Cognitive performance is associated with worse prognosis in patients with heart failure with reduced ejection fraction.

AIMS: Heart failure (HF) is a complex clinical syndrome with multiple comorbidities. Cognitive impairment, stress, anxiety, depression, and lower quality of life are prevalent in HF. Herein, we explore the interplay between these parameters and study their value to predict major adverse cardiovascular events (MACEs) and health-related quality of life (HrQoL) in patients with HF with reduced ejection fraction using guideline recommended assessment tools. METHODS AND
RESULTS: We conducted a longitudinal study using a sample of 65 patients from two hospitals. A battery of tests was applied to assess cognition [Montreal Cognitive Assessment (MoCA)], stress (Perceived Stress Scale-10), anxiety, and depression (Hospital Anxiety and Depression Scale) at baseline. MACEs were registered using clinical records. HrQoL was estimated using the Kansas City Cardiomyopathy Questionnaire (KCCQ). A descriptive statistical analysis was conducted, and multiple linear and Cox regression models conducted to determine the predictive value of neurocognitive parameters and HrQoL in MACE. Both MoCA [hazard ratio = 0.906 (0.829-0.990); P = 0.029] and KCCQ scores were predictors of MACE, but not of overall mortality. Anxiety, depression, and stress scores did not predict MACE. However, anxiety (β = -0.326; P = 0.012) and depression levels (β = -0.309; P = 0.014) were independent predictors of the KCCQ score.
CONCLUSIONS: The MoCA score and HrQoL were predictors of MACE-free survival. Anxiety and depression were good predictors of HrQoL, but not of MACE-free survival.

Introduction

Patients with heart failure (HF) have a worst health‐related quality of life (HrQoL) when compared with healthy populations and patients with other chronic diseases. HF has a multisystemic impact, including mental health. Several studies show that HF is associated with a high prevalence of neuropsychological disorders such as depression, anxiety, psychological stress, and cognitive impairment (CI). Evidence suggests that some of these factors are associated with prognosis. For example, patients with a higher prevalence of depressive symptoms are at higher risk of mortality and poorer HF outcomes. , Recently, the OPERA‐HF study investigated the prognostic value of psychosocial factors in HF and showed that moderate‐to‐severe depression, anxiety, and CI increase the risk of repeated admissions. Despite the importance of these findings, the sample was composed of both patients with reduced and preserved ejection fraction, which adds heterogeneity to the analysis. Besides, the tool used to assess cognitive performance on this trial was not amongst the ones suggested by the European Society of Cardiology. In fact, using different tools to assess cognitive performance and the heterogeneity of the populations studied may justify the wide range of CI prevalence reported in HF (25% to 80% across studies). , , , The European Society of Cardiology guidelines for the diagnosis and treatment of acute and chronic HF suggest the screening of CI using specific instruments such as the Montreal Cognitive Assessment (MoCA) or the Mini‐Mental State Examination. While these two tools were validated in clinical studies, MoCA identifies clinical relevant CI more often than Mini‐Mental State Examination. , However, no previous study has followed HF with reduced ejection fraction (HFrEF) patients over time using guideline‐recommended tools to assess both CI and HrQoL.

In this context, the current report aims to further explore the interplay between neuropsychological parameters, HrQoL, and prognosis in a sample of ambulatory patients with HFrEF.

Methods

We performed a longitudinal observational study using a convenience sample composed of 65 outpatients with HFrEF under follow‐up in the cardiology department of two hospitals in the North of Portugal (Guimarães and Barcelos).

The inclusion criteria were (i) HF patients under follow‐up in an outpatient clinic, (ii) clinically stable and under maximum tolerated medical treatment without medication changes in the last 6 months, and (iii) left ventricular ejection fraction <40% in transthoracic echocardiogram (Simpson biplane method). The exclusion criteria were (i) hospitalization for HF decompensation in the previous 6 months, (ii) history of clinical stroke, and (iii) CI due to neuropsychiatric diseases or individuals unable to complete the questionnaires (severe visual or auditory deficits).

The participants were submitted to a battery of four scales/questionnaires to assess depressive and anxiety symptoms, cognitive function, perceived stress, and quality of life. Herein, we analyse the health outcomes of these patients during a 2 year follow‐up (July 2016 to June 2018).

The battery of scales/questionnaires was composed by MoCA, , for cognitive performance evaluation; Kansas City Cardiomyopathy Questionnaire (KCCQ) , for quality of life, Perceived Stress Scale‐10 , to assess psychological stress and the Hospital Anxiety and Depression Scale (HADS) , to assess anxiety and depression. Previously, validated Portuguese versions of each questionnaire were used.

MoCA is a cognitive screening instrument with greater sensitivity than the Mini‐Mental State Examination to detect neurocognitive deficits, namely, mild CI. It assesses different cognitive domains: attention, concentration, executive functions, memory, language, visual‐constructional skills, conceptual thinking, calculations, and orientation; its maximum score is 30 points. We use the cut‐off for mild CI according to the MoCA Portuguese version, as validated and normalized by Freitas and colleagues. ,

Regarding the HADS‐A and HADS‐D scales, we used the following cut‐offs: normal if value <8, abnormal if >10, and borderline if 8–10. For Perceived Stress Scale‐10, values above 19 indicate high levels of perceived stress.

A clinical interview was also performed to assess the New York Heart Association (NYHA) class and overall symptoms. Clinical data were retrieved from the clinical records of the patients.

For the follow‐up, the primary endpoint was major adverse cardiovascular event (MACE)‐free survival time. MACE was defined as the composite of cardiovascular (CV) hospitalizations, CV emergency department admissions, and CV deaths. A more comprehensive characterization was also performed including the total number of hospitalizations, the total number of emergency department visits, and the current clinical status based on the clinical record of the last outpatient clinic visit. During the follow‐up period, medical treatment changes were allowed according to the clinical judgement of the assistant cardiologists. A systematic recording of medical compliance was not performed.

All participants signed the informed consent and the study received approval from the Ethics Committees of both hospitals. The Declaration of Helsinki, the Council of Europe's Convention on Human Rights, and Biomedicine and the Council for International Organizations of Medical Science's guidelines were strictly followed.

Results

Baseline evaluation

The cohort was composed of 83% male participants, with an average age of 63 years old and 5.6 years of school (Table 1). The aetiology of HF was ischemic disease in 63% of the patients (52% had a previous myocardial infarction). About 62% were in NYHA class II at the beginning of the follow‐up. Anxiety had been previously diagnosed in 17% of patients and depression in 9%. On the cognitive evaluation, 92% of the patients were classified as having mild CI. The scores in anxiety and depressive scales reached the clinical cut‐off of abnormal in 45% and 48% of the patients. Thirty‐seven per cent of patients presented a ‘high level of perceived stress’.

TABLE 1

Characterization of the cohort: medical background and heart failure

Characteristic		N (%)
Gender (male)		54 (83.10%)
Age (years)		M = 62.92 (SD = 11.25)
Years of school		M = 5.57 (SD = 3.35)
None		1 (1.5%)
1–2		0 (0.0%)
3–4		41 (63.1%)
5–8		13 (20.0%)
9–12		6 (9.2%)
13+		4 (6.2%)
Living alone		3 (4.6%)
Smoking status		9 (13.8%)
Alcohol consumption		29(23.8%)
Physical activity		19 (29.2%)
Height(m)		M = 1.66 (SD = 0.081)
Weight (kg)		Mdn = 72 (IR = 16)
BMI (kg/m2)		26.43 (SD = 5.22)
Normal weight		27 (41.5%)
Abdominal perimeter (cm)		M = 98.85 (SD = 12.4)
Medical background
Diabetes mellitus		29 (44.6%)
Hypertension		37 (56.9%)
Dyslipidaemia		54 (83.1%)
Atrial fibrillation		33 (50,7%)
Myocardial infarction		34 (52.3%)
Depression		6 (9.2%)
Anxiety		11 (16.9%)
Thyroid disease		6 (9.2%)
COPD/asthma		15 (23.1%)
Sleep disorders
OSA		3 (4.6%)
Cheyne–Stokes		4 (6.2%)
Heart failure aetiology (ischemic)		41 (63.0%)
Time of disease (>1 year)		58 (89.4%)
NYHA functional class
I		12 (18.50%)
II		40 (61.50%)
III		13 (20.00%)
Pro‐BNP (pg/mL)		M = 1294 (IR = 3593)
Rhythm (atrial fibrillation)		10 (15.40%)
Left bundle branch block		48 (73.80%)
Transthoracic echocardiogram
LVEF		M = 28.6 (SD = 7.6)
35–40%		16 (24.6%)
<35%"/>		49 (75.4%)
LA diameter		M = 45.4 (SD = 6.2)
E/E' ratio		M = 14.1 (SD = 6.9)
E/A ratio		M = 69.7 (SD = 25.6)
LV end diastolic diameter		M = 62.4 (SD = 7.8)
LV end diastolic volume		Mdn = 137 (IR = 64)
Mitral regurgitation
No/Mild		51 (84.60%)
Moderate/severe		10 (15.40%)
Tricuspid regurgitation
No		56 (91.8%)
Moderate		5 (8.20%)
Right ventricular dysfunction		48 (73.8%)
Heart failure pharmacologic therapy
ACEI or ARA		65 (100%)
Beta‐adrenergic blockers		60 (92.3%)
Aldosterone antagonists		53 (81.5%)
Diuretics		46 (70.8%)
Nitrates		5 (77%)
Digoxin		18 (27.7%)
Ivabradine		7 (10.8%)
Other pharmacologic therapy
Warfarin		17 (26.2%)
Acetylsalicylic acid		36 (55.4%)
DAOCs		16 (24.6%)
OADs		20 (30.8%)
Insulin		12 (18.5%)
Statin		49 (75.4%)
Fibrate		2 (3.1%)
PPI		38 (58.5%)
Electronic device
None		27 (41.5%)
ICD		21 (32.3%)
CRT		17 (26.2%)

ACEI, angiotensin‐converting‐enzyme inhibitor; ARA, angiotensin II receptor antagonists; BMI, body mass index; CRT, cardiac resynchronization therapy; df, degrees of freedom; DOACs, direct‐acting oral anticoagulants; EF, ejection fraction; ICD, implantable cardioverter defibrillator; IR, interquartile range; LVEF, left ventricular ejection fraction; M, mean; MCI, mild cognitive impairment; Mdn, median; NYHA, New York Heart Association; OAD, oral antidiabetic drugs; Pro‐BNP, pro‐brain‐type natriuretic peptide; PPI, proton pump inhibitor; phi, phi coefficient; r, Pearson's correlation presented as r(df); rs, Spearman's rho, presented as rs(df); SD, standard deviation; * P < 0.001

MoCA, Montreal Cognitive Assessment, mild cognitive impairment cut‐off: <24, 7 points, adjusted to the Portuguese population for all ages and years of school.

HADS, Hospital Anxiety and Depression Scale, Anxiety (HADS‐A) and Depression (HADS‐D) subscales, pathological score if >10 points, borderline score if ranging 8 to 10 points.

PSS‐10, Perceived Stress Scale, score ≥20 indicative of high level of perceived stress.

One of the objectives of this work was to understand how the neuropsychological parameters are related to each other. These results are displayed in Table 2. Given the importance of gender, age, and school years in anxiety, depression, and cognitive performance, these parameters were included in the analysis. We found a significant association between anxiety and depression (r(64) = 0.559; P < 0.001); perceived stress and anxiety (r(64) = 0.357; P < 0.001); and perceived stress and depression (r(64) = 0.252; P = 0.045). There was also a significant correlation between age and school years (r(65) = −0.611; P < 0.001); and between depression and school years (r(64) = 0.229; P = 0.048). MoCA scores were significantly associated with the number of school years (rMoCA(64) = 0.436; P < 0.001) and age (rMoCA(64) = −0.556; P < 0.001). Psychometric scores, such as anxiety, depression, and perceived stress, as well as KCCQ scores, demonstrated no association with cognitive performance.

TABLE 2

Correlations between neuropsychological parameters, age, gender, and school year

Parameter	HADS‐Anxiety	HADS‐Depression	MoCA	PSS10	Age	Gender
HADS‐Depression	0.559 ** (64)
MoCA	−0.037 (63) b	−0.178 (63) b
PSS‐10	0.357 ** (64)	0.252 * (64)	0.0112 (63) b
Age	−0.042 (64) b	0.158 (64)	−0.556 ** (64)	−0.011 (65) b
Gender a	−0.148 (64) b	−0.225 (64) b	0.17 (64) b	−0.061 (65) b	−0.114 (65) b
Years of scholarity	−0.006 (64) b	−0.229 * (64)	0.436 ** (64)	0.12 (64) b	−0.611 ** (65)	0.206 (65) b

HADS, Hospital Anxiety and Depression Scale; MoCA, Montreal Cognitive Assessment; PSS‐10, Perceived Stress Scale.

P < 0.05.

P < 0.001.

Gender, reference category: female.

Spearman's rho.

Major adverse cardiovascular event‐free survival and mortality

As shown in Table 3, 40% of the patients had at least one MACE during the follow‐up period. From all MACE, 14.5% were CV hospitalizations, 69% were CV emergency department visits, and 15.4% were CV‐related deaths. The average time for the first MACE event was 255 days. The overall mortality rate of our cohort was 13.8%; in the first year of follow‐up, the mortality rate was 7.7%, and in the second year, it was 6.2%.

TABLE 3

Characteristics of the MACE

Characteristic
Days to event	Mean (SD)
Days to MACE	255.00 (214.84)
Days to first hospitalization	267.27 (219.76)
Days to first CV hospitalization	289.00 (230.68)
Days to first ED event	193.85 (157.39)
Days to first ED CV event	300.80 (231.36)
Days to death	312.78 (195.24)

CV, cardiovascular; ED, emergency department; SD, standard deviation; MACE, major adverse cardiovascular event.

As previously referred on the Methods section, we used Cox regression models to search for predictors of MACE‐free survival. We used a stepwise forward selection method that included the following variables: age, gender, years of school, MoCA, HADAS‐A, HADS‐D, and KCCQ. On the univariate analysis, MoCA and KCCQ were the only variables significantly associated with MACE‐free survival (Table 4). These two were combined on a model that was statistically significant for predicting MACE‐free survival. In this model, a 1‐point increase in MoCA score was associated with a decreased risk of MACE [hazard ratio (HR) = 0.906; 95% confidence interval (CI) = 0.829–0.990; P = 0.029] independently of the KCCQ score. While significant, the association between KCCQ score and MACE‐free survival was weak (HR = 0.986; CI = 0.973–0.999; P = 0.031). The model did not survive correction for further addition of other variables. It is important to highlight that we used MoCA scores as a continuous variable, which makes it independent of the categorization of the clinical entity (mild CI or dementia).

TABLE 4

Cox‐regressions for MACE‐free survival and the neuropsychological parameters scores

Model	MACE‐free survivala
	HR (95% CI)	P value
Univariable Cox regression model
MoCA	0.894 (0.819–0.977)	0.012
KCCQ	0.982 (0.969–0.996)	0.011
Multivariable Cox regression model
MoCA	0.906(0.829–0.990)	0.029
KCCQ	0.986 (0.973–0.999)	0.031
Final overall model (χ2 = 12.220; df = 2; P = 0.002)

CI, confidence interval; HR, hazard ratio; KCCQ, Kansas City Cardiomyopathy Questionnaire; MACE, major adverse cardiovascular event; MoCA, Montreal Cognitive Assessment;.

There was no association between MACE‐free survival and age (P = 0.101), gender (P = 0.734, HADS‐A (P = 0.665), HADS‐D (P = 0.391), and years of school (P = 0.709).

We used a similar approach to search for predictors of mortality. On the univariate analysis, age and KCCQ were the only variables associated with mortality. However, KCCQ lost significance when adjusted for age. As so, age persisted as the main predictor of mortality (HR = 1.091; CI = 1.012–1.177; P = 0.017).

Health‐related quality of life evaluation

Health‐related quality of life was assessed by the KCCQ. Its overall summary score mean was 64.3 ± 24.7 points; the domains, ‘self‐efficacy’ and ‘knowledge’, ‘symptoms frequency’, and ‘symptoms burden’, showed the higher scores (means of 87.5; 72.9 and 71.5, respectively).

To better understand the strength of the relationship between KCCQ and the other battery test scores, we performed a Pearson's correlation (Table 5) that demonstrated a significant correlation between KCCQ levels and anxiety (R(65) = −0.390), depression (R(65) = −0.378), and perceived stress (R(64) = −0.333) levels.

TABLE 5

Inferential analysis to explore the correlation between KCCQ score and neuropsychological parameters and MACE‐free survival

A. Pearson correlation coefficients between neuropsychological parameters and KCCQ score
	KCCQ Pearson coefficient (R); (N)
MoCA	0.074 (64)
HADS‐A	−0.390 ** (65)
HADS‐D	−0.378 ** (65)
PSS‐10	−0.333 ** (64)
Age	−0.049 (65)
Years of school	0.082 (65)

CI, confidence interval; HADS, Hospital Anxiety and Depression Scale; HR, hazard ratio; KCCQ, Kansas City Cardiomyopathy Questionnaire; MACE, major adverse cardiovascular event; MLRM, Multivariate Linear Regression Model; MoCA, Montreal Cognitive Assessment; PSS‐10, Perceived Stress Scale; SE, standard error.

P < 0.05.

P < 0.001.

Adjusted for age.

The Pearson's correlation demonstrates a significant correlation between KCCQ levels and anxiety (R(65) = −0.390), depression (R(65) = −0.378) and perceived stress (R(64) = −0.333) levels. The MLRM was able to explain 38.8% of the variance of the KCCQ score (F(3.60) = 12.704, P < 0.001). Also, both anxiety (β = −0.326; P = 0.012) and depression (β = −0.309; P = 0.014) were independently associated with worst quality of life. On the longitudinal evaluation, KCCQ was a predictor of MACE‐free survival when adjusted to age.

To determine if anxiety, depression, and perceived stress are predictors of HrQoL, we performed a multilinear regression model. As shown in Table 5, this model was able to explain 38.8% of the variance of the KCCQ score (F(3.60) = 12.704, P < 0.001). Also, both anxiety (β = −0.326; P = 0.012) and depression (β = −0.309; P = 0.014) scores were independently associated with worst HrQoL.

Discussion

In this study, we explored the value of neuropsychological assessment tools to predict events and quality of life in patients with HF. Our data show that both lower MoCA and KCCQ scores are predictors of MACE‐free survival. These results are in line with other studies showing that lower scores in MoCA and KCCQ are associated with poor health outcomes in adults with HF. , , , , A possible explanation for this finding is that lower MoCA and KCCQ scores reflect greater severity of HF and consequently worst outcome; another explanation, especially for the predictive value of MoCA, is that mild CI may be associated with poorer self‐care and therapeutic compliance, although we did not explore this topic on our study. Recently, the OPERA‐HF study found that psychosocial variables (presence of frailty, moderate‐to‐severe depression, and moderate‐to‐severe anxiety) were independently associated with both the first and recurrent adverse events in patients with HF. In that study, the presence of CI was independently associated only with an increased risk of recurrent events. However, a direct comparison with our study cannot be performed for several reasons. First, the tool used on OPERA‐HF to assess cognitive performance is different from the ones herein used; second, the cohort of OPERA‐HF included patients with preserved ejection fraction, and finally, the events accounted for the referred study were not only CV events, unlike the criteria used in our study that was restricted to MACE. Another noteworthy finding of our study was the very high prevalence of CI (92%). This may be explained not only by the severity of HF on our cohort (patients with reduced ejection fraction) but also by the high prevalence of alcohol consumption and low‐level of academic education.

The pathophysiology of CI in HF is not completely understood, but some studies suggest that the reduction in cerebral blood flow, alterations of cerebrovascular reactivity, and modification of baseline levels of blood pressure play an important role. Neuroanatomical changes such as reduced regional cortical thickness, smaller grey matter volume, and reduction of hippocampus volume , have been well documented. Of notice, an important finding of the current study was the lack of significant association between MoCA score and clinical parameters (such as NYHA class, Pro‐BNP and left ventricular ejection fraction). The same holds for the HrQoL component. The perception of patients of their own HrQoL was not associated with cognitive performance. There are controversial findings in the literature about this topic with some studies reporting a negative impact of CI in quality of life, while others show no association.

The present study did not find any significant association between CI and depression, anxiety, or stress, which is contrary to findings from other studies, where a consistent association between depression and cognition has been reported. , One possible explanation for this lack of association may be that in patients with HFrEF, cardiac dysfunction is the dominant cause for CI, diluting the effects of other factors such as education or depression. However, a different study design is needed to address this question (p.e. case–control study).

On our cohort, anxiety and depressive symptoms were not associated with CI or MACE but were correlated with the worst quality of life reinforcing the importance of screening and managing anxious and depressive symptoms in patients with HF. The efficacy of pharmacological and non‐pharmacological strategies to treat depression is yet to be completely determined in patients with HF. Further studies are needed to clarify the best approach to tackle neuropsychological disorders in these patients.

The current study presents limitations. The major limitation is the size of the cohort, although it should be highlighted that this study population represents a subgroup of patients with HFrEF under follow‐up in a hospital outpatient setting. While this brings specificity to the current study, it makes it impossible to determine if the results can be extrapolated to other populations of patients, such as those with HF with preserved or mid‐range EF or who do not attend follow‐up at the hospital. Another limitation was the descriptive nature of the study, which does not allow determining the mechanisms underlying the predictive value of MoCA and KCCQ scores for MACE‐free survival time. Finally, external factors in patient selection may have influenced the results, such as the willingness to participate in the study.

In summary, our study concludes that the prevalence of mild CI is very high in patients with HFrEF and, of greater clinical relevance, MoCA, and KCCQ scores predict MACE‐free survival. On the contrary, anxiety and depression performed poorly to predict MACE‐free survival but were good predictors of quality of life in this cohort. Overall, these results highlight the importance of neuropsychological and HrQoL parameters in the management of patients with HFrEF.

Statistical analysis

A descriptive analysis was conducted. For quantitative variables, the average and standard deviation, (if normal distribution) and the median and IR (if non‐normal distribution) were used. For categorical variables, the absolute and relative (%) frequencies are presented. A two‐sample t‐test was used to study the differences between provenience hospital groups and between cohorts, for quantitative variables. To study differences between groups and association between categorical variables, the χ 2 was used, or the Fisher's exact test, when one of χ 2 test assumptions was violated. Pearson's χ 2 test was used whenever possible.

To measure the strength of associations between quantitative variables, the Pearson correlation was used, or Spearman's rho if any assumption was violated.

To determine which variables predicted MACE‐free survival time in this sample, we used Cox regression analysis. To determine which variables predicted KCCQ, we performed a univariate analysis followed by a multiple linear regression model.

Statistical differences were considered significant when P value < 0.05. Data were analysed using the software IBM SPSS version 24.0.

Conflict of interest

None declared.

Funding

This work was done within the project (NORTE‐01‐0124‐FEDER‐000021), supported by the Portuguese North Regional Operational Program (ON.2 ‐ O Novo Norte), under the National Strategic Reference Framework (QREN), through the European Regional Development Fund (FEDER); by the project (NORTE‐01‐0145‐FEDER‐000013), supported by the Northern Portugal Regional Operational Programme (NORTE 2020), under the Portugal 2020 Partnership Agreement, through the European Regional Development Fund (FEDER); and by FEDER funds, through the Competitiveness Factors Operational Programme (COMPETE).