Introduction

Anemia is a common comorbidity in patients with heart failure (HF) and an independent risk factor for long-term outcomes.1,2,3,4,5,6,7,8 In addition to anemia, renal dysfunction (RD) is a significant risk factor in both acute and chronic HF.9,10,11 HF, anemia, and RD interact negatively, resulting in the cardio-renal-anemia syndrome (CRAS).12,13,14,15 However, the therapeutic strategies for anemia and RD in HF have not been established.

In patients with chronic kidney disease (CKD) who had hemoglobin (Hb) levels below 10 g/dL had a worse prognosis compared to those with Hb levels between 10 and 12 g/dL. On the other hand, administration of erythropoiesis-stimulating agents (ESAs) targeting Hb ≥ 13 g/dL increased the risk of cardiovascular events.16,17,18 Recent studies demonstrated no significant difference in renal outcomes between patients with Hb 9–11 g/dL and 11–13 g/dL.19,20 These findings raise the possibility that the appropriate cutoff for anemia may be around 10 g/dL. In fact, a recent Japanese guideline for CKD recommends maintaining Hb levels between 10–12 g/dL with Hb < 10 g/dL as the therapeutic target. With respect to HF, the RED-HF trial demonstrated that administration of an ESA targeting Hb ≥ 11 g/dL did not improve the prognosis of HF patients and instead increased thrombosis in HF patients.21 As in CKD, maintaining Hb levels in the normal range may not be beneficial in HF patients with anemia. However, the impact of the severity of anemia, especially Hb < 10 g/dL, associated with RD in HF patients has not been fully elucidated. In particular, the therapeutic target range of Hb levels in HF remains unknown.

The status of RD and anemia can change during hospitalization for HF. Therefore, their management is critical not only in outpatients with HF but also in patients hospitalized for HF. Most studies on HF and anemia have evaluated Hb levels at the time of HF admission. To our knowledge, only a few studies have reported short-term changes in Hb levels in HF.22,23 The information on changes in anemia severity associated with RD during HF hospitalization may be useful in the management of CRAS.

The Japanese Cardiac Registry of Heart Failure in Cardiology (JCARE-CARD) prospectively studied the characteristics and treatments of a large sample of patients hospitalized with HF at teaching hospitals in Japan from January 2004 to June 2005 and the outcomes including death and rehospitalization were followed until 2007.24 The JCARE-CARD database is useful for studying the transitional status of anemia and RD during HF hospitalization. The aim of the present study was to analyze the prognostic impact of anemia severity, especially Hb < 10 g/dL, associated with RD on mortality and rehospitalization, as well as the changes in RD and anemia, in patients due to worsening HF by using the JCARE‐CARD database.

Results

Impact of admission anemia severity on long-term outcomes in HF patients with RD

In the JCARE database, 1,608 patients were identified to have creatinine and Hb levels at admission. Of these, 569 patients had normal renal function (eGFR ≥ 60 ml/min/1.73 m2) and 1,039 patients had RD (eGFR < 60 ml/min/1.73 m2) (Table 1).

Table 1 Patient characteristics according to severity of anemia in the presence or absence of RD.
Full size table

In patients with normal renal function, 344 (60%), 169 (30%), and 56 (10%) had non-anemia (Hb ≥ 13.0 g/dL in males, ≥ 12.0 g/dL in females), mild anemia (Hb ≥ 10.0 g/dL, < 13.0 g/dL in males, ≥ 10.0 g/dL, < 12.0 g/dL in females), and moderate/severe anemia (Hb < 10 g/dL), respectively (Table 1). According to the severity of anemia, patients were older and more likely to be female and have valvular heart disease, higher creatinine levels, smaller LV diameters, and higher a LVEF. Consistently, patients with moderate/severe anemia had more HFpEF (Supplementary Fig. 1). Hb levels in non- (14.2 ± 1.4 g/dL vs. 14.4 ± 1.5 g/dL, P = 0.119), mild (11.4  ± 0.8 g/dL vs. 11.4 ± 0.8 g/dL, P = 0.983), and moderate/severe anemia groups (8.6 ± 1.1g/dL vs. 8.8 ±  1.4 g/dL, P = 0.171) were similar between patients with and without RD.

The mean post-discharge follow-up was 698 ± 349 days. The composite of all-cause death or HF rehospitalization and HF rehospitalization did not differ between these groups (Fig. 1A and C). All-cause death was higher in patients with mild and moderate/severe anemia than in those without anemia (Fig. 1B). In Cox hazard analysis, the composite of all-cause death or HF rehospitalization and HF rehospitalization in mild anemia group (aHR 0.970, 95% CI 0.619–1.497) and moderate/severe anemia group (aHR 1.424, 95% CI 0.758–2.560) was compared with the non-anemia group (Table 2).

Fig. 1
figure 1

Long-term outcomes according to severity of anemia in the presence or absence of RD on admission among HF patients. Kaplan–Meier, event-free curves free from all-cause death or rehospitalization (A and D), all-cause death (B and E), and rehospitalization due to worsening HF (C and F) in HF patients with and without RD (eGFR < 60 ml/min/1.73 m2) according to Hb levels on admission; normal Hb (Hb ≥ 13.0 g/dL in males, ≥ 12.0 g/dL in females), mild anemia (Hb ≥ 10.0 g/dL, < 13.0 g/dL in males, ≥ 10.0 g/dL, < 12.0 g/dL in females), moderate severe anemia (Hb < 10.0 g/dL).

Full size image
Table 2 HRs for Outcomes According to Severity of Anemia in the Presence or Absence of RD.
Full size table

Among patients with RD, 372 (41%), 326 (36%), and 196 (23%) had non-anemia, mild anemia, or moderate/severe anemia, respectively. Like those with normal renal function, patients with lower Hb levels were older and likely to be more female and had valvular heart disease, higher creatinine levels, smaller LV diameters, higher LVEF (Table 1), and more HFpEF (Supplementary Fig. 1). However, these patients were less likely to be prescribed beta-blockers, ACEi/ARB, and MRA. The mean post-discharge follow-up was 584 ± 376 days. The composite of all-cause death or HF rehospitalization, all-cause death, and HF rehospitalization were higher in patients with mild anemia than in patients without anemia and the highest in patients with moderate/severe anemia among groups (Fig. 1D–F). In Cox hazard analysis, the composite of all-cause death or HF rehospitalization and HF rehospitalization was higher in moderate/severe anemia group (aHR 1.717, 95% CI 1.257–2.343, P < 0.001) compared with non-anemia group (Table 2). The same trends were observed for all-cause death and HF rehospitalization (Table 2).

These results suggest that the severity of anemia has a greater impact on long-term outcomes in HF patients with RD than in those with normal renal function.

Impact of moderate/severe anemia associated with RD on long-term outcomes in HF patients

A recent guideline for CKD defines moderate/severe anemia (Hb < 10 g/dL) as a therapeutic target in patients with renal anemia. To investigate the impact of moderate/severe anemia associated with RD on admission in HF patients, we analyzed four groups: non-RD and non/mild anemia, RD and non/mild anemia, non-RD and moderate/severe anemia, and RD and moderate/severe anemia (Table 3). Patients with RD and moderate/severe anemia were older, more female, and more likely to have s severe NYHA functional class, previous HF admission, ischemic and valvular heart diseases, hyperuricemia, previous myocardial infarction, higher BNP levels, and lower prescription rates of beta-blocker, ACEi/ARB, and MRA. On the other hand, patients with RD had the highest prevalence of HFpEF among all groups. Hb levels were comparable in patients with non-RD and moderate/severe anemia and those with RD and moderate/severe anemia (8.8 ± 1.4 g/dL vs. 8.6 ± 1.1 g/dL, P = 0.171) as described above. On the other hand, patients with RD and moderate/severe anemia had lower eGFR levels compared with those with RD and non/mild anemia (30.2 ± 15.0 g/dL vs. 41.6 ± 12.6 g/dL, P < 0.001).

Table 3 Patient characteristics according to the presence of RD and/or moderate/severe anemia.
Full size table

In-hospital death was occurred in 91 patients (non/mild anemia:11, RD:45, moderate/severe anemia, RD and moderate/severe anemia:31). In a multivariate logistic analysis, BMI and systolic BP were negatively associated with in-hospital death, while RD and RD with moderate/severe anemia were positively associated with it (odds ratio (OR) 5.412, 95%CI 2.007–15.431, P < 0.001) (Supplementary Table 1). The composite of all-cause death or HF rehospitalization increased in the order of non-RD and non/mild anemia, RD and non/mild anemia, non-RD and moderate/severe anemia, and RD and moderate/severe anemia (Fig. 2A). The same pattern was observed in all-cause death and HF rehospitalization (Fig. 2B and C). In Cox hazard analysis, the composite of all-cause death or HF rehospitalization (aHR 2.120, 95% CI 1.559–2.881, P < 0.001), all-cause death (aHR 3.340, 95% CI 2.154–5.251, P < 0.001), and HF rehospitalization (aHR 2.212, 95% CI 1.559–2.881, P < 0.001) were the highest in patients with RD and moderate/severe anemia (Table 4).

Fig. 2
figure 2

Long-term outcomes according to the presence of RD and/or moderate/severe anemia on admission among HF patients. Kaplan–Meier, event-free curves free from all-cause death or rehospitalization (A), all-cause death (B), and rehospitalization due to worsening HF (C) in HF patients with/without RD (eGFR < 60 ml/min/1.73 m2) and no/mild anemia (Hb ≥ 10.0 g/dL) or moderate/severe anemia (Hb < 10.0 g/dL). Group 1; non-RD and non/mild anemia, Group 2; RD and non/mild anemia, Group 3; non-RD and moderate/severe anemia, Group 4; RD and moderate/severe anemia.

Full size image
Table 4 HRs for outcomes according to the presence of RD and/or moderate/severe anemia.
Full size table

We also analyzed the impact of RD and moderate/severe anemia in each HF type. These outcomes were greatest in patients with RD and moderate/severe anemia in HF with reduced ejection fraction (HFrEF) (Supplementary Fig. 2), HF with mildly reduced EF (HFmrEF) (Supplementary Fig. 3), and HF with preserved EF (HFpEF) (Supplementary Fig. 4). The presence of RD and moderate/severe anemia was independently associated with the composite of all-cause death or HF rehospitalization (Supplementary Table 2).

These findings suggest that RD and moderate/severe anemia have a synergistic effect on poor outcomes in HF patients.

Changes in RD and anemia during HF hospitalization

During HF hospitalization, the status of RD and anemia can change. Thus, we examined changes in RD and anemia from admission to discharge in each group (non-RD and non/mild anemia defined as group 0 [383 patients], RD and non/mild anemia defined as group 1 [606 patients], non-RD and moderate/severe anemia defined as group 2 [45 patients], and RD and moderate/severe anemia defined as group 3 [211 patients]) (Fig. 3A). Most patients in each group did not change their status (non-RD and non/mild anemia: 88%, RD and non/mild anemia: 83%, non-RD and moderate/severe anemia:83%, and RD and moderate/severe anemia: 75%). In non-RD and non/mild anemia group at admission, 12% of patients had moved on to other groups at discharge. In RD and non/mild anemia group on admission, 12% transitioned to non/mild anemia, 5% progressed to moderate/severe anemia or RD with moderate/severe anemia. In non-RD and moderate/severe anemia patients on admission, 7%, 8%, and 1% transitioned to RD and moderate/severe anemia, non-RD and moderate/severe anemia, and non-RD and non/mild anemia, respectively. In RD and moderate/severe anemia patients on admission, 14%, 8%, and 1% moved on to non-RD and moderate/severe anemia, RD and non/mild anemia, and non-RD and non/mild anemia, respectively. Overall, 6% of patients’ condition worsened, while 10% improved during their HF hospitalization.

Fig. 3
figure 3

Changes in RD and anemia during HF hospitalization. (A) Changes in RD and anemia from admission to discharge. Group 1; non-RD and non/mild anemia, Group 2; RD and non/mild anemia, Group 3; non-RD and moderate/severe anemia, Group 4; RD and moderate/severe anemia. (B) Kaplan–Meier, event-free curves free from all-cause death or rehospitalization according to changes in RD and anemia.

Full size image

In Kaplan–Meier analysis (Fig. 3B), the composite of all-cause death or HF rehospitalization was lowest in patients who moved from group 0 to group 0, and highest in those who moved from group 3 to group 3. Surprisingly, whereas patients who transitioned from group 0 to group 3 tended to have worse outcomes, those who transitioned from group 3 to other groups had better outcomes.

In the logistic regression model, predictors of transition to Group 3 were ischemic etiology, valvular etiology, systolic BP, Hb (OR 0.678, 95%CI 0.489–0.941, P = 0.020) and eGFR levels (OR 0.962, 95%CI 0.938–0.986, P = 0.020) (Table 5). On the other hand, transition to Group 0 was predicted by Hb (OR 1.316, 95%CI 1.148–1.509, P < 0.001), eGFR (OR 1.040, 95%CI 1.022–1.058, P < 0.001) and LVEF (OR 1.025, 95%CI 1.005–1.046, P = 0.012) (Table 5).

Table 5 Multivariate Predictors of Group Transition.
Full size table

We then investigated the association between the increase in Hb levels and prognosis in patients with non/mild and moderate/severe anemia, respectively. We analyzed 1,026 patients who had Hb values on admission and discharge. Of 839 patients with non/mild anemia, 72 had increased Hb values during hospitalization, but 767 did not (ΔHb: 0.47 ± 0.32 vs -0.25 ± 0.56, p < 0.0001). There was no trend toward improved prognosis with improved anemia (Supplementary Fig. 5A). Of 187 cases of moderate/severe anemia, whereas 27 had increased Hb levels, 160 did not (ΔHb: 0.45 ± 0.29 vs -0.47 ± 1.40, p < 0.001). Patients with increased Hb levels during hospitalization tended to have a better prognosis, although not significantly, compared to those without it (Supplementary Fig. 5B).

Discussion

In this study, we found that the severity of anemia was more strongly associated with poor outcomes in HF patients with RD than in those without RD. Furthermore, moderate/severe anemia (Hb < 10 g/dL) and RD (eGFR < 60 ml/min/1.73 m2) had a similar and additive effect on the prognosis of HF patients. In addition, changes in anemia and RD levels during hospitalization were associated with post-discharge outcomes in HF patients. These findings provide important insights into the management of HF patients with anemia.

In general, the presence of cardiovascular risk factors tends to be more frequent in males. However, patients with anemia and/or RD were more likely to be females than males (Tables 1 and 3). This may be due to the fact that females are more likely to be older than males in this study. In addition, the lower prescription rate of beta-blocker might also be associated with age.

Numerous studies have shown that anemia is strongly associated with a poor prognosis in HF patients.1,2,3,4,5,6,7,8 However, the Japanese guideline on diagnosis and treatment of acute and chronic HF in 2017 did not describe an effective treatment for anemia in HF.25 In particular, the optimal therapeutic range for Hb levels remains unknown. A recent Japanese guideline for CKD recommends Hb < 10 g/dL as the therapeutic target. Therefore, in this study, we focused on the impact of Hb < 10 g/dL in HF patients. Mild anemia (Hb ≥ 10.0 g/dL, < 13.0 g/dL in males, ≥ 10.0 g/dL, < 12.0 g/dL in females) affected all-cause mortality but not the composite of all-cause death or HF rehospitalization and HF rehospitalization among HF patients without RD. On the other hand, mild anemia was associated with the composite outcome, all-cause mortality, and HF rehospitalization in HF patients with RD (Fig. 1 and Table 2). One of the reasons for this difference could be the small number of patients without RD compared to those with RD. However, at least, this study shows the relevance of anemia in RD rather than anemia in normal renal function.

While patients with moderate/severe anemia had the worst prognosis regardless of RD, the risk of adverse outcomes was higher in patients with RD than in those without RD (Table 2). In addition, Hb levels in each anemia severity class were comparable in patients with and without RD. In particular, Hb levels in moderate to severe anemia without/with RD were same (8.8 ± 1.4 g/dL and 8.6 ± 1.1 d/dL, respectively) (Table 1). These findings suggest that RD exacerbates adverse effects of anemia in HF patients regardless of Hb levels. Anemia and CKD are known to have an additive effect on poor outcomes in HF patients.12,15 However, the impact of anemia severity has not been understood. In this study, moderate/severe anemia and RD had a similar impact on the composite of all-cause death or HF rehospitalization, all-cause death, and HF rehospitalization (Table 4). In addition, moderate/severe anemia and RD had an additive negative effect on long-term outcomes in HF patients (Table 4). Furthermore, the combination of moderate/severe anemia and RD was linked to poorer outcome in all HF types (HFrEF, HFmrEF, and HFpEF). The worsening effects of the combination of moderate/severe anemia and RD are thought to be complicated. For example, the depleting treatment with a good response leads to a worsening of renal function and an improvement in anemia. However, in this study, only a few HF patients had improved anemia. This may be because the changes in Hb values during hospitalization is small. In fact, of 839 patients with non/mild anemia, 72 had increased Hb values during hospitalization, but 767 did not (ΔHb: 0.47 ± 0.32 vs -0.25 ± 0.56, p < 0.0001). Of 187 cases of moderate/severe anemia, whereas 27 had increased Hb levels, 160 did not (ΔHb: 0.45 ± 0.29 vs -0.47 ± 1.40, p < 0.001). In general, the period hospitalization in Japan is longer than that in western countries. We speculate that Hb values increase immediately after depletive treatment, and then other factors, such as worsened renal dysfunction, negatively affect anemia status during hospitalization. From another perspective, eGFR levels in HF patients with moderate/severe anemia and RD are likely to be lower than in patients with RD alone. Therefore, moderate/severe anemia in HF patients with RD requires intensive treatment. Several previous studies have demonstrated that elevated Hb levels at or above normal are detrimental not only in CKD but also in HF patients.16,17,18,21,26 Based on ours and previous findings, a therapeutic range of Hb at 10–12 g/dL may be appropriate even in HF patients, as recommended by the Japanese CKD guideline in 2023.

The timing of treatment for anemia is also important. Hb levels may fluctuate with RD during HF hospitalization. A rapid increase in Hb during the early phase of HF hospitalization was independently associated with a favorable outcome.22 Patients with acute Hb drop in levels during HF hospitalization had poor outcomes.23 However, the baseline Hb levels in these studies were 12.6 g/dL and 11.4 g/dL, respectively. Short-term changes in Hb levels in patients with moderate/severe anemia have not been investigated. Our findings have implications for this point. Changes from non/mild to moderate/severe anemia, with or without RD, were associated with a poor prognosis (Fig. 3B). In addition, Hb and eGFR levels predict worsening and improving anemia and RD (Table 3). Intriguingly, LVEF was an independent positive predictor of improving anemia and RD, suggesting that anemia and RD may be more likely to improve in patients with HFpEF. Patients with increased Hb levels during hospitalization tended to have a better prognosis than those without it (Supplementary Fig. 5B). The lack of statistically significant differences may be due to the sample size. Among patients from group 0 to group 1, 2, or 3, there was only one patient who transited group 0 to group 3 during hospitalization (data not shown). The other patients transited from group 0 to group 1 or 2. Thus, we cannot fully elucidate the characteristics of patients who transited group 0 to group 3. We need to investigate this issue in larger cohorts. However, our findings indicate that the prevention and treatment of moderate/severe anemia during hospitalization may be effective in improving outcomes for HF patients.

Regardless of the presence or absence of anemia, intravenous iron therapy was associated with improved exercise capacity and a lower risk of HF hospitalization and cardiovascular death in HF patients.27,28 However, intravenous iron therapy for HF was uncommon in 2004–2005. Thus, intravenous iron administration appears to have little or no effect in this study. Recently, hypoxia-inducible factor prolyl hydroxylase (HIF-PH) inhibitors have been used to treat renal anemia.29,30 A randomized controlled study is currently underway to determine the safety and efficacy of a HIF-PH inhibitor in patients with HF and renal anemia is ongoing.31 The effect of anemia treatment with HIF-PH inhibitors on prognosis in HF patients with and without renal anemia should be prospectively investigated in the future.

Study limitations

This study has several potential limitations that should be acknowledged. First, this study did not collect data on anemia treatment, such as ESAs, iron supplements, or blood transfusions. The effects of these treatments could not be excluded. Second, serum iron, ferritin, transferrin saturation, and erythropoietin levels were not determined in this study. Iron deficiency and renal anemia could not be properly diagnosed. The etiology of anemia might also have a significant prognostic impact, rather than just the Hb values. However, numerous previous studies have demonstrated that Hb levels are negatively correlated with prognosis in HF patients without investigating the cause of anemia.1,2,3,4,5,6,7,8 Like these studies, we focused on the severity of anemia in this study. Third, the JCARE-CARD is a previous study, in which patients were registered from January 2004 to June 2005 and the outcomes, including death and rehospitalization, were followed until 2007. At this time, neither sodium-glucose co-transporter 2 (SGLT2) inhibitors, ARNI (valsartan/sacubitril), ivabradine, nor HIF-PH inhibitors were used. Thus, the effect of improved medical management on patients with anemia and RD remains uncertain in this study. Extrapolation of our results to current clinical practice requires caution. Nonetheless, this study demonstrates the impact of moderate/severe anemia associated with CKD on HF patients, in addition to the relationship between these diseases. Furthermore, our data may provide reference points or a basis for investigating the effectiveness of improved treatments in the future. Forth, nutritional indicators such as albumin might affect the prognosis of HF patients with or without RD. However, our database did not contain serum albumin levels. Fifth, the exact cut-off point of moderate-severe anemia in terms of its impact on heart failure is still unknown. To elucidate this point, we need further investigations (for example, the cubic-spline analysis of clinical outcomes and Hb values) in a larger cohort. Finally, there may be unmeasured biases that affect the results. In particular, the effect of changes in Hb and eGFR over the follow-up period on long-term outcomes is unclear. To address these concerns, further investigation in larger-scale, more recent studies is needed. Furthermore, prospective studies are required to determine optimal Hb levels and how to improve anemia to improve prognosis in HF patients.

Conclusion

Moderate/severe anemia associated with RD had a negative impact on the prognosis of HF patients. The management of anemia and RD, particularly during HF hospitalization, is critical for HF patients. Further studies are needed to elucidate the impact of the etiology of anemia and the effect of recent advances in treatment in HF patients with anemia and RD.

Methods

Patient selection

Clinical data was collected from the JCARE‐CARD database. The baseline data includes (i) demographics; (ii) cause of HF; (iii) precipitating cause; (iv) co‐morbidities; (v) complications; (vi) clinical status; (vii) laboratory data; (viii) electrocardiographic and echocardiographic findings; and (ix) treatment including discharge medications. The discharge laboratory data were also collected. Follow‐up data was collected at least 1 year after the index admission. Patients with both Hb and creatinine levels at admission were selected.

Definition of anemia and renal dysfunction

Anemia was defined as a Hb level of < 13.0 g/dl in males and < 12.0 g/dl in females, based on World Health Organization criteria.2 Furthermore, mild and moderate/severe anemia were defined as Hb levels of ≥ 10.0 g/dl, < 13.0 g/dl in males and < 12.0 g/dl in females, and < 10.0 g/dl in both genders as same as previous studies.32,33,34 RD was defined as an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m2 based on stages of CKD.10.

Outcomes

The primary outcome of this study was a composite of all‐cause death or HF rehospitalization. Secondary outcomes were all‐cause death, HF rehospitalization, and in-hospital death.

Statistical analysis

Patients with mild anemia (Hb ≥ 10.0 g/dl, < 13.0 g/dL in males; Hb ≥ 10.0 g/dl, < 12.0 g/dL in females) and moderate/severe anemia (Hb < 10 g/dL) were compared with those without anemia (Hb ≥ 13.0 g/dL in males, ≥ 12.0 g/dL in females) according to RD. Patient characteristics, such as age, gender, New York Heart Association (NYHA) functional class, previous HF admission, echocardiographic data, co‐morbidities, previous procedures, and therapeutic agents, were compared using Pearson χ2 test for categorical variables and Student’s t‐test, or Wilcoxon rank sum test for continuous variables, as appropriate. Results were presented as mean ± SD or median with interquartile range (IQR). The cumulative incidence of outcomes was estimated using the Kaplan–Meier method and a Cox proportional hazards model including clinically relevant covariates. Adjusted hazard ratios (aHRs) were calculated using the Cox regression model and presented with a 95% confidence interval (CI) and P value. Covariates were selected based on previous studies investigating prognostic factors for HF patients.35,36 In detail, demographics (age, sex, BMI), previous HF admission, causes of heart failure (ischemic, hypertensive, dilated cardiomyopathy, valvular heart disease), medical history (hypertension, stroke, smoking, chronic atrial fibrillation or flutter), NYHA functional class, LVEF, and medication use (ACE inhibitor or ARB, β-blocker, MRA, digitalis) were selected. In-hospital death rates and predictors of group transition were assessed using the logistic regression model. In addition to factors shown above, blood pressure and heart rate were included as covariates. All tests were two-tailed with P < 0.05 indicating statistical significance. All analyses were conducted using JMP Pro (version 14.2.0, SAS Institute, Cary, North Carolina).

Ethical statement

This study protocol was designed to ensure compliance with the Declaration of Helsinki and the Guidelines for Epidemiological Research published by the Japanese Ministry of Health, Labour and Welfare. The original study protocol was approved by the Institutional Review Board (IRB) of Kyushu University. Because this study investigated a nationwide administrative database (unconnectable anonymized data), we could not obtain informed consent from each patient. The IRB of Kyushu University approved the waiver of informed consent. As a result, we announced our research on our website for a certain period of time instead of obtaining informed consent.