Skip Navigation

European Journal of Heart Failure 2004 6(6):813-820; doi:10.1016/j.ejheart.2004.08.003
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Fonseca, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fonseca, C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© 2004 European Society of Cardiology

Evaluation of the performance and concordance of clinical questionnaires for the diagnosis of heart failure in primary care

Cândida Fonsecaa,*,1, António G. Oliveirab,2, Teresa Motac,3,4, Fernando Matiasd,5, Humberto Moraisd,5, Catarina Costad,5, Fátima Ceiae,1,6 and on behalf of the EPICA Investigators7

a Department of Internal Medicine, Medical Sciences School, New University of Lisbon, Lisbon, Portugal
b Datamedica Ltd., Lisbon, Portugal
c Department of Cardiology, Medical Sciences School, New University of Lisbon, Lisbon, Portugal
d EPICA Working Group, Lisbon, Portugal
e Department of Medical Therapeutics, Medical Sciences School, New University of Lisbon, Lisbon, Portugal

* Corresponding author. Current address. R. Salvador Barata Feyo no. 1 r/c D.to 2780-355 Oeiras, Portugal. Tel.: +351 21 443 81 61; fax: +351 21 301 7958.. E-mail address: candidafonseca{at}netc.pt


   Abstract
Top
 Notes
 Abstract
1. Introduction
 2. Methods
 3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
 References
 
Aim: To validate and estimate the performance statistics and concordance of seven clinical questionnaires for the diagnosis of chronic heart failure (HF).

Methods: Community-based epidemiological survey of patients aged >25 years attending a random sample of primary health care centers in Portugal. Heart failure was identified according to the Guidelines of the European Society of Cardiology (ESC).

Results: A total of 5434 subjects evaluated by 365 investigators were eligible for analysis, 551 of which had criteria for heart failure. Overall, the questionnaires had high specificity, usually above 90%, but low sensitivity, providing an increase in the likelihood of heart failure from 4.3% pre-test to 25–35% post-test in most cases. The Göteborg questionnaire was the most balanced regarding sensitivity (84%) and specificity (81%) but this may reflect its use of prescription of digoxin or diuretics as diagnostic criteria for diagnosis. The Walma, Framingham and NHANES-I questionnaires performed similarly (Sensitivity: 63%, Specificity: 93%), while the Boston and the Gheorghiade questionnaires had a somewhat lower sensitivity (55%). Concordance was good between the Boston, Framingham, Gheorghiade, NHANES-I and Walma questionnaires.

Conclusions: This study evaluated seven clinical questionnaires for the diagnosis of heart failure in the community. Their low sensitivity impairs their usefulness as diagnostic instruments, but their high specificity makes them useful for the identification of patients with fluid retention and/or exercise intolerance from non-cardiac causes.

Key Words: Heart failure • Epidemiology • Diagnosis • Echocardiography • Left ventricular dysfunction

Received July 6, 2004; Accepted August 25, 2004


   1. Introduction
Top
 Notes
 Abstract
 1. Introduction
2. Methods
 3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
 References
 
For several decades, large-scale epidemiological and therapeutic studies have been faced with the problem of reaching an agreed definition of heart failure (HF). Variations in the diagnostic criteria used make it difficult to analyse and to compare data. This leads to gaps in knowledge, with particular regard to heart failure in the community. In order to resolve this problem, the European Society of Cardiology (ESC) has issued recommendations for the diagnosis of heart failure [1]. These guidelines provide an operational definition with standardised criteria that have the merit of being accepted by all. According to the ESC guidelines for the diagnosis of heart failure, it is necessary to have not only appropriate symptoms but also objective proof of cardiac dysfunction at rest and, in doubtful cases, an appropriate response to therapy directed towards treating the syndrome. However, the recommendations recognised that they could not define the severity of symptoms, signs or cardiac dysfunction required, which remains a major limitation. Heart failure remains a syndrome requiring a large subjective element from both the patient and clinicians perspective for its diagnosis.

Heart failure is a syndrome. It is not therefore surprising that symptoms and signs, when assessed in isolation, have low predictive value for the diagnosis. It is to be expected that combinations of signs and symptoms will prove to be more useful and reliable instruments for the diagnosis of heart failure than each variable by itself. It is only a small step from the description of these combinations of signs and symptoms to the production of scores to assess the data from history, physical examination and, in some cases, chest X-ray and therapy. The Framingham [2], Göteborg [3,4], NHANES [5], Boston [6], Duke [7], Walma [8], and Georghiade [9] questionnaires and scores appeared and came into use as screening instruments for the selection of patients with heart failure for therapeutic trials and epidemiological studies. However, despite the accepted use of these diagnostic instruments, only some of them have been objectively validated, using as gold standards such as an elevated pulmonary capillary wedge pressure or impaired cardiac function on imaging [6].

The need for objective proof of cardiac dysfunction at rest, as a requirement for the diagnosis of heart failure according to the Guidelines of the European Society of Cardiology, has raised diagnostic standards but requires resources [10]. Echocardiograms, the simplest means of proving cardiac dysfunction at rest, are a relatively scarce resource in primary care and many hospital laboratories are already working at full capacity. It is often difficult to offer an echocardiogram to everyone with suspected heart failure. Structured questionnaires could be a simple but effective means of screening for heart failure in order to target those who should undergo further tests.

With regard to the diagnostic accuracy of the questionnaires and clinical scores, there are few validation studies and virtually nothing is known about the concordance between these instruments for diagnosing the disease, which reduces their usefulness as a diagnostic instrument in clinical practice and in clinical and epidemiological studies. To our knowledge, they have rarely [7] been validated using as the "gold standard" clinical data confirmed by objective evidence of ventricular dysfunction at rest, in accordance with the current standards of the European Society of Cardiology, nor in large population studies.

The Portuguese EPICA (EPidemiologia da Insuficiência Cardíaca e Aprendizagem) study [11] was designed primarily to estimate the prevalence of chronic heart failure in the community, as defined by the European Society of Cardiology. A secondary goal of this study was to validate and to evaluate the diagnostic performance of the seven clinical questionnaires mentioned above, i.e., the Framingham, Boston, NHANES-I, Georghiade, Duke, Göteborg and Walma questionnaires. These particular questionnaires were selected because, over the years, they have been the most commonly used in the diagnosis of heart failure in large therapeutic and epidemiological studies.


   2. Methods
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
 References
 
The methodology of the EPICA study has been described in detail elsewhere [11,11a].

A structured questionnaire was prepared, which contained all the questions from the seven instruments under assessment, using the translation/back-translation method. This questionnaire was administered by the primary care physician to every subject included in the study. The primary care physician also collected details from the subjects medical history, such as cardiovascular risk factors and current therapy, before proceeding to a complete physical examination.

Patients were considered to have possible heart failure if they scored more than two points in the sum of categories I and II (history and physical examination) of the Boston questionnaire or were receiving loop or thiazide diuretics for heart failure, and were referred for further assessment of cardiac function. In accordance with the recommendations of the European Society of Cardiology [1], the following diagnostic tests were performed: blood and urine tests, electrocardiogram, chest X-ray, peak flow assessment and echocardiogram, in order to confirm cardiac dysfunction at rest and to exclude or diagnose other causes for the signs and symptoms. Chest X-ray and electrocardiographic examinations were reported by the performing physicians on structured forms that included all the items contained in the questionnaires under evaluation. The diagnosis of heart failure was established according to the echocardiographic criteria adopted by the EPICA study, as reported previously [11].

Many questionnaires include data from the chest X-ray and electrocardiogram. However, because the design of the study established that only symptomatic individuals would have these examinations performed, only the sensitivity of the questionnaires could be estimated. In order to be able to estimate the specificity and predictive values of each questionnaire under evaluation, a random sample of all subjects who did not have criteria for further clinical investigation had a chest X-ray and electrocardiogram.

2.1. Statistical analysis
Population estimates of common diagnostic indexes of the different questionnaires (sensitivity, specificity, positive and negative predictive values) were obtained by projection of the sample estimates over the age and gender distribution of the general population determined by the 1991 population census, and over the distribution of symptomatic and asymptomatic individuals estimated from this study. In the calculation of population estimates, allowance was made for the clustering of cases within primary sampling units. For questionnaires having a continuous score (Boston, NHANES-I, Gheorghiade and Walma), the performance at different cut-off values was estimated by the same method and used to construct ROC (receiver operating characteristic) curves. For the analysis of concordance between the seven questionnaires, Cohen's inter-observer reliability kappa coefficient was used [12]. All calculations were made with Stata Release 7.0 for Windows (Stata, College Station, TX).


   3. Results
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
4. Discussion
 5. Conclusions
 Acknowledgments
 References
 
A total of 6300 subjects were enrolled in the EPICA study. The distribution by age group was: 675 between 25 and 49 years, 661 between 50 and 59 years, 1273 between 60 and 69 years, 1698 between 70 and 79 years, and 1993 over 80 years. A total of 5434 subjects were eligible for analysis. A detailed account of the reasons for exclusion of 866 subjects was presented in a separate paper [11]. The population valid for analysis included 2025 males (37.3%) and 3409 females with a mean age of 68.1±15.1 years.

A total of 1058 subjects obtained a score of more than two points in the sum of categories I and II of the Boston questionnaire (707 individuals, 263 males and 444 females) or were under therapy for heart failure with diuretics (351 individuals, 121 males and 230 females). Echocardiographic examinations were performed in 1022 of these individuals, the 36 who did not have the examination, already had a definitive diagnosis of heart failure. The presence of cardiac dysfunction at rest was identified by echocardiography in 551 cases (208 males, 343 females).

From the remaining group of 4376 subjects, 363 subjects were randomly selected. The age distribution of this sample was 69.9±11.1 years, 149 (41.1%) were male and 214 female.

Table 1 presents the population estimates of sensitivity, specificity and positive and negative predictive values of each of the heart failure questionnaires. For the Boston and the Göteborg questionnaires, the diagnostic indexes are presented for each of the proposed cut-offs. These results validate the use of all of these questionnaires for the diagnosis of chronic heart failure. Overall, all questionnaires showed a high specificity with moderate to low sensitivity. Post-test probabilities of chronic heart failure are, in most cases, between 25% and 35%, a significant increase from the 4.3% pre-test probability. Fig. 1 shows a direct comparison of the performances of the heart failure questionnaires, showing that the instrument with the greatest sensitivity was the Göteborg questionnaire at the grades 1 and 2 cut-off. It is also clear from the figure that the Framingham, NHANES-I and Walma questionnaires perform very similarly, that the Boston and Gheorghiade questionnaires have almost identical performance, and that the Duke questionnaire, as well as the Boston and Göteborg questionnaires at the highest cut-off, have much lower sensitivity than the previous instruments.


View this table:
[in this window]
[in a new window]

  		Table 1 Performance statistics of clinical questionnaires for the diagnosis of chronic heart failure

 


Figure 1
View larger version (17K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Plot of the sensitivity and specificity of the seven questionnaires for the diagnosis of heart failure. HF: heart failure.

 
ROC curves were constructed for those questionnaires that produced continuous scores (Boston, Gheorghiade, NHANES-I and Walma) and are presented in Fig. 2. The graph suggests that the cut-offs selected by the authors of all questionnaires were designed to sacrifice sensitivity in favour of greater specificity, and that it is possible to obtain more balanced instruments with higher overall predictive value by reducing the cut-off by one point, i.e., a score of ≥3 for the Boston questionnaire and ≥2 for the other instruments.


Figure 2
View larger version (18K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2 ROC (receiver operating characteristics) curves of the four questionnaires for diagnosis of heart failure that have numeric scores. Circles represent possible scores of each questionnaire, starting with ≥0 in the upper right corner.

 
Table 2 shows the results of the concordance analysis. The Boston, Framingham, Gheorghiade, NHANES-1 and Walma questionnaires have kappa coefficients near or above 0.60, which indicates good concordance between any two of those questionnaires. The Göteborg and Duke questionnaires show poor concordance with virtually all other instruments.


View this table:
[in this window]
[in a new window]

  		Table 2 Pair wise concordance analysis of the chronic heart failure questionnaires

 
Fig. 3 shows the sensitivity and specificity of each item present on one or more questionnaires, divided for legibility in four groups: dyspnoea, clinical history, physical findings, and chest X-ray and ECG. Many items cluster in regions with about the same sensitivity and specificity. The figure also explains how these questionnaires work. Basically, each questionnaire is made from one item from each cluster, combining items with high sensitivity and low specificity with items with low sensitivity and high specificity. Assigning points to each item according to its sensitivity and specificity creates the scoring system, such that items with higher specificity and low sensitivity have higher scores, while items with higher sensitivity and lower specificity have lower scores.


Figure 3
View larger version (20K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3 Plot of the sensitivity and specificity of the items included in all the questionnaires for diagnosis of heart failure. For the sake of legibility, the items were classified in the following groups, clockwise from top left: symptoms of shortness of breath and exercise intolerance, clinical history, chest X-ray and ECG findings, and physical examination. * includes "dyspnoea on walking on the level", "dyspnoea at ordinary pace on the level" and "stops for breath when walking at own pace".

 

   4. Discussion
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
5. Conclusions
 Acknowledgments
 References
 
Structured questionnaires and clinical scores are the most commonly used methods for diagnosis of heart failure in therapeutic and epidemiological studies. However, the validity and concordance between these methods when used in the general population are unknown. Additionally, not all the questionnaires have been objectively validated.

Of the seven questionnaires under analysis in this work, only three, the Framingham, Göteborg and NHANES-I questionnaires, were designed for application in epidemiological studies, the others having been developed for use in drug trials [2-4]. The first study to use this method was the Framingham study. The Framingham questionnaire [2] defines a series of major and minor criteria. These refer to symptoms, signs and abnormalities on the chest X-ray. The diagnosis of congestive heart failure (CHF) is made in the presence of two major criteria or one major and two minor criteria.

The instrument used in the Göteborg study [3] is based on signs, symptoms and atrial fibrillation on the ECG. The subjects are assessed on their history of dyspnoea and on treatment for heart failure with digitalis and/or diuretics, and awarded points according to a respiratory score and a cardiac score. The combination of these scores classifies individuals with either absent, latent or manifest CHF [4]. While the respiratory score is objectively validated, the cardiac score is not.

The Boston questionnaire [6] is the instrument for diagnosis of heart failure that has been most commonly used in epidemiological studies and therapeutic trials. The questionnaire was developed based on clinical criteria which had been validated using a pulmonary capillary wedge pressure >12 mm Hg in a group of individuals who had undergone elective right heart catheterisation as the ‘gold standard’ for the diagnosis of CHF. This questionnaire includes symptoms, signs and findings from the chest X-ray. The subjects are scored in each of these three categories, each one with a maximum of four points, making a total possible score of 12 points. The total score classifies patients in terms of probability of CHF: certain (8-12 points), possible (5-7 points) or unlikely (0-4 points). The Boston questionnaire was modified by Remes et al. [13] in a study of the prevalence of CHF in Finland. The modifications consist of a reformulation of one of the questions, on the degree of dyspnoea, and replacing the value of the cardiothoracic index with the determination of cardiac volume estimated by measurements obtained from the chest X-ray [14]. The Duke questionnaire [7] bases the diagnosis of heart failure on the presence of either an S3 gallop on auscultation or on cardiomegaly as seen on chest X-ray (cardiothoracic index >0.48). These criteria were obtained through multivariate analysis of a series of clinical characteristics obtained from a large number of individuals, with heart failure, defined as the presence of a left ventricular end-diastolic pressure >15 mm Hg.

More recently, the NHANES-I study [5] developed a new instrument based on the Framingham questionnaire, introducing modifications in the criteria suggested by the results of the Duke study. The new criteria were reformulated to include the physical examination used in the study; the questions evaluating symptoms of dyspnoea and exercise intolerance were modified to conform to the WHO dyspnoea questionnaire; and the scoring system was developed on the basis of the system used in the Framingham and Boston questionnaires. This instrument thus sets out to integrate the existing questionnaires.

The Walma [8] and Gheorghiade [9] questionnaires, both developed to assess heart failure in clinical practice in patients receiving therapy with, respectively, diuretics and digitalis, are essentially attempts to simplify the Boston questionnaire in order to make it more easily applicable to outpatients.

Reports in the literature on the diagnostic accuracy of the clinical scores are few and far between, and those that exist were conducted on populations that are not representative of the general population with heart failure. As the performance of these diagnostic instruments obviously depends on the characteristics of the population in which they are applied and on the prevalence of the disease in that population, comparison of the present study with the existing data is difficult.

Remes et al. [15] assessed the performance of the Boston questionnaire in 113 individuals aged between 45 and 74 years with clinically suspected heart failure, taking as the ‘gold standard’ a clinical diagnosis by a cardiologist. The sensitivity of the Boston questionnaire was 80% and its specificity was 92%, which is considerably better than the results from our study. Marantz et al. [16] published data on a subset of 407 individuals with suspected heart failure or myocardial ischaemia referred for radioisotopic assessment of systolic ventricular function from a large trial. The study assessed the performance of the Framingham, Duke and Boston questionnaires, taking as gold standard an ejection fraction of ≤0.40 determined by radionuclide angiography. The sensitivity of the three questionnaires was 63%, 73% and 50%, respectively, and the specificities of the Boston and Framingham scores were respectively 63% and 78%. With the exception of the Duke questionnaire, the sensitivities are similar to those found in our study. It should be noted that the study by Marantz et al. was on a group of patients who were largely unrepresentative of the population with heart failure: 91% were hospitalised, almost all were on digitalis, beta-adrenergic blockers or diuretics, and only patients with left ventricular systolic dysfunction were included.

Both Remes et al. and Marantz et al. assessed the characteristics of the questionnaires in high-risk patients, usually selected because of existing symptoms and signs of heart failure, and their study populations had a much higher prevalence of heart failure than the 4.3% prevalence in the Portuguese population. This may explain the higher sensitivity and lower specificity they found, compared to our results. To summarise this point, the patients in the EPICA study were less symptomatic than patients with heart failure in hospital-based studies, but existing questionnaires are designed to capture information on signs and symptoms of fluid overload and of reduced effort tolerance.

Mosterd et al. [17] assessed the performance of six of the seven scores studied in this work (they did not include the Duke score), in a population of non-hospitalised individuals. They selected 54 subjects from among the first 308 enrolled in the Rotterdam study [18], with the pre-determined aim of obtaining significant numbers of both patients with heart failure and patients with respiratory disease, as respiratory failure is one of the main causes of misdiagnosis of heart failure. With the exception of the Göteborg questionnaire, which had a sensitivity of 80%, all the diagnostic instruments had a sensitivity of 100% for definite heart failure, using the judgement of a cardiologist as the gold standard. Apart from the Walma and Göteborg questionnaires, with sensitivities of 82% and 76%, respectively, all other showed lower sensitivity for the diagnosis of definite or possible heart failure (Framingham 71%, Boston 41%, Gheorghiade 47% and NHANES 52%). The authors themselves report certain limitations of the study, particularly a higher prevalence of heart failure and of chronic pulmonary disease than in the general population, due to the way in which this population was selected and to the small number of individuals studied, which led to wide confidence intervals. They therefore pointed out the possibility that the sensitivity, specificity and predictive values of the tests could be different for the general population, a possibility that we have now confirmed for the Portuguese population.

With regard to the agreement between the various diagnostic instruments, studies are even fewer in number. Good agreement between the Boston and Framingham questionnaires was reported by Remes et al. [13], who showed an agreement of 87% between these two scores in 113 individuals aged between 45 and 74 years with suspected CHF. In another paper [15], the same authors assessed the performance of the Boston questionnaire in the same population, taking as standard reference clinical diagnosis by a cardiologist; agreement between the two methods was 88%. Percent agreement, however, is not a good method to evaluate concordance because of its dependence on the prevalence of a condition. To our knowledge, there are no other studies comparing the various questionnaires.

Our study was based on a community survey using a probability sample, subjects being included in the study irrespective of symptomatic status. Therefore, our study design eliminates the bias introduced by basing the selection of subjects on criteria that are identical or related to items in the questionnaires, a situation that happened in all previous studies. Additionally, the final criteria for the diagnosis of heart failure, the findings from echocardiography, were objective and independent of the subject's symptoms or physical signs. This, in turn, eliminates the bias introduced by using the items in the instruments for the diagnosis of heart failure, as what happens in those studies that use the cardiologists' diagnosis as the gold standard.

Our study may be biased because individuals were selected from people attending primary health care centers in the community, instead of the general population. However, we are convinced that the potential impact in the estimates is very small because the vast majority of the subjects attended the health care center for reasons unrelated to cardiac conditions, usually for primary or secondary prevention evaluations or for minor health problems. The decision to use the people attending primary health care centers in the community as the sampling frame was due to the low feasibility of conducting a study requiring several ancillary diagnostic exams in addition to a clinical evaluation in the general population, without a high number of dropouts and given relatively limited research resources. Possible biases introduced by differences in the demographic structure between study participants and the general population was corrected by post-stratifying the analysis.

The control group referred for investigation was relatively small. However, the sample was used to estimate only specificities. Because specificities were very high overall, this sample size was adequate to provide estimates with a very small error, usually within 1% point. As a consequence of this small sampling error, the estimates of predictive values were also not adversely affected.

Another possible limitation of our study was the use of some of the questions from the Boston questionnaire to screen for subjects who would undergo further laboratory evaluation. This may have resulted in somewhat inflated estimates of sensitivity for this questionnaire and, because many other questionnaires are based on the Boston questionnaire, their estimates of sensitivity could have been inflated as well. Our design is justified by the need to adopt objective and reproducible criteria for defining symptoms and signs of heart failure, a condition for the diagnosis of the syndrome according to the European Society of Cardiology. Furthermore, for operational reasons, we needed to predict the approximate number of ancillary examinations that would be performed, and there were data from the literature that would allow us to make that prediction as a function of the score on the Boston questionnaire. On the other hand, a large number of asymptomatic subjects who were receiving diuretics also underwent laboratory evaluation, which helps to compensate for the potential inflation of the sensitivity estimates.

Our results show that the Göteborg questionnaire has a more balanced relationship between sensitivity and specificity than the other questionnaires. However, the performance of this questionnaire seems to be highly dependent on the inclusion of data on diuretic administration. This means that the performance of this questionnaire is tied to prevailing treatment practices and, with the introduction of new drugs or treatment approaches such as ACE inhibitors and beta-blockers, may not be valid in the future. The Walma, Framingham and NHANES-I questionnaires perform approximately the same, with sensitivities around 60% and specificities over 90%. The Walma questionnaire has a considerable advantage over the other two instruments in that it does not require radiographic or ECG data. The Gheorghiade questionnaire and the Boston questionnaire with the cut-off set on more than four points perform almost identically, with about 55% sensitivity and 95% specificity. The Göteborg questionnaire at the higher cut-off and the Duke questionnaire, although easy to administer, have no other advantages over the previous instruments. Finally, the Boston questionnaire at the higher cut-off has very high specificity, over 99%, but the lowest sensitivity of all.

It would seem to make sense to attempt to increase the sensitivity of the scores by changing the values of the cut-off points used. When we tested this possibility in the heart failure population in the EPICA study, we found that a minimal increase in the sensitivity of the test led to a considerable reduction in its PPV, with a significant increase in the number of false positives.

The analysis of concordance between the questionnaires revealed an overall low concordance although the Boston, Framingham, Gheorghiade, NHANES-1 and Walma questionnaires have a kappa statistic within the good concordance range. The results still are somewhat lower than expected, since the latter three questionnaires derive from the other two.

We also present population estimates of sensitivity and specificity of the symptoms, signs and ECG and chest X-ray findings included in all the questionnaires evaluated. This analysis provides insight on the rational for the development of these questionnaires.


   5. Conclusions
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 5. Conclusions
Acknowledgments
 References
 
The utility of the clinical questionnaires appears limited for clinical diagnostic purposes due to their inadequate sensitivity. They do have a high negative predictive value and therefore, as with natriuretic peptides [19-21], could be used to identify patients who do not require further investigation. With a PPV of around 30% (compared to a pre-test probability of 4.6%), patients who test positive by questionnaire would have about a one in three chance of having an important underlying cardiac problem. This would be highly effective compared to screening measures for other malignant diseases. Currently, echocardiography is the cornerstone of investigation, as it provides a comprehensive, non-invasive assessment of global cardiac function and structure.

In the near future, the natriuretic peptides appear to have a great potential for clinical use and may be used to screen for the presence of heart failure in the community. Brain natriuretic peptide (BNP) has been the most studied candidate for routine diagnosis. It may even have an important role in screening of the wider population in order to distinguish patients with heart failure and left ventricular systolic dysfunction from those with heart failure and preserved ventricular systolic function. As this is an inexpensive, probably cost-effective test, BNP might be used immediately after or in conjunction with clinical screening to determine which patient should have an echocardiogram.


   Acknowledgments
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
References
 
The EPICA Project is supported by Servier Research Group, and has the Scientific Sponsorship of the Portuguese Society of Cardiology and of the Working Group on Heart Failure of the European Society of Cardiology. We are especially grateful to Prof. John G. Cleland for his advice. The authors are indebted to the EPICA investigators, without whom, this work would not have been possible.


   Notes
Top
 Notes
Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
 References
 
1 Serviço de Medicina, Hospital S. Francisco Xavier, 1400 Lisboa, Portugal. Back

2 Datamedica, R. Garcia de Orta 70, 2 D, 1200 Lisboa, Portugal. Back

3 Serviço de Cardiologia, Hospital Pulido Valente, 1750 Lisboa, Portugal. Back

4 R. do Loreto no. 34-3°, 1200 Lisboa, Portugal. Back

5 Grupo de Investigaço EPICA, Av. António Augusto de Aguiar 128, 1050 Lisboa, Portugal. Back

6 Av. Grão Vasco 47-1°-Esq, 1500-336 Lisboa, Portugal. Back

7 EPICA Investigators and Steering Committee Members are listed in the appendix (doi:10.1016/j.ejheart.2004.09.002). Back


   References
Top
 Notes
 Abstract
 1. Introduction
 2. Methods
 3. Results
 4. Discussion
 5. Conclusions
 Acknowledgments
 References
 

  1. The Task Force on Heart Failure of the European Society of Cardiology. Guidelines for the diagnosis of heart failure. Eur. Heart. J. (1995) 16:741–751.[Free Full Text]
  2. McKee P.A., Castelli W.P., McNamara P.M., et al. The natural history of congestive heart failure: the Framingham study. N. Engl. J. Med. (1971) 26:1441–1446.
  3. Eriksson H., Caidahl K., Larsson B., et al. Cardiac and pulmonary causes of dyspnoea. A validation of a scoring test for clinical-epidemiological use. The Study of Men Born in 1913. Eur. Heart J. (1987) 8:1007–1014.[Abstract/Free Full Text]
  4. Eriksson H., Svardsudd K., Caidhal K., et al. Early heart failure in the population. The Study of Men Born in 1913. Acta Med. Scand. (1988) 223:197–209.[Web of Science][Medline]
  5. Schocken D.D., Arrieta M.I., Leaverton P.E., Ross E.A. Prevalence and mortality of congestive heart failure in the United States. J. Am. Coll. Cardiol. (1992) 20:301–306.[Abstract]
  6. Carlson K.J., Lee D.C.-S., Goroll A.H., Leahy M., Johnson R.A. An analysis of physicians' reasons for prescribing long-term digitalis therapy in outpatients. J. Chronic. Dis. (1985) 38:733–739.[CrossRef][Web of Science][Medline]
  7. Harlan W.r., Oberman A., Grimm R., Rosati R.A. Chronic congestive heart failure in coronary artery disease: clinical criteria. Ann. Intern. Med. (1977) 86:133–138.[Abstract/Free Full Text]
  8. Walma E.P., Hoes A.W., Prins A., Boukes F.S., van der Does E. Withdrawing long-term diuretic therapy in the elderly: a study in general practice in the Netherlands. Fam. Med. (1993) 25:661–664.[Medline]
  9. Gheorghiade M., Beller G.A. Effects of discontinuing maintenance digoxin therapy in patients with ischemic heart disease and congestive heart failure in sinus rhythm. Am. J. Cardiol. (1983) 51:1243–1250.[CrossRef][Web of Science][Medline]
  10. McMurray J., Hart W., Rhodes G. An evaluation of the economic costs of heart failure to the National Health Service in the United Kingdom. Br. Med. Econom. (1993) 6:99–110.
  11. Ceia F., Fonseca C., Mota T., Morais H., Matias F., de Sousa A., et al. on behalf of the EPICA Investigators. Prevalence of chronic heart failure in Southwestern Europe: the EPICA study. Eur. J. Heart Fail. (2002) 4:531–539.[Abstract/Free Full Text]
  12. Fonseca C., Morais H., Mota H., et al. The diagnosis of heart failure in primary care: value of symptoms and signs. Eur. J. Heart Fail. (2004) 6:797–802. [in this issue].
  13. Cohen J. A coefficient of agreement for nominal scales. Educ. Psychol. Meas. (1960) 20:37–46.[CrossRef][Web of Science]
  14. Remes J., Reunanen A., Aromaa A., Pyorala K. Incidence of heart failure in eastern Finland: a population-based surveillance study. Eur. Heart J. (1992) 13:588–593.[Abstract/Free Full Text]
  15. Jonsell S.A. A method for the determination of the heart size by teleroentgenograph. Acta Radiol. (1939) 20:325–340.
  16. Remes J., Miettinen H., Reunanen A., Pyorala K. Validity of clinical diagnosis of heart failure in primary health care. Eur. Heart J. (1991) 12:315–321.[Abstract/Free Full Text]
  17. Marantz P.R., Tobin J.N., Wassertheil-Smoller S., et al. The relationship between left ventricular systolic function and congestive heart failure diagnosed by clinical criteria. Circulation (1988) 77:607–612.[Abstract/Free Full Text]
  18. Mosterd A., Deckers J.W., Hoes A.W., et al. Classification of heart failure in population based research: an assessment of six heart failure scores. Eur. Heart J. (1997) 13:491–502.
  19. Hofman A., Grobbee D.E., de Jong P.T., Van den Ouweland F.A., et al. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur. J. Epidemiol. (1991) 7:403–422.[CrossRef][Web of Science][Medline]
  20. Cleland J.G.F., Goode K. Natriuretic peptides for heart failure. Fashionable? Useful? Necessary? Eur. J. Heart Fail. (2004) 6:253–255.[Free Full Text]
  21. McDonagh T.A., Holmer S., Raymond I., Luchner A., Hildebrant P., Dargie H.J. NT-pro BNP and the diagnosis of heart failure: a pooled analysis of three European epidemiological studies. Eur. J. Heart Fail. (2004) 6:269–273.[Abstract/Free Full Text]
  22. Seino Y., Ogawa A., Yamashita T., et al. Application of NT-proBNP and BNP measurements in cardiac care: a more discerning marker for the detection and evaluation of heart failure. Eur. J. Heart Fail. (2004) 6:295–300.[Abstract/Free Full Text]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 StrokeHome page
P. M. Pullicino, L. A. McClure, V. G. Wadley, A. Ahmed, V. J. Howard, G. Howard, and M. M. Safford
Blood Pressure and Stroke in Heart Failure in the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study
Stroke, December 1, 2009; 40(12): 3706 - 3710.
[Abstract] [Full Text] [PDF]

Home page
 J. Am. Soc. Nephrol.Home page
A. Kottgen, S. D. Russell, L. R. Loehr, C. M. Crainiceanu, W. D. Rosamond, P. P. Chang, L. E. Chambless, and J. Coresh
Reduced Kidney Function as a Risk Factor for Incident Heart Failure: The Atherosclerosis Risk in Communities (ARIC) Study
J. Am. Soc. Nephrol., April 1, 2007; 18(4): 1307 - 1315.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Fonseca, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fonseca, C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?