The validation of a new quality of life questionnaire for patients with congestive heart failure–an extension of the Cardiac Health Profile

Abstract

Objective. To study the validity and reliability of the new disease-specific extension of the CHP, Cardiac Health Profile congestive heart failure (CHPchf), used to assess QoL specifically for patients with heart failure. Methods. Eighty-three consecutive patients with chronic heart failure were enrolled. Criteria and construct validity were studied by correlating the Global Mean Score (GMS) for the CHPchf with the total sum score of the Minnesota Living with Heart Failure Questionnaire (MLHFQ) and other measures predicted to correlate with QoL. Reliability was evaluated on the basis of internal consistency. Results. Significant correlations for CHPchf were found with MLHFQ (r=0.76, p=<0.0001), the Boston Heart Failure Score (BHFS) (p<0.001), the generic second section of the CHP, maximal workload at exercise test (p<0.05) and the New York Heart Association (NYHA) classification (p<0.01). Cronbach′s α was calculated to 0.82. Conclusions. CHPchf was shown to be valid and reliable. CHPchf has favourable psychometric properties. Further studies are needed to explore its discriminative abilities.

• Quality of life
• heart failure
• validation studies

Measurement of Quality of Life (QoL) is today regarded as essential in clinical evaluations 1, 2. Recent studies show that QoL is important when aiming to monitor clinical change 3 and that it predicts morbidity as well as mortality 4, increasing recognition of the concept even more. There is general agreement that QoL assessments should cover certain domains, in essence physical status, functional abilities, general well-being, social interactions, psychological/emotional status and somatic symptoms 5–9. Besides being valid and reliable, a questionnaire assessing QoL must be able to detect differences between patients and to respond to changes over time for a given patient or patient population. Disease-specific instruments focus on aspects of the health status that are specific for the disease of primary interest. Such an instrument is more responsive and therefore clinically more relevant 9. Only a few validated disease-specific measuring instruments exist; the Minnesota Living with Heart Failure Questionnaire (MLHFQ) is probably the most widely used 10, 11. A Swedish version has recently been validated 12. Although its validity and reliability have been demonstrated in several studies, it may lack sensitivity 13. Factor analysis has shown that several items do not constitute meaningful clusters 11, 14 which could explain this lack of sensitivity. To improve the psychometric properties they may need to be excluded 15. Unpublished results indicate a lack of unidimensionality (meaning that the items in the test do not seem to measure the same thing) and that it differentiates between the sexes and different age groups in a way that is not adequate (K. Franzén, pers. comm., May 18, 2006).

The Cardiac Health Profile (CHP) was developed and validated to assess QoL in patients with heart disease in general; it includes a disease-specific section for assessing QoL in patients with angina pectoris 16. It is in use and has shown favorable psychometric properties 17–22. The CHP is made up of three parts: Part I assesses subjectively reported functional ability, part II measures generic QoL for patients with heart disease in general; and part III measures disease-specific QoL. Different versions are accordingly used, depending on the type of heart disease in the study population. An advantage of the CHP is that different heart conditions can be compared in terms of generic QoL, and QoL can be related to the patient's subjectively reported functional capacity. This report presents data regarding validity and reliability for the CHP's disease-specific part for congestive heart failure (CHPchf).

Material and methods

Definition

Disease-specific QoL is defined as how the patient's disease influences his/her subjective perception of the degrees of physical, psychological and social well-being.

Study population

Any patients with congestive heart failure were included that did not meet the following exclusion criteria:

1. Patients not fluent with the Swedish native language;

2. Patients with dementia, psychiatric disease or other condition making it impossible to participate.

The study was approved by the local Ethics Committee of Gothenburg University (R 412-98).

The questionnaire

The questionnaire consists of three parts. The first concerns the patient's subjectively reported degree of fatigue, palpitation or dyspnea in connection with physical activity according to the NYHA classification 23. It thus constitutes a redirection of the original NYHA classification, towards the patient instead of the physician. NYHA III is divided into an A and a B category, as shown in Table I. The second part consists of 16 questions asking the respondent to indicate his/her situation by placing an X on a Visual Analogous Scale (VAS), a 100 mm long line with verbal “anchors” expressing extremes at either end. This part, which has been validated earlier, constitutes the generic part 16. The third part (CHPchf) consists of ten questions requesting the same type of response as in part II. To further illustrate how the verbal anchors fit into the Visual Analogous Scale: Item number reads “Are you often tired?”. The patient can then choose to place his/her mark anywhere from the very left extreme of the line where the words “no never” are written, to the opposite right extreme with the words “yes always”. Low CHP scores (where the patient places the mark to the left) indicate a good QoL, high scores a poor QoL. Standardised mean scores of the second and third parts are obtained by summing the responses and by measuring the distance in millimetres from the left end to the X mark for each item. The total is divided by the number of answered items to arrive at the global mean score (GMS). The questions in the disease-specific part are shown in Table II.

Table I.  Classification of cardiovascular disability according to the New York Heart Associations functional classification in the Cardiac Health Profile.

NYHA I:	Ordinary physical activity does not cause undue fatigue, palpitation or dyspnea.
NYHA II:	I am comfortable at rest. Ordinary physical activity results in fatigue, palpitation or dyspnea.
NYHA III A:	I am comfortable at rest. Less than ordinary physical activity (such as stair climbing) causes fatigue, palpitation or dyspnea.
NYHA III B:	I am comfortable at rest. Very slight physical activity (such as dressing yourself or washing up) causes fatigue, palpitation or dyspnea.
NYHA IV:	I am unable to carry on any physical activity without discomfort. Fatigue, palpitation or dyspnea may be present at rest.

NYHA: New York Heart Association.

Table II.  Questions in the CHPchf

1	Are you often tired?
2	Do your feet or calves swell?
3	Do you get out of breath easily?
4	Are your legs often tired?
5	Do you sometimes fear dying?
6	Do you have a good appetite?
7	Has your heart problem affected your outlook on life?
8	Are you at any time affected by a feeling of pressure in your chest?
9	Is your heart problem more of a difficulty with varying weather, such as cold or windy weather?
10	Are you easily tired in the afternoon?

Item selection

An original pool of about 17 items was carefully selected by Bert Andersson and Peter Währborg, both cardiologists with a long clinical experience, which proposed items relevant in the evaluation of quality of life in patients with congestive heart failure. This was done through discussions with patients in the clinic where the patients were actively asked to express the experienced impact of congestive heart failure on their quality of life. The original pool of items was then tried on approximately 30 patients.

The final ten questions were selected by the authors, based on the following criteria:

1. the item should mirror the study's operational definition of quality of life;

2. the item should be easy to understand; and

3. the item should have good psychometric qualities i.e. be related to the overall result and be normally distributed (implying that it is discriminative).

Procedure and laboratory methods

The European Guidelines 24 for the diagnosis of heart failure were applied to ensure correct recognition of the presence of heart failure.

The Cardiac Health Profile and a Swedish version of the Minnesota Living with Heart Failure Questionnaire (MLHFQ) were handed out consecutively to patients attending the out-patient clinic, the cardiology ward and the heart-transplantation ward. The patients were given approximately 10 min to complete the CHPchf. Patients that answered less than 75% of the questions were excluded. The MLHFQ is a 21-item scale in Likert format for assessing the disease's impact on various aspects of life quality 11 with a maximum possible score of 105. The clinical history was recorded and the Boston Heart Failure Score (BHFS) 25 was calculated. NYHA class was recorded and a chest x-ray was performed for calculation of the BHFS. Left Ventricular Ejection Fraction (LVEF) was calculated by means of echocardiography, using the Teichholz formula 26 or the Simpson formula 27. Patients, who were physically capable to do so, performed a graded exercise test on an ergometer bicycle, starting at a work load of 20 or 30 W and increasing the load by 10 W each minute.

Power calculation

The estimation of the study population was based on finding a probable significant correlation between CHPchf and the MLHFQ, indicating good content validity.

Statistical methods

We used Statview, SAS institute software version 5.0.1.0. P-values < 0.05 were regarded as significant. Standard descriptive statistics were applied in Tables III and IV and Figures 12 and 4 except for the correlations between different items and the Global Mean Score of CHPchf which was assessed using Pearson′s correlation coefficient (r). Simple regression analysis and ANalysis Of VAriance (ANOVA) were performed in Table V and Figure 3 respectively. Reliability was evaluated on the basis of internal consistency, i.e. Cronbach′s α 28. Cronbach′s α indicates internal consistency, i.e. the extent to which a set of test items can be treated as measuring a single model parameter and is applied by a random assortment of scale items into multiple split-half comparisons.

Figure 1.  Frequency distribution for Global Mean Score in the CHPchf. CHPchf: Cardiac Health Profile congestive heart failure, GMS: Global mean score, n: number.

Figure 2.  Frequency distribution for total sum score in the Minnesota Living with Heart Failure Questionnaire. n: number, MLHFQ: Minnesota Living With Heart Failure Questionnaire.

Figure 3.  Correlation between the Global Mean Score of the CHPchf and the total sum score of MLHFQ. GMS: Global Mean Score, MLHFQ: Minnesota Living With Heart Failure Questionnaire.

Table III.  Characteristics of patient population (n = 83).

Age (years, SD)	61±14
Gender (%)
Female	20
Male	80
Clinical history (%)
Ischemic heart disease	44
- verified by angiogram	25
- myocardial infarction	36
Hypertension	20
Diabetes	6
Valvular heart disease	23
Cardiomyopathy	42
Other	6
Left ventricular ejection fraction (10)^* (%, SD)	30±14
Maximal work load on exercise test (22) (Watt, SD)	109±41
Boston heart failure score (10) (M, SD)	6.2±3.3
NYHA class (%)
I	6
II	24
IIIA	35
IIIB	21
IV	13
Present level of treatment (%)
Outpatient clinic	59
Cardiology ward	6
Transplantation ward	34
Presence of angina pectoris (%)	15
CCS II	64
CCS III	36

*: missing data, CCS: Canadian Cardiovascular Society, M: Mean, NYHA: New York Heart Association, SD: Standard deviation.

Table IV.  Descriptive data for each item in CHPchf.

Item	n (%)	Mean	SD	SEM	Correlation with total
1	79 (95)	54.9	28.1	3.1	0.86
2	80 (96)	31.3	34.1	3.8	0.64
3	81 (98)	63.3	27.7	3.1	0.77
4	81 (98)	42.6	29.4	3.3	0.62
5	82 (99)	19.4	21.4	2.4	0.42
6	82 (99)	29.3	28.3	3.1	0.53
7	82 (99)	67.7	29.7	3.3	0.77
8	82 (99)	36.5	31.7	3.5	0.68
9	79 (95)	43.3	36.6	4.1	0.55
10	80 (96)	55.4	28.6	3.2	0.65
GMS	83 (100)	44.4	20.3	2.2	–

GMS: global mean score, n and%: response frequency, SD: standard deviation, SEM: standard error of mean.

Table V.  Correlation between GMS for CHPchf with GMS for the generic part of the CHP and clinical variables.

Variable	r-value	F-value	p
Global Mean Score
CHP, generic part	0.74	94.5	<0.0001
Left ventricular ejection fraction	0.09	0.6	n.s.
Boston Heart Failure Score	0.43	16	<0.001
Max work load on ergometer bicycle	0.44	4.8	<0.05
NYHA classification	0.48	3.9	<0.01

CHP: Cardiac Health Profile, NYHA: New York Heart Association.

Results

Eighty-seven consecutive patients with chronic heart failure from the out-patient clinic, a cardiology ward and the heart-transplantation ward at Sahlgrenska University Hospital were enrolled in the study from March to August 1999. Four patients were excluded after inclusion as they failed to reach the lower level of 75% with regard to the proportion of questions needed to be answered.

Descriptive data

Patient characteristics concerning clinical history, LVEF, maximum workload in graded exercise test and BHFS are shown in Table III. GMS for the generic second part of the CHP was in this group 36±16, which is almost identical with the group of angina pectoris patients studied in the validation study for part II (36±17) and significantly different from the value for a healthy control group (23±11) 16. Descriptive data for each item of the CHPchf are shown in Table IV. The highest scores were for items 7 (Has your heart problem affected your outlook on life?) and 3 (Do you get out of breath easily?). The lowest scores were for items 5 (Do you sometimes fear dying?) and 6 (Do you have a good appetite?). The frequency distribution for the Global Mean Score for CHPchf in the population is shown in Figure 1. The frequency distribution for MLHFQ is shown in Figure 2, with a mean score of 41±24.

Key measurement properties

Three aspects were assessed: validity, reliability and sensitivity.

Validity

Content validity was addressed by the procedure described for item selection. Both the original pool of items and the final ten items in the questionnaire were selected by clinicians experienced in the field. Criterion and construct validity was studied by comparing the GMS for the CHPchf with the total sum score of MLHFQ (Figure 3). A highly significant correlation was found (r = 0.76, p = < 0.0001). Correlations with clinical variables are shown in Table V. Significant correlations were found with the BHFS (p < 0.001), the generic second part of the CHP, maximal workload at exercise test (p < 0.05) and NYHA classification (p < 0.01). There was no significant correlation with LVEF.

Reliability

Reliability was addressed by analysing internal consistency, i.e. the extent to which different items within the questionnaire (part III) are assessing the same content or characteristic. Cronbach′s α was applied and the overall Cronbach α coefficient was 0.82.

Sensitivity

Sensitivity was approached in a limited way, since no intervention was performed in this study. The GMS in relation to NYHA class is shown in Figure 4. The mean of the GMS for patients in NYHA class I was 19±12, in NYHA class II 40±23, in NYHA class IIIA 43±21, in NYHA class IIIB 58±18 and in NYHA class IV 51±9.

Figure 4.  Global Mean Score in CHPchf in relation to NYHA class. CHPchf: Cardiac Health Profile congestive heart failure, GMS: Global Mean Score, NYHA: New York Heart Association.

Discussion

The present study concerns a new health-related and disease-specific questionnaire for patients with chronic heart failure, the CHPchf. Its validation was based on clinically well-known manifestations of the disease. Evaluation of criteria and construct validity is based on established methods for assessing somatic consequences of heart failure, such as physical capacity (maximum work load in exercise test), BHFS and the NYHA classification. The questionnaire showed significant correlations with these criteria but not with LVEF. A correlation with LVEF was likewise not found in earlier studies aiming at evaluating quality of life in patients with chronic heart failure 29. The relationship of physical capacity to quality of life is disputed in the literature. Some reports argue that it is closely correlated 30, others that it is not 29. There is, however, evidence of an increased QoL after physical training programmes and when physical capacity has improved after pharmacological treatment 31, 32.

However, the questionnaire also contains items not related to somatic consequences of chronic heart failure. The scores were highest for the question concerning the patient's “outlook on life”.

Surprisingly low scores were found for the item concerning “fear of dying”. This might reflect good coping abilities in the patients or suppression/denial of the disease. The Minnesota Living with Heart Failure Questionnaire is probably the most commonly used questionnaire for assessing QoL in this group of patients. A comparison showed a convincingly good correlation between the two questionnaires.

Limitations

This study has a number of limitations. Given the fact that the patient sample constitutes of a high proportion of men (80%) and that a relatively large part was included from the transplantation ward (34%) lowering the mean age of the study population (61 years), the results may not be generalisable to the congestive heart failure population in general. However, all NYHA classes (Table III) are well represented with the majority of patients in NYHA class II and IIIA. The exercise test was not performed by 22 of the patients, which might interfere with the results. No systematic drop-out was found. The drop-out rate is simply a result of practical limitations in the daily clinical work. As shown in Figure 4, there is a trend break between NYHA IIIb and NYHA IV. The reason may related to the fact that dimensioning of the study was not assessed with the aim to elucidate norm values for different NYHA classes but only to be able to correlate the global mean score of CHPchf and MLHFQ and was thus rather small. Even if the global mean score CHPchf was correlated to NYHA classes, it was not possible to permit conclusions for each class especially NYHA IV as it contained only 13 patients. A less direct indication of sensitivity than comparing NYHA classes constitutes a comparison of the frequency distributions for CHPchf (Figure 1) and MLHFQ. In Figure 2 it is evident that the former is bell shaped, with the highest scores around the centre of the x-axis, whereas for MLHFQ the distribution is flat and skewed to the left. Data from a Swedish material show a similar picture (U. Alehagen, pers. comm., May 18, 2006). This may indicate a difference in sensitivity, as a psychometric scale should be designed so that the frequency distribution is normal 7. Moreover, the Swedish version of the MLHFQ used in this study had not been validated before. This is a limitation because translation into another language has been shown to alter an instrument's psychometric properties 33. However, we compared our version with a Swedish version that was recently validated 12 and found they were similar.

CHPchf was shown to be valid and reliable. CHPchf has favourable psychometric properties. Future studies aiming at assessing QOL before and after an intervention or in a controlled setting are needed to explore its discriminative abilities.

Conflicts of interests

None declared. Approval date by the local ethics committee: 21/10/98, Registration number: R41298