REVIEW ARTICLE. A Systematic Review of the Diagnostic Accuracy of Natriuretic Peptides for Heart Failure

Transcription

1 REVIEW ARTICLE A Systematic Review of the Diagnostic Accuracy of Natriuretic Peptides for Heart Failure Jenny A. Doust, BMBS, FRACGP; Paul P. Glasziou, FRACGP, PhD; Eva Pietrzak, PhD; Annette J. Dobson, PhD Background: The diagnosis of heart failure is difficult, with both overdiagnosis and underdiagnosis occurring commonly in practice. Natriuretic peptides have been proposed as a possible test for assisting diagnosis. We assessed the diagnostic accuracy of brain natriuretic peptide (BNP), including a comparison with atrial natriuretic peptide (ANP). Methods: Electronic searches were conducted of MEDLINE and EMBASE from January 1994 to December 2002 and handsearches of reference lists of included studies. We included studies that assessed the diagnostic accuracy of BNP against echocardiographic or clinical criteria or that compared the diagnostic accuracy of BNP with ANP. Two reviewers assessed studies for inclusion and quality and extracted the relevant data. A meta-analysis was performed by pooling the diagnostic odds ratios for studies that used a common reference standard. Results: Twenty studies were included. For the 8 studies (n=4086) that measured BNP against the criterion of left ventricular ejection fraction of 40% or less (or equivalent), the pooled diagnostic odds ratio was 11.6 (95% confidence interval, ). The pooled diagnostic odds ratio was greater, 30.9 (95% confidence interval, ), in the 7 studies (n=2374) that measured BNP against clinical criteria (generally a consensus view using all other clinical information). The diagnostic odds ratio was similar in studies conducted in general practice and in hospital settings. Three studies compared BNP with N-terminal ANP, a precursor form of ANP, and pooling of the results of these studies showed BNP to be a more accurate marker of heart failure than NT-ANP. Conclusions: Brain natriuretic peptide is an accurate marker of heart failure. Use of a cutoff value of 15 pmol/l achieves high sensitivity, and BNP values below this exclude heart failure in patients in whom disease is suspected. As the diagnostic odds ratio for BNP is greater when assessed against clinical criteria than against left ejection fraction alone, BNP may also be detecting patients with diastolic heart failure. Arch Intern Med. 2004;164: From the Centre for General Practice (Drs Doust, Glasziou, and Pietrzak) and Division of Epidemiology and Social Medicine (Dr Dobson), School of Population Health, University of Queensland, Herston, Australia. The authors have no relevant financial interest in this article. THE RISING PREVALENCE 1 AND cost 2 of heart failure and increasing treatment options 3 have made the accurate diagnosis of heart failure increasingly important. Heart failure is difficult to diagnose correctly, with both overdiagnosis and underdiagnosis occurring commonly in practice. Studies that have investigated the prevalence of echocardiographic abnormalities in populations have found that at least half of patients with significant left ventricular dysfunction on echocardiogram are asymptomatic or have not previously been diagnosed as having heart failure. 4,5 In the Rotterdam Study, a population-based cohort study of chronic disease in persons aged 55 years or older, 60% of people with a left ventricular ejection fraction of 25% or less were asymptomatic. 6 Conversely, patients with signs and symptoms suggestive of heart failure are frequently found not to have the disease when measured against more objective criteria. One study investigated all suspected new cases of heart failure referred by general practitioners during a 15-month period against a consensus decision of 3 cardiologists, including the results of further investigations such as echocardiogram. Against this definition, only 30% of the referred patients had heart failure. 7 The diagnosis of heart failure is hindered if access to further investigations is limited. It is especially difficult for general practitioners, who are faced with many patients at high risk of the disease and who must make decisions regarding appropriate further investigation, treatment, and referral. General practitioners in the United Kingdom identified a lack of confidence in establishing an accurate diagnosis as a major barrier to treating patients with heart 1978

2 failure. 8 Both in general practice and in emergency departments, many patients presenting with symptoms of heart failure have comorbidities that may also account for their symptoms. The myocardium releases neuropeptides that serve to maintain circulatory homeostasis. A-type (atrial) natriuretic peptide (ANP) is secreted primarily by the atrial myocardium in response to dilation, and B-type (brain) natriuretic peptide (BNP) is secreted by the ventricles in response to end-diastolic pressure and volume. 9 Both ANP and BNP and various precursor forms, such as N-terminal (NT) ANP, have been evaluated as potential diagnostic tests for heart failure. We performed a systematic review of the literature and meta-analysis to quantify the diagnostic accuracy of BNP for the diagnosis of heart failure and to compare it with the diagnostic accuracy of ANP. METHODS We searched MEDLINE and EMBASE from January 1994 to December 2002 for all studies of the diagnostic accuracy of natriuretic peptides for heart failure. The search strategy for MEDLINE is available fom the authors. The reference lists of primary studies and review articles identified by the search were checked for further relevant studies. We included all studies that compared the diagnostic accuracy of natriuretic peptides against a reference standard and where the results were reported so that a 2 2 table of results could be constructed. Several studies that examined the association between natriuretic peptide levels and heart failure were excluded, as were casecontrol studies. Six studies involving overlapping or duplicate cohorts of patients were also excluded. Two reviewers (J.A.D. and E.P.) assessed independently the quality of each study and extracted data. Disagreements were resolved by consensus or by consulting a third reviewer (P.P.G.). Each reviewer extracted the data to construct a 2 2 table for every cutoff point that was published in each study. To allow for changes in the cutoff points used both within and between studies, the diagnostic odds ratio (DOR)={[Sensitivity/ (1 Sensitivity)]/[(1 Specificity)/ Specificity]} was chosen as the measure of diagnostic accuracy. If there was more than one cutoff point within a study, we took the average of the DORs for each cutoff point. This was done by taking the average of the natural logarithm of the odds ratio and the average of the variance of the natural logarithm of the odds ratio for cutoff points within a study (unweighted, because the study size was the same in each case), and back-transforming to calculate the average DOR and confidence interval (CI). The studies were grouped so that a DOR was calculated against each reference standard, using a DerSimonian and Laird randomeffects model 10 on a logarithmic scale with a corresponding test of heterogeneity. Where possible, the positive and negative likelihood ratios were calculated where studies had used similar cutoff levels and the same reference standard, again using a DerSimonian and Laird random-effects model. An unweighted least-squares regression model using the method of Moses et al 11 was used to assess whether the odds ratio was independent of the cutoff point. The model is D=a+bS, where the difference D is defined as the logit true-positive rate minus the logit falsepositive rate, which is equivalent to the log of the DOR, and the sum S is defined as the logit true-positive rate plus the logit false-positive rate, which is a measure in the variation of the threshold. For each study, the values D and S were calculated, then the regression line was fitted. An unpaired, 2-tailed t test was used to see whether the slope of the line was significantly different from zero, which would imply that the overall diagnostic accuracy of the test varied with the cutoff point used. In studies that compared the diagnostic accuracy of BNP with ANP, the estimated area under the curve (AUC) for each study was pooled by an inverse variance method. The diagnostic accuracy of the 2 natriuretic peptides was assessed by comparing the difference between the 2 pooled AUCs divided by the variance of the AUC, AUC BNP AUC ANP var BNP + var ANP with a standard normal distribution. 12 This method will underestimate the true measure of difference, as the data are derived from paired study designs, but would require individual patient data to estimate the true difference. RESULTS We identified 20 studies with published data that met the inclusion criteria of the review (Table 1). The quality of the studies was generally high, with most satisfying the following criteria (with the number reporting each criterion in parentheses): (1) patients were a consecutive series or random sample (15/20); (2) the index and reference tests were assessed independently and blinded to the other test result (15/20); (3) all patients received both tests (16/20); (4) the methods for performing both tests were described (20/20); (5) the characteristics of the study population were described (20/20); and (6) there was no time delay between the measurement of the 2 tests (14/20). The details of the first 2 quality criteria are also shown in Table 1. The DORs of BNP against each reference standard are shown in Figure 1. In the 8 studies that measured BNP against a left ventricular ejection fraction of 40% or less, the pooled DOR was 11.6 (95% CI, ). The results of the studies were consistent, with no evidence of heterogeneity. With less restrictive echocardiographic criteria, the DOR was smaller and the degree of heterogeneity greater, as would be expected with studies with a more imperfect reference standard and varying levels for the reference threshold. Using a reference standard of left ventricular ejection fraction of 40% or less and pooling studies that used a cutoff between 14 and 19 pmol/l gave an estimated positive likelihood ratio of 4.1 (95% CI, ) and a negative likelihood ratio of 0.35 (95% CI, ). In the 7 studies that measured BNP against a consensus clinical opinion, generally using all other diagnostic information available, the DOR was greater (30.9; 95% CI, ). The degree of heterogeneity between the studies was also low. This result suggests that BNP and clinical diagnosis are in greater agreement than BNP and left ventricular function, assuming no other differences between the 2 groups of studies. The results were heavily weighted by the results of one study, however, the Breathing Not Properly study reported by Maisel et al in This was a multicenter study with 1586 patients that assessed the diagnostic accuracy of BNP in patients presenting to 7 emergency departments against the diagnosis by 2 cardiologists who had access to all clinical data, includ- 1979

3 Table 1. Accuracy of Brain Natriuretic Peptides for Diagnosis of Heart Failure by Diagnostic Standard Prevalence, % Quality: Series,* Blind Reference Test Criteria Cutoff Closest to 15 pmol/l Source Setting, Location N Index Test Reference Standard, LVEF 30% McDonagh et al, 13 MONICA study, , NR LVEF 30% BNP, Peninsula Glasgow, Scotland Reference Standard, LVEF 40% Bettencourt et al, 14 Patients day 4 and 5 after MI, , + LVEF 40% BNP, Shionogi Portugal Choy et al, Patients day 3 after MI, , + LVEF 40% BNP, Peninsula Dundee, Scotland Valli et al, , + LVEF 40% BNP, CIS Bio radionuclide ventriculography Female Framingham Study , + LVEF 40% and/or BNP, Shionogi FS 22% Male Framingham Study LVEF 40% and/or FS 22% BNP, Shionogi Hutcheon et al, Patients referred to day hospital with suspected cardiovascular disease, Dundee , + Qualitative assessment of LVSD BNP, Peninsula Landray et al, Patients referred to hospital NR, NR Qualitative BNP, Shionogi clinic with suspected heart failure, Oxford, England assessment of LVSD Smith et al, General practice patients aged 70 to 84 y, England , + Qualitative assessment of LVSD BNP, Peninsula McGeoch et al, Yamamoto et al, Yamamoto et al, Luchner et al, Krishnaswamy et al, Osca et al, General practice patients being treated for heart failure, Christchurch, New Zealand echocardiography to assess ventricular function, Mayo Clinic, Rochester, Minn cardiac catheterization, Mayo Clinic MONICA study, Augsburg, Germany echocardiogram, San Diego, Calif Female Framingham Study Male Framingham Study Patients admitted for heart failure Reference Standard, LVEF 45%-55% , + LVEF 45% BNP, Christchurch assay TP, FN, TN, FP, , + LVEF 45% BNP, Shionogi , + LVEF 45% BNP, Shionogi , + Fractional shortening 28% BNP, Shionogi NR, + LVEF 50% or BNP, Biosite global hypokinesis or wall motion abnormality , + LVEF 50% and/or BNP, Shionogi FS 29% , + LVEF 50% and/or BNP, Shionogi FS 29% , NR LVEF 55% BNP, Shionogi (continued) ing results of radiology and echocardiography. In this study, using a cutoff level of 14.4 pmol/l, the positive likelihood ratio was 2.6 (95% CI, ) and the negative likelihood ratio was 0.05 (95% CI, ). In the 2 studies that measured BNP against echocardiographic criteria for both systolic and diastolic heart failure, the DOR (37.7; 95% CI, ) was again greater than in studies that measured only systolic function, but the results of the 2 studies were different. Of the studies that investigated BNP vs systolic or systolic plus diastolic function, 7 studies were conducted in general practice or community settings and 11 were conducted in hospital settings. The DOR from the studies in the 2 settings showed similar results (Figure 1). As these groups of studies pool results using different reference standards, they both showed highly statistically significant levels of heterogeneity. 1980

4 Table 1. Accuracy of Brain Natriuretic Peptides for Diagnosis of Heart Failure by Diagnostic Standard (cont) Source Setting, Location N Cowie et al, Davis et al, Hobbs et al, Hobbs et al, Incidence cases of heart failure in general practice, Hillingdon, London, England Patients admitted for acute dyspnea, Christchurch General population aged 45 y, England Patients at high risk of heart failure Prevalence, % Quality: Series,* Blind Reference Test Criteria Reference Standard, Clinical Diagnosis , + ESC criteria NR, NR Committee of 3 physicians and radiologist , + ESC criteria , + ESC criteria Hobbs et al, Patients taking diuretics , + ESC criteria Hobbs et al, Maisel et al, Castro et al, Krishnaswamy et al, Lubien et al, Krishnaswamy et al, Bettencourt et al, Patients with existing diagnosis of heart failure BNP study, patients attending emergency department with dyspnea (United States, France, Norway) Patients referred with suspected heart failure and LVEF 40% echocardiogram, San Diego echocardiogram with normal LV function, San Diego echocardiogram, San Diego Patients day 4 and 5 after MI, Portugal , + ESC criteria NR, + 2 cardiovascular clinicians Reference Standard, Diastolic Failure , NR Symptoms and signs of HF and abnormal pattern of mitral inflow or atrial fibrillation and no other cause of symptoms NR, + Diastolic dysfunction = impaired relaxation or restrictive or pseudonormal pattern , + Diastolic dysfunction = impaired relaxation or restrictive or pseudonormal pattern Reference Standard, Systolic or Diastolic Failure NR, + Systolic or diastolic dysfunction , + LVEF 40% or LVEDD index 3.2 cm/m 2 + abnormal relaxation or restrictive pattern Index Test Cutoff Closest to 15 pmol/l TP, FN, TN, FP, BNP, Peninsula BNP, Christchurch assay NT-proBNP, Roche NT-proBNP, Roche NT-proBNP, Roche NT-proBNP, Roche BNP, Biosite BNP, Shionogi BNP, Biosite BNP, Biosite BNP, Biosite BNP, Shionogi Abbreviations: BNP, brain natriuretic peptide; ESC, European Society of Cardiology; FN, false negative; FP, false positive; FS, fractional shortening; HF, heart failure; LVEDD, left ventricular end-diastolic diameter; LVEF, left ventricular ejection fraction; LVSD, left ventricular systolic dysfunction; MI, myocardial infarction; MONICA, Multinational Monitoring of Trends and Determinants in Cardiovascular Disease; NT, N-terminal; TN, true negative; TP, true positive. *Consecutive series or random sample of a consecutive series of patients. The results of the index and the reference test were read independently and blinded to the result of the other test; plus sign indicates fulfilled; minus sign, not fulfilled; and NR, not reported. Manufacturer information is as follows: Peninsula Laboratories Europe Ltd (Peninsula), St Helens, England; Shionogi, Osaka, Japan; CIS Bio International (CIS Bio), Gif-sur-Yvette, France; Biosite Incorporated (Biosite), San Diego, Calif; and Roche Diagnostics (Roche), Mannheim, Germany. 1981

5 The data for the individual studies and the curve derived from the pooled odds ratio for studies comparing BNP with a reference standard of left ventricular ejection fraction of 40% or less are shown in Figure 2. This figure illustrates the variation seen between studies in BNP vs LVEF 40% (Studies = 8, n = 4086) BNP vs LVEF 45%-55% (Studies = 7, n = 7808) BNP vs Clinical Criteria (Studies = 7, n = 2374) BNP vs Diastolic (Studies = 3, n = 759) BNP vs Systolic + Diastolic (Studies = 2, n = 504) BNP in General Practice (Studies = 7, n = 5583) BNP in Hospital (Studies = 11, n = 3453) BNP After MI (Studies = 2, n = 176) DOR = 11.6 ( ), χ2 7= 6.9, P =.44 DOR = 5.6 ( ), χ2 6 = 19.4, P <.01 DOR = 30.9 ( ), χ2 6 = 5.98, P =.43 DOR = 28.3 ( ), χ2 2 = 128.4, P <.001 DOR = 37.7 ( ), χ2 1= 5.6, P =.02 DOR = 10.8 ( ), χ2 6= 34.0, P <.001 DOR = 12.2 ( ), χ 10 2 = 48.8, P <.001 DOR = 9.4 ( ), χ2 1= 0.5, P = Diagnostic Odds Ratio Figure 1. Pooled diagnostic odds ratio (DOR) (95% confidence interval), with 2 test for homogeneity. BNP indicates brain natriuretic peptide; LVEF, left ventricular ejection fraction; and MI, myocardial infarction. The abscissa is on a log scale. terms of the sensitivity and specificity for various cutoff points. The Moses et al 11 regression model was estimated for studies that used a reference standard of left ventricular ejection fraction of 40% or less and clinical diagnosis. These both showed a significant negative relationship, implying that the diagnostic accuracy decreases with increases in the cutoff. 32 Six studies compared the diagnostic accuracy of BNP and ANP. The results of these studies are Sensitivity DOR = 11.6 AUC = Specificity Figure 2. Estimates of diagnostic accuracy from individual studies that measured brain natriuretic peptide vs left ventricular ejection fraction of 40% or less plotted in receiver operating characteristic curve space, ie, sensitivity vs 1 specificity. Results from different cutoff points within the same study are joined, with the numbers next to the circles being the cutoff point (in picomoles per liter). The pooled diagnostic odds ratio (DOR) is used to estimate the plot of summary receiver operating characteristic curve and the area under the curve (AUC). The area of the circles is proportional to the size of the studies. Table 2. Studies That Compared BNP and ANP Source McDonagh et al, Choy et al, Cowie et al, Cowie et al, Setting MONICA study, Glasgow, Scotland Patients day 3 after MI, Dundee, Scotland Female Framingham Male Framingham Female Framingham Female Framingham (reference = LVEF 40%) Incident cases of suspected heart failure in general practice, Hillingdon, London, England Incident cases of suspected heart failure in general practice, Hillingdon, London Reference Standard Used Form of ANP Result for ANP* Result for BNP* LVEF 30% NT-ANP AUC = 0.75 ( ) AUC = 0.88 ( ) LVEF 40% ANP OR = 1.45 ( ) OR = 8.57 ( ) LVEF 40% NT-ANP AUC = 0.88 ( ) AUC = 0.85 ( ) LVEF 40% NT-ANP AUC = 0.78 ( ) AUC = 0.79 ( ) LVEF 50% NT-ANP AUC = 0.54 ( ) AUC = 0.56 ( ) LVEF 50% NT-ANP AUC = 0.70 ( ) AUC = 0.72 ( ) ESC criteria (3 cardiovascular ESC criteria (3 cardiovascular NT-ANP OR = ( ) OR = ( ) ANP OR = ( ) OR = ( ) Abbreviations: ANP, atrial natriuretic peptide; AUC, area under the curve; BNP, brain natriuretic peptide; ESC, European Society of Cardiology; LVEF, left ventricular ejection fraction; MI, myocardial infarction; MONICA, Multinational Monitoring of Trends and Determinants in Cardiovascular Disease; NT, N-terminal; OR, odds ratio. *The 95% confidence intervals are given in parentheses. 1982

6 shown in Table 2. 7,13,15,17 The BNP was generally more accurate as a diagnostic marker of heart failure than ANP. The AUCs for the receiver operating characteristic curves were available for 3 studies that compared NT-ANP with BNP. The pooled AUC for the 3 studies that compared NT-ANP with echocardiogram was 0.78 (95% CI, ). The pooled AUC for BNP in the same studies was 0.84 (95% CI, ). The test statistic for the difference between the 2 measures shows a marginally statistically significant difference with a z score of 1.66 (P=.048). This calculation underestimates the true significance because the original data were from paired study designs. COMMENT The results of these 20 studies show that BNP is accurate in the diagnosis of heart failure. However, there is considerable variation in the estimates of diagnostic accuracy between studies, and this variation does not seem to be accounted for by differences in the clinical setting or the type of test used. It would be helpful to understand the sources of this variation before recommending the routine clinical use of the test. Measurement of BNP is cheaper and is potentially more accessible than echocardiography. Results of BNP testing can be obtained within 20 minutes of blood collection. 33,34 As echocardiography provides additional information that may be important in the clinical treatment of patients with heart failure, the most likely use of BNP will be in the ambulatory care setting to determine which patients require further testing with echocardiogram. The cutoff level will therefore need to be sufficiently low that patients who have heart failure are not excluded from further testing. In the Breathing Not Properly study, combining BNP with clinical judgment improved diagnostic accuracy across the entire range of diagnostic certainty. Even when the clinician had a high degree of certainty in his or her diagnosis, combining this judgment with the BNP result improved diagnostic accuracy. 35 Measurement of BNP may play a role in the diagnosis of patients with diastolic heart failure. The diagnostic accuracy of BNP was greater when the definition of disease used as the reference standard included patients who were diagnosed as having heart failure but who had preserved left ventricular systolic dysfunction. This raises the question of whether BNP could be a better marker of disease, prognosis, and response to treatment than left ventricular function and whether comparison with echocardiographic criteria of left ventricular function may, in fact, be a comparison with a silver standard. To date, almost all of the trials of treatment for heart failure have been conducted in patients with impaired left ventricular function Although there is a group of patients who have heart failure with preserved left ventricular systolic function (diastolic heart failure), it is not clear how these patients could be identified or how patients chosen by alternative criteria would respond to treatment. 39 Using a cutoff value of 15 pmol/l achieves high sensitivity, and BNP values below this are able to exclude heart failure in patients in whom disease is suspected. Early studies have also shown a potential role for BNP in monitoring the response to treatment. 40,41 Future research will need to further define BNP s role in clinical management, and to determine whether it is better at predicting prognosis and response to treatment than echocardiography. Accepted for publication March 23, We thank Jon Deeks, Senior Medical Statistician, Centre for Statistics in Medicine, Oxford, England, for comments on an earlier draft of this article. Correspondence: Jenny A. Doust, BMBS, FRACGP, Centre for General Practice, School of Population Health, University of Queensland, Herston Road, Herston Qld 4006, Australia (j.doust@uq.edu.au). REFERENCES 1. Davies M, Hobbs F, Davis R, et al. Prevalence of left-ventricular systolic dysfunction and heart failure in the Echocardiographic Heart of England Screening study: a population based study. Lancet. 2001;358: Redfield M. Heart failure: an epidemic of uncertain proportions. N Engl J Med. 2002;347: Redfield MM, Jacobsen SJ, Burnett JC Jr, Mahoney DW, Bailey KR, Rodeheffer RJ. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA. 2003;289: Morgan S, Smith H, Simpson I, et al. Prevalence and clinical characteristics of left ventricular dysfunction among elderly patients in general practice setting: cross sectional survey. BMJ. 1999; 318: McDonagh TA, Morrison CE, Lawrence A, et al. Symptomatic and asymptomatic left-ventricular systolic dysfunction in an urban population. Lancet. 1997;350: Mosterd A, Hoes AW, de Bruyne MC, et al. Prevalence of heart failure and left ventricular dysfunction in the general population: the Rotterdam Study. Eur Heart J. 1999;20: Cowie MR, Struthers AD, Wood DA, et al. Value of natriuretic peptides in assessment of patients with possible new heart failure in primary care. Lancet. 1997;350: Fuat A, Hungin AP, Murphy JJ. Barriers to accurate diagnosis and effective management of heart failure in primary care: qualitative study. BMJ. 2003;326: Baughman KL. B-type natriuretic peptide: a window to the heart. N Engl J Med. 2002;347: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7: Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data-analytic approaches and some additional considerations. Stat Med. 1993; 12: Pepe M. The Statistical Evaluation of Medical Tests for Classification and Prediction. Oxford, England: Oxford University Press; 2003: McDonagh TA, Robb SD, Murdoch DR, et al. Biochemical detection of left-ventricular systolic dysfunction. Lancet. 1998;351: Bettencourt P, Ferreira A, Pardal-Oliveira N, et al. Clinical significance of brain natriuretic peptide in patients with postmyocardial infarction. Clin Cardiol. 2000;23: Choy AM, Darbar D, Lang CC, et al. Detection of left ventricular dysfunction after acute myocardial infarction: comparison of clinical, echocardiographic, and neurohormonal methods. Br Heart J. 1994;72: Valli N, Georges A, Corcuff JB, Barat JL, Bordenave L. Assessment of brain natriuretic peptide in patients with suspected heart failure: comparison with radionuclide ventriculography data. Clin Chim Acta. 2001;306: Vasan RS, Benjamin EJ, Larson MG, et al. Plasma natriuretic peptides for community screening for left ventricular hypertrophy and systolic dysfunction: the Framingham Heart Study. JAMA. 2002; 288: Hutcheon SD, Gillespie ND, Struthers AD, McMurdo ME. B-type natriuretic peptide in the diagnosis of cardiac disease in elderly day hospital patients. Age Ageing. 2002;31: Landray MJ, Lehman R, Arnold I. Measuring brain natriuretic peptide in suspected left ventricular systolic dysfunction in general practice: crosssectional study. BMJ. 2000;320: Smith H, Pickering RM, Struthers A, Simpson I, Mant D. Biochemical diagnosis of ventricular dysfunction in elderly patients in general practice: observational study. BMJ. 2000;320:

7 21. McGeoch G, Lainchbury J, Town GI, Toop L, Espiner E, Richards AM. Plasma brain natriuretic peptide after long-term treatment for heart failure in general practice. Eur J Heart Fail. 2002;4: Yamamoto K, Burnett JC Jr, Bermudez EA, Jougasaki M, Bailey KR, Redfield MM. Clinical criteria and biochemical markers for the detection of systolic dysfunction. J Card Fail. 2000;6: Yamamoto K, Burnett JC Jr, Jougasaki M, et al. Superiority of brain natriuretic peptide as a hormonal marker of ventricular systolic and diastolic dysfunction and ventricular hypertrophy. Hypertension. 1996;28: Luchner A, Burnett JC Jr, Jougasaki M, et al. Evaluation of brain natriuretic peptide as marker of left ventricular dysfunction and hypertrophy in the population. J Hypertens. 2000;18: Krishnaswamy P, Lubien E, Clopton P, et al. Utility of B-natriuretic peptide levels in identifying patients with left ventricular systolic or diastolic dysfunction. Am J Med. 2001;111: Osca J, Quesada A, Arnau MA, et al. Brain natriuretic peptide: diagnostic value in heart failure [in Spanish]. Rev Esp Cardiol. 2002;55: Davis M, Espiner E, Richards G, et al. Plasma brain natriuretic peptide in assessment of acute dyspnoea. Lancet. 1994;343: Hobbs F, Davis R, Roalfe A, Hare R, Davies M, Kenkre J. Reliability of N-terminal pro-brain natriuretic peptide assay in diagnosis of heart failure: cohort study in representative and high risk community populations. BMJ. 2002;324: Maisel AS, Krishnaswamy P, Nowak RM, et al. Rapid measurement of B-type natriuretic peptide in the emergency diagnosis of heart failure. N Engl J Med. 2002;347: Castro A, Dias P, Pereira M, et al. Type B natriuretic peptide in the diagnosis of heart failure with preserved systolic function [in Portuguese]. Rev Port Cardiol. 2001;20: Lubien E, DeMaria A, Krishnaswamy P, et al. Utility of B-natriuretic peptide in detecting diastolic dysfunction: comparison with Doppler velocity recordings. Circulation. 2002;105: Boyko EJ. Re: Meta-analysis of Pap test accuracy. Am J Epidemiol. 1996;143: Kalra PR, Struthers AD. More evidence for bedside BNP in heart failure assessment. Int J Cardiol. 2002;86: Del Ry S, Giannessi D, Clerico A. Plasma brain natriuretic peptide measured by fully-automated immunoassay and by immunoradiometric assay compared. Clin Chem Lab Med. 2001;39: McCullough PA, Nowak RM, McCord J, et al. B- type natriuretic peptide and clinical judgment in emergency diagnosis of heart failure: analysis from Breathing Not Properly (BNP) Multinational Study. Circulation. 2002;106: Cardiac Insufficiency Bisoprolol Study II (CIBIS-II): a randomised trial. Lancet. 1999;353: Pfeffer MA, Braunwald E, Moye LA, et al, SAVE Investigators. Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction: results of the survival and ventricular enlargement trial. N Engl J Med. 1992;327: Packer M, Coats AJ, Fowler MB, et al. Effect of carvedilol on survival in severe congestive heart failure. N Engl J Med. 2001;344: Cheitlin MD. Can clinical evaluation differentiate diastolic from systolic heart failure? if so, is it important? Am J Med. 2002;112: Troughton RW, Frampton CM, Yandle TG, Espiner EA, Nicholls MG, Richards AM. Treatment of heart failure guided by plasma aminoterminal brain natriuretic peptide (N-BNP) concentrations. Lancet. 2000;355: Murdoch DR, McDonagh TA, Byrne J, et al. Titration of vasodilator therapy in chronic heart failure according to plasma brain natriuretic peptide concentration: randomized comparison of the hemodynamic and neuroendocrine effects of tailored versus empirical therapy. Am Heart J. 1999; 138: