The Relationship Between N-terminal Pro-brain Natriuretic Peptide and Risk for Hospitalization and Mortality is Curvilinear in Patients With Chronic Heart Failure

Morten Schou, MD; Finn Gustafsson, MD, PhD; Pernille Corell, MD; Caroline N. Kistorp, MD, PhD; Andreas Kjaer, MD, PhD, DrMSci; Per R. Hildebrandt, MD, DrMSci

Am Heart J.  2007;154(1):123-129.  ©2007 Mosby, Inc.
Posted 07/20/2007

Abstract and Introduction

Abstract

Background: N-terminal pro-brain natriuretic peptide (NT-proBNP) carries prognostic information in patients with chronic heart failure and predicts risk for mortality and cardiovascular events. It is unknown whether NT-proBNP predicts risk for hospitalization for any cause. Furthermore, a clinically useful algorithm for risk stratification based on NT-proBNP as a continuous variable has not yet been described. We therefore evaluated NT-proBNP as a risk marker for mortality and hospitalization and developed a simple algorithm for risk stratification based on NT-proBNP as a continuous variable.
Methods: Data from 345 patients with chronic heart failure were collected prospectively in our heart failure clinic, and the patients were followed for 28 months (median). Seventy patients died, and 201 patients were hospitalized. Cox proportional hazard models for mortality and hospitalization were constructed with NT-proBNP as a dichotomous (median 1381 pg/mL) and a continuous variable (log2NT-proBNP).
Results: Patients with supramedian levels of NT-proBNP had a 2.40-fold (95% CI 1.40-4.10) increased risk for mortality and 1.71- fold (95% CI 1.24-2.36) increased risk for hospitalization. The effect of doubling NT-proBNP on adjusted hazard ratios was 1.56 (95% CI 1.32-1.85) for mortality and 1.19 (95% CI 1.09-1.31) for hospitalization. We observed a curvilinear relationship between NT-proBNP and risk for mortality and hospitalization in the whole range of NT-proBNP.
Conclusions: N-terminal pro-brain natriuretic peptide predicts risk for hospitalization and mortality. A simple algorithm indicates that every time NT-proBNP is doubled, estimated hazard ratio for death increases by a factor of 1.56 (56%) and by a factor of 1.19 (19%) for hospitalization. Finally, the relationship between NT-proBNP and risk is curvilinear if NT-proBNP is considered as a continuous variable.

Introduction

Risk stratification in patients with chronic heart failure is based on a variety of clinical and laboratory variables. Indeed, several predictors of prognosis have been identified, including age, New York Heart Association (NYHA) class, left ventricular ejection fraction, renal function, body mass index (BMI), and comorbidities such as atrial fibrillation, diabetes mellitus, and ischemic heart disease[1]; but a single measurement of plasma concentrations of N-terminal pro-brain natriuretic peptide (NT-proBNP) has shown promise for simple and reliable risk stratification in patients with chronic heart failure. However, NT-proBNP data usually follow a log-normal distribution, and appropriate statistical analyses are therefore mostly calculated from log(NT-proBNP) or log10(NT-proBNP). Consequently, the association between NT-proBNP and risk for mortality may be difficult to implement quantitatively into clinical practice. A simple algorithm that quantitatively describes the relationship between hazard ratio and NT-proBNP as a continuous variable in a clinically useful way has not been described to date.

Previously, NT-proBNP has been shown to predict heart failure and mortality in patients with left ventricular dysfunction, overall mortality, sudden cardiac death, death from pump failure, cardiovascular events, combined mortality and morbidity, and hospitalization due to heart failure in patients with chronic heart failure.[2-7] However, hospitalization for any cause is a highly relevant end point in an outpatient heart failure clinic because patients are often hospitalized for dehydration, electrolyte disturbances, problems with comorbidity, and infections as complication to their chronic heart failure disease. Furthermore, most data on the prognostic value of NT-proBNP in heart failure have been derived from hospitalized patients or patients enrolled in randomized clinical trials. Currently, available guidelines recommend that patients are followed in specialized heart failure outpatient clinics, and indeed a large share of patients in several countries are now being managed in nurse-based or multidisciplinary heart failure clinics.[8-10] Hence, it seems of particular interest to investigate in detail the prognostic value of NT-proBNP in a population of patients with heart failure recruited from a nurse-based heart failure clinic.

The aim of the present study was therefore to evaluate NT-proBNP for risk stratification for mortality and hospitalization for any cause in patients with chronic heart failure and to establish a simple algorithm describing the relationship between NT-proBNP modulated as a continuous variable and risk (hazard ratio) for use in clinical practice in heart failure clinics.

Methods

Study Population

All of the patients (N = 345) in the present study were included prospectively from our specialized heart failure clinic at Frederiksberg University Hospital, Copenhagen, Denmark.[11] The clinic has been operating since 1999. Patients with known systolic chronic heart failure (left ventricular ejection fraction of <0.45 by echocardiography) can be referred to the clinic directly either by general practitioners or by the departments of internal medicine or cardiology at the hospital.

At the baseline visit, all patients were examined by a physician, and the following information was obtained: medical history, medication, physical examination, NYHA classification based on patient information, measurements of height and weight, resting blood pressure and heart rate, x-ray of the heart and lungs, and electrocardiogram. All data were collected in the database program Hjerterplus (Frederiksberg, Denmark).[11] Data in the present study were collected from January 2000 to April 2005.

Venous blood samples to determine S-sodium, S-potassium, S-creatinine, and B-hemoglobin were drawn from all patients, and if the patients accepted, NT-proBNP was also analyzed. Testing for NT-proBNP was approved by the ethical committee of Copenhagen (KF 01-019699), and informed consent was obtained according to the Helsinki Declaration II.

Laboratory Analysis

After a minimum 8-hour overnight fast and 15-minute rest, venous blood was drawn into EDTA (NT-proBNP) and heparin tubes (sodium, potassium, and creatinine). It was either (1) promptly centrifuged at 4°C (3000 rpm in 10 minutes), and plasma was analyzed the same day; or (2) stored as frozen plasma at -80°C in aliquots.

Plasma concentrations of sodium and potassium (Integra 700, Diamond Diagnostics, Holliston, MA) and creatinine[12] and hemoglobin (Sysmex XE 2100 TOA Medical Electronics, Kobe, Japan) were analyzed on the same day.

Plasma concentrations of NT-proBNP were analyzed on the Elecsys 2010 platform[13] (Roche Diagnostics, Basel, Switzerland) by a double-antibody "sandwich" method (Eclia sandwich assay) with intra- and interassay variation coefficients of <3%. The analytical range is from 5 to 35000 pg/mL. Analysis was performed on the same day or on frozen samples (first 197 patients). Storing plasma samples at -80°C does not affect the plasma level of NT-proBNP.[14]

Estimated glomerular filtration rate (eGFR) was calculated from the 4-component modification of diet in renal disease (MDRD) equation incorporating age, race, sex, and serum creatinine level (x): eGFR = 186 • (serum creatinine in milligrams per deciliter)-1.154 • (age in years)-0.203. For women and for African Americans (none in our cohort), the product of the equation has to be multiplied by a correction factor of 0.742 and 1.21, respectively.[15]

Information on vital status and hospitalization was obtained from the Danish Personal Register (Copenhagen, Denmark), which registers all deaths within 2 weeks and hospitalization within 6 months. Hospitalization was defined as hospitalization of >24 hours at a medical department. Follow-up ended on July 1, 2005. No patients were lost to follow-up. Median follow-up time was 887 days (range 90-2080 days).

Statistical Analysis

The purpose of the present cohort study was to test the hypothesis that NT-proBNP carries independent prognostic information concerning mortality in patients with chronic heart failure. Number of needed events were calculated to 71, and sample size was then calculated to 284 to 355 based on an expected mortality rate of 20% to 25% after 900 days (median) of follow-up and an expected hazard ratio of 2.0 for NT-proBNP (greater than vs less than the median) for mortality (power 80%, significance level of 0.05).

Baseline characteristics of the patients, grouped according to the median of NT-proBNP, are presented as percentage for dichotomous variables and means (median) and ranges for continuous variables. Baseline characteristics were compared between median with the use of X2 test for discrete variables and unpaired t tests (parametric) and Mann-Whitney U test (nonparametric) for continuous variables, as appropriate. In Cox proportional hazards model, the associations between NT-proBNP and mortality and hospitalization for any cause were examined. N-terminal pro-brain natriuretic peptide was treated both as a dichotomous and continuous variable. Survival curves were generated by means of Kaplan-Meier estimates, and differences in survival and hospitalization were compared with the use of the log-rank test. Multivariate models were fitted with the use of available clinical covariates. Of the 345 patients, 70 died within the follow-up period. Six covariates were in a univariate analysis associated with mortality and hospitalization and were therefore chosen in the main models. Covariates were chosen if they were associated with mortality and hospitalization in a univariate analysis. Rather than renal failure, eGFR was chosen because the continuous variable was associated with mortality and hospitalization, whereas the dichotomous variable was not. A backward elimination model was applied. The assumptions underlying the Cox proportional hazards model (proportional hazards, lack of interaction, and linearity of continuous variables) were tested and found valid unless otherwise indicated. N-terminal pro-brain natriuretic peptide was log transformed. All logarithms were proportional [logy(x) = log(x) / log(y)], and NT-proBNP was log transformed. To get the effect of doubling plasma concentrations of NT-proBNP on hazard ratio, log2 transformation was used. P < .05 was considered significant (2-sided). Analyses were made using Statistical Analysis Software (SAS 9.1, Cary, NC).

Results

Baseline Characteristics

Patient characteristics divided according to the median of NT-proBNP are presented in Table 1 . Patients with supramedian levels of NT-proBNP were older, had lower BMI, lower left ventricular ejection fraction, more frequently had a history of atrial fibrillation and renal failure, and were more frequently treated with loop diuretics and digoxin than patients with inframedian levels of NT-proBNP.

Seventy (20%) patients died within the follow-up period, and 201 (58%) were hospitalized (>24 hours at a medical department) for any cause.

N-terminal Pro-brain Natriuretic Peptide as a Continuous Variable

When NT-proBNP was entered in the model as a continuous variable, a curvilinear relationship was seen between estimated hazard ratios for mortality and hospitalization for any cause and log2(NT-proBNP) with log2(125) used as reference (Figure 1A and B) ("Discussion" section). In the adjusted model for NT-proBNP, each doubling of the value was associated with a hazard ratio of 1.56 (95% CI 1.32-1.85, P < .0001) for mortality and 1.19 (95% CI 1.09-1.31, P = .0002) for hospitalization for any cause ( Table 2 ).

Figure 1. 

A, Estimated hazard ratio for hospitalization according to NT-proBNP at baseline (log2 scale). The estimated hazard ratio (solid line) is shown with the 95% confidence limits (dotted lines). B, Estimated hazard ratio for mortality according to NT-proBNP at baseline (log2 scale). The estimated hazard ratio (solid line) is shown with the 95% confidence limits (dotted lines).

     

N-terminal Pro-brain Natriuretic Peptide as a Dichotomous Variable

Kaplan-Meier estimates of survival and hospitalization for all subjects according to the median of NT-proBNP are shown in Figure 2A and B. In an adjusted Cox proportional hazard model with NT-proBNP as a dichotomous variable, the adjusted hazard ratios for death and hospitalization for any cause of patients with supramedian levels compared with those with inframedian levels of NT-proBNP were 2.40 (95% CI 1.40-4.10) and 1.71 (95% CI : 1.24-2.36), respectively ( Table 2 ).

Figure 2. 

A, Kaplan-Meier plot showing freedom from hospitalization in patients with chronic heart failure according to the median of NT-proBNP at baseline (log-rank P = .0002). B, Kaplan-Meier plot showing survival in patients with heart failure according the median of NT-proBNP at baseline (log-rank P = .0001).

     

Discussion

The novel findings of this study are that NT-proBNP predicts risk for hospitalization for any cause and mortality in outpatients with heart failure followed in a dedicated heart failure clinic and that a simple algorithm indicates that a doubling in plasma concentrations is associated with an increase in hazard ratio by a factor of 1.56 (56%) for mortality and by a factor 1.19 (19%) for hospitalization for any cause. Finally, the relationship between NT-proBNP and risk for mortality and hospitalization is curvilinear if NT-proBNP is considered as a continuous variable.

Continuous Variable and Simple Algorithm

If NT-proBNP data are used for prognostication as a continuous variable, log transformation is mandatory. log2 transformation is an attractive transformation of NT-proBNP data, because the effect of an increase equal to 1 on the log scale (doubling in pg/mL) is easily transformed back to picogram per milliliter, which is used in clinical practice. Adjusted hazard ratios for log2NT-proBNP was 1.56 (56%, 95% CI 1.32-1.85) for mortality and 1.19 (19%, 95% CI 1.09-1.31) for hospitalization ( Table 2 ). Furthermore, patients with NT-proBNP levels of <125 pg/mL had low risk for mortality and hospitalization (1 of 19 died, and 3 of 19 were hospitalized), and 125 pg/mL is a proposed cutoff limit for left ventricular systolic dysfunction in the primary care.[16] Therefore, 125 pg/mL is an appropriate reference for risk calculation in patients with chronic heart failure and left ventricular systolic dysfunction. Thus, if NT-proBNP is considered as a continuous variable and 125 pg/mL is used as reference, a simple algorithm would be that estimated hazard ratios are increased by a factor of 1.56 (adjusted hazard ratio) for mortality and by a factor of 1.19 for hospitalization every time NT-proBNP is doubled in relation to 125 pg/mL: hazard ratioEstimated = hazard ratioAdjusted[log2(x) - log2(125)] ⇔ hazard ratioEstimated = hazard ratioAdjustedlog2(x/ 125). The interindividual relationship between estimated hazard ratio and NT-proBNP in the observed range (22-35000 pg/mL) is curvilinear (Figure 2A and B). This is parallel to the interindividual relationship between eGFR and estimated hazard ratios reported from the VALIANT study.[17] It should be noted that if a reference of >125 pg/mL was chosen or if the range had been smaller, a linear trend would have been observed; the curvilinear relationship would have been overlooked. It may be argued that 121 pg/mL (95% confidence limit in the control group in the Val-HeFT study[3]) or 109 pg/mL (lower limit of the upper tertile in the study by McKie et al[18] in patients without heart failure in the community) is a more appropriate reference. However, if these values are used, reference doubling will be complicated in clinical practice; hence, we chose 125 pg/mL. In the Val-HeFT database,[3] patients with NT-proBNP of <173 pg/mL were chosen as reference, and patients were divided into deciles. There was also a curvilinear trend between adjusted hazard ratios and the deciles in the Val-HeFT study.[3] In clinical practice, the ranges of the deciles described from the Val-HeFT cohort[3] may, however, be difficult to apply, whereas the effect of doubling NT-proBNP in relation to 125 pg/mL (250, 500, 1000, 2000, 4000, 8000, etc) may be more clinically applicable.

It should be noted that the curvilinear relationship between NT-proBNP and estimated hazard ratios is the interindividual relationship at a certain point in time. The intraindividual relationship may differ, and our calculated algorithm should therefore not be used for monitoring purposes. However, the algorithm may be useful when NT-proBNP is analyzed to identify the risk profile of patients with chronic heart failure at a certain point in time.

Dichotomous Variable

In our cohort, patients with supramedian levels of NT-proBNP had a 2.40-fold (95% CI 1.40-4.10) increased risk for mortality and a 1.71-fold (95% CI 1.24-2.36) increased risk for hospitalization for any cause ( Table 2 ) (Figure 1A and B). Our adjusted hazard ratio for mortality is consistent with the adjusted hazard ratios for mortality identified in the Val-HeFT study[3] (adjusted hazard ratio 2.07, 95% CI 1.76-2.46) and in the COPERNICUS study[6] (adjusted hazard ratio 2.17, 95% CI 1.33-3.54), whereas the adjusted hazard ratio for hospitalization for any cause in our study seems a little lower than the adjusted hazard ratio for hospitalization for heart failure in the Val-HeFT study[3] (adjusted hazard ratio 2.66, 95% CI 2.19-3.22) and the adjusted hazard ratio for the composite end point in the COPERNICUS study[6] (adjusted hazard ratio 2.11, 95% CI 1.54-2.90). Therefore, NT-proBNP can predict risk for hospitalization for any cause but not with the same power as risk for hospitalization for heart failure. This, of course, is an indication of the cardiac specificity of the peptide. We observed a median of NT-proBNP of 1381 pg/mL at baseline in an outpatient heart failure clinic. Gardner et al[5] observed a median of 1490 pg/mL in patients with chronic heart failure on a transplantation list, and in the COPERNICUS[6] and Val-HeFT[3] studies the median were 1767 pg/mL and 895 pg/mL, respectively, in large study populations. In all of the mentioned cohorts, patients with NT-proBNP levels more than the median had increased risk for mortality. However, the low areas of the receiver operating characteristic curves in the Val-HeFT study[3] demonstrated that NT-proBNP should be considered as a continuous variable rather than as a dichotomous variable concerning risk stratification in patients with chronic heart failure; hence, we created the simple algorithm based on NT-proBNP as a continuous variable.

If the median of NT-proBNP identified in different cohorts is used to separate the patients into 2 groups, patients with NT-proBNP levels greater than the median may have an increased frequency of atrial fibrillation, renal failure, a trend toward an increased frequency of ischemic cardiomyopathy, higher age, lower left ventricular ejection fraction, and lower BMIs ( Table 1 ). These risk markers have all in larger cohorts been associated with mortality.[19-24]

Limitations

In the heart failure clinic, up-titration and withdrawal of angiotensin-converting enzyme inhibitors, angiotensin II blockers, ß-blockers, and spironolactone are done over time; and changes in dosages of medication may in theory affect our results. The number of patients in our single-center database is limited, the range of the CIs on our results should be noted, and the parameter estimates have to be interpreted with some caution. However, our data and parameter estimates are consistent with the results from the large Val-HeFT database[3] (N = 3916).

Perspectives

At present it is recommended that all patients with chronic heart failure are offered follow-up in a multidisciplinary program in a heart failure clinic,[1] but the optimal intensity and duration of such programs are not known.[25] In particular, it is unknown whether some patients benefit from indefinite clinic follow-up or other measures such as access to telephone service. Our data suggest that NT-proBNP measurement can be used to identify high-risk patients, who may benefit from more intensive follow-up or indefinite enrollment in a multidisciplinary heart failure clinic, but this hypothesis must be tested in a clinical trial. Furthermore, based on these data, we have calculated a simple algorithm that quantitatively describes the interindividual relationship between NT-proBNP as a continuous variable and estimated hazard ratio in a clinical useful way. This study is therefore novel to provide a quantitative understanding of the interindividual relationship between NT-proBNP as a continuous variable in the whole of its range expressed in picogram per milliliter and estimated hazard ratios.

Conclusion

N-terminal pro-brain natriuretic peptide predicts risk for hospitalization for any cause and mortality in an outpatient heart failure clinic. A simple algorithm that uses NT-proBNP as a continuous variable indicates that every time plasma concentrations are doubled relative to 125 pg/mL (250, 500, 1000, 2000, etc), the estimated hazard ratio increases by a factor of 1.56 (56%) for mortality and by a factor of 1.19 (19%) for hospitalization for any cause, resulting in curvilinear interindividual relationships in the whole range of NT-proBNP.


Table 1. Baseline Characteristics According to NT-proBNP Median (N = 345)


  <1381 pg/mL >1381 pg/mL P
n = 172 n = 173
Demographic and clinical variables
Age (y) 69 [35-92] 75 [47-91] <.0001
Males 74 65 .0855
Females 26 35 .0855
Height (cm) 174 [149-192] 172 [147-192] .0221
Weight (kg) 82 [42-138] 74 [39-120] <.0001
BMI (kg/m2) 27 [16-43] 25 [15-37] <.0001
Duration of CHF (m) 6 [0-180] 6 [0-168] .8765
Diabetes 20 28 .1080
Renal failure (eGFR <60 mL/[min • m2]* 25 44 .0002
ECG   <.0001
Sinus rhythm 85 53 
Atrial fibrillation 11 40 
Other 4 7 
Cardiovascular and renal variables
LVEF 33 [13-45] 30 [10-45] .0006
Heart rate (beat/min) 72 [50-108] 72 [41-112] .8998
Systolic arterial pressure (mm Hg) 135 [85-212] 130 [90-200] .1435
eGFR (mL/[min•1.73 m2]) 70 [34-130] 62 [22-115] .0005
NYHA class (%)   .1686
1 16 10 
2 61 66 
3 22 22 
4 1 2 
Etiology (%)   .0676
Ischemic cardiomyopathy 51 60 
Idiopathic dilatated cardiomyopathy 14 12 
Hypertensive cardiomyopathy 10 7 
Alcoholic cardiomyopathy 8 5 
Valvular cardiomyopathy 3 9 
Other cardiomyopathy or unknown 14 7 
Laboratory tests
Sodium (mmol/L) 138 [127-148] 138 [118-146] .3274
Potassium (mmol/L) 4.0 [3.2-6.0] 4.0 [2.7-5.2] .9723
NT-proBNP (pg/mL) 542 [22-1377] 3071 [1381-35000] <.0001
Medication (%)
ACE inhibitor/AII antagonist 80 75 .2564
ß-Blockade 50 50 .8716
Spironolactone 17 15 .4834
Digoxin 12 25 .0010
Loop diuretics 60 70 <.0001
Potassium supplement 30 47 .0007

ECG, Electrocardiogram; SR, sinus rhythm; AF, atrial fibrillation; CHF, chronic heart failure; LVEF, left ventricular ejection fraction; ACE, angiotensin-converting enzyme; AII, angiotensin II.
*Definition according to The National Kidney Foundation[17]: glomerular filtration rate of <60 mL/(min•1.73 m2) for >3 months.

 

Table 2. Adjusted Cox Proportional Hazard Models (Mortality and Hospitalization)


  Adjusted hazard ratio 95% CI P
Model 1: NT-proBNP as continuous variable (mortality)
log2(NT-proBNP) 1.56 1.32-1.85 <.0001
LVEF (1% point change) 1.04 1.01-1.07 .0062
Model 2: NT-proBNP as a dichotomous variable (mortality)
NT-proBNP (>1381 pg/mL) 2.40 1.41-4.10 .0013
NYHA class 1.60 1.04-2.45 .0318
Age (1-y increase) 1.04 1.01-1.07 .0064
Model 3: NT-proBNP as continuous variable (hospitalization)
log2(NT-proBNP) 1.19 1.09-1.31 .0002
Model 4: NT-proBNP as dichotomous variable (hospitalization)
NT-proBNP (>1381 pg/mL) 1.71 1.24-2.36 .0011

Six variables were included in the main models: NT-proBNP, eGFR, age, BMI, NYHA class, and LVEF. Eliminated variables: model 1 (step): (1) BMI, (2) eGFR, (3) age, (4) NYHA class; model 2 (step): (1) BMI, (2) eGFR, (3) LVEF; model 3 (step): (1) LVEF, (2) eGFR, (3) age, (4) NYHA class, (5) BMI; model 4 (step): (1) LVEF, (2) eGFR, (3) age, (4) BMI, (5) NYHA class.

 



References

  1. Swedberg K, Cleland J, Dargie H, et al. Guidelines for the diagnosis and treatment of chronic heart failure: executive summary (update 2005): The Task Force for the Diagnosis and Treatment of Chronic Heart Failure of the European Society of Cardiology. Eur Heart J. 2005;26:1115-1140.
  2. Doust JA, Pietrzak E, Dobson A, et al. How well does B-type natriuretic peptide predict death and cardiac events in patients with heart failure: systematic review. BMJ. 2005;330:625.
  3. Masson S, Latini R, Anand IS, et al. Direct comparison of B-type natriuretic peptide (BNP) and amino-terminal proBNP in a large population of patients with chronic and symptomatic heart failure: the Valsartan Heart Failure(Val-HeFT) data. Clin Chem. 2006;52:1528-1538.
  4. Richards AM, Doughty R, Nicholls MG, et al. Plasma N-terminal pro-brain natriuretic peptide and adrenomedullin: prognostic utility and prediction of benefit from carvedilol in chronic ischemic left ventricular dysfunction. Australia-New Zealand Heart Failure Group. J Am Coll Cardiol. 2001;37:1781-1787.
  5. Gardner RS, Ozalp F, Murday RJ, et al. N-terminal pro-brain natriuretic peptide. A new gold standard in predicting mortality in patients with advanced heart failure. Eur Heart J. 2003;24:1735-1743.
  6. Hartmann F, Packer M, Coats AJ, et al. Prognostic impact of plasma N-terminal pro-brain natriuretic peptide in severe chronic congestive heart failure: a substudy of the Carvedilol Prospective Randomized Cumulative Survival (COPERNICUS) trial. Circulation. 2004;110:1780-1786.
  7. Richards AM, Nicholls MG, Espiner EA, et al. B-type natriuretic peptides and ejection fraction for prognosis after myocardial infarction. Circulation. 2003;107:2786-2792.
  8. Gustafsson F, Ulriksen H, Villadsen H, et al. Prevalence and characteristics of heart failure clinics in Denmark -- design of the Danish heart failure clinics network. Eur J Heart Fail. 2005;7:283-284.
  9. Stromberg A, Martensson J, Fridlund B, et al. Nurse-led heart failure clinics in Sweden. Eur J Heart Fail. 2001;3:139-144.
  10. Jaarsma T, Tan B, Bos RJ, et al. Heart failure clinics in the Netherlands in 2003. Eur J Cardiovasc Nurs. 2004;3:271-274.
  11. Galatius S, Gustafsson F, Nielsen PH, et al. An integrated approach to diagnosis and therapeutic management of patients with systolic heart failure in the Copenhagen metropolitan area. Am Heart J. 2002;144:E2.
  12. Junge W, Wilke B, Halabi A, et al. Determination of reference intervals for serum creatinine, creatinine excretion and creatinine clearance with an enzymatic and a modified Jaffe method. Clin Chim Acta. 2004;344:137-148.
  13. Sokoll LJ, Baum H, Collinson PO, et al. Multicenter analytical performance evaluation of the Elecsys proBNP assay. Clin Chem Lab Med. 2004;42:965-972.
  14. Olsen MH, Hansen TW, Christensen MK, et al. N-terminal pro brain natriuretic peptide is inversely related to metabolic cardiovascular risk factors and the metabolic syndrome. Hypertension. 2005;46:660-666.
  15. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461-470.
  16. Gustafsson F, Steensgaard-Hansen F, Badskjaer J, et al. Diagnostic and prognostic performance of N-terminal proBNP in primary care patients with suspected heart failure. J Card Fail. 2005;11:S15-S20.
  17. Anavekar NS, McMurray JJ, Velazquez EJ, et al. Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction. N Engl J Med. 2004;351:1285-1295.
  18. McKie PM, Rodeheffer RJ, Cataliotti A, et al. Amino-terminal pro-B-type natriuretic peptide and B-type natriuretic peptide: biomarkers for mortality in a large community-based cohort free of heart failure. Hypertension. 2006;47:874-880.
  19. Corell P, Gustafsson F, Markenvard J, et al. Prevalence and prognostic significance of atrial fibrillation in outpatients with systolic heart failure. Eur J Heart Fail. 2007;9:258-265.
  20. Felker GM, Shaw LK, O'Connor CM. A standardized definition of ischemic cardiomyopathy for use in clinical research. J Am Coll Cardiol. 2002;39:210-218.
  21. Gustafsson F, Torp-Pedersen C, Seibaek M, et al. Effect of age on short and long-term mortality in patients admitted to hospital with congestive heart failure. Eur Heart J. 2004;25:1711-1717.
  22. Gustafsson F, Torp-Pedersen C, Brendorp B, et al. Long-term survival in patients hospitalized with congestive heart failure: relation to preserved and reduced left ventricular systolic function. Eur Heart J. 2003;24:863-870.
  23. Gustafsson F, Kragelund CB, Torp-Pedersen C, et al. Effect of obesity and being overweight on long-term mortality in congestive heart failure: influence of left ventricular systolic function. Eur Heart J. 2005;26:58-64.
  24. Gustafsson I, Brendorp B, Seibaek M, et al. Influence of diabetes and diabetes-gender interaction on the risk of death in patients hospitalized with congestive heart failure. J Am Coll Cardiol. 2004;43:771-777.
  25. Jaarsma T, Van Der Wal MH, Hogenhuis J, et al. Design and methodology of the COACH study: a multicenter randomised coordinating study evaluating outcomes of advising and counselling in heart failure. Eur J Heart Fail. 2004;6:227-233.
Acknowledgements

The excellent assistance in the heart failure clinic by the nurses Per H Nielsen, Hanne Bartholdy, Anna Marie Jensen, Birgitte Carlsen, and Louise Fly Jensen is acknowledged.

Reprint Address

Morten Schou, MD, Department of Cardiology and Endocrinology, Clinic E Frederiksberg University Hospital Ndr. Fasanvej 57-59 DK-2000 Frederiksberg, Denmark.


Morten Schou, MD,a Finn Gustafsson, MD, PhD,b Pernille Corell, MD,a Caroline N. Kistorp, MD, PhD,a Andreas Kjaer, MD, PhD, DrMSci,c,d Per R. Hildebrandt, MD, DrMScie

aDepartment of Cardiology and Endocrinology, Clinic E, Frederiksberg University Hospital, Frederiksberg, Denmark
bDepartment of Cardiology, The Heart Centre, Rigshospitalet University Hospital, Copenhagen, Denmark
cDepartment of Clinical Physiology and Nuclear Medicine, The PET Centre, Rigshospitalet University Hospital, Copenhagen, Denmark
dCluster for Molecular Imaging, University of Copenhagen, Copenhagen, Denmark
eDepartment of Cardiology, Roskilde University Hospital, Roskilde, Denmark

Dr Hildebrandt has received honoraria for lectures on N-terminal pro-brain natriuretic peptide from Roche Diagnostics, Basel, Switzerland. Dr Schou is supported by research grant 200207135A-321 from the Copenhagen Hospital Corporation, Copenhagen, Denmark.