Article ID: CJ-23-0129
Background: Acute decompensated heart failure (ADHF) has a poor prognosis and common comorbidities may be contributory. However, evidence for the association between dementia and clinical outcomes in patients with is sparse and it requires further investigation into risk reduction.
Methods and Results: We assessed the clinical profiles and outcomes of 1,026 patients (mean age 77.8 years, 43.2% female) with ADHF enrolled in the CURE-HF registry to evaluate the relationship between investigator-reported dementia status and clinical outcomes (all-cause death, cardiovascular (CV) death, non-CV death, and HF hospitalization) over a median follow-up of 2.7 years. In total, dementia was present in 118 (11.5%) patients, who experienced more drug interruptions and HF admissions due to infection than those without dementia (23.8% vs. 13.1%, P<0.01; 11.0% vs. 6.0%, P<0.01, respectively). Kaplan-Meier analysis revealed that dementia patients had higher mortality rates than those without dementia (log-rank P<0.001). After multivariable adjustment for demographics and comorbidities, dementia was significantly associated with an increased risk of death (adjusted hazard ratio, 1.43; 95% confidence interval, 1.06–1.93, P=0.02) and non-CV death (adjusted hazard ratio, 1.65; 95% confidence interval, 1.04–2.62, P=0.03), but no significant associations between dementia and CV death or HF hospitalization were observed (both, P>0.1).
Conclusions: In ADHF patients dementia was associated with aggravating factors for HF admission and elevated risk of death, primarily non-CV death.
Heart failure (HF) is a growing healthcare problem in aging populations and is associated with a significant risk of recurrent cardiovascular (CV) events, including death and rehospitalization for worsening HF.1–3 Acute decompensated HF (ADHF) causes millions of hospitalizations annually, and despite therapeutic advances, its prognosis remains poor. Comorbidities are common in elderly patients with HF and may contribute to increased risk of poor outcomes.4,5 In a Japanese HF cohort, dementia was highly prevalent in patients aged ≥80 years,6 and cognitive impairment is highly prevalent in heart disease: >50% of HF cases in previous reports.7,8 Cognitive function is a strong and independent predictor of readmission or death in patients with HF, and even mild cognitive impairment predicts a greater risk of adverse outcomes in patients with HF, independent of other factors.8,9 Although dementia has become increasingly prevalent in elderly patients with HF, most guidelines do not provide any specific recommendation regarding its management.10 Furthermore, there is a paucity of data regarding the association of dementia with aggravating factors of HF hospitalization and clinical outcomes, including non-CV events, in these patients. Therefore, we conducted the present study to assess the prevalence of dementia and clarify its association with adverse events in ADHF patients of a Japanese HF cohort.
This study was a post-hoc analysis of the Clue of Risk Stratification in the Elderly Patients with Heart Failure (CURE-HF) registry, which was a prospective multicenter cohort study conducted in Nagano Prefecture, Japan. In the CURE-HF registry, 1,036 consecutive patients hospitalized with a primary diagnosis of ADHF and discharged after treatment at 13 institutions between July 2014 and August 2019 were enrolled. Exclusion criteria were age <20 years, follow-up unfeasible, unable to provide informed consent, in-hospital death, and acute coronary syndromes. After admission, medical treatment was initiated at the discretion of the investigator at each local site. Baseline clinical data, including demographics, medical history, laboratory data, and echocardiography findings, were assessed during the compensated stage of HF. Follow-up data were obtained from hospital charts through direct contact with patients or from referring physicians. To ensure an accurate assessment of clinical events, additional information was obtained from visits or telephone contact with living patients or family members and from medical records obtained from other hospitals, as necessary, between June 2021 and August 2021. These events were fully anonymized before the analysis by investigators who were blinded to the participants. The study was approved by each participating institution’s review board or ethics committee. All study participants provided written informed consent prior to enrollment. The study was performed in accordance with the tenets of the Declaration of Helsinki and registered with the University Hospital Medical Information Network (UMIN 000024470).
Of the 1,036 patients with ADHF, we excluded 10 patients with missing follow-up data. The primary outcome was all-cause death, and the secondary outcomes were CV death, non-CV death, and HF rehospitalization.
The presence of dementia was based on a case report form completed by site investigators at study entry, medical interview or the use of relevant drugs. Infections, drug interruptions and overwork as aggravating factors for HF hospitalization were also identified on the report form by the local-site investigators at the study entry. CV death was defined due to HF, acute myocardial infarction, cerebral infarction, CV hemorrhage (non-stroke intracranial hemorrhage or non-traumatic vascular rupture), and sudden cardiac death.11 Non-CV death was defined as having gastrointestinal, infectious, orthopedic/rheumatologic, cancer, renal, pulmonary, endocrine, neurological/psychiatric, and urologic/gynecologic cause.12 Rehospitalization for HF was defined as documented worsening of HF requiring hospitalization;11 that is, the patient had to show signs and symptoms of HF and require treatment with intravenous diuretics. Prior HF hospitalization was defined as a previous diagnosis of HF according to the Framingham criteria,13 and a history of hospitalization for worsening HF. Hypertension was defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or ongoing therapy for hypertension. Dyslipidemia was defined as a serum total cholesterol concentration ≥220 mg/dL, low-density lipoprotein cholesterol level ≥140 mg/dL, or the need for treatment with lipid-lowering agents. Diabetes mellitus was defined as hemoglobin (Hb) A1c ≥6.5%, fasting glucose ≥126 mg/dL, random plasma glucose ≥200 mg/dL, or a clinical history of oral hypoglycemic agent and/or insulin use. The estimated glomerular filtration rate (eGFR) was calculated using the Modification of Diet and Renal Disease equation coefficients modified for Japanese patients.14 In the Japanese long-term care insurance system, people aged ≥65 years who satisfied the eligibility criteria, as well as those aged 40–64 years with age-related diseases, can receive long-term care services, irrespective of income level and availability of informal care provided by the family. Eligibility is assessed using a 74-item questionnaire based on activities of daily living, and there are 7 levels of long-term care need certificates: support levels 1 and 2, and care need levels 1 (least disabled) to 5 (most disabled). The availability of family caregiving is not taken into consideration.15
Statistical AnalysisContinuous variables are summarized as mean±standard deviation or median with interquartile range. Categorical variables are expressed as frequencies and percentages. Clinical data were compared between categories (dementia vs. non-dementia) using the unpaired Student’s t-test or Mann-Whitney U test. Clinical factors associated with dementia were assessed using backward stepwise regression models (P value threshold=0.05), including age, sex, body mass index (BMI), hypertension, dyslipidemia, diabetes mellitus, prior coronary artery disease (CAD), prior stroke, peripheral artery disease, atrial fibrillation (F), chronic obstructive pulmonary disease (COPD), smoking status, New York Heart Association (NYHA) functional class, left ventricular ejection fraction (LVEF), B-type natriuretic peptide (BNP), albumin, Hb, and eGFR. The relationships between continuous variables were assessed using Spearman’s correlation. Poisson models were used to estimate the incidence rates. Kaplan-Meier curves were calculated from baseline (the date of discharge) to the incidence of all-cause death, CV death, non-CV death, and first HF hospitalization and compared using the log-rank test. A multivariate Cox proportional hazards regression analysis was performed to estimate the association between dementia categories and subsequent clinical events by 16 covariates for clinical risk factors of HF (age, sex, BMI, NYHA functional class at discharge, prior hospitalization for HF, drug interruption, long-term care insurance, hypertension, diabetes mellitus, history of CAD, history of stroke, AF, BNP, albumin, Hb, and eGFR), providing an estimated hazard ratio (HR) with 95% confidence interval (CI). We also used Fine–Gray regression models to account for competing risks providing sub-distribution HRs (SHRs).16 Specifically, competing risks were specified as non-CV death, CV death, and all-cause death when assessing the associations with dementia and CV death, non-CV death, and HF rehospitalization, respectively. We conducted subgroup analyses to evaluate the effect of dementia on the incidence of all-cause death within subpopulations of age (<80 vs. ≥80 years), sex, NYHA functional class (I–II vs. III–IV), prior history of stroke, serum albumin level (<3.5 vs. ≥3.5 mg/dL), and LVEF (<50% vs. ≥50%). All statistical analyses were performed using SPSS version 28.0 (SPSS, Chicago, IL, USA) and STATA version 14.1 (Stata Corp, College Station, TX, USA).
We assessed a total of 1,026 patients (follow-up rate: 99.0%) and categorized them according to the presence of dementia: with dementia (n=118, 11.5%) and without dementia (n=908, 88.5%) (Figure 1). Their baseline clinical characteristics are listed in Table 1. The mean age was 77.8 years, and 43.2% were female. Patients with dementia were significantly older (85.7±5.9 years vs. 76.7±12.9 years; P<0.01), had lower BMI (20.5±4.0 vs. 21.7±4.4; P<0.01), more likely to be female (63.6% vs. 42.1%; P<0.01), and more likely to have a history of stroke than those without (28.0% vs. 13.0%; P<0.01), but they did not have a higher prevalence of AF. There were no significant differences in the prevalence of hypertension, dyslipidemia, diabetes mellitus, prior CAD, or COPD. Patients with dementia more often received long-term care insurance than those without dementia (61.9% vs. 19.3%; P<0.01). The difference in HF medications between patients with and without dementia was in the use of angiotensin-converting enzyme inhibitors (ACEIs) and β-blockers, which were prescribed less often to patients with dementia than to those without dementia (ACEIs: 29.7% vs. 43.2%, P<0.01; β-blockers: 56.8% vs. 74.1%, P<0.01). Patients with dementia received less CV medications than those without dementia (mean total number of CV medications: 4 vs. 6, P<0.01). Patients with dementia had lower albumin levels than those without dementia. As for the causes of HF decompensation, infections and drug interruptions were more common for patients with dementia, and overwork was more common in patients without dementia (Table 1, Figure 2). The determinants of dementia were assessed using multivariable logistic regression models, and advanced age, female sex, worse NYHA functional class at discharge, history of stroke, and lower albumin levels were significantly associated with dementia (Table 2).
Flowchart of the post-hoc analysis of the Clue of Risk Stratification in the Elderly Patients with Heart Failure (CURE-HF) registry.
| Overall (n=1,026) |
Dementia (n=118) (11.5%) |
Non-dementia (n=908) (88.5%) |
P value | |
|---|---|---|---|---|
| Age | 77.8±12.6 | 85.7±5.9 | 76.7±12.9 | <0.01 |
| Female sex, n (%) | 457 (45.3) | 75 (63.6) | 382 (42.1) | <0.01 |
| BMI, kg/m2 | 21.6±4.4 | 20.5±4.0 | 21.7±4.4 | <0.01 |
| NYHA functional class (III or IV), n (%) | 220 (21.1) | 35 (29.7) | 185 (20.4) | 0.02 |
| Prior HF hospitalization, n (%) | 321 (30.9) | 41 (34.7) | 281 (31.0) | 0.54 |
| Etiology of HF | ||||
| Ischemic heart disease, n (%) | 269 (25.8) | 27 (22.9) | 242 (26.7) | 0.44 |
| Dilated heart disease, n (%) | 117 (11.2) | 6 (5.1) | 111 (12.2) | 0.02 |
| Hypertrophic heart disease, n (%) | 34 (3.3) | 3 (2.5) | 31 (3.4) | 0.79 |
| Valvular heart disease, n (%) | 309 (29.7) | 41 (34.7) | 268 (29.5) | 0.24 |
| Atrial fibrillation, n (%) | 380 (36.5) | 52 (44.1) | 328 (36.1) | 0.11 |
| Hypertensive heart disease, n (%) | 189 (18.2) | 29 (24.6) | 160 (17.6) | 0.08 |
| Comorbidities and lifestyle factors | ||||
| Hypertension, n (%) | 648 (62.2) | 85 (72.0) | 563 (62.0) | 0.03 |
| Dyslipidemia, n (%) | 261 (25.1) | 28 (23.7) | 233 (25.7) | 0.74 |
| Diabetes mellitus, n (%) | 296 (28.4) | 33 (28.0) | 263 (29.0) | 0.91 |
| Hyperuricemia, n (%) | 186 (17.9) | 22 (18.6) | 164 (18.1) | 0.90 |
| Atrial fibrillation, n (%) | 526 (50.5) | 64 (54.2) | 462 (50.9) | 0.56 |
| History of coronary artery disease, n (%) | 248 (23.8) | 22 (18.6) | 226 (24.9) | 0.17 |
| Stroke, n (%) | 151 (14.5) | 33 (28.0) | 118 (13.0) | <0.01 |
| COPD, n (%) | 56 (5.4) | 2 (1.7) | 54 (5.9) | 0.06 |
| Smoking, n (%) | 380 (37.5) | 27 (22.8) | 363 (40.0) | <0.01 |
| Living alone, n (%) | 156 (15.0) | 15 (12.7) | 141 (15.5) | 0.50 |
| Long-term care insurance, n (%) | 247 (23.7) | 73 (61.9) | 174 (19.3) | <0.01 |
| CV medications | ||||
| ACEIs, n (%) | 427 (41.0) | 35 (29.7) | 392 (43.2) | <0.01 |
| ARBs, n (%) | 309 (29.7) | 36 (30.5) | 273 (30.1) | 0.92 |
| ACEIs/ARBs, n (%) | 735 (71.6) | 71 (60.2) | 664 (73.1) | <0.01 |
| β-blockers, n (%) | 739 (71.0) | 67 (56.8) | 672 (74.1) | <0.01 |
| MRA, n (%) | 564 (54.2) | 64 (54.2) | 500 (55.1) | 0.92 |
| Calcium-channel blockers, n (%) | 325 (31.2) | 42 (35.6) | 283 (31.2) | 0.35 |
| Loop diuretics, n (%) | 868 (83.4) | 96 (81.4) | 772 (85.0) | 0.34 |
| Furosemide, n (%) | 459 (44.1) | 49 (41.5) | 410 (45.2) | 0.49 |
| Azosemide, n (%) | 390 (37.5) | 42 (35.6) | 348 (38.3) | 0.51 |
| Tolvaptan, n (%) | 219 (21.0) | 19 (16.1) | 200 (22.0) | 0.15 |
| Digitalis, n (%) | 66 (6.3) | 5 (4.2) | 61 (6.7) | 0.42 |
| Amiodarone, n (%) | 73 (7.0) | 6 (5.1) | 67 (7.4) | 0.45 |
| Other antiarrhythmics, n (%) | 83 (8.0) | 13 (11.0) | 70 (7.7) | 0.21 |
| Pimobendane, n (%) | 66 (6.3) | 14 (11.9) | 52 (5.7) | 0.02 |
| Nitrates, n (%) | 90 (8.6) | 8 (6.8) | 82 (9.0) | 0.49 |
| Statins, n (%) | 279 (26.8) | 25 (21.2) | 254 (28.0) | 0.13 |
| Antiplatelets, n (%) | 353 (33.9) | 43 (36.4) | 310 (34.1) | 0.81 |
| Aspirin, n (%) | 242 (23.2) | 21 (17.8) | 221 (24.3) | 0.13 |
| Warfarin, n (%) | 272 (26.1) | 26 (22.0) | 246 (27.1) | 0.27 |
| DOAC, n (%) | 323 (31.5) | 41 (34.7) | 282 (31.1) | 0.46 |
| SGLT2 inhibitors, n (%) | 40 (3.8) | 4 (3.4) | 36 (4.0) | 0.90 |
| Non-CV medications | ||||
| Proton pump inhibitors, n (%) | 569 (54.7) | 59 (50.0) | 510 (58.2) | 0.24 |
| Oral hypoglycemic agents, n (%) | 193 (18.5) | 17 (14.4) | 176 (19.4) | 0.21 |
| Insulin, n (%) | 42 (4.0) | 3 (2.5) | 39 (4.3) | 0.27 |
| Total number of medications | 12 [8, 16] | 10 [7, 15] | 12 [8, 17] | 0.06 |
| Total number of CV medications | 5 [4, 8] | 4 [3, 6] | 6 [4, 7] | <0.01 |
| Total number of non-CV medications | 6 [3, 10] | 6 [3, 10] | 6 [3, 11] | 0.39 |
| Laboratory and echocardiography data | ||||
| Hemoglobin (g/dL) | 12.2±2.4 | 11.4±2.0 | 12.4±2.4 | <0.01 |
| Albumin (g/dL) | 3.4±0.5 | 3.2±0.4 | 3.5±0.5 | <0.01 |
| Creatinine (mg/dL) | 1.3±0.8 | 1.3±0.8 | 1.3±0.8 | 0.94 |
| eGFR (mL/min/1.73 m2 surface area) | 46 [33, 59] | 39 [30, 55] | 46 [34, 59] | 0.01 |
| Sodium (mEq/L) | 139±3.6 | 139±4.1 | 139±3.5 | 0.15 |
| Potassium (mEq/L) | 4.3±0.5 | 4.3±0.6 | 4.3±0.5 | 0.31 |
| Uric acid (mEq/L) | 7.1±2.0 | 6.9±2.3 | 7.1±2.0 | 0.35 |
| BNP (pg/mL) | 290 [139, 527] | 305 [116, 532] | 290 [141, 527] | 0.84 |
| HbA1c (%) | 6.1±0.9 | 6.1±0.9 | 6.1±0.9 | 0.39 |
| LDL-C (mg/dL) | 97.0±31.6 | 100±31.3 | 96.5±31.6 | 0.29 |
| LVEF (%) | 48.4±16.1 | 52.4±14.8 | 47.9±16.1 | <0.01 |
| LV interventricular septal thickness (mm) | 11.0±2.5 | 11.3±2.4 | 11.0±2.5 | 0.14 |
| LV posterior wall thickness (mm) | 10.8±2.2 | 11.0±2.1 | 10.8±2.1 | 0.25 |
| Left atrial dimension (mm) | 45.1±9.0 | 43.7±9.5 | 45.4±8.9 | 0.07 |
| HFpEF (LVEF ≥50%), n (%) | 485 (47.2) | 65 (55.1) | 420 (46.3) | 0.08 |
| HFmrEF (40% ≤ LVEF < 50%), n (%) | 195 (19.0) | 22 (18.6) | 173 (19.1) | 0.52 |
| HFrEF (LVEF <40%), n (%) | 346 (33.7) | 31 (26.2) | 315 (35.0) | 0.08 |
| Precipitating factors for ADHF | ||||
| Medical reasons | ||||
| Myocardial ischemia, n (%) | 87 (8.4) | 8 (6.8) | 79 (8.7) | 0.60 |
| Atrial fibrillation, n (%) | 243 (23.3) | 31 (26.3) | 212 (23.3) | 0.49 |
| Anemia, n (%) | 81 (7.8) | 10 (8.5) | 71 (7.8) | 0.86 |
| Infection, n (%) | 147 (14.1) | 28 (23.7) | 119 (13.1) | <0.01 |
| Poor control of blood pressure, n (%) | 118 (11.3) | 11 (9.3) | 107 (11.8) | 0.54 |
| Non-medical reasons | ||||
| Salt/Water overdose, n (%) | 149 (14.3) | 18 (15.3) | 131 (14.4) | 0.78 |
| Drug interruption, n (%) | 68 (6.5) | 13 (11.0) | 55 (6.1) | 0.049 |
| Overwork, n (%) | 222 (21.3) | 8 (6.8) | 214 (23.6) | <0.01 |
Values are presented as number (%), mean±standard deviation, or median [25th, 75th percentiles]. ACEI, angiotensin-converting enzyme inhibitor; ADHF, acute decompensated heart failure; ARB, angiotensin receptor blocker; BMI, body mass index; BNP, B-type natriuretic peptide; COPD, chronic obstructive pulmonary disease; DOAC, direct oral anticoagulant; DPP, dipeptidyl peptidase; eGFR, estimated glomerular filtration rate; HF, heart failure; HFmrEF, heart failure with mid-range ejection fraction; HFpEF, heart failure preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NSAIDs, non-steroidal anti-inflammatory drugs; NYHA, New York Heart Association; SGLT, sodium-glucose co-transporter.
Precipitating factors for heart failure hospitalization.
| Variable | Adjusted OR (95% CI) |
|Z| score | P value |
|---|---|---|---|
| Age | 1.08 (1.05–1.11) | 5.18 | <0.001 |
| Female sex | 2.00 (1.29–3.10) | 3.11 | 0.002 |
| NYHA functional class at discharge | 1.37 (1.04–1.79) | 2.27 | 0.023 |
| History of stroke | 2.26 (1.38–3.69) | 3.24 | 0.001 |
| Albumin, per 1 g/dL decrease | 1.86 (1.17–2.95) | 2.63 | 0.009 |
Variables in the multivariate logistic regression models were selected by backward stepwise regression model using a cutoff P value of 0.05. Initial covariates included age, sex, NYHA functional class at discharge, prior hospitalization for HF, hypertension, diabetes mellitus, history of coronary artery disease, history of stroke, BNP, albumin, hemoglobin, eGFR and atrial fibrillation. The Z-score was measured in terms of standard deviation from the mean (expressed per 1 SD). CI, confidence interval; OR, odds ratio. Other abbreviations as in Table 1.
During a median follow-up period of 2.7 years (interquartile range 1.7–3.6 years days), there were 399 all-cause deaths (incidence rate: 14.9/100 patient-years), 229 CV deaths (incidence rate 8.6/100 person-years), 170 non-CV deaths (incidence rate 6.3/100 person-years), and 392 first HF hospitalization events (incidence rate 18.4/100 person-years) (Table 3). Kaplan-Meier analysis revealed that patients with dementia had higher mortality rates than those without dementia (log-rank: P<0.001). The incidence of CV death (log-rank: P<0.001) and non-CV death (log-rank: P<0.001) were also significantly higher in patients with dementia than in those without dementia However, there were no between-group differences in the incidence of rehospitalization for HF (log-rank: P=0.22) (Figure 3). Table 3 presents the univariable- and multivariable-adjusted Cox proportional hazards analyses for all-cause death, CV death, non-CV death, and rehospitalization for HF categorized by the presence of dementia. In the crude model, dementia compared with non-dementia was associated with a higher risk of all-cause death (HR, 2.49; 95% CI, 1.92–3.23, P<0.001), CV death (HR, 2.15; 95% CI, 1.51–3.06, P<0.001), and non-CV death (HR, 3.04; 95% CI, 2.08–4.45, P<0.001). After multivariate adjustments for clinical risk factors of HF, dementia remained significantly associated with increased all-cause death (HR, 1.43; 95% CI, 1.06–1.93, P=0.02) and non-CV death (HR, 1.65; 95% CI, 1.04–2.62, P=0.03). Dementia was associated with non-CV death even when accounting for competing risk of CV death (Supplementary Table). Conversely, dementia was not associated with the risk of CV death (HR, 1.28; 95% CI, 0.85–1.91, P=0.23) when adjusted for clinical risk factors, and a consistent result was observed when accounting for competing risk of non-CV death (Supplementary Table; adjusted SHR, 1.10; 95% CI, 0.71–1.71; P=0.67). Dementia, compared with non-dementia, was not associated with a higher risk of rehospitalization for HF in either the Cox proportional hazard model or sub-distribution model (i.e., competing risk analysis).
| Presence of dementia | Overall (n=1,026) |
Non-dementia (n=908) Reference |
Dementia (n=118) HR (95% CI) P value (vs. Non-) |
|---|---|---|---|
| All-cause death | (399 events) | (328 events) | (71 events) |
| Event rate (per 100 patient-years, 95% CI) | 14.9 [13.5, 16.4] | 13.3 [11.9, 14.8] | 33.3 [26.4, 42.0] |
| Unadjusted (n=1,026) | Reference | 2.49 (1.92–3.23) P<0.001 |
|
| Adjusted (n=894) | Reference | 1.43 (1.06–1.93) P=0.02 |
|
| CV death | (229 events) | (192 events) | (37 events) |
| Event rate (per 100 patient-years) | 8.6 [7.5, 9.7] | 7.8 [6.8, 9.0] | 17.4 [12.6, 24.0] |
| Unadjusted (n=1,026) | Reference | 2.15 (1.51–3.06) P<0.001 |
|
| Adjusted (n=894) | Reference | 1.28 (0.85–1.91) P=0.23 |
|
| Non-CV death | (170 events) | (136 events) | (34 events) |
| Event rate (per 100 patient-years) | 6.3 [5.5, 7.4] | 5.5 [4.7, 6.5] | 16.0 [11.4, 22.4] |
| Unadjusted (n=1,026) | Reference | 3.04 (2.08–4.45) P<0.001 |
|
| Adjusted (n=894) | Reference | 1.65 (1.04–2.62) P=0.03 |
|
| Hospitalization for HF | (392 events) | (350 events) | (42 events) |
| Event rate (per 100 patient-years) | 18.4 [16.7, 20.3] | 17.9 [16.1, 19.9] | 24.6 [18.2, 33.3] |
| Unadjusted (n=1,026) | Reference | 1.22 (0.89–1.69) P=0.22 |
|
| Adjusted (n=894) | Reference | 0.82 (0.57–1.17) P=0.82 |
Missing data: BNP (n=98), albumin (n=47). Multivariate model was adjusted for 16 covariates: age, sex, BMI, NYHA functional class at discharge, prior hospitalization for HF, drug interruption, long-term care insurance, hypertension, diabetes mellitus, history of coronary artery disease, history of stroke, atrial fibrillation, BNP, albumin, hemoglobin, and eGFR. CV, cardiovascular; HR, hazard ratio. Other abbreviations as in Tables 1,2.
Kaplan-Meier curves by the presence of dementia: (A) all-cause death, (B) cardiovascular (CV) death, (C) non-CV death, and (D) hospitalization for heart failure (HF).
Supplementary Table shows the competing risk analysis for clinical outcomes. The presence of dementia remained significantly associated with an increased risk of non-CV death (adjusted SHR, 1.68; 95% CI, 1.07–2.63; P=0.023), whereas the presence of dementia was not associated with an increased risk of CV death (adjusted SHR, 1.10; 95% CI, 0.71–1.71; P=0.67) or HF rehospitalization (adjusted SHR, 0.75; 95% CI, 0.51–1.09; P=0.13).
We evaluated the effect of dementia on the incidence of all-cause death within subpopulations of age (<80 vs. ≥80 years), sex, NYHA functional class (I–II vs. III–IV), past history of stroke, serum albumin level (<3.5 vs. ≥3.5 mg/dL), and LVEF (<50% vs. ≥50%). As shown in Figure 4, there were no significant interactions between dementia and sex, NYHA functional class at discharge, past history of stroke, and LVEF for all-cause death (all P for interaction >0.05). In contrast, we observed significant interactions between the presence of dementia with age (<80 vs. ≥80 years) and serum albumin level (<3.5 vs. ≥3.5 mg/dL) for all-cause death (P=0.029 and 0.026, respectively), suggesting modification of the effect of dementia on all-cause death by age and lower albumin level.
Forest plot of all-cause death in subgroups of non-dementia (reference) vs. dementia. CI, confidence interval; HR, hazard ratio; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association.
In this post-hoc analysis of the CURE-HF database, the prevalence of dementia was 11.5% among patients with ADHF. We demonstrated significant differences in precipitating factors for HF hospitalization between patients with and without dementia. Infections and drug interruptions were more common in patients with dementia than in those without. Furthermore, dementia was strongly associated with a heightened risk of death and non-CV death, but there was no significant association between the presence of dementia and CV death or HF rehospitalization.
The prevalence of dementia varies according to the definition, assessment of cognitive function by dementia scales such as the Montreal Cognitive Assessment, and study population. A Japanese systematic review showed that the prevalence of all types of dementia ranged from 2.9% to 12.5% and has been gradually increasing.17 The participants of most studies in that review were aged ≥65 years and living at home.16 In a cohort of Japanese elderly (≥80 years old) HF patients, the prevalence of dementia was 17.0% (19/112 patients),5 which is consistent with our finding. In contrast, a prior review article revealed that dementia was common in HF patients, with a prevalence ranging between 25% and 75%,4 which was higher than that in our study. We did not assess the presence of dementia using a validated scale, which may have contributed to these differences. We found that age, female sex, NYHA functional score at discharge, history of stroke, and lower albumin levels were significantly associated with dementia in our multivariable logistic regression models. Another study revealed that in elderly patients with HF (n=270, mean age: 72.5 years), NYHA class IV decreased the Mini Mental State Examination score by an average of 2.31 points compared with NYHA classes I and II.18 A meta-analysis in Japan reported that the prevalence of dementia was not modified by sex (male sex: coefficient: −0.01, 95% CI: −0.06–0.03; compared with female sex),19 which was concordant with our results. In our study, the prevalence of stroke was significantly higher in patients with dementia than in those without. The incidence of CV events for vascular dementia in men is higher than tin women, but its prevalence varies.20 Rocca et al reported that females are at a greater risk of developing Alzheimer’s disease,21 and this may have contributed to our finding that both female sex and prior history of stroke were linked to the presence of dementia in our cohort; however, its consistency in HF remain unknown.
The present study demonstrated that HF admissions due to infection and drug interruption were more prevalent in patients with dementia than in those without dementia, and overwork was less prevalent in patients with dementia than in those without dementia. Several registries have investigated the presence of precipitating factors in patients with ADHF.22,23 Acute coronary syndromes, arrhythmias (especially AF), infections (in particular. airway infections), uncontrolled hypertension, and non-compliance with dietary recommendations and drug prescriptions are the most commonly identified precipitating factors in HF patients.22,23 Hawkins et al reported that compared with no cognitive impairment, medical adherence significantly worsened by 8 percentage points (78% to 70%, P=0.017) in patients with mild cognitive impairment.24 Another prospective cohort study reported that in 577 patients (mean age=71 years, 44% female) hospitalized for HF in the USA and Canada, memory impairment was associated with poorer adherence to self-care practices.25 Drug interruption or poor adherence to medication are common causes of ADHF in elderly patients.26 We consider that our data explain the higher prevalence of medication schedule interruptions in patients with HF and dementia than in those without dementia. Furthermore, a previous meta-analysis of 57 controlled trials showed that medication adherence interventions were found to significantly reduce mortality risk among patients with HF (risk ratio, 0.89; 95% CI, 0.81‒0.99).27 Because elderly patients with HF tend to have complex medication regimens and more difficulties adhering to them, educating both patients and their families is necessary to prevent this problem. The Japanese long-term care insurance system is important for supporting elderly patients with dementia.1–3 In our study, patients with dementia received long-term care insurance more often, and in the KICKOFF registry, a prospective, multicenter cohort of patients with ADHF in Osaka, Japan, the composite endpoint, including all-cause death and HF rehospitalization, increased significantly in patients without long-term care insurance and those with care level 1–5, whereas there was no significant difference between patients without long-term care insurance and patients requiring support level 1 or 2.28 These findings highlight areas for further studies to evaluate the role of multidisciplinary interventions, including disease management and personalized care support under long-term care insurance, on clinical outcomes in Japanese ADHF patients with dementia.
We found that dementia was associated with an elevated risk of all-cause and non-CV death in patients with ADHF, whereas CV death and HF rehospitalization were not. A previous study showed that the presence of dementia was associated with an increased risk of HF hospitalization or CV death,27 discordant with our findings. Although the reasons for these differing results are unclear, a potential contributor may be the variability in ages and event rates between studies. The median age in the CURE-HF registry and the previous study27 was 81 and 74 years, respectively. Additionally, variability in mortality and HF rehospitalization rates might have contributed to the observed differences in the results of the 2 studies. Generally, the event rate in HF rehospitalization is higher than all-cause death in HF patients.4 Older patients with HF die more often and the first clinical outcome tends to be all-cause death rather than HF rehospitalization compared with younger patients. Indeed, the event rate in all-cause death and HF rehospitalization among the dementia patients was 33.3 (100 patient-years) and 24.6 (100 patient-years), respectively. Furthermore, older patients with HF are more likely to die due to non-CV reasons than younger patients.12 One potential explanation for why patients with dementia have a higher incidence of non-CV death may be the association between malnutrition and infection. Lower albumin levels strongly correlated with the dementia in this study, and 34 dementia patients died of non-CV causes and 12 patients died of infections (35.3%), which was the main cause of non-CV death in patients with dementia. Infections, particularly airway infections, are one of the most commonly identified precipitating factors.23 The GREAT registry revealed that ADHF caused by infection was independently associated with a higher 90-day mortality rate.22 Previous study showed that influenza vaccination was associated with a reduced risk for all-cause death (HR: 0.81, 95% CI: 0.67–0.97; P=0.015) in propensity-adjusted models,29 suggesting that the prevention of pneumonia and influenza with vaccines may be beneficial in reducing serious adverse events in HF patients with dementia. There are limited data on the cause of HF admissions (precipitating factors) and mode of death among ADHF patients with dementia, but in our study we found significant differences in aggravating factors for HF admission due to dementia and the presence of dementia was associated with an increased risk of death, primary driven by non-CV death. Dementia and HF may frequently coexist owing to shared disease processes and pathological mechanisms. The presence of one may lead to and/or worsen the other. This bidirectional relationship between dementia and HF acts as a vicious cycle that increases the burden of both diseases. Thus, our findings underscore the importance of dementia assessment to identify HF patients at increased risk for future adverse events.
Study LimitationsFirst, the presence of dementia, infection, drug interruption and overwork were defined by yes/no checkboxes filled in by local-site investigators according to medical interviews. We did not evaluate cognitive function using dementia scales, such as the Montreal Cognitive Assessment. There were not prespecified definitions of infection, drug interruption, and overwork as precipitating factors for HF hospitalization. The primary precipitating factor for ADHF was not assessed systematically; local-site investigators were requested to complete a multiple-choice form. Second, we did not have detailed data on the cause of dementia, such as Alzheimer’s disease, which is the most common cause of primary (degenerative) dementia, and we were unable to assess whether the etiology modifies the effect of dementia on CV outcomes. Third, our study was a post-hoc analysis, and unmeasured confounding factors might have affected the results, regardless of the multivariable adjustment. Fourth, the results were prone to selection bias because patients with severe dementia who did not consent to participate in this study and those who did not provide written informed consent before enrollment were excluded from this cohort. Most cases of dementia in our cohort might have been mild to moderate, which may limit the generalizability and power to assess the relationship between the presence of dementia and clinical outcomes in patients with ADHF. Finally, our sample size was relatively small; therefore, we did not assess whether our results were relevant. Despite these limitations, our exploratory findings underscore the importance of assessing dementia in HF patients to mitigate the risk of future adverse events.
In conclusion, precipitating factors for HF hospitalization differed significantly according to the presence of dementia in patients after ADHF. Dementia was strongly associated with an elevated risk of all-cause death, primarily non-CV deaths. Our post-hoc exploratory findings suggest an important role in the prevention of infection and multidisciplinary interventions in medication reviews in this population.
The authors thank all participants (patients, caregivers, and staff) in the CURE-HF Registry, and Minako Aono and Mebae Kobayashi for their invaluable contribution.
K. Kuwahara is a member of Circulation Journal’s Editorial Team. Other authors have nothing to declare.
The present study was approved by the Institutional Review Board of the Shinshu University School of Medicine (approval no. 4237)
The de-identified participant data will not be shared.
Please find supplementary file(s);
https://doi.org/10.1253/circj.CJ-23-0129
