European Heart Journal Advance Access published online on July 23, 2008
European Heart Journal, doi:10.1093/eurheartj/ehn334
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
NT-proBNP has a high negative predictive value to rule-out short-term cardiovascular events in patients with diabetes mellitus
1 Department of Cardiology, Medical University Vienna, Vienna, Austria
2 Research Institute for Health Care Management and Economics, University of Economics and Business Administration, Vienna, Austria
3 Third Department of Medicine, Hietzing Hospital, Vienna, Austria
4 Department of Endocrinology, Medical University, Vienna, Austria
Received 24 January 2008; revised 24 June 2008; accepted 27 June 2008.
* Corresponding author. Tel: +43 1 40400 4614, Fax: +43 1 40400 6278, Email: richard.pacher{at}meduniwien.ac.at
 |
Abstract |
|---|
 Top Abstract IntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Aims: This study evaluated the predictive value of NT-proBNP for patients with diabetes mellitus and compared the prognostic aptitude of this neurohumoral marker to traditional markers of cardiovascular events.
Methods and results: A prospective observational study was conducted in 631 diabetic patients. The composite endpoint consisted of unplanned hospitalization for cardiovascular events or death within the observation period of 12 months. Of all variables analysed (age, gender, history of hypertension, ischaemic heart disease/any cardiac disease, smoking, duration of diabetes, body mass index, blood pressure, New York Heart Association-class, Dyspnoea score, Minnesota Living with Heart Failure Questionnaire, LDL-cholesterol, HbA1c, creatinine, glomerular filtration rate), the logarithm of NT-proBNP gave the most potent information in a stepwise Cox regression analysis (P < 0.0001). Bootstrapping with 500 samples supports this result in 95% samples. The negative predictive value of a normal value (<125 pg/mL) of NT-proBNP for short-term cardiovascular events in diabetic patients is 98%.
Conclusion: We have demonstrated a strong and independent correlation between NT-proBNP and short-term prognosis of cardiovascular events for patients with diabetes mellitus. With a high negative predictive value it can identify individuals who are not at intermediate risk for cardiovascular events. NT-proBNP proved to be of higher predictive value than traditional cardiovascular markers, in this unselected cohort.
Key Words: Natriuretic peptides • Diabetes mellitus • Cardiovascular risk • Prognosis
|
Introduction |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Although is widely recognized that the absolute risk of cardiovascular events varies among individuals with diabetes mellitus, there is a lack of reliable short-term predictors to guide timely and individualized management.
The increased risk of cardiovascular disease in patients with diabetes mellitus is well established since the robust association has been shown by the Framingham heart study,1 a correlation that has been confirmed in several subsequent trials.2,3 Many variables, such as HbA1c, blood pressure, several lipid parameters, and markers of kidney function have been put forward to serve as markers for risk stratification to identify individual patients with excessive risk in a timely manner.4–7 To some extent, all variables are of limited value, as they are good predictors for long-term prognosis; but there is a lack of data on risk estimation for the clinically more relevant short-term prognosis. The need for an individualized and timely risk assessment was also highlighted in a recent statement from the American Heart Association and the American Diabetes Association.8
Natriuretic peptides, such as brain natriuretic peptide (BNP), atrial natriuretic peptide (ANP), or their inactive N-terminal precursors NT-proBNP and NT-proANP are released from myocardial cells in response to volume expansion and increased wall tension.9 They are well-established rule-out tools for cardiac disease in unselected communities.10–12 Furthermore, their value for risk stratification in populations with known heart disease has been proven for long-term as well as for short-term outcome.13
Data about the prognostic value of NT-proBNP in diabetic patients are limited.14–18 Again, data for short-term prognosis, which are clinically more relevant, are lacking entirely in this distinct population, as observation periods studied so far lasted from 2 to 15 years. Moreover, there is no direct comparison with the known markers mentioned above.
Early identification of cardiovascular risk is of vital importance in a comprehensive diabetes management, since it allows early, targeted interventions. Although never investigated, it is of great clinical significance for a timely and individualized management to recognize those individuals among the very large population of asymptomatic diabetic patients who might have an increased risk of cardiovascular events in their imminent future, and at the same time identify those individuals who are not at risk for the time being.
We hypothesized that NT-proBNP is superior at providing prognostic information about short-term cardiovascular risk for patients with diabetes mellitus compared with the above-mentioned traditional predictors of cardiovascular events. We proved our hypothesis in an unselected cohort of diabetic patients.
|
Research design and methods |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The study population consisted of diabetic patients treated at the diabetes outpatient clinic of the Vienna General Hospital. All patients attending this tertiary care centre between January 1, 2006 and February 17, 2007 were invited to participate in our prospective observational study. The only exclusion criterion was refusal to participate. The refusal rate was <10%. A total of 631 consecutive patients were recruited. A certified nurse took a compiled medical history for each patient, with a special focus on cardiovascular disease, in order to obtain information about concomitant diseases and current treatment. All patients were asked to complete the Minnesota Living with Heart Failure Questionnaire (MLHFQ) and the Dyspnoea score chart. An electrocardiogram was recorded and later analysed for the presence of atrial fibrillation, bundle branch block, or any cardiovascular disease by a cardiologist. Blood pressure was measured in every patient. Blood was drawn from an antecubital vein. Kidney function was determined by measurement of serum-creatinine and glomerular filtration rate (GFR) was calculated by the Cockroft–Gault formula. To obtain risk markers for cardiovascular disease, cholesterol (especially LDL) from fasting samples was measured and NTpro-BNP was determined by a commercially available kit (Roche Diagnostics). Moreover, HbA1c, as a marker of glucose metabolism was determined.
Endpoints
Based on the short observation period, a composite endpoint consisting of unplanned hospitalization for cardiovascular disease or death was chosen as the primary endpoint in this study. Secondary endpoints were death, unplanned cardiovascular hospitalization, hospitalization due to heart failure, and all-cause hospitalization. All patients were traced through the national registry during 2007. All patients were monitored for a fixed time period of 12 months for outcome. Mortality data were obtained from the Austrian Central Office of Civil Registration (Zentrales Melderegister). If a patient had died before February 17, 2008 the date of death was recorded. Hospital reports about hospitalization were obtained from the regional hospital data network (Krankenanstaltenverbund). Information about hospitalizations for cardiovascular disease was obtained from hospital files by a cardiologist, unaware of the results at index time. As this was a one-point analysis, no patient was lost of follow-up.
The study was conducted in accordance with the Helsinki II declaration and was approved by the ethics committee of our institution. All participants gave a written informed consent.
Statistical analysis
Continuous variables are expressed as means ± standard deviation; categorical variables are presented as frequencies and percentages.
Sample size calculation was based on an expected log hazard ratio of 0.5 and an expected event rate of 10%. For alpha = 0.05 and power >0.9 sample size of 600 patients was obtained.
Two stepwise Cox regression models were calculated to identify independent variables in order to predict the combined endpoint of unplanned hospitalization for cardiovascular events and death over time. P-value for entering the stepwise model was set at 0.05 and 0.10 for exclusion. A stepwise approach was used to determine the most potent single predictor independent of the number of events out of a large number of variables. All results of the regression model are presented using hazard ratios. Hazard ratios are given for increase per unit. Five hundred bootstrap repetitions are done for both Cox regression models, repeating the variable selection for each sample using the same entering and exclusion rules. It was counted how often a variable was entered into the Cox regression models. The overall predictive accuracy (D) of the Cox regression models was calculated for each model by correlating the prognostic index of each patient with the observed survival time. The prognostic index is defined as the linear combination of regression coefficients and the values of the covariables. The accuracy of a bootstrap model is calculated for the prediction within the actual bootstrap sample (Dboot) and for its usefulness for prediction using the whole original data set (Dorig). The difference between the two predictive accuracies is called optimism in the fit from the bootstrap sample; and it is averaged over the 500 samples. The described procedure was done for two different sets of variables. One model uses all 17 variables available from our study. One smaller model (8 variables) is based on all variables with a significant predictive influence in the single variable Cox regression models. Proportional hazards assumption was assessed and satisfied for all variables based on partial residual plots and on time interaction tests. For all continuous variables squared values and the logarithm were calculated. Only for NT-proBNP and serum-creatinine the logarithm outperforms the untreated data within single variable Cox regression models. Therefore, the logarithms of NT-proBNP and serum-creatinine were added as two additional variables into both variable sets.
Variables as follows were included in the 17 variables model: age (years), gender (1/0), body mass index (kg/m2), history of any heart disease (0/1), ischaemic heart disease (IHD) (0/1), hypertension (0/1), history of smoking (0/1), systolic blood pressure (mmHg), HbA1c (%), LDL-cholesterol (mg/dL), serum-creatinine (mg/dL), GFR (mL/min), NT-proBNP (pg/mL), New York Heart Association (NYHA)-class (1–4), MLHFQ (0–100), Self-assessment Dyspnoea score (1–10), and duration of diabetes (years).
Variables in the reduced model: age (years), history of any heart disease (0/1), IHD (0/1), serum-creatinine (mg/dL), GFR (mL/min), NT-proBNP (pg/mL), NYHA-class (1–4), and MLHFQ (0–100).
Survival was calculated using the Kaplan–Meier method, where patients were divided into those with NT-proBNP values above or below the cut-point of 125 pg/mL.
Receiver operating characteristic (ROC) analysis was calculated to assess the predictive power of NT-proBNP.
A P < 0.05 was considered significant in all analysis. SPSS 15.0 software (SPSS, Chicago, IL) and GChaos 13.2 statistical software written in C++ by one of the authors were used for all statistical analysis.
|
Results |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
Description of the total study population
Demographics and outcome
The total study population comprised 631 patients with diabetes mellitus, which were included consecutively into the study. Characteristics of this population correspond to a usual collective of unscreened diabetic patients, mean age being 58 ± 14 years, mean duration of diabetes being 9 ± 10 years (Table 1).
|
A total of 42% of patients were treated with insulin, 60% with oral antidiabetics, 15% of patients were receiving both drugs. The mean value of HbA1c was 8.0 ± 1.6%. 23% of patients had a history of cardiovascular disease. Mean blood pressure values were 143 ± 22 mmHg systolic and 85 ± 13 diastolic. Twenty-five per cent of patients were treated with a beta-blocker, 54% with a RAAS antagonist, 20% with a calcium channel blocker. Mean LDL-cholesterol values were 112 ± 38 mg/dL, 34% of patients were treated with statins. The gender ratio (male/female) was 55/45%. Only <5% of patients were defined as type I diabetes patients.
Of the entire collective, 44 (7.0%) patients reached the composite endpoint (39 unplanned hospitalizations for cardiovascular disease and 7 deaths—2 deaths occurring after an unplanned CV hospitalization) during the observation period of 12 months.
The reasons for hospitalization were classified as follows: chronic heart failure 13; coronary artery disease 8; atrial fibrillation 4, carotid artery disease 6; peripheral artery occlusive disease 8.
Stepwise Cox regression model
Both stepwise Cox regression models demonstrate that the logarithm of NT-proBNP gives the most potent information to predict future events (Table 2). Both models are finally based on the logarithm of NT-proBNP (best Wald statistic) and on the MLHFQ (second best Wald statistic). The model initially started with 17 variables plus the logarithms of serum-creatinine and NT-proBNP and adds at last the duration of diabetes. The small variable set includes the age as third best predictor. Bootstrap testing supports the importance and robustness of the logarithm of the NT-proBNP (Table 3). For the large variable set 92% of the bootstrap samples include the logarithm of the NT-proBNP. In the small variable set it is included in >95% of the 500 models.
|
|
Optimism for the model based on the large variable set is 0.05, resulting in a corrected predicted accuracy of D = 0.31. Optimism for the model from the small data set is 0.02 also resulting in a corrected predicted accuracy of D = 0.31.
Kaplan–Meyer analysis
The Kaplan–Meyer analysis for NT-proBNP showed differences between patients with values above and below the cut-points of 125 pg/mL. The difference was statistically significant (P < 0.0001), through the observation period (Figure 1).
|
Receiver operating characteristic curve
The area under the ROC curve with respect to the combined endpoint unplanned cardiovascular hospitalization and death was 0.785 for NT-proBNP in our study population. Sensitivity, specificity, negative predictive value, positive predictive value, and accuracy for different values of NT-proBNP are depicted in Figure 2.
|
|
Discussion |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
In the present study, we have demonstrated a strong and independent correlation between plasma NT-proBNP levels and short-term prognosis of cardiovascular events for patients with diabetes mellitus. Patients with low levels of NT-proBNP (<125 pg/mL) had an excellent short-time prognosis. This was true despite the fact that they had less background therapy (data not shown). At the same time, NT-proBNP provided more concise information about cardiovascular risk in this diabetic population compared with traditional markers.
Traditional predictors of increased risk
Many variables such as HbA1c, glucose management, blood pressure, LDL-cholesterol, and kidney function have been evaluated for their potential to predict outcome in diabetic patients and to identify those that need more aggressive management. All of these markers have proved valuable for long-term prognosis, but not for the assessment of imminent threat of major cardiovascular events for diabetic patients. The American Heart Association and the American Diabetes Association recently extensively discussed and questioned the individual predictive role of blood pressure, lipids, or glucose management on outcome.8
The reason for the limited aptitude of these traditional markers to predict short-term events might be that they are not functional markers of cardiovascular health but mediators of cardiovascular injury. There is a well-studied dose–effect relationship over time for these modifiable risk factors for large cohorts, and targeted, multifactorial interventions should be undertaken to reduce cardiovascular long-term risk as outlined, for example, in the Steno-2 Study.19 Notwithstanding this important role, the information about the current, immediate risk for the individual patient is uncertain.
An additional explanation for the shortcomings of cardiovascular markers to predict outcome in this setting might be attributed to the ‘heart failure paradox’. Several studies have linked obesity, hypercholesterolaemia, higher blood pressure, and even higher HbA1c levels, which are well-known risk factors for coronary artery disease, to improved survival of patients with heart failure.20–24 The mechanisms of this ‘reverse epidemiology’ are not quite clear.25 However, traditional risk markers of cardiovascular disease might be understood as risk modifiers in advanced heart failure, which is the endgame of cardiac disease and a metabolically taxing condition. Given that heart failure patients make up a relevant part of any population of diabetic patients,26 clinical trials trying to assess the prognostic accuracy of these markers should get mixed results.
Natriuretic peptides as prognostic markers
The use of natriuretic peptides such as ANP and BNP and their precursors NT-proBNP and NT-proANP for the assessment of cardiovascular risk is firmly established in cardiovascular guidelines.27 Increasing plasma levels of natriuretic hormones are associated with the development of cardiac arrhythmias and increasing haemodynamic instability.28 Higher NT-proBNP levels have been demonstrated to be associated with advanced patient age, renal impairment, cardiac arrhythmias, and systolic and diastolic dysfunction.29 Natriuretic peptide may therefore reflect an integral of risk factors resulting in the current functional cardiovascular status of individual patients.
NT-proBNP and diabetes mellitus
Data about the prognostic accuracy of natriuretic peptides for patients with diabetes mellitus are limited. Several studies have examined the relationship between BNP/NT-proBNP and long-term cardiovascular outcome. Christoffersen et al.18 outlines some of the challenges of interpreting natriuretic peptides in diabetic patients. In Tarnows et al.15 study only relatively young (<66 years) type 2 diabetic patients were included. In this observational study, increased NT-proBNP levels were found to be predictive of overall and cardiovascular mortality for a follow-up period of 15 years. An important observation in this population was that the predictive accuracy of NT-proBNP is independent of GFR, and the level of albumin excretion rate. In Gaedes et al. study,16 microalbuminuretic type 2 diabetic patients were enrolled. A NT-proBNP level above the median was associated with an increased risk of cardiovascular disease during a follow-up of 
7.8 years. In Bhallas et al. study,17 some of the predominantly male patients were referred on the basis of clinical suspicion of cardiac dysfunction, and the rest recruited from a diabetic clinic. Follow-up to show an association between increased BNP and cardiac and all-cause mortality was 2.3 years.
The results of our total study population, which are, unlike the preceding studies, an unselected collective of consecutive patients attending a tertiary care centre, confirm previous observations. It adds new information about NT-proBNP predictive potential, in the clinically more relevant, short-term design. Bootstrapping clearly demonstrate the robustness of this finding. NT-proBNP is the only variable, which provides significant independent information irrespective of the chosen population. All other variables such as history of any heart disease, kidney function, LDL-cholesterol, or HbA1c strongly depend on the predefined study population. Although results of ROC curves have to be interpreted with caution,30 excellent negative predictive values or accuracies can be achieved, dependent on the chosen cut-point.
Implications of risk stratification with NT-proBNP
In the present study, we have identified a subset of diabetic patients who appear to be at a dramatically increased risk of short-term cardiovascular events. This group, which can be readily recognized by increased NT-proBNP levels, might need a more extensive work up. This rapid determination of increased risk would allow targeted interventions and more aggressive management to prevent hospitalization or death.
Limitations: our study is based on a data set with more than 600 patients, which seems to be sufficiently big enough for robust statistics. As the observation time is limited to 1 year, there is a certain risk of over-fitting the stepwise regression models. We tried to avoid this risk by using bootstrapping methods. Additionally, we limited the number of predictive variables in our second model. The role of NT-proBNP seems to be robust. Additional mechanisms could be relevant but cannot be identified based on the limited observation time.
As already mentioned above, interpretations of results from ROC curves have to take some limitations into account. One of the most important is the rank order statistic on which ROC curves are based, which means that the amount of differences between the risks of patients is not reflected in the area under the curve. Notwithstanding these limitations, the ROC curve is the most widely used statistical tool in cardiovascular literature.30
|
Conclusions |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
NT-proBNP, a neurohumoral prognostic marker could be part of a comprehensive, timely and individualized cardiovascular risk assessment of patients with diabetes mellitus. With a high negative predictive value it can safely identify those individuals who are not at intermediate risk for cardiovascular events.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
The authors wish to thank three anonymous reviewers for helpful comments and suggestions.
Conflict of interest: none declared.
|
Footnotes |
|---|
These authors contributed equally to this work 
|
References |
|---|
|
TopAbstractIntroductionResearch design and methodsResultsDiscussionConclusionsAcknowledgementsReferences |
|---|
- Kannel WB, McGee DL. Diabetes and cardiovascular risk factors: the Framingham study. Circulation (1979) 59:8–13.
[Abstract/Free Full Text] - Stamler J, Vaccaro O, Neaton JD, Wentworth D. Diabetes, other risk factors, and 12-yr cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care (1993) 16:434–444.[Abstract]
- Almdal T, Scharling H, Jensen JS, Vestergaard H. The independent effect of type 2 diabetes mellitus on ischemic heart disease, stroke, and death: a population-based study of 13,000 men and women with 20 years of follow-up. Arch Intern Med (2004) 164:1422–1426.
[Abstract/Free Full Text] - Khaw KT, Wareham N, Bingham S, Luben R, Welch A, Day N. Association of Hemoglobin A1c with cardiovascular disease and mortality in adults: The European prospective investigation into cancer in Norfolk. Ann Intern Med (2004) 141:413–420.
[Abstract/Free Full Text] - Turner RC, Millns H, Neil HAW, Stratton IM, Manley SE, Matthews DR, Holman RR. Risk factors for coronary artery disease in non-insulin dependent diabetes mellitus: United Kingdom Prospective Diabetes Study (UKPDS: 23). BMJ (1998) 316:823–828.
[Abstract/Free Full Text] - Knobler H, Zornitzki T, Vered S, Oettinger M, Levy R, Caspi A, Faraggi D, Livschitz S. Reduced glomerular filtration rate in asymptomatic diabetic patients. J Am Coll Cardiol (2004) 44:2142–2148.
[Abstract/Free Full Text] - Pambianco G, Costacou T, Orchard TJ. The prediction of major outcomes of type 1 diabetes: a 12-year prospective evaluation of three separate definitions of the metabolic syndrome and their components and estimated glucose disposal rate: the Pittsburgh Epidemiology of Diabetes Complications Study experience. Diabetes Care (2007) 30:1248–1254.
[Abstract/Free Full Text] - Buse JB, Ginsberg HN, Bakris GL. Primary prevention of cardiovascular diseases in people with diabetes mellitus: a scientific statement from the American Heart Association and the American Diabetes Association. Diabetes Care (2007) 30:162–172.
[Abstract/Free Full Text] - Levin ER, Gardner DG, Samson WK. Natriuretic peptides. N Engl J Med (1998) 30:321–328.
- Mc Donagh TA, Robb SD, Murdoch DR, Morton JJ, Ford I, Morrison CE, Tunstall-Pedoe H, McMurray JJV, Dargie HJ. Biochemical detection of left-ventricular systolic dysfunction. Lancet (1998) 351:9–13.[CrossRef][Web of Science][Medline]
- Nakamura M, Endo H, Nasu M, Arakawa N, Segawa T, Hiramori K. Value of plasma B type natriuretic peptide measurement for heart disease screening in a Japanese population. Heart (2002) 87:131–135.
[Abstract/Free Full Text] - Wang TJ, Larson MG, Levy D, Benjamin EJ, Leip EP, Omland T, Wolf PA, Ramachandran SV. Plasma natriuretic peptide levels and risk of cardiovascular events and death. N Engl J Med (2004) 350:655–663.
[Abstract/Free Full Text] - Koglin J, Pehlivanli S, Schwaiblmair M, Vogeser M, Cremer P, von Scheidt W. Role of brain natriuretic peptide in risk stratification of patients with congestive heart failure. J Am Coll Cardiol (2001) 38:1934–1941.
[Abstract/Free Full Text] - Dawson A, Jeyaseelan S, Morris AD, Struthers AD. B-type natriuretic peptide as an alternative way of assessing total cardiovascular risk in patients with diabetes mellitus. Am J Cardiol (2005) 96:933–934.[CrossRef][Web of Science][Medline]
- Tarnow L, Gall MA, Hansen BV, Hovind P, Parving HH. Plasma N-terminal pro-B-type natriuretic peptide and mortality in type 2 diabetes. Diabetologia (2006) 49:2256–2262.[CrossRef][Web of Science][Medline]
- Gaede P, Hildebrandt P, Hess G, Parving HH, Pedersen O. Plasma N-terminal pro-brain natriuretic peptide as a major risk marker for cardiovascular disease in patients with type 2 diabetes and microalbuminuria. Diabetologia (2005) 48:156–163.[CrossRef][Web of Science][Medline]
- Bhalla MA, Chiang A, Epshteyn VA, Kazanegra R, Bhalla V, Clopton P, Krishnaswamy P, Morrison LK, Chiu A, Gardetto N, Mudaliar S, Edelmann SV, Henry RR, Maisel AS. Prognostic role of b-type natriuretic peptide levels in patients with type 2 diabetes mellitus. Prognostic role of b-type natriuretic peptide levels in patients with type 2 diabetes mellitus. J Am Coll Cardiol (2004) 44:1047–1052.
[Abstract/Free Full Text] - Christoffersen C, Hunter I, Jensen A, Goetze J. Diabetes and the endocrine heart. Eur Heart J (2007) 28:2427–2429.
[Free Full Text] - Gaede P, Vedel P, Larsen N, Jensen GVD, Parving HH, Pedersen O. Multifactorial intervention and cardiovascular disease in patients with type 2 diabetes. N Engl J Med (2003) 348:383–393.
[Abstract/Free Full Text] - Horwich TB, Fonarow GC, Hamilton MA, MacLellan WR, Woo MA, Tillisch JH. The relationship between obesity and mortality in patients with heart failure. J Am Coll Cardiol (2001) 38:789–795.
[Abstract/Free Full Text] - Fonarow GC, Srikanthan P, Costanzo MA, Cintron GB, Lopatin M. An obesity paradox in acute heart failure: analysis of body mass index and inhospital mortality for 108927 patients in the acute decompensated heart failure registry. Am Heart J (2007) 153:74–81.[Web of Science][Medline]
- Lee TT, Chen J, Cohen DJ, Tsao L. The association between blood pressure and mortality in patients with heart failure. Am Heart J (2006) 151:76–83.[CrossRef][Web of Science][Medline]
- Rauchhaus M, Clark AL, Doehner W, Davos C, Bolger A, Sharma R, Coats AJ, Anker SD. The relationship between cholesterol and survival in patients with chronic heart failure. J Am Coll Cardiol (2003) 42:1933–1940.
[Abstract/Free Full Text] - Eshaghian S, Horwich TB, Fonarow GC. An unexpected inverse relationship between HbA1c levels and mortality in patients with diabetes and advanced systolic heart failure. Am Heart J (2006) 151:91.e1–91.e6.
- Kalantar-Zadeh K, Block G, Horwich T, Fonarow GC. Reverse epidemiology of conventional cardiovascular risk factors in patients with chronic heart failure. J Am Coll Cardiol (2004) 43:1439–1444.
[Abstract/Free Full Text] - Nichols GA, Gullion CM, Koro CE, Ephross SA, Brown JB. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care (2004) 27:1879–1884.
[Abstract/Free Full Text] - Swedberg K, Cleland J, Dargie H, Drexler H, Follath F, Komajda M, Tavazzi L, Smiseth O, Gavazzi A, Haverich A, Hoes A, Jaarsma T, Korewicki J, Levy S, Linde C, Lopez-Sendon JL, Nieminen MS, Pierard L, Remme WJ. The Task Force for the diagnosis and treatment of CHF of the European Society of Cardiology: Guidelines for the diagnosis and treatment of chronic heart failure: full text (update 2005). Eur Heart J (2005) 26:1115–1140.
[Free Full Text] - Gottlieb SS, Kukin ML, Ahern D, Packer M. Prognostic importance of atrial natriuretic peptides in patients with chronic heart failure. J Am Coll Cardiol (1989) 13:1534–1539.[Abstract]
- Omland T, Persson A, Ng L, O’Brien R, Karlsson T, Herlitz J, Hartford M, Caidahl K. N-terminal pro-b-type peptide and long-term mortality in acute coronary syndrome. Circulation (2002) 106:2913–2918.
[Abstract/Free Full Text] - Cook N. Use and Misuse of the Receiver Operating Characteristic Curve in Risk Prediction. Circulation (2007) 115:928–935.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Maier, M. Clodi, S. Neuhold, M. Resl, M. Elhenicky, R. Prager, D. Moertl, G. Strunk, A. Luger, J. Struck, et al. Endothelial Markers May Link Kidney Function to Cardiovascular Events in Type 2 Diabetes Diabetes Care, October 1, 2009; 32(10): 1890 - 1895. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||