Article ID: CJ-18-1087
Background: The natural course of heart failure (HF) is typically associated with repeated hospitalizations, and subsequently, patient prognosis deteriorates. However, the precise relationship between repeated admissions for HF and long-term prognosis remains unknown.
Methods and Results: We analyzed data from 1,730 consecutive acute HF patients registered in the West Tokyo Heart Failure (WET-HF) registry between June 2005 and April 2014 (median age, 76 years). Patients were divided into 3 groups according to the number of previous HF admissions at the time of the index admission (0, n=876 [55.4%]; 1, n=425 [26.9%]; ≥2, n=279 [17.7%] previous admissions). A history of multiple previous admissions was an independent predictor for all-cause death and HF readmission in reference to a history of a single previous admission (hazard ratio (HR), 1.53; 95% confidence interval (CI) 1.10–2.13; HR, 1.90 95% CI, 1.47–2.44, respectively) or no previous admissions (HR, 1.37, 95% CI, 1.01–1.85; HR, 2.83, 95% CI, 2.19–3.65, respectively). On the other hand, a history of a single previous admission was an independent predictor for HF readmission in reference to a history of no previous admissions (HR, 1.51, 95% CI, 1.18–1.92), but not for all-cause death (HR, 0.89, 95% CI, 0.66–1.20).
Conclusions: Based on a contemporary multicenter HF registry, a history of multiple previous HF admissions was revealed as an independent, strong risk factor of adverse events following the index admission. The number of hospitalizations could be a simple and important surrogate indicating subsequent adverse events in patients with HF.
Heart failure (HF) is associated with repeated HF admissions and high mortality rates in conjunction with increased economic burden on the healthcare system.1–3 In addition to the use of medications and devices, there is accumulating evidence of the benefits of multimodal interventions including lifestyle modifications. The initiation of a nurse-led, multidisciplinary, home-based intervention decreased all-cause deaths and readmissions after hospitalization for HF.4,5 Patients who received a follow-up appointment before discharge were also strongly protected against readmission for HF.6 However, to improve the prognosis and cost-effectiveness in HF patients under limited medical resources, identification of the appropriate timing of these interventions using a simple surrogate measure for the prognosis of HF has been the subject of much research.
Editorial p ????
HF admission could be a simple, alarming sign of high risk in patients with HF and also an optimization opportunity for management;7 if the number of HF admissions could be a more informative sign of subsequent worse clinical outcomes, the number of admissions could be an intriguing hint about the timing of discussions on intervention options for HF.8 Previous data from healthcare utilization databases in the early 2000 s revealed that all-cause death increased after each HF hospitalization compared with after a first hospitalization for HF.9 However, the effect of the number of HF admissions on the subsequent long-term clinical outcome remains unknown in an era characterized by the widespread use of guideline-based treatment, owing to a lack of evidence from well-defined, contemporary HF cohorts that evaluate patients’ detailed backgrounds as well as adjudicated clinical outcomes, including HF hospitalization.
Accordingly, the objective of this study was to analyze the clinical characteristics of patients with multiple previous admissions as well as the clinical effect of multiple admissions on long-term, event-free survival following the index admission in patients with acute HF (AHF) who were registered in a contemporary Japanese HF registry, the West Tokyo Heart Failure (WET-HF) database.
We analyzed data from 1,730 consecutive AHF patients registered in the WET-HF registry between June 2005 and April 2014. The median and interquartile age was 76 (66–83), and 1,087 (62.8%) patients were male. The details of the WET-HF have been described previously.10,11 Briefly, data were extracted from a prospective, multicenter registry designed to collect the clinical background and outcomes of patients hospitalized with AHF as the primary cause of admission. Before the launch of WET-HF, information on the objective of the present study, its social significance, and an abstract were provided for clinical trial registration to the University Hospital Medical Information Network (UMIN000001171). This study was conducted at 4 high-volume hospital centers within the Tokyo metropolitan area (2 university hospitals, 2 tertiary referral hospitals). Participating hospitals were instructed to record and register data from consecutive hospital visits for AHF.10,11 AHF was defined according to the Framingham AHF criteria.12 In addition, the site investigators were instructed to prospectively document the details of AHF during the registration period. Patients who refused to participate in the study or presented with HF with acute coronary syndrome were excluded from registration. The present study was approved by each center’s ethics review committee, and all the patients provided informed consent.
Figure 1 is a flowchart of the study’s design. Within the cohort, we excluded 71 patients (4.1%) whose history of previous HF admission could not be determined, 60 patients (3.4%) who died in hospital, and 19 patients (1.1%) who were lost during follow-up. For the remaining 1,580 patients, we categorized them into 3 groups based on the number of previous HF admissions: no previous admissions, single previous admission, and multiple previous admissions. For each group, baseline data included (1) age, sex, and body mass index (BMI); (2) cause of HF; (3) medical history; (4) prior procedures; (5) vital signs and physical examination; (6) laboratory data; (7) echocardiographic data; and (8) medication use on admission and at discharge.
Flowchart of the study (A) and distribution of the number of previous heart failure admissions (B). AHF patients were divided into 3 groups according to the number of previous heart failure admissions: no previous admissions, single previous admission, multiple previous admissions. AHF, acute heart failure; WET-HF, West Tokyo Heart Failure.
A follow-up survey, using medical charts or telephone review, was performed, and the patients who were lost to follow-up were censored at the date of last contact.13 The median follow-up period of survivors was 943 days (interquartile range, 648–1352 days). Information regarding specific outcomes was obtained from participating cardiologists and investigators, including (1) all-cause death, (2) readmission for worsening HF, and (3) a composite of all-cause death and readmission for worsening HF. The treating physicians in each participating hospital made decisions regarding readmission for HF according to the usual standard of care. In most cases, patients were readmitted when clinical signs of decompensation, such as orthopnea or lower extremity edema, were present. Additionally, these events were adjudicated through chart abstractions performed by the site investigators.
Statistical AnalysisNormally distributed data are expressed as mean±standard deviation, non-parametric data as median (interquartile range), and categorical data as percentage. Independent continuous variables were compared with Kruskal-Wallis tests, and categorical variables with Pearson’s chi-square test. Survival curves were calculated using Kaplan-Meier estimates and were analyzed with the log-rank test. A multivariate Cox proportional hazard model was used to identify independent predictors of each endpoint. Models were adjusted for age, sex, BMI, and other prognostic values, including the levels of hemoglobin (Hb), estimated glomerular filtration rate (eGFR), echocardiographic left ventricular ejection fraction (LVEF), history of atrial fibrillation (AF), hypertension, mineralocorticoid-receptor antagonist at discharge, implantable cardioverter defibrillator (ICD), and cardiac resynchronization therapy (CRT). These variables were chosen according to clinical relevance and the results of previous studies.14 Interactions between a history of multiple previous HF admissions and the aforementioned clinically important variables were estimated using Cox regression. The effect of the interaction between the number of previous HF admissions and other confounding factors on the composite endpoint of all-cause death and HF readmission was examined by adding an interaction term to the statistical model. Each subgroup was divided on the basis of age, sex, BMI at hospital discharge, eGFR at hospital discharge, anemia (serum Hb level <13 g/dL in male patients or <12 g/dL in female patients at hospital discharge), LVEF, AF, and systolic BP. Results are reported as hazard ratio (HR), 95% confidence interval (CI), and P values. HR values for outcomes in patients who had experienced multiple previous HF admissions were compared with those of patients who had experienced either no previous HF admission or a single previous HF admission. All statistical analysis was performed using SPSS version 23 (SPSS, Inc., Chicago, IL, USA).
Of the 1,580 patients, those who had not experienced a previous HF admission (no previous admission group) accounted for 55.4% (n=876). Patients who experienced a single (single previous admission group) or multiple (multiple previous admission group) previous HF admissions accounted for 26.9% (n=425) and 17.7% (n=279), respectively. In the multiple previous admissions group, those who experienced 2, 3, 4 or ≥5 previous admissions accounted for 8.8%, 4.5%, 2.0%, and 2.4%, respectively (Figure 1A,B).
BP and heart rate on hospital admission were lower, but the prevalence of dilated cardiomyopathy was higher as the number of previous admissions increased (Table 1). Hb concentration, eGFR and C-reactive protein (CRP) levels were lower at discharge in the multiple previous admissions group. Left ventricular and left atrial diameters were larger, LVEF was lower, and the prescription rates of renin-angiotensin system inhibitors, β-blockers, mineralocorticoid antagonists, and oral anticoagulants on admission increased as the number of previous admissions increased (Table 2). The NYHA functional class did not differ among the none, single, and multiple previous admission groups (NYHA III or IV on admission: 84%, 85%, and 82%, P=0.62, at discharge; 24%, 26%, and 25%, P=0.73).
| None n=876 (55.4%) |
Single n=425 (26.9%) |
Multiple n=279 (17.7%) |
P value | |
|---|---|---|---|---|
| Age (years) | 76 (66–83) | 76 (64–82) | 75 (63–81) | 0.23 |
| Female | 38% | 36% | 34% | 0.42 |
| BMI (kg/m2) | 21.6 (19.3–23.9) | 21.2 (18.9–23.9) | 21.5 (19.1–23.7) | 0.38 |
| Systolic BP (mmHg) | 139 (120–163) | 132 (115–152) | 121 (107–144) | <0.001 |
| Diastolic BP (mmHg) | 80 (69–97) | 76 (62–90) | 70 (60–86) | <0.001 |
| Heart rate (beats/min) | 92 (75–112) | 90 (76–106) | 81 (68–102) | <0.001 |
| NYHA | 0.65 | |||
| II | 16% | 15% | 18% | |
| III | 33% | 33% | 29% | |
| IV | 51% | 52% | 53% | |
| Medical history | ||||
| Hypertension | 73% | 63% | 54% | <0.001 |
| Diabetes | 35% | 40% | 37% | 0.24 |
| Dyslipidemia | 39% | 36% | 42% | 0.34 |
| Smoking | 46% | 39% | 40% | 0.027 |
| COPD | 5% | 4% | 6% | 0.33 |
| Atrial fibrillation | 42% | 48% | 55% | 0.001 |
| Stroke | 11% | 16% | 10% | 0.025 |
| Dialysis | 4% | 4% | 5% | 0.83 |
| Home oxygen therapy | 2% | 5% | 8% | <0.001 |
| Permanent pacemaker | 4% | 11% | 11% | <0.001 |
| ICD | 1% | 5% | 10% | <0.001 |
| CRT | 0.1% | 2% | 7% | <0.001 |
| Etiology of HF | ||||
| Ischemic | 26% | 36% | 30% | 0.001 |
| Valvular | 26% | 20% | 22% | 0.039 |
| Dilated | 12% | 21% | 33% | <0.001 |
| Clinical profile | ||||
| PND | 36% | 29% | 35% | 0.06 |
| Orthopnea | 37% | 35% | 34% | 0.71 |
| Jugular venous distention | 43% | 41% | 50% | 0.068 |
| Edema | 68% | 60% | 63% | 0.012 |
| Cold extremities | 21% | 23% | 19% | 0.55 |
| Rales | 55% | 55% | 48% | 0.085 |
| 3rd heart sound | 37% | 45% | 50% | <0.001 |
Categorical values are expressed as percentage, and continuous variables are expressed as median (interquartile range). BMI, body mass index; BP, blood pressure; COPD, chronic obstructive pulmonary disease; CRT, cardiac resynchronization therapy; HF, heart failure; ICD, implantable cardioverter defibrillator; NYHA, New York Heart Association; PND, paroxysmal nocturnal dyspnea.
| None n=876 (55.4%) |
Single n=425 (26.9%) |
Multiple n=279 (17.7%) |
P value | |
|---|---|---|---|---|
| Laboratory data | ||||
| Hemoglobin (g/dL) | 12.1 (10.7–13.9) | 11.7 (10.4–13.7) | 11.6 (10.3–13.2) | 0.006 |
| BUN (mg/dL) | 21.4 (16.0–30.4) | 25.0 (17.7–36.2) | 28.1 (19.0–41.2) | <0.001 |
| Creatinine (mg/dL) | 0.97 (0.74–1.30) | 1.10 (0.90–1.50) | 1.20 (0.92–1.70) | <0.001 |
| eGFR (ml/min/1.73 m2) | 53.2 (38.4–68.9) | 45.7 (30.9–60.2) | 41.0 (28.9–56.7) | <0.001 |
| Sodium (mEq/L) | 139 (137–141) | 139 (136–141) | 138 (136–140) | 0.053 |
| Potassium (mEq/L) | 4.3 (4.0–4.6) | 4.3 (4.0–4.7) | 4.4 (4.1–4.7) | 0.081 |
| BNP (pg/mL) | 266 (123–548) | 247 (137–587) | 312 (163–693) | 0.085 |
| CRP (mg/dL) | 0.53 (0.17–1.47) | 0.41 (0.11–1.10) | 0.28 (0.08–0.79) | <0.001 |
| Echocardiographic data | ||||
| LVDd (mm) | 51 (45–58) | 53 (45–61) | 57 (49–65) | <0.001 |
| LVDs (mm) | 38 (30–47) | 41 (31–51) | 46 (35–56) | <0.001 |
| LVEF (%) | 45 (33–59) | 41 (31–56) | 35 (23–52) | <0.001 |
| LAD (mm) | 42 (37–47) | 44 (39–49) | 47 (42–53) | <0.001 |
| Medication use on admission | ||||
| ACEI or ARB | 37% | 52% | 60% | <0.001 |
| β-blocker | 30% | 50% | 67% | <0.001 |
| MRA | 9% | 23% | 41% | <0.001 |
| Loop diuretic | 31% | 54% | 70% | <0.001 |
| Tolvaptan | 1% | 2% | 5% | <0.001 |
| Calcium blocker | 32% | 24% | 24% | 0.025 |
| Oral anticoagulant | 30% | 44% | 54% | <0.001 |
| PDE3 inhibitor | 1% | 2% | 1% | 0.14 |
| Medication use at discharge | ||||
| ACEI or ARB | 70% | 68% | 67% | 0.58 |
| β-blocker | 78% | 75% | 80% | 0.25 |
| MRA | 33% | 46% | 48% | <0.001 |
| Loop diuretic | 74% | 76% | 83% | 0.017 |
| Tolvaptan | 2% | 1% | 5% | 0.014 |
| Calcium blocker | 38% | 26% | 25% | <0.001 |
| Oral anticoagulant | 48% | 49% | 58% | 0.014 |
| PDE3 inhibitor | 0.2% | 2% | 2% | 0.005 |
Categorical values are expressed as percentage, and continuous variables are expressed as median (interquartile range). ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BUN, blood urea nitrogen; BNP, B-type natriuretic peptide; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; LAD, left atrial dimension; LVDd, left ventricular internal dimension in diastole; LVDs, left ventricular internal dimension in systole; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid-receptor antagonist; PDE3, phosphodiesterase 3.
The log-rank test demonstrated that the multiple previous admission group had significantly higher rates of all-cause death and HF readmission, as well as the composite endpoint of all-cause death and HF readmission after 3 years, compared with the single and no previous admission groups (Figure 2A–C). Although the single previous admission group had a significantly higher rate of HF readmission and the composite endpoint than did the no previous admission group, the single previous admission group did not have a significant difference in the occurrence of all-cause death compared with the no previous admission group (Figure 2A–C). Thus, the difference in the composite endpoint between the single and no previous admission groups was mainly because of the differences in HF admissions.
Kaplan-Meier survival curves for all-cause death (A), HF readmission (B), and the composite endpoint of all-cause death and HF readmission (C), according to the number of previous HF admissions. HF, heart failure.
The Cox proportional hazard model demonstrated that the adjusted HR for the composite endpoint was 2.27 (95% CI, 1.81–2.84, P<0.001) for multiple vs. no previous admissions, 1.82 (95% CI, 1.45–2.29, P<0.001) for multiple vs. single previous admissions, and 1.27 (95% CI, 1.03–1.57, P=0.026) for single vs. no previous admissions (Table 3, Supplementary Table). The adjusted HR for all-cause death was 1.37 (95% CI, 1.01–1.85, P=0.041) for multiple vs. no previous admissions and 1.53 (95% CI, 1.10–2.13, P=0.012) for multiple vs. single previous admissions. There was no significant association in the risk of all-cause death when the single and no previous admission groups were compared with each other (adjusted HR 0.89, 95% CI, 0.66–1.20, P=0.46). In the subgroup analysis comparing the multiple and no previous admission groups, the association of multiple previous admissions with worse clinical outcomes was consistent among subgroups. Sex affected their association (P value for interaction = 0.048), and the association of multiple previous admissions with higher rates of composite endpoints was more significant among female patients (Figure 3A). There was no significant interaction in the various subgroup analysis comparing the multiple and single previous admission groups (Figure 3B).
| Multiple vs. None | Multiple vs. Single | Single vs. None | |||||||
|---|---|---|---|---|---|---|---|---|---|
| P value | HR | 95% CI | P value | HR | 95% CI | P value | HR | 95% CI | |
| A. All-cause death | |||||||||
| Unadjusted | <0.001 | 1.77 | 1.36–2.31 | 0.001 | 1.72 | 1.26–2.34 | 0.80 | 1.04 | 0.79–1.36 |
| Adjusted | 0.041 | 1.37 | 1.01–1.85 | 0.012 | 1.53 | 1.10–2.13 | 0.46 | 0.89 | 0.66–1.20 |
| B. HF readmission | |||||||||
| Unadjusted | <0.001 | 3.09 | 2.48–3.84 | <0.001 | 2.03 | 1.60–2.58 | <0.001 | 1.54 | 1.23–1.93 |
| Adjusted | <0.001 | 2.83 | 2.19–3.65 | <0.001 | 1.90 | 1.47–2.44 | 0.001 | 1.51 | 1.18–1.92 |
| C. All-cause death+HF readmission | |||||||||
| Unadjusted | <0.001 | 2.56 | 2.11–3.11 | <0.001 | 1.91 | 1.54–2.37 | 0.003 | 1.35 | 1.11–1.64 |
| Adjusted | <0.001 | 2.27 | 1.81–2.84 | <0.001 | 1.82 | 1.45–2.29 | 0.026 | 1.27 | 1.03–1.57 |
Models were adjusted for age, sex, BMI, eGFR, hemoglobin, LVEF, atrial fibrillation, hypertension, MRA at discharge, ICD, and CRT. CI, confidence interval; HR, hazard ratio. Other abbreviations as in Tables 1,2.
Adjusted hazard ratio of the composite endpoint in each subgroup, comparing between (A) the “no previous admission” group and the “multiple previous admissions” group and (B) the “single previous admission” group and the “multiple previous admissions” group. Af, atrial fibrillation; BMI, body mass index; BP, blood pressure; eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction.
In this study, the major findings were as follows: (1) a history of multiple previous HF admissions was an independent predictor for all-cause death and HF readmission compared with a history of a single or no previous admissions, even after adjustment for known prognostic factors; and (2) a history of a single previous HF admission was an independent predictor for HF readmission, but not all-cause death, compared with no previous admission history.
Previous HF admission has been reported as a strong risk factor for death of patients admitted for AHF.15 Bello et al advocated that previous HF hospitalization was a strong predictor of adverse cardiac outcomes in both HF with reduced EF and HF with preserved EF patients enrolled in the Candesartan in Heart failure: Reduction in Mortality and morbidity (CHARM) trials.16 However, the significance of the number of previous HF admissions was not evaluated in those studies. Furthermore, these data from a randomized-control study cannot be generalized to the general population of HF patients because of several exclusions that limited enrollment of the elderly or coexistence of comorbidities. One of the strengths of our study is that we focused on the number of HF hospitalizations as an important factor for prognosis, using a contemporary multicenter HF registry. We demonstrated that a history of multiple previous admissions was an independent determinant of subsequent survival and this relationship was universal among several HF subpopulations. These findings suggested that whether a patient has multiple previous HF admissions could be a simple and important surrogate marker of subsequent adverse events.
In addition to biomarkers such as kidney function17 and B-type natriuretic peptide level,18 the effect of a patient’s background has been analyzed to elucidate the optimal timing for multimodal interventions for HF. For instance, depending on age or the extent of comorbidities, there was a different effect of a nurse-led multidisciplinary home-based intervention on prognosis across the spectrum of patients with chronic heart disease.19 In our study, the number of previous admissions, an administrative parameter, showed a possible strong association with subsequent HF hospitalization and all-cause death. This simple measure, which identifies patients at high risk, may facilitate the medical team’s decision making; that is, changing prescriptions, nursing care, and the application of intensive healthcare support system (frequent outpatient follow-up, rehabilitation programs, etc.), as well as aiding in the targeting of devices (mechanical heart support).20,21
Although elucidation of the mechanism underlying the independent association of multiple previous admissions with a higher risk of subsequent worse prognosis is beyond the scope of this study, this association may likely reflect progressive HF deterioration during repeated hospitalizations. It was recently suggested that myocardial damage during HF hospitalization, owing to hemodynamic instability or in-hospital treatment such as inotropes, could be an important therapeutic target from the perspective of myocyte preservation.22 Although the ideal therapy to prevent subsequent death or frequent HF readmission following repeated HF hospitalization was not elucidated in this study, large cohort or randomized-control trials should investigate this issue in the near future. Interestingly, a history of multiple, but not single, previous admissions had a higher risk of all-cause death than no previous admissions in our Japanese study cohort from 2005 to 2014. Contrary to our findings, Setoguchi et al9 revealed that all-cause death increased after each HF hospitalization among patients with a first hospitalization for HF in a Canadian healthcare utilization dataset from 2000 to 2004; a history of a single previous admission was also associated with a higher risk of all-cause death than for no previous admissions.9 The 30-day all-cause mortality in patients with no previous HF admission was 12%, and the 1-year mortality was 34% in this Canadian cohort, whereas in our cohort mortality was 1.5% and 10.1%, respectively. Therefore, it is plausible that the better prognosis in our study cohort blunted the effect of a single previous admission history on prognosis. Accumulating evidence suggests major regional variations in the management and outcomes in patients with HF.3,23 As compared with those in Western registries, data from 3 Japanese large-scale acute HF registries, including ours, show that relatively longer hospital stays are associated with infrequent readmissions and lower 1-year mortality.24 Therefore, the characteristics and prognostic effect of admission for HF in Japan could be different from those in Western countries. The relationship between readmission and subsequent death is still inconclusive, and further international studies are needed to elucidate whether the relationship between the number of previous HF hospitalizations and subsequent event-free survival differs according to regional variations or differences in health insurance systems.
Understanding the relationship between multiple previous HF admissions and subsequently worse clinical outcomes could have significant clinical implications. First, our results could be introduced to the educational program for HF patients, especially for patients with no previous HF admissions (first admission for de novo HF) or previous single HF admission patients (second admission), warning them that their risk of death increases following a third HF admission. Sharing this information could motivate patients to adhere to their recommended lifestyle modifications, such as drug adherence, diet restriction, and regular exercise, in order to prevent readmission. Second, it can be used to discuss the best timing for considering a living-will and advanced care planning. Healthcare staff need time for individualized discussions with patients and their families about prognosis and decisions regarding future management if there is limited time remaining.25 The best timing for such individualized discussions is when HF patients with multiple previous admissions (i.e., ≥2 admissions) are readmitted despite optimal medical therapy and good adherence with lifestyle modifications.
Study LimitationsFirst, our survey was an observational cohort study and despite covariate adjustment, unmeasured or unknown variables may have influenced the outcomes. Second, this analysis included only patients who survived their previous hospitalization and lived to be enrolled in the WET-HF registry, and patients with the most severe HF could have died before enrollment. Although the risk of all-cause death did not differ between patients with de novo HF and those with a single previous HF admission, selection bias could influence this relationship. Third, the interval between the previous and index admissions was not assessed in this registry. It is interesting that Pocock et al revealed that previous hospitalization for HF within 6 months, but not beyond 6 months, was an independent predictor of all-cause death in a large contemporary population with chronic HF.26 We need to investigate the effect of the time of readmission on subsequent clinical outcomes.
HF patients with multiple previous HF admissions had significantly higher risk of all-cause death and HF readmission within 3 years than patients with no or single previous HF admissions. Patients with multiple previous HF admissions should be carefully monitored following hospital discharge to improve their long-term clinical outcomes. Interventions, including advanced therapies, multidisciplinary care, and advanced care planning, should be targeted to these vulnerable patients.
This study was supported by Grant-in-Aid for Young Scientists (JPSS KAKENHI, 18K15860 [Y. Shiraishi]), Grant-in-Aid for Scientific Research (23591062, 26461088 [T.Y.], 17K09526 [T.K.]), Health Labour Sciences Research Grant (14528506 [S.K.]), the Sakakibara Clinical Research Grant for Promotion of Sciences (2012, 2013, 2014 [T.Y.]), and Grant from the Japan Agency for Medical Research and Development (201439013C [S.K.]).
None.
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-18-1087
