www.elsevier.com/locate/ejheart

European Journal of Heart Fa

Prognostic importance of cognitive impairment in chronic heart failure

patients: Does specialist management make a difference?

Skye N. McLennan a, Sue A. Pearson a, Janette Cameron b, Simon Stewart a,c,*

a University of South Australia, Division of Health Science, Australiab Deakin University, School of Nursing, Australia

c Division of Health Science, University of Queensland, School of Medicine, Australia

Received 4 June 2005; received in revised form 3 September 2005; accepted 10 November 2005

Available online 28 February 2006

Abstract

Background: Cognitive impairment is common among chronic heart failure (CHF) patients.

Aims: To determine the prognostic significance of cognitive impairment in patients participating in a randomized study of a CHF

management program (CHF-MP).

Methods: CHF patients were randomized to a CHF-MP (n =100) or usual care (n =100). Baseline cognition was assessed using the Mini

Mental Status Examination (MMSE). Five-year all-cause mortality, and combined death-or-readmission, were compared on the basis of the

presence (MMSE 19–26) or absence (MMSE >26) of cognitive impairment.

Results: 27 patients (13.5%) had cognitive impairment and, on an adjusted basis, were more likely to die (96.3% versus 68.2%. RR 2.19,

95% CI 1.41 to 3.39: P <0.001) and/or experience an unplanned hospitalization (100% versus 94%. RR 1.44, 95% CI 1.06 to 1.95:

P=0.019). Cognitively impaired patients had a similar (non-significant) adjusted risk of death-or-readmission in both the CHF-MP (RR 1.40,

95% CI 0.63 to 3.11: P=0.403) and in usual care (RR 1.38, 95% CI 0.75 to 2.53: P=0.305). In the usual care cohort, cognitive impairment

was associated with a greater (non-significant), adjusted risk of death (RR 1.61, 95% CI 1.10 to 4.92: P=0.122). In the CHF-MP, adjusted

risk of death was significantly higher for cognitively impaired patients (RR 2.33, 95% CI 1.10 to 4.92: P=0.027).

Conclusion: These data suggest that ‘‘mild’’ cognitive impairment is of prognostic importance in CHF: even when a CHF-MP has been

applied.

D 2006 European Society of Cardiology. Published by Elsevier B.V.

Keywords: Chronic heart failure; Cognitive impairment; Prognosis; Management

1. Introduction

In response to the significant burden imposed by large

numbers of older, high-risk patients with chronic heart

failure (CHF) on the hospital sector [1,2] there has been

an increasing interest in developing and applying effective

CHF-specific management programs (CHF-MPs). These

1388-9842/$ - see front matter D 2006 European Society of Cardiology. Publish

doi:10.1016/j.ejheart.2005.11.013

* Corresponding author. School of Nursing and Midwifery, University Of

South Australia, City East Campus, North Terrace, Adelaide, SA 5000,

Australia. Tel.: +61 08 8302 1115, +61 04 3830 2111 (Mobile); fax: +61 08

8302 2337.

E-mail address: simon.stewart@unisa.edu.au (S. Stewart).

programs employ a systematic but individualized approach

to apply gold-standard pharmacotherapy and non-pharma-

cologic intervention strategies to optimize the management

of this complex condition. Over the last decade more than

30 randomized trials of CHF-MPs have been conducted.

Recent meta-analyses have confirmed their potential to

reduce unacceptably high morbidity and mortality rates

[3–6]. Worldwide, these programs have become part of

the gold-standard management of CHF following acute

hospitalization.

Despite the obvious benefits of applying CHF-MPs, data

from these randomized studies show that even in patients

exposed to optimal management, morbidity and mortality

ilure 8 (2006) 494 – 501

ed by Elsevier B.V.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501 495

rates remain high. Clearly, some hospitalizations in CHF are

unavoidable, and even desirable for clinical stabilization.

Moreover, CHF is an inevitably fatal condition in most cases.

However, anecdotal evidence suggests that there is some

heterogeneity as to the impact of CHF-MPs on subsequent

health care behaviors at an individual level. Distinguishing

which patients gain the most (or least) benefit from these

programs is highly desirable in order to either exclude poor

responders for referral to better-tailored treatment, or to

modify the interventions to better cater to high risk patients.

Given that CHF-MPs rely on enhancing patient self-care

behaviors to improve health outcomes [7], one of the factors

most likely to modulate the impact of CHF-MP’s is cognitive

impairment [8]. While dementia is the most extreme and well

recognised form of cognitive impairment, less severe

cognitive impairment can develop independently from

dementia. Mild cognitive impairment can manifest as

difficulties with memory, attention, concentration and/or

problem solving. As such, even milder forms of cognitive

impairment in CHF have the potential to impair an

individual’s ability to make a judgement call on seeking

medical assistance and/or adhere to prescribed therapy in a

safe and effective manner [9,10]. For example, it has been

demonstrated that CHF patients with mild cognitive impair-

ment identified using the Mini Mental Status Exam (MMSE)

are less inclined or able to attend scheduled medical

appointments [11].

Although it has been shown that more severe cognitive

impairment is prognostically significant in CHF [12,13],

there is minimal data examining whether less severe

cognitive dysfunction is also prognostically important in

this setting. The impact of mild cognitive impairment is of

particular relevance to health care professions attempting to

improve the management and outcomes associated with

CHF, not only because of its high prevalence [8] and its

potential to ‘‘blunt’’ the impact of otherwise effective CHF-

MP’s, but because unlike dementia, it is more likely to go

undetected in the absence of systematic screening.

1.1. Study hypotheses

It is within this context that we prospectively tested the

following null hypotheses in 200 typically elderly patients,

without obvious signs of cognitive impairment at baseline

(e.g. dementia) involved in randomized study of a home-

based CHF-MP [14] followed-up for a prolonged period:

1. There is no relationship between baseline cognitive

function status (as determined by the MMSE [15]) and

either long-term event-free survival or all-cause mortality

when adjusting for potential confounders.

2. If a relationship between cognitive function and health

outcomes does exist, the CHF-MP will have the same

effect on long-term health outcomes in patients with

cognitive impairment (as defined by an MMSE score of

�26) as it does on those with intact cognition.

2. Methods

The current study involves the prolonged follow-up, and

prospectively planned secondary analysis, of baseline data

derived from a group of typically old patients with CHF,

participating in a previously reported randomized controlled

trial of a home-based CHF-MP [14].

2.1. Participants

As described in more detail previously [14], we

consented and randomized 200 CHF patients recruited from

a tertiary referral hospital in South Australia to either usual

care or a multidisciplinary, home-based CHF-MP. The initial

study with prolonged follow-up was approved by the

institution’s Ethics of Human Research Committee, and

conformed to the principles outlined in the Declaration of

Helsinki. Patients were eligible for inclusion in the study if

they had a history of at least one admission for acute

decompensated heart failure, a confirmed left ventricular

ejection fraction of �55% as determined by echocardiog-

raphy, and chronic exercise intolerance at hospital discharge

defined as New York Heart Association (NYHA) Class II,

III or IV. Alternatively, they were excluded if they had a

terminal malignancy, a recorded diagnosis of dementia, or

were not being discharged to their own home. It is important

to note that in effect the combination of: (i) systematic, in-

hospital screening for patients with dementia, conducted by

a gerontology health care team, (ii) the need for informed

consent directly from the patient and (iii) the active

recruitment of patients being discharged to their own

homes, predicated that the majority of patients would be

assessed as being cognitively intact. As such, although we

obtained MMSE scores from all patients at baseline, they

were not used to exclude study involvement.

2.2. Baseline measures

2.2.1. Baseline clinical status

A comprehensive range of clinical, demographic and

psychosocial variables (see Table 1) were collated through

patient interviews and by reviewing medical records imme-

diately prior to discharge [14]. A number of these variables

deserve particular mention as they are of potential importance

in the development of cognitive impairment. For example, a

low left ventricular ejection fraction (LVEF) could potentially

result in poor cerebral perfusion, and has been associated with

poor cognitive performance in previous research [16], as have

atrial fibrillation [17] and hypertension [18]. In addition, the

cumulative load of serious comorbid illnesses, which was

calculated using the Charlson Index of Comorbidity [19],

may also impact on cognitive capacity.

2.2.2. Baseline cognitive function

Consistent with the pragmatic assessment of cognitive

function in the clinical setting, baseline cognitive impair-

Table 1

Baseline characteristics of chronic heart failure patients according to

cognitive status

Cognitively intact

(n =173)

Cognitive

impairment

(n =27)

Treatment and cognitive status

Randomized to CHF-DMP 83 (48%) 17 (63%)

MMSE score 29.6T0.84 24.7T1.95*

Demographic profile

Male 104 (60.1%) 16 (59.3%)

Age 74.8T8.2 80.7T6.7*

<8 years schooling 89 (51.4%) 22 (81.5%)a

English is second language 57 (33%) 7 (26%)

Weight (kg) 71.2T15.5 73.0T15.3

Live alone 57 (33%) 11 (41%)

Home support 78 (45%) 12 (46%)

Heart failure profile

Mean LVEF (%) 0.369T0.11 0.372T0.10

Duration of CHF (months) 33.07T28.68 24.58T27.34

Prior HF admission 88 (51%) 15 (55%)

Prior admission in last

6 months

170 (98%) 27 (100%)

Systolic BP at discharge

(mm Hg)

122.5T20.4 123.9T19.3

Diastolic BP at discharge

(mm Hg)

66.5T11.3 67.8T11.2

Comorbidity

Charlson comorbidity index 3.08T1.4 3.15T1.4Atrial fibrillation 48 (28%) 8 (30%)

Acute pulmonary oedema 91 (53%) 13 (48%)

NIDDM 53 (42%) 11 (41%)

Blood profile

Sodium (mmol/L) 138.3T3.4 138.0T3.6

Potassium (mmol/L) 4.0T0.5 4.1T0.6

Platelet count (/dL) 223.3T80.1 249.5T88.5White cell count (/dL) 9.1T3.0 9.5T3.8

Albumin (g/L) 38.4T4.2 39.1T4.5

Urea (mmol/L) 12.4T6.6 13.3T12.5

Hemoglobin (g/dL) 13.2T1.7 13.3T1.8Medications

Diuretic 166 (96%) 27 (100%)

ACE inhibitor 122 (71%) 20 (74%)

Nitrate 128 (74%) 23 (85%)

Beta blocker 46 (27%) 11 (41%)

Digoxin 112 (65%) 19 (70%)

Warfarin 41 (24%) 5 (19%)

Calcium channel blocker 61 (35%) 11 (41%)

Anti-hypertensive 112 (65%) 20 (74%)

Left bundle branch block 42 (24%) 6 (22%)

CHF-DMP=Chronic Heart Failure Disease Management Program;

NIDDM=non-insulin dependent diabetes mellitus; ACE=angiotensin-

converting enzyme.a P <0.05.

* P <0.001.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501496

ment was identified using the MMSE [15], which was

administered just prior to hospital discharge (i.e. once

subjects were judged to be clinically stable). The MMSE is a

brief screening tool, which assesses aspects of short-term

memory, orientation, concentration, and visuospatial skills.

It provides an overall rating of global functioning ranging

from 0 to 30, with higher scores indicating better

performance [15]. A cut-off between 21 and 24 has

traditionally been used to identify patients with probable

dementia [20,21]. For the purposes of this study, we were

interested in the effects of less severe impairment, and we

therefore adopted a higher cut-off score of �26: a thresholdstill demonstrated to be of prognostic significance in less

acutely ill cohorts [22]. As expected, only 7 patients (3.5%)

recorded an MMSE score of 18–24 and, therefore,

‘‘probable dementia’’. A further 20 patients recorded an

MMSE score of 25 and 26. Overall, therefore, a total of

27 patients (13.5%) were prospectively designated as

‘‘cognitively impaired’’.

2.2.3. Multidisciplinary, home-based intervention

(CHF-MP)

As described in greater detail in the original report [14],

patients in the intervention arm (n =100) received two visits

by a specialist heart failure nurse in addition to the usual

care administered through the hospital. During the first visit,

which occurred prior to discharge, patients were counselled

about compliance with their treatment regimen and the need

to report any signs of clinical deterioration. A total of

89 patients consented to a second visit by the same nurse

7–10 days after discharge. Patients who withdrew their

consent to be visited at home did not differ with respect

to any clinical or demographic characteristics. During the

second visit, an assessment was made of patients’ knowl-

edge of, and compliance with medications, and their

available social supports. Patients demonstrating poor

knowledge or compliance received tailored intervention

involving one or more of the following: additional counsel-

ling and information, reminder aides, a flexible diuretic

regimen and/or exercise program, referral to a community

pharmacist, nurse and/or social worker for ongoing regular

review. In addition, during this visit patients were examined

for signs of clinical deterioration or adverse effects of

prescribed medication. A detailed report and long-term plan

was sent to all patients’ primary care physicians and

cardiologists. Additional telephone support, via patient-

initiated calls and routine follow-up at 3 and 6 months was

also provided. Repeat home visits were made for patients

who survived an unplanned readmission within 6 months of

their index hospitalization. Importantly, a low MMSE was

not specifically used to tailor the study intervention.

2.3. Study follow-up and endpoints

All 200 patients were subject to 5-year follow-up (median

34 months, interquartile range 9 to 60 months, when taking

into account fatal events) with censoring of all morbid and

fatal events at July 31, 2004. All inpatient and outpatient

hospital activity was monitored through the institution’s

computerized medical records system, individual case

records, and contact with primary care physicians to deter-

mine the status of surviving patients. Official records of the

time and location of all deaths in the region were used to

compile mortality data.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501 497

2.4. Statistical analysis

Analysis was conducted in several stages. Initial univar-

iate analyses were used to determine variables to be entered

in multiple regression models (described below). Chi square

tests were used for categorical data, Student t-tests for

continuous variables and Mann–Whitney U-tests for non-

normally distributed data. Step-wise, multiple-logistic re-

gression analysis was then used to determine potential

independent correlates of cognitive impairment (MMSE

score�26) with calculation of adjusted odds ratios (OR)

and 95% confidence intervals (CIs).

Two unadjusted Kaplan Meier Survival Curves were

initially constructed from actuarial life-tables of event-free

survival and all-cause mortality and analysed with the

log-rank and Breslow tests, respectively, to determine

differences in the number and timing of events for

patients with cognitive impairment relative to cognitively

intact patients. Cox Proportional Hazards Models with

adjusted relative risks (RR) and 95% CIs were also

constructed (with entry of variables at a significance level

of P <0.01 from initial univariate analysis and, if

necessary, forced entry of cognitive impairment) in order

to determine the potential independent correlates of all-

cause mortality and event-free survival. Using the same

methods, four additional adjusted survival curves were

then generated to compare the all-cause mortality and

event-free survival of patients with and without cognitive

impairment on a treatment group-specific basis: separate

survival curves generated for the usual care group and

CHF-MP groups.

All analyses were performed with SPSS version 12.0 and

on an intention-to-treat basis.

1.0

43210 50.0

0.8

0.6

0.4

0.2

Year of follow-up

Eve

nt-f

ree

surv

ival

Cognitively Intact (n = 173)

MMSE ≤ 26 RR 1.44 (95% CI 1.06 - 1.95): P = 0.019

Beta-blocker 0.63 (0.44 - 0.88): P = 0.007Charlson Index 1.13 (1.02 - 1.25): P = 0.022LVEF < 30% 3.74 (1.35 - 7.10): P = 0.018CHF-MP 0.61 (0.39 - 0.96): P = 0.030

Cognitively Impaired (n = 27)

Fig. 1. Kaplan Meier survival curves for event-free survival: cognitively

impaired versus cognitively intact patients.

3. Results

3.1. Baseline characteristics

Table 1 (summary of baseline characteristics according to

cognitive status) shows that this was a typically elderly

cohort of patients with moderate to severe heart failure and

significant comorbidity. Through randomization, 17 of the

27 patients with cognitive impairment were allocated to

the CHF-MP arm of the trial, and 10 to usual care. Uni-

variate analysis of baseline characteristics indicated that

cognitively impaired patients were similar to cognitively

intact patients with respect to all clinical and demographic

variables except age and level of schooling: patients with

cognitive impairment being significantly older (P <0.001)

and less likely to have high school education (P <0.01).

Multivariate analysis confirmed the independent correlation

between cognitive impairment and advancing age (adjusted

OR 1.12; 95% CI 1.04 to 1.20 per year: P=0.002) and less

than 8 years of formal schooling (adjusted OR 4.89; 95% CI

1.33 to 17.1: P=0.017).

3.2. Cognitive status and long-term outcomes

3.2.1. Event-free survival

During 5-year follow-up, all 27 patients with cognitive

impairment experienced an unplanned readmission or died

within 42 months of their index hospitalization compared to

163 (94%) cognitively intact patients. Median event-free

survival was 4 months (IQR 1 to 8) and 7 months (IQR 2 to

20), respectively, for the two groups (P=0.009). On an

adjusted basis, the following variables were independently

associated with a greater risk for this composite endpoint:

greater comorbidity, more severe left ventricular systolic

dysfunction and cognitive impairment. Conversely, those

patients assigned to the CHF-MP and prescribed a beta-

blocker at baseline were more likely to remain event free.

Fig. 1 shows the unadjusted Kaplan Meier survival curves

for event-free survival according to baseline cognitive

status: the inset showing the results of the multivariate

analysis. As such, those with cognitive impairment at

baseline had a significant 1.4-fold increased risk of being

admitted or dying relative to cognitively intact patients when

adjusting for potential confounders (P=0.019).

3.2.2. All-cause mortality

During 5-year follow-up, 26 of 27 (96.3%) patients

found to be cognitively impaired at baseline died, compared

to 118 of 173 (68.2%) cognitively intact patients. Median

survival times were 11 months (IQR 4 to 39) in subjects

with cognitive impairment compared to 36 months (IQR 11

to 60) for the remainder of this patient cohort (P <0.001).

On an adjusted basis, the following variables were

independently associated with a greater risk of a fatal event:

greater comorbidity, more severe left ventricular systolic

dysfunction and impaired cognition. Once again, those

patients assigned to the CHF-MP were more likely to

survive to 5 years. Fig. 2 shows the unadjusted Kaplan

Meier survival curves for all-cause mortality according to

baseline cognitive status: the inset showing the results of the

multivariate analysis. As such, those with cognitive impair-

1.0

43210.0

0.8

0.6

0.4

0.2

Year of follow-up

Surv

ival

(%

)

Cognitively Intact (n = 173)

4 53210

MMSE ≤ 26 RR 2.19 (95% CI 1.41- 3.39): P<0.001

Charlson Index 1.15 (1.03 -1.29): P = 0.014LVEF< 30% 3.50 (2.35 - 7.42): P<0.001CHF-MP 0.67 (0.47 - 0.94): P = 0.020

Cognitively Impaired (n = 27)

Fig. 2. Kaplan Meier survival curves for all-cause mortality: cognitively impaired versus cognitively intact patients.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501498

ment were more than two-fold more likely to die during

study follow-up when adjusting for potential confounders

(P <0.001).

3.3. Effect of the CHF-MP on health outcomes based on

baseline cognitive status

Having established that cognitive function was indepen-

dently associated with poor long-term outcomes, we then

separately analysed the prognostic impact of cognitive

impairment on long-term health outcomes in patients

managed via the usual care and CHF-MP arms of the study.

3.3.1. Event-free survival

Fig. 3 shows the adjusted event-free survival curves

(separate Cox Proportional Hazard Models) for the two

treatment groups according to cognitive status. Curves from

both analyses are plotted on the same axis. In both treatment

groups, those with cognitive impairment appeared to fare

worse (approximate 1.4-fold increased risk in both groups)

than cognitively intact patients. Neither group-specific

analysis found a significant relationship in this regard, most

probably due to Type-II error. While group numbers were

Year of follow-up

Cognitively Impaired (UC)Cognitively Impaired (CHF-MP)

Rest of UC cohort

Usual Care: RR 1.40 (0.63 to 3.11): P = 0.403

1.0

0.8

0.6

0.4

0.2

0.00 1 2 3 4 5

Rest of CHF-MP cohort

Usual Care: RR 1.40 (0.63 to 3.11): P = 0.403

Eve

nt-f

ree

surv

ival

Usual Care: 3.11): P = 0.403

Independent effect of Cognitive Impairment:

Usual Care: RR 1.38 (0.75 to 2.53): P = 0.305RR 1.40 (0.63 to 3.11): P = 0.403CHF-MP:

Fig. 3. Adjusted event-free survival curves for patients assigned to usual

care (UC) or the CHF-MP based on cognitive status at baseline.

too small to allow for between-group statistical analyses, the

adjusted event-free survival curves represented in Fig. 3

suggest that the beneficial effects of the CHF-MP were

indeed ‘‘blunted’’ in those patients with cognitive impair-

ment; their risk being similar to and even worse than those

exposed to usual care.

3.3.2. All-cause mortality

Fig. 4 shows the adjusted all-cause mortality curves

(once again separate Cox Proportional Hazard Models) for

the two treatment groups according to cognitive status. In

both treatment groups, those with cognitive impairment

appeared to fare worse than cognitively intact patients. In

the CHF-MP treatment group, this relationship reached

statistical significance. For patients in the CHF-MP, patients

with cognitive impairment were more than two-fold more

likely to die than their cognitively intact counterparts

(P=0.027). Consistent with the pattern observed in relation

to event-free survival, the adjusted survival curves repre-

sented in Fig. 4, not withstanding the non-significant within-

group comparison, strongly suggest that the beneficial

effects of the CHF-MP on survival were ‘‘blunted’’ in those

43210

Independent effect of Cognitive Impairment:

Usual Care: RR 1.61 (0.88 to 2.95): P = 0.122 CHF-MP: RR 2.33 (1.10 to 4.92): P = 0.027

Independent effect of Cognitive Impairment:

Usual Care: RR 1.61 (0.88 to 2.95): P = 0.122 CHF-MP: RR 2.33 (1.10 to 4.92): P = 0.027

Rest of CHF-MP cohort

Rest of UC cohortCognitively Impaired (UC)Cognitively Impaired (CHF-MP)

5

Year of follow-up

1.0

0.8

0.6

0.4

0.2

0.0

All-

caus

e m

orta

lity

Fig. 4. Adjusted all-cause mortality curves for patients assigned to usual

care (UC) or the CHF-MP based on cognitive status at baseline.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501 499

patients with cognitive impairment; their risk of death being

similar to and even worse than those exposed to usual care.

4. Discussion

Accumulating evidence indicates that a significant pro-

portion of CHF patients suffer from cognitive impairment

[8]. During prolonged follow-up, we found that patients who

had mildly impaired cognition at baseline experienced

significantly reduced event-free survival and overall life

expectancy. Two large studies have previously reported

higher mortality rates in CHF patients with cognitive

impairment, however in both of these publications, only

patients with severe cognitive impairment were included in

analysis, and follow-up periods were limited to 12 months or

less [12,13]. The results of this study suggest, for the first

time, that even mild to moderate forms of cognitive

impairment, which can be easily missed in the acute care

setting, are predictive of negative clinical outcomes in

typically elderly CHF patients over a prolonged period.

Further exploratory analysis focusing on the potential

modulating effects of an otherwise beneficial home-based

CHF-MP suggested that cognitively impaired patients failed

to gain the same benefits in terms of prolonged event-free

survival and, particularly, overall survival relative to

cognitively intact patients: adjusted event rates for cogni-

tively impaired patients in the CHF-MP group being similar

to, or even worse than, those exposed to usual care. Once

again this represents the first time such observations have

been made.

Pending further research, the results of this study suggest

that all CHF patients should be screened with a reliable but

pragmatic screening tool such as the MMSE to detect even

mild forms of cognitive impairment. Additional surveillance

and long-term support should be carefully considered to

decrease their risk of premature morbidity and mortality:

clearly, this is usually the role of a CHF-MP. However, our

additional analyses suggest that this broad approach may not

be suitable for patients with even mild forms of cognitive

impairment.

Why would patients exposed to this particular form of

CHF-MP derive little or no benefit if suffering from

cognitive impairment? Both the high mortality rates and

the poor response to CHF-MP observed in patients with

cognitive dysfunction could be related to a reduced capacity

for self-care. The intervention model applied in this study,

like other successful CHF-MPs, placed a heavy emphasis on

providing patients with information and strategies aimed at

increasing their capacity for self-management. Patients who

had cognitive impairment may have been less able to

comprehend and assimilate this information and/or less able

to recall or implement strategies when changes in their

condition occurred. Although no research to date has

directly examined the effect of poor cognitive function on

CHF patients’ adherence to prescribed treatment regimens,

adherence to both pharmacological and non-pharmacolog-

ical interventions is low among CHF patients as a whole

[23,24]. This may in part be a reflection of the high

prevalence of cognitive dysfunction in the CHF population.

An alternative explanation for our results, both in terms

of poor response to the CHF-MP and the higher rate of

morbidity/mortality overall, is that impaired cognition is a

strong, surrogate marker of disease progression or concur-

rent cerebrovascular disease, both of which lead to

premature mortality. If patients with cognitive impairment

are indeed suffering from a more severe phase of their

illness, then even interventions that successfully improve

self-management and adherence may have limited impact on

subsequent health outcomes. Our results did not provide any

specific support for this hypothesis. We failed to observe

any associations between cognition and clinical indicators of

disease severity. Findings from other research have been

mixed. Some studies have reported modest relationships

between cognitive impairment and symptom duration,

LVEF, symptom severity, hemodynamic pressure variables

and cardiovascular risk factors while others have failed to

do so [8]. The available research data does not provide a

conclusive explanation for the relationship between cogni-

tion and clinical outcomes, but it appears likely that

physiological, cognitive and behavioral factors may interact

in a cyclical manner to affect health outcomes.

In exploring possible explanations for the current

findings, consideration must also be given to limitations

of the study methodology, and in particular the way in

which cognitive function was measured. The MMSE has

been criticized because it lacks sensitivity in the detection of

very mild forms of cognitive impairment [21]. An alterna-

tive approach to identifying and categorising low-level

cognitive impairment might have been to apply the more

rigorous diagnostic criteria for Mild Cognitive Impairment

(MCI) currently used in clinical settings and epidemiolog-

ical research to identify people who exhibit significant and

measurable cognitive deficits, but do not meet diagnostic

criteria for dementia [25]. Accurately diagnosing MCI

according to this conceptualisation requires comprehensive

neuropsychological and functional assessment. The MMSE

was not designed for this purpose, and thus can only provide

a general indication of cognitive capacity. To assist health

care administrators who may be weighing the costs and

benefits of a brief assessment using a screening tool such as

the MMSE, against a more comprehensive examination,

future research should investigate whether the more precise

definition of MCI and the incorporation of more compre-

hensive assessment tools offer better predictive information.

In this study, older patients with less education were

more frequently identified as cognitively impaired. A

relationship between MMSE scores and age and education

has been frequently reported in other research [20,21]. It is

therefore possible that we mislabelled some older less

educated subjects as cognitively impaired due to measure-

ment bias. Analysis was complicated further because age

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501500

and education also correlated with the outcome variables

(mortality and event-free survival), raising the potential for

confounding. However, we used regression analysis to test

the independent effects of age, education and MMSE scores,

and in each case MMSE scores proved to be a better

predictor of morbidity/mortality than either age or educa-

tion: even when forced into a model with these two

variables mild cognitive impairment was still associated

with a significant increased risk of suffering a hospitaliza-

tion or death (RR 1.62, 95% CI 1.02 to 2.55: P=0.039) or

death alone (RR 1.89, 95% CI 1.18 to 2.97: P=0.008).

During the study, we measured cognition only once and

we do not know whether cognitive function improved or

declined post-discharge. In all probability more patients

developed cognitive dysfunction during 5-year follow-up,

and these effects were not accounted for in our results. The

extent of cognitive impairment was also lower than in

comparable patient cohorts [8] with only 13.5% of our

patients affected. This potential selection bias is likely to be a

reflection of our exclusion of patients with a documented

diagnosis of dementia, or who were to be discharged to

institutional care or were clearly unable to consent to

participate in the study. While the majority of studies in this

area have included subjects more representative of the total

CHF population, the subjects in this study cohort more

accurately represent patients likely to be streamed to CHF-

MPs.

Despite its limitations, this study has important clinical

implications. Previously, cognitive impairment has been

overlooked as an important factor in the clinical manage-

ment of CHF. Results of this study indicate not only that

CHF patients with even mild forms of cognitive impairment

may be at increased risk of mortality and hospital

readmission, but that CHF-MPs, a treatment model with

substantial support for its efficacy, may not improve the

prognosis of this high-risk group. Before recommendations

for changes to clinical practice can be made, the results of

this preliminary analysis need to be replicated in a

prospective study using more sensitive neuropsychological

measures of cognitive performance, and assessing additional

clinical factors such as depression which may provide not

only a better understanding of the aetiology of this

condition, but also direction for effective treatment plan-

ning. Until better data on the aetiology, progression and

mechanisms of impact of cognitive dysfunction are avail-

able, at the very least, all patients with CHF should be

screened for cognitive impairment and, if present, should be

reviewed carefully for more intensive management/surveil-

lance with less reliance on promoting self-care activities to

improve health outcomes.

Acknowledgements

SS is supported by the National Heart Foundation and

National Health and Medical Research Council of Australia.

References

[1] Stewart S, MacIntyre K, McCleod MC, Bailey AE, McMurray JJV.

Trends in heart failure hospitalisations in Scotland, 1990–1996: an

epidemic that has reached its peak? Eur Heart J 2000;22:209–17.

[2] Haldeman GA, Croft JB, Giles WH, Rashidee A. Hospitalization of

patients with heart failure national hospital discharge survey 1985–

1995. Am Heart J 1999;137:352–60.

[3] McAlister FA, Stewart S, Ferrua S, McMurray JJ. Multidisciplinary

strategies for the management of heart failure patients at high risk for

admission: a systematic review of randomized trials. J Am Coll

Cardiol 2004;44(4):810–9.

[4] Gonseth J, Guallar-Castillon P, Banegas JR, Rodriguez-Artalejo F.

The effectiveness of disease management programmes in reducing

hospital readmission in older patients with heart failure: a systematic

review and meta-analysis of published reports. Eur Heart J 2004;25:

1570–1595.

[5] Gwadry-Sridhar FH, Flintoft V, Lee DS, Lee H, Guyatt GH. A

systematic review and meta-analysis of studies comparing readmission

rates and mortality rates in patients with heart failure. Arch Intern Med

2004;164(21):2315–20.

[6] Phillips CO, Wright SM, Kern DE, Singa RM, Shepperd S, Rubin HR.

Comprehensive discharge planning with postdischarge support for

older patients with congestive heart failure: a meta-analysis. JAMA

2004;291(11):1358–67.

[7] Stewart S, Blue L. Specialist Nurse Intervention in Chronic Heart

Failure: From Research to Practice. London’ BMJ Books; 2004.

[8] Bennett SJ, Sauve MJ. Cognitive deficits in patients with heart failure:

a review of the literature. J Cardiovasc Nurs 2003;18(3):219–42.

[9] Putzke JD, Williams MA, Daniel FJ, Bourge RC, Boll TJ. Activities of

daily living among heart transplant candidates: neuropsychological

and cardiac function predictors. J Heart Lung Transplant 2000;19(10):

995–1006.

[10] Zuccala G, Onder G, Pedone C, Cocchi A, Carosella L, Cattel C,

et al. Cognitive dysfunction as a major determinant of disability in

patients with heart failure: results from a multicentre survey. J Neurol

Neurosurg Psychiatry 2001;70(1):109–112.

[11] Ekman I, Fagerberg B, Skoog I. The clinical implications of cognitive

impairment in elderly patients with chronic heart failure. J Cardiovasc

Nurs 2001;16(1):47–55.

[12] Zuccala G, Pedone C, Cesari M, et al. The effects of cognitive

impairment on mortality among hospitalized patients with heart

failure. Am J Med 2003;115(2):97–103.

[13] Rozzini R, Sabatini T. Cognitive impairment and mortality in elderly

patients with heart failure. Am J Med 2004;116:137–8.

[14] Stewart S, Marley JE, Horowitz JD. Effects of a multidisciplinary,

home-based intervention on planned readmissions and survival among

patients with chronic congestive heart failure: a randomised controlled

study. Lancet 1999;354:1077–83.

[15] Folstein MF, Folstein SE, McHugh PR. Mini-mental state: a

practical method for grading the cognitive state of patients for the

clinician. J Psychiatr Res 1975;12(3):189–98.

[16] Zuccala G, Cattel C, Manes-Gravina E, Di Niro MG, Cocchi A,

Bernabei R. Left ventricular dysfunction: a clue to cognitive impair-

ment in older patients with heart failure. J Neurol Neurosurg Psychiatry

1997;63(4):509–12.

[17] Ott A, Breteler M, de Bruyne M, van Harskamp F, Grobbee DE,

Hofman A. Atrial fibrillation and dementia in a population-based

study: the Rotterdam Study. Stroke 1997;28(2):316–21.

[18] Papademetriou V. Hypertension and cognitive function. Blood

pressure regulation and cognitive function: a review of the literature.

Geriatrics 2005;60(1):20–2.

[19] Charlson ME, Pompei P, Ales KL, McKenzie CR. A new method of

classifying prognostic comorbidity in longitudinal studies: develop-

ment and validation. J Chronic Dis 1987;40(5):373–83.

[20] Brayne C. The mini-mental state examination, will we be using it in

2001? Int J Geriatr Psychiatry 1998;13:285–90.

S.N. McLennan et al. / European Journal of Heart Failure 8 (2006) 494–501 501

[21] Tombaugh TN, McIntyre NJ. The mini-mental state examination: a

comprehensive review. Am Geriatr Soc 1992;40:922–35.

[22] Guessekloo J, Westendorp RGJ, Remarque EJ, Lagaay AM, Heeren

DL, Knook DL. Impact of mild cognitive impairment on survival

in very elderly people: cohort study. Br Med J 1997;315(7115):

1053–1054.

[23] Cline CMJ, Bjorck-Linne AK, Israelsson BYA, Willenheimer RB,

Erhardt LR. Non-compliance and knowledge of prescribed medica-

tion in elderly patients with heart failure. Eur J Heart Fail 1999;1(2):

145–149.

[24] Carlson B, Riegel B, Moser DK. Self-care abilities of patients with

heart failure. Heart Lung 2001;30(5):351–9.

[25] Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Whalund L,

et al. Mild cognitive impairment—beyond controversies, towards a

consensus: report of the International Working Group on Mild

Cognitive Impairment. J Intern Med 2004;256(3):240–6.