We approached a consecutive sample of 63 adults aged 60 years or over admitted to a cardiology emergency service (Heart Institute of São Paulo, Brazil) with primary diagnosis of CHF according to Carlson's criteria of heart failure . All subjects were classified as functional class IV (New York Heart Association, 1979) at the time of admission to hospital  and had left ventricular ejection fraction lower than 45%. Thirteen subjects with a positive medical history of stroke during the previous 6 months, current diagnosis of depressive episode (ICD-10), myocardial infarction during the previous 12 weeks, severe anaemia (Hb lower than 10), orthopaedic problems that interfered with the subjects' ability to walk, illiteracy and severe visual impairment (unable to read font size 20) were excluded from further assessment.
We also recruited a sample of 30 controls amongst the patients attending a geriatric outpatient service at the same hospital. All subjects were aged 60 years or over and had a left ventricular ejection fraction greater than 65%. Subjects with positive medical history of stroke, current diagnosis of depressive episode (ICD-10), myocardial infarction during the previous 12 weeks, severe anaemia (Hb lower than 10), orthopaedic problems that interfered with the subjects' ability to walk, illiteracy and severe visual impairment (unable to read font size 20) were not included in the control group.
Each study participant was examined by either a cardiologist or geriatrician. The clinical assessment was designed to clarify the diagnosis of CHF  and the functional capacity of patients . The left ventricular ejection fraction (EF) was calculated from data derived from 2-dimensional echocardiography according to the formula: EF = (final diastolic volume – final systolic volume) / final diastolic volume. The correlation between the calculated EF and direct volumetric measures are greater than 0.9 and the coefficient of variance of the method ranges from 3 to 6% . Finally, CHF subjects were asked to perform the 6-minute walk test in a flat surface – this is a sensitive test of functional capacity that produces, as a score, the total distance covered in metres .
Mental state examination and cognitive assessment
All participants were assessed with the Cambridge Examination for Mental Disorders of the Elderly (CAMDEX). The CAMDEX is a semi-structured interview used to evaluate the present mental state of older adults, as well as their past medical and family history. This interview covers a range of signs and symptoms that provide clinically relevant information necessary for the diagnosis of mental disorders according to ICD-10 and DSM-III-R, including dementia and depression (which was an exclusion criteria in the present study). The cognitive component of the CAMDEX, known as the CAMCOG, includes several tests designed to assess orientation, language, memory, praxis, attention, abstract thinking and calculation. Scores range from 0 to 107 and include the 19 Mini-Mental State Examination (MMSE) items. Both the CAMCOG and MMSE produce valid and reliable measures of general cognitive abilities [15, 16].
The cognitive performance of subjects was further evaluated with four tasks that measure, directly or indirectly, attentional skills and visuomotor abilities: Digit Span (forwards component only), Digit Symbol Test, Letter Cancellation Test, and Trail Making A and B. The Digit Span is part of the Wechsler Adult Intelligence Scale-Revised (WAIS-R)  and provides a measure of span and immediate recall. The test consists of seven pairs of a random sequence of 2, 3, 4, 5, 6, 7, and 8 digits that the examiner reads aloud at the rate of one per second (auditory attention). The span of the subject is the maximum number of paired digits that they are able to recall correctly. The Digit Symbol Test is also a subtest of the WAIS-R . It consists of four rows containing, in all, 100 small blank squares, each paired with a randomly assigned number from one to nine. Above these rows is a printed key that pairs each number with a different nonsense symbol. Following a practice trial, the task is to fill in the blank spaces with the symbol that is paired to the number above the blank space as quickly as possible for 90 seconds. The score is the number of squares filled in correctly. The Letter Cancellation Test consists of six 52-character rows in which the two target characters are randomly interspersed approximately 18 times in each row, as depicted by Lezak . Subjects are asked to strike the target characters through as quickly as possible until they finish all the rows. In the present study, the score of participants was recorded as the number of correctly identified target characters in 100 seconds (which has been shown to be the median time necessary to complete the task). Finally, subjects were asked to complete the Trail Making Test A and B . In the part A of the test, the subject is asked to connect consecutively numbered circles, whereas on part B they have to alternate between consecutive numbers and letters. The final mark for each part of the test is the number of seconds that the subject takes to complete the task.
Single photon emission tomography (SPECT)
A convenience sub-sample of 6 patients with CHF who consented and were considered to be clinically capable of going through the procedures involved in a SPECT assessment, were recruited with the aim of measuring brain blood perfusion using 99mTc-HMPAO. Images were recorded by a gamma-camara (double-head Sophycamara model) and processed by a NXT computer. The minimum resolution of this method is 6.1 mm per pixel. The tracer (28 mCi/4 ml) was inject into the blood stream after the patient had been resting for 15 minutes in an environment with low sensory stimulation (closed eyes and ear plugs were used). Immediately after the injection, subjects were asked to perform a mental task: consecutively add 3 starting from 1. The SPECT images were recorded 15 minutes after the injection (patient lying down with closed eyes). Subjects were scanned within 48 hours of their admission to the hospital.
The clinical treatment of patients was tailored to suit each patient's individual needs, and included diet restrictions (such as salt intake), as well as the use of diuretics, ACE inhibitors, cardiotonic medication (such as digoxin) and anti-arrhythmic drugs. The endpoint of clinical treatment was improvement of functional status from class-IV to class-I or II over the subsequent 6 weeks. Subjects with CHF were asked to repeat the 6-minute walk test and all participants to perform the following cognitive tests: Digit Span, Digit Symbol Test, Letter Cancellation Test, and Trail Making Test A and B.
Single photon emission tomography (SPECT)
Six subjects were reassessed with SPECT using the same procedure described for the baseline assessment.
The data were analysed using the statistical package 'STATA-release 6.0'. Contingency tables using Pearson's chi-square were utilised for the analyses of categorical variables. Fisher's Exact Test was calculated when the expected number within 2 or more cells was 5 or less. Between group comparisons of continuous numerical variables (in practical terms) was performed using Student's t-test. Ordinal numerical variables (for example, digit span score) were compared using Mann-Whitney test. Similarly, intra-group comparisons were performed with the Wilcoxon test. Spearman correlation coefficient was used to investigate the strength of the association between EF and CAMCOG scores. Ninety-five percent confidence limits (CI) were estimated for means and odds ratio.