and 5696 (20.0%) died before progression to ESRD requiring dialysis. The number of events and incidence rates of long-term dialysis and the composite outcome are listed in Table 2. A comparison of incidence rate ratios of long-term dialysis and death between ACEI/ARB users and nonusers can be found in the Supplement (eTable 2).
The Nelson-Aalen curves for the adjusted cumulative hazards of long-term dialysis (Figure 2A) and dialysis or death (Figure 2B) among the ACEI/ARB users compared with nonusers were both significant (P < .001). This finding indicated that ACEI/ARB users had a lower risk of initiating long-term dialysis or of developing the composite outcome.
As shown in Table 2, the incidence of dialysis decreases from 75.6 per 100 patient-years to 70.1 per 100 patient-years, indicating that Quiz Ref IDamong patients with predialysis advanced CKD, 5.5% of ACEI/ARB users each year could be prevented from starting long-term dialysis compared with nonusers. In univariate Cox analysis, we found that treatment with an ACEI/ARB in patients with stage 5 CKD significantly reduced the risk for long-term dialysis and the composite outcome, with an identical unadjusted HR of 0.93 (95% CI, 0.91-0.96; P < .001) (Table 2). The beneficial effects of ACEI/ARB use on long-term dialysis (HR, 0.94 [95% CI, 0.91-0.97]; P < .001) and the composite outcome of dialysis or death (0.94 [0.92–0.97]; P < .001) persisted after adjustment for various potential confounders.
Table 2 also presents the association between different types of ACEIs/ARBs used and development of long-term dialysis and/or death. Overall, ACEIs/ARBs as a class of drugs or as monotherapy could reduce the risks of long-term dialysis and the composite outcome. However, concurrent ACEI and ARB users experienced neutral effects on long-term dialysis and the composite outcome compared with ACEI/ARB nonusers.
We conducted a series of stratified analyses to test the reliability of our analyses (Figure 3). The reduced HRs of long-term dialysis and dialysis or death among patients with stage 5 CKD in favor of ACEI/ARB use were consistent across almost all patient subgroups except those with cancer and those without previous visits to nephrologists.
In sensitivity analyses, we found that the estimated effects of ACEI/ARB use were similar whether we changed the entry and observation periods, redefined the exposure time for ACEIs/ARBs, restricted analysis to patients receiving ESA therapy persistently, or selected a group of propensity-based matched ACEI/ARB nonusers. The results of sensitivity analyses can be found in the Supplement (eTables 3-9). Furthermore, if we count those patients who died within 90 days of starting dialysis as having a renal event (723 [5.1%] of ACEI/ARB users vs 691 [4.8%] of nonusers), we could redefine renal outcome as dialysis initiation without recovery within 90 days to account for the high risk of death within the first 90 days of dialysis. The results shown in the Supplement (eTable 10) are similar to those illustrated in Table 2, indicating that our findings in this study are robust.
More hyperkalemia-associated hospitalizations were found among ACEI/ARB users than among nonusers (9.2% vs 6.7%; P < .001]) (Supplement [eTable 11]). Quiz Ref IDA relative risk of 1.31 (95% CI, 1.21-1.43; P < .001) was found, corresponding to a risk increment of 31% owing to ACEI/ARB use. However, those who developed hyperkalemia did not significantly increase the risk of predialysis death (HR, 1.03 [95% CI, 0.92-1.16]; P = .30 [data not shown]).
This study is the first, to our knowledge, to demonstrate that ACEI/ARB use reduced the HR of initiation of dialysis and the composite outcome of dialysis or death by 6% among patients with hypertension and predialysis stage 5 CKD. Furthermore, ACEI/ARB use in most subgroups of patients could have similar risk reductions. We estimate that, every year, ACEI/ARB use could prevent 5.5% of the patients with advanced CKD from commencing long-term dialysis. Our study not only extends the current knowledge in the field but also represents the consistency and generalizability of the effectiveness of ACEI/ARB use in patients with stages 1 through 4 CKD to those with stage 5 CKD.
Because of concerns about a rise of serum creatinine levels, physicians may hesitate to prescribe ACEIs/ARBs in patients with advanced CKD. Suissa et al14 found that ACEIs are renoprotective in the early stage of therapy, but this effect may be reversed with longer-term ACEI use. Ahmed et al15 further found that GFR improved from 16.3 to 26.6 mL/min/1.73 m2 within 12 months after ACEI/ARB therapy was stopped in 52 elderly patients with stages 4 and 5 CKD. In the United States, 62% to 68% of patients with stages 3 to 5 CKD had used ACEIs or ARBs during the 2 years before ESRD, but rates of use fell to 42% within 3 months before dialysis.16
In contrast, ACEI/ARB use reduced the decline of residual renal function in patients undergoing dialysis in an observational study17 and in clinical trials.18,19 These results corroborate our findings that treatment with ACEIs/ARBs resulted in an approximate 6% reduction in the HR (0.94) for the necessity of long-term dialysis and for the composite outcome in patients with predialysis stage 5 CKD. Previous clinical trials provided strong evidence that ACEI/ARB use reduced the risk of ESRD compared with placebo in patients with nondiabetic and diabetic stage 3 CKD by 56%3 and 28%,1 respectively, and that ACEIs reduced the risk by 40% in nondiabetic stage 4 CKD.5 This discrepancy in the reduced risk of commencing dialysis from stages 3 and 4 to stage 5 CKD could be explained primarily by the fact that the decision on the initiation of dialysis is not based on GFR alone. Several measured or unmeasured clinical conditions may contribute to the decision and thus dilute the renal beneficial effects. For instance, in our study, ACEI/ARB users had higher incidence rates of hyperkalemia-associated hospitalization, and hyperkalemia increased the decision to commence long-term dialysis. In addition, the ACEI/ARB users were prone to have more comorbidities, such as diabetes mellitus and cardiovascular disease. This finding might attenuate the dialysis-saving effect of ACEI/ARB use.
Our analysis concentrated on cardiorenal protection as a whole and found that ACEI/ARB use might moderately reduce the incidence of dialysis in patients with advanced CKD without increasing the risk of death. However, increased activity of the renin-angiotensin-aldosterone system is only one of many risk factors contributing to cardiovascular abnormalities in CKD. Clinical trials of losartan potassium1 and irbesartan2 demonstrated that these ARBs did not differ from other classes of antihypertensives in the overall incidence of cardiovascular events. Most patients with CKD do not survive to ESRD development, mainly owing to cardiovascular mortality.20 To substantially improve the cardiovascular outcomes in patients with advanced CKD, focus should be on managing the multidimensional risk factors.21
Combination ACEI/ARB therapy is more antiproteinuric in the short term than is ACEI or ARB alone.22,23 However, a recent meta-analysis showed no significant difference in mortality risk and ESRD development between ACEI plus ARB combinations and monotherapy in patients with early CKD (stages 1-3).24 Along the same lines, our study found no renal benefit in the combination group.
Previous clinical trials demonstrated that moderate hyperkalemia was associated with the use of an ACEI/ARB in patients with CKD.25 As expected, the higher rate of hyperkalemia-associated hospitalization among ACEI/ARB users than among nonusers (9.2% vs 6.7%) was observed in the present study. Investigators have reported that patients with CKD who have a serum potassium level greater than 6.0 mEq/L (conversion to millimoles per liter is 1:1) did not exhibit apparent electrocardiographic or cardiovascular manifestations.26 The presence of other electrolyte disturbances influences the cardiac membrane potential, which may contribute to the reduced sensitivity to cardiac toxicity of hyperkalemia in the CKD population.27 Consequently, although the incidence rate of hyperkalemia-associated hospitalization was increased among ACEI/ARB users, it did not have a significant effect on the predialysis mortality in our study.
Our study was notable for its large sample size and its nationally representative nature. By selecting those who survived longer than 90 days after the ESA prescription and observing them from the end of this window, we are able to control for survival bias.28 However, some precautions are needed in interpreting the results. First, patients with acute-on-chronic renal failure who had transient creatinine levels of more than 6 mg/dL were possibly included in this database. Therefore, we restricted the analysis to patients receiving ESA therapy persistently at 2 or more consecutive ambulatory care visits, and the results were not materially changed. Furthermore, use of a serum creatinine level as a major inclusion criterion is also likely to result in selection of women with more advanced CKD than men in the cohort; thus, the insignificant composite outcome in women as shown in Figure 3 could indicate that any benefit of renin-angiotensin-aldosterone system blockade becomes undetectable at an extremely low GFR. Second, we assumed the onset of predialysis stage 5 CKD as the first day of an ESA prescription in patients with an existing diagnosis of CKD. An indication for a late or early prescription of an ESA associated with ACEI/ARB therapy can introduce bias. However, we are confident that any misclassification of the timing of this proxy measure for stage 5 CKD is nondifferential between ACEI/ARB users and nonusers, which would tend to bias the results toward the null hypothesis. Quiz Ref IDThird, our study was observational, so it cannot prove causality. However, to ensure an adequate statistical power (α = .05 and 1 − β = .8, no loss to follow-up), at least 15 000 patients with predialysis ESRD should be enrolled to examine a relative risk reduction of 6%. Apart from the impracticability of conducting such a large-scale randomized clinical trial, emerging evidence29,30 also suggests that well-designed observational studies could yield comparable outcomes. Fourth, although some important prognostic factors, such as the underlying cause of ESRD, biochemical data, and health-related behaviors (eg, smoking), are not available in this study, we believe that consistency in the series of subgroup analyses ensures the robustness of our study results. Finally, the generalizability of our data is limited to the population of patients with predialysis CKD and stable hypertension who were receiving ESA therapy. These results may not be applicable to all patients with predialysis CKD.
In conclusion, our findings expand the existing knowledge in the field and provide clinicians with new information about the effectiveness and safety of ACEI/ARB use when CKD progresses to pre-ESRD. In such a condition, withholding ACEI/ARB therapy is unwarranted and may hasten the onset of ESRD. Our study does not support concomitant use of ACEI and ARB in predialysis stage 5 CKD. In these patients and those whose course of treatment involves an ACEI/ARB, physicians should also increase their alertness to prevent hyperkalemia.
Accepted for Publication: September 22, 2013.
Corresponding Authors: Chih-Cheng Hsu, MD, DrPH, Division of Geriatrics and Gerontology, Institute of Population Health Sciences, National Health Research Institutes, No. 35, Keyan Road, Zhunan Town, Miaoli County, Taiwan 35053 (firstname.lastname@example.org); and Der-Cherng Tarng, MD, PhD, Institutes of Physiology and Clinical Medicine, National Yang-Ming University, 201, Section 2, Shih-Pai Road, Taipei, Taiwan 11217 (email@example.com).
Published Online: December 16, 2013. doi:10.1001/jamainternmed.2013.12700.
Author Contributions: Drs C.-C. Hsu and Tarng had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: C.-C. Hsu, Tarng.
Acquisition of data: T-W. Hsu, C.-C. Hsu, Tarng.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: T-W. Hsu, Liu.
Critical revision of the manuscript for important intellectual content: Hung, Kuo, Chang, Chen, C.-C. Hsu, Tarng.
Statistical analysis: Liu, Chang.
Obtained funding: C.-C. Hsu, Tarng.
Administrative, technical, or material support: T-W. Hsu, Hung, Kuo, Chang, Chen, C.-C. Hsu.
Study supervision: C.-C. Hsu, Tarng.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the National Science Council, the Taipei Veterans General Hospital, the National Health Research Institutes, and the National Yang-Ming University.
Role of the Sponsors: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.