Mode of death in elderly and super‐elderly patients with acute heart failure: Insights from Japanese heart failure registry

Abstract Background In Japan, both the prevalence of the elderly and super‐elderly and those of acute heart failure (AHF) have been increasing rapidly. Methods This registry was a prospective multicenter cohort, which enrolled a total of 1253 patients with AHF. In this study, 1117 patients' follow‐up data were available and were categorized into three groups according to age: <75 years old (nonelderly), 75–84 years old (elderly), and ≥ 85 years old (super‐elderly). The endpoint was defined as all‐cause death and each mode of death after discharge during the 3‐years follow‐up period. Results Based on the Kaplan–Meier analysis, a gradually increased risk of all‐cause death according to age was found. Among the three groups, the proportion of HF death was of similar trend; however, the proportion of infection death was higher in elderly and super‐elderly patients. After adjusting for potentially confounding effects using the Cox and Fine–Gray model, the hazard ratio (HR) of all‐cause death increased significantly in elderly and super‐elderly patients (HR, 2.60; 95% confidence interval [CI], 1.93–3.54 and HR, 5.04; 95% CI, 3.72–6.92, respectively), when compared with nonelderly patients. The highest sub‐distribution HR in detailed mode of death was infection death in elderly and super‐elderly patients (HR, 4.25; 95% CI, 1.75–10.33 and HR, 10.10; 95% CI, 3.78–27.03, respectively). Conclusions In this population, the risk of all‐cause death was found to increase in elderly and super‐elderly. Elderly patients and especially super‐elderly patients with AHF were at a higher risk for noncardiovascular death, especially infection death.


| INTRODUCTION
In 2018, life expectancy in Japan was 87.3 and 81.2 years old in females and males, respectively. 1 Japan has the highest proportion of elderly and super-elderly people in the world. Both the prevalence of the elderly and super-elderly and those of heart failure (HF) have been increasing rapidly in most countries worldwide. 2,3 Improvements in the treatment of HF are shown in many guidelines, and patients with HF have better outcomes. 4,5 However, many patients with acute HF (AHF) were older and had many comorbidities, and previous large randomized clinical trials had a limitation of a direct adoption for all patients in real-world settings because of various exclusion criteria. 6,7 Therefore, information for detailed prognosis in elderly and superelderly patients are limited.
Although well-established medical and support therapies improved survival discharge and prolonged prognosis in patients with AHF, morbidity and mortality are still high especially in elderly patients with AHF. [8][9][10] Some single-center studies showed detailed mode of death in elderly and super-elderly patients with AHF. 11,12 Elderly and super-elderly patients with AHF had the tendency to have noncardiovascular death and infection death. However, few multicenter prospective studies describing the detailed mode of death during long-term follow-up period still exist. Further data and understanding about the detailed mode of death in patients with AHF are necessary.
The Kitakawachi Clinical Background and Outcome of Heart Failure (KICKOFF) Registry was a prospective, multicenter registry of Japanese patients with AHF. 13 A total of 13 hospitals in Osaka, Japan, participated in the registry. This study aimed to use the database to evaluate the detailed mode of death in patients with AHF according to age especially focusing on elderly and super-elderly patients.

| Study design
The AHF patients' data from the KICKOFF Registry, which were registered between April 2015 and August 2017, were analyzed. A total of 13 hospitals, consisting of one cardiovascular center and 12 small-or medium-sized hospitals in the north of Kitakawachi (Hirakata City, Neyagawa City, and Katano City) and Yawata, participated in the registry. 13 The patients were diagnosed with HF based on the Framingham criteria when the presence of at least two of the major criteria, or one major and two minor criteria, in each hospitalization was identified. 14 This registry has no exclusion criteria. The KICKOFF Registry is registered with the University Hospital Medical Information Network Clinical Trials Registry in Japan (UMIN000016850). The registry design has been described in detail elsewhere. 7,13 The study protocol was in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the ethics committee of the Hirakata Kohsai Hospital (Osaka, Japan). All participants provided written informed consent before their enrollment in the study. The study did not affect any treatment or any other methods of outpatient care.

| Patient and outcome data definitions
Among the 1253 patients enrolled in this study, 1118 were alive at discharge, and one patient was excluded because of missed follow-up data. Finally, 1117 patients were available and categorized into three groups according to age: <75 years old (nonelderly), 75-84 years old (elderly), and ≥ 85 years old (super-elderly) with reference to previous studies. 3,11,12 The other definitions of each comorbidity were described in our previous paper. 13 Follow-up data were primarily collected by reviewing hospital records, and additional follow-up information was obtained via telephone or mail contact with the patients or their relatives collected at 6 months, 1 year, 2 years, and 3 years after discharge. All-cause death and detailed mode of death during the follow-up period were evaluated as adverse outcome. All-cause death was divided into two groups: cardiovascular death and noncardiovascular death. Cardiovascular death was defined as death due to HF, sudden death, acute coronary syndrome (ACS), stroke or intracranial hemorrhage, and other cardiac death causes. HF death was defined as a death that occurred as a result of worsening or intractable HF. Sudden death was defined as an unexplained death of a previously stable patient. ACS included acute myocardial infarction and unstable angina. Noncardiovascular death was defined as death due to infection, cancer, renal or liver failure, and other causes. The mode of death in each patient was defined as the critical condition that initiated the train of events leading directly to death. Only one underlying mode of death was applied to each death event. The mode of death was classified as unknown when the mode of death could not be classified, or the detailed information of death was lacking.

| Statistical analysis
The clinical baseline characteristics were compared among the three groups using Cochran-Armitage tests and Dunnett's tests, for categorical variables and continuous variables, respectively. Continuous variables are expressed as a mean ± SD or interquartile range, and categorical variables are expressed as numbers and percentages. Crude mode of death rates per 100 person-years with 95% confidence interval (CI) was calculated in each group. The Kaplan-Meier method was applied to evaluate the cumulative incidences of all-cause death, cardiovascular death, noncardiovascular death, HF death, and infection death. The differences were evaluated by conducting a log-rank test and the Bonferroni method for multiple comparisons.
Moreover, multivariate analysis was conducted using a Cox proportional hazard model to evaluate the association among the three groups and the incidence of the all-cause death. Hazard ratio (HR) with 95% CI was also calculated. Potentially confounding effects were also adjusted in multivariable models that were considered to be associated with the clinical outcomes, including sex and comorbidities (yes/no): history of HF, coronary artery disease (CAD), valvular disease, cardiomyopathy, hypertension, diabetes mellitus, atrial fibrillation, chronic kidney disease (CKD), and stroke. Furthermore, to T A B L E 1 Baseline characteristics of the study population as stratified according to age Note: Categorial data are presented as number (%). Continuous data are presented as mean ± standard deviation (SD) or median (Q1 -Q3).
Abbreviations: BMI, body mass index; BNP, brain natriuretic peptide; eGFR: estimated glomerular filtration rate; LVEF, left ventricle ejection fraction; MRA, mineralocorticoid receptor antagonist; RAS, reninangiotensin system. evaluate the effect of clinical factors, the covariates were selected as follows: sex, eGFR, BNP and prescriptions at discharge (yes/no): renin-angiotensin system (RAS) inhibitors, β-blocker, diuretic, mineralocorticoid receptor antagonist (MRA), calcium-channel blocker, oral inotropic agent, digitalis and oral diabetic agent. Included variables were assumed to be time independent from the discharge. 10,15 Competing risk analysis was conducted to obtain a more insight- the lowest proportion of cardiovascular death (43% vs. 48% in nonelderly and 47% in elderly). and super-elderly, respectively. In nonelderly, the annual incidence rate of cardiovascular death was higher than that of noncardiovascular death; however, in elderly and super-elderly, the annual incidence rate of cardiovascular death was lower than that of noncardiovascular death. In all groups, the highest incidence rate of the detailed mode of death was HF death.
During the follow-up period, the cumulative event rates of all patients were compared among the three groups. Using the Kaplan-Meier method, a significantly lower rate of the all-cause death for nonelderly than for the other groups was obtained ( Figure 2). As regards the detailed mode of death, significantly decreasing rates of cardiovascular death, noncardiovascular death, HF death, and infection death with decreasing age groups were obtained.
In the Cox proportional hazard model (

| DISCUSSION
In this prospective registry of AHF patients in Japan, the elderly and super-elderly patients had a significantly greater risk of reaching allcause death after discharge among AHF patients. Additionally, this study revealed that noncardiovascular death, especially infection death, was a more common risk in elderly and super-elderly AHF patients than cardiovascular death. These differences were independently maintained despite adjustments for differences in comorbidities and prescriptions. To our knowledge, this is the first prospective report to examine the association between super-elderly and prognosis and include information that is helpful for preventative approaches and management for patients following AHF.
This study found that the risk of all-cause death in AHF patients with elderly was 2.5-fold higher, and that with super-elderly was more than fivefold higher than that with nonelderly. Interestingly, a gradual increased risk was found more in noncardiovascular death than in cardiovascular death. Moreover, the major noncardiovascular death was caused by infection death. Especially, the risk of infection death in AHF patients with super-elderly was tenfold higher than that with nonelderly. These results were obtained from previous studies. 10,11 The NARA-HF study, a single-center retrospective study, reported that no significant differences were observed in cardiovascular death between patients aged <75 years and those aged ≥75 years with AHF, but a risk of noncardiovascular death, mostly caused by infection, in those aged ≥75 years was significantly higher than in those aged <75 years. 11

| CONCLUSION
In this population, the risk of all-cause death was found to have a gradually increase in elderly and super-elderly patients, when compared with nonelderly patients. Elderly patients, especially superelderly patients, with AHF were at a higher risk for noncardiovascular death, especially infection death. They should be managed to prevent not only following cardiovascular events but also infections using medical resources and their supporters.

CONFLICT OF INTEREST
The authors declare that there is no conflict of interest.

DATA AVAILABILITY STATEMENT
The deidentified participant data will be shared on a request basis.
Please directly contact the corresponding author to request data sharing. The data contained the baseline data follow-up data of patients and the study protocol in Japanese. The data is available immediately.