A Multiparametric Clinical and EchocardiographicScore to Risk Stratify Patients with Chronic SystolicHeart Failure: Derivation and Testing

Paolo Fontanive, M.D.,* Mario Miccoli, Ph.D.,† Anca Simioniuc, M.D., Ph.D.,‡ Marco Angelillis, M.D.,‡Vitantonio Di Bello, M.D.,‡ Angelo Baggiani, M.D.,† Maria Grazia Bongiorni, M.D.,* Mario Marzilli, M.D.,‡and Frank Lloyd Dini, M.D.‡

*Cardiovascular Diseases Unit 2, Cisanello Hospital, Pisa, Italy; †Department of Experimental PathologyM.B.I.E., Epidemiology Unit, University of Pisa, Pisa, Italy; and ‡Cardiovascular Diseases Unit 1, University ofPisa, Pisa, Italy

Although echo Doppler and biomarkers are the most common examinations performed worldwide inheart failure (HF), they are rarely considered in risk scores. In outpatients with chronic HF and left ven-tricular ejection fraction (LVEF) �45%, data on clinical status, echo Doppler variables, aminoterminalpro-type B natriuretic peptide (NT-proBNP), estimated glomerular filtration rate (eGFR), and drug thera-pies were combined to build up a multiparametric score. We randomly selected 250 patients to producea derivation cohort and 388 patients were used as a testing cohort. Follow-up lasted 29 � 23 months.The univariable predictors that entered into the multivariable Cox model were as follows: furosemidedaily dose >25 mg, inability to tolerate angiotensin converting enzyme (ACE) inhibitors, inability totolerate b-blockers, age >75 years, New York Heart Association (NYHA) >2, eGFR<60 mL/min,NT-proBNP plasma levels above the median, tricuspid plane systolic excursion (TAPSE) �14 mm, LVend-diastolic volume index (LVEDVi) >96 mL/m2, moderate-to-severe mitral regurgitation (MR) andLVEF <30%. The scores of prognostic factors were obtained with the respective odds ratio divided bythe lower odd ratio: 4 points for furosemide dose, 3 points for age, NT-proBNP, LVEDVi, TAPSE, 2 pointsfor inability to tolerate b-blockers, inability to tolerate ACE inhibitors, NYHA, eGFR<60 mL/min, moder-ate-to-severe MR, 1 point for LVEF. The multiparametric score predicted all-cause mortality either in thederivation cohort (68.4% sensitivity, 79.5% specificity, area under the curve [AUC] 78.7%) or in thetesting cohort (73.7% sensitivity, 71.3% specificity, AUC 77.2%). All-cause mortality significantlyincreased with increasing score both in the derivation and in the testing cohort (P < 0.0001). In conclu-sion, this multiparametric score is able to predict mortality in chronic systolic HF. (Echocardiography2013;30:1172-1179)

Several recent studies have indicated that,despite advances in management of heart failure(HF), prognosis of patients over a 10-year periodhas not changed substantially.1 The variablesrelated to adverse outcome include demographicdata, etiology, comorbidities, clinical, radiological,hemodynamic and biochemical parameters andexercise testing.2–4 Although echo Dopplerparameters have emerged as independent pre-dictors of prognosis in patients with chronicsystolic HF,5–8 most of the scoring systems whichare currently utilized to risk stratify patients withchronic systolic HF do not include echo Doppler

variables with the exception of left ventricularejection fraction (LVEF).9,10 In addition, althoughcirculating levels of natriuretic peptides are rec-ognized as important biomarkers in the prognos-tic stratification of patients with chronic HF,11,12

quite a few studies have included them amongthe variables that are currently comprised in themodels for outcome prediction. Thus, the aim ofthis study was to develop and validate a multi-parametric HF score that included the contribu-tion of prognostic markers, such as echo Dopplervariables and natriuretic peptide levels that arecommonly evaluated in clinical practice in anoutpatient clinic.

Methods:Study Population:Outpatients referred for transthoracic echocardi-ography were evaluated at the echo Doppler labs

Address for correspondence and reprint requests: Frank LloydDini, M.D., F.E.S.C., Cardiovascular Diseases Unit 1, Cardiac,Thoracic and Vasculars Department, Azienda UniversitariaOspedaliera Pisana, Via Paradisa 2, 56100 – Pisa, Italy.Fax: 0039 (050) 995308;Emails: f.dini@ao-pisa.toscana.it, frank.dini@tiscali.it

1172

© 2013, Wiley Periodicals, Inc.DOI: 10.1111/echo.12255 Echocardiography

of the Cardiac, Thoracic and Vascular Depart-ment of the University of Pisa between January2002 and June 2010. The inclusion criteria wereas follows: history of chronic HF, LVEF �45%and etiology due to ischemic or hypertensiveheart disease or idiopathic cardiomyopathy.Patients were excluded if they had primary valvu-lar disease, mechanical valve prosthesis, unstableangina, recent myocardial infarction (within3 months), or coexisting terminal diseases.Written informed consent was obtained fromall patients participating in the study.

Echo Doppler:Transthoracic two-dimensional and echo Dopplerexamination was performed with an AcusonSequoia C256 ultrasound instrument (Acuson,Mountain View, CA, USA) with 2nd-harmonicimaging and a 3.5 MHz transducer. Recordingswere made with the patient in left lateral decub-itus during quiet respiration. Continuous single-lead ECG monitoring was maintained during thestudy. Left ventricular (LV) volumes and LVEFwere calculated from apical two- and four-cham-ber views using the modified Simpson’s rule. LVvolume indices and LV mass index were calcu-lated.13 Right ventricular systolic function wasevaluated by M-mode echocardiography usingthe tricuspid annular plane systolic excursion(TAPSE).14 Pulsed-wave Doppler mitral E-wavedeceleration time (DT) was measured and LV dia-stolic function was classified by the presence orabsence of restrictive mitral filling pattern,defined by an E-wave DT <140 msec.15 Mitralregurgitation severity was graded according tothe vena contracta method. Patients were con-sidered to have moderate-to-severe mitral regur-gitation (MR) if they had a vena contracta width�0.5 cm in the parasternal long-axis view.16

The estimated pulmonary artery systolic pressurewas obtained by the sum of the Doppler-derivedtranstricuspid gradient and the estimated rightatrial pressure, as assessed by the inspiratory col-lapse of the inferior vena cava.17

Clinical Assessment and Measurement ofBiological Variables:Medical history, physical examination, laboratoryresults, and medical management of HF wereregistered at the time of the index echocardio-gram. The presence and severity of decompen-sated HF score were assessed according to ascoring system based on Framingham criteria byan experienced cardiologist at the time of theindex echocardiogram (Table I).18 The estimatedglomerular filtration rate (GFR) was calculated onthe basis of the Cockcroft-Gault formula.19

According to the estimated GFR, patients weredivided into 2 groups: those with normal or

relatively preserved renal function (estimated GFR�60 mL/min) and those with renal dysfunction(estimated GFR <60 mL). Aminoterminal pro-type B natriuretic peptide (NT-proBNP) plasmalevels were measured by an Elecsys 2010 analyzer(Roche Diagnostics, Basel, Switzerland) (measur-ing range, 5 to 35 000 pg/mL) using a chemilu-minescent immunoassay kit (Roche Diagnostics,Grenach-Wyhlen, Germany).

Follow-Up:Patients were followed up after the baseline echoDoppler evaluation. The primary endpoint of thestudy was all-cause mortality. Survival data wereobtained through follow-up visits of patients and,in the case of a missed visit, vital status was veri-fied through telephone contacts, local authorityregistry, and hospital records. In the absence ofconfirmed mortality, survival times were basedon the last contact (clinical visit or phone call).

Statistics:Continuous variables are expressed as mean �standard deviation or median and interquartilerange. As the first step, we performed the univari-able analysis to define which variables we neededto include in the multivariable model. For thispurpose, we used the Shapiro–Wilk test to checknormality of data distribution to assess whetheror not to perform parametric tests (Student’st-test or Mann–Whitney test). With regard to thedichotomic variables, we used the chi-squaretest. For the multivariable logistic regressionswe identified categorical variables according toliterature data: estimated GFR, LVEF, LV volumeindices, LV mass, mitral restrictive filling pattern,moderate-to-severe MR and TAPSE, or medianvalues: plasma NT-proBNP. The preliminaryphase was concluded with a statistical power

TABLE I

Scoring System Based on Framingham Criteria for the Dis-crimination of Patients into Those with Decompensated HeartFailure and Those Stabilized by Therapy. A Total Score of 2.0

or More Would Indicate Clinically Overt Heart Failure

Major criteria Score Minor Criteria Score

Paroxysmal nocturnaldyspnea

1.0 Orthopnea 0.5

Hepatojugular refluxpositive

1.0 Reduction in exercisetolerance

0.5

Third sound present 1.0 Resting sinustachycardia

0.5

Basal crackles 1.0 Jugular venouspressure >4 cm

0.5

1.0 Hepatomegaly 0.51.0 Peripheral edema 0.5

1173

Risk Stratification by Heart Failure Score

analysis (ex-post) to estimate if the sample sizewas appropriate for the specific tests. The 1-bvalues of the significant variables were >0.8,assuring a low risk of type II error in the studyand showing an acceptable sample size. Then,we conducted four multivariable regressions inthe derivation cohort considering differentgroups of factors based on homogeneous char-acteristics: (1) clinical variables, (2) echo Dopplervariables, (3) biological variables, (4) pharmaco-logical variables. The variables with significantinteraction (r > 0.5) were excluded from themultivariable regression. The scores of prognosticfactors were obtained with the respective oddsratio divided by the lower odd ratio. With theaim of testing the new score we selected a newgroup of patients recruited with another samplesize analysis (testing cohort). In an attempt toinvestigate the incremental prognostic contribu-tion of echocardiographic and Doppler indexes,an interactive stepwise procedure was performedand variables were included in the model in thesame order in which they are actually consideredin clinical practice. Therefore, clinical and phar-macological variables, including estimated GFR,(basic model) were firstly analyzed, and the glo-bal chi-square was calculated. Further steps werecreated after the addition of NT-proBNP andecho Doppler variables to the independent pre-dictors at first step. The incremental prognosticvalue of the added variables was assessed bycomparison of the global chi-square at each step.Finally, probabilities of survival free from all-causemortality were estimated by the use of the Kap-lan–Meier method. Survival curves were com-pared by the log-rank test. Data were enteredinto an Excel database and were subsequentlylocked, imported into SPSS for Windows (version17.0, SPSS Inc., Chicago, IL, USA) and R-2.12.1.

Results:An echo Doppler estimate of pulmonary arterysystolic blood pressure (PASP) was feasible in 552(88%) patients. TAPSE was measured in allpatients. E-wave DT was measured in 513 (80%)patients. Information on NT-proBNP and esti-mated GFR was available in all the study patients.Relevant demographic, clinical, echo Doppler andbiochemical characteristics of the patient popula-tion are summarized in Table II. We did not findsignificant differences between the variables ofthe derivation group and the variables of the test-ing group. During a mean follow-up time of29 months (range 1.0–90 months), 178 patientsdied (derivation: 79, testing: 99); overall mortalityrate was 28% (derivation: 32%, testing: 26%).

In the derivation cohort, parameters thatwere associated with increased risk of death in

the univariable and multivariable models areillustrated in Table III. The following univariablepredictors of increased risk were entered into themultivariable Cox regression model: furosemidedaily dose >25 mg, inability to tolerate ACEinhibitors, inability to tolerate b-blockers, age>75 years, New York Heart Association (NYHA)class >2, eGFR <60 mL/min, NT-proBNP plasmalevels above the median, TAPSE �14 mm, LVend-diastolic volume index >96 mL/m2, moderate-to-severe MR and LVEF <30%. The score of singlevariables was obtained as previously described: 4points for furosemide dose, 3 points for age>75 years, NT-proBNP above the median, LVEDVi>96 mL/m2, TAPSE �14 mm, 2 points for inabil-ity to tolerate b-blockers, inability to tolerate ACEinhibitors, NYHA class >2, estimated GFR<60 mL/min, moderate-to-severe MR, 1 point forLVEF <30%. Combining these measurements,each patient was assigned a total score rangingfrom 0 to 26.

The relationship between the event rate andthe score in the entire study population isdepicted in Figure 1. The ability of the multipara-metric clinical and echocardiographic HF score toidentify the risk of all-cause mortality wasassessed by the receiver operator characteristic(ROC) curve analysis. The risk score was able topredict the outcome either in the derivationcohort (68.4% sensitivity, 79.5% specificity, areaunder the curve [AUC] 78.7%) or in the testingcohort (73.7% sensitivity, 71.3% specificity, AUC77.2%). In the overall study population, this HFrisk score better predicted survival by ROC analy-sis than the Muerte Subita en Insuficiencia Cardi-aca study (MUSIC) risk score and CongestiveHeart Failure (3CHF) risk score, as it is apparent inFigure 2. In the interactive stepwise procedureapplied to the prediction of all-cause mortality,the addition of NT-proBNP improved the chi-square value from 38.8 to 39.9 (P < 0.0001).The addition of echo Doppler variables movedthe chi-square up to 55.4 (P < 0.0001).

Kaplan–Meier survival curves for patients ofthe derivation and in testing cohorts stratifiedaccording to tertiles of HF score (<7, 7–13, >13)are presented in Figure 3. As shown, patientswith HF score >13 had the lowest survival rate,patients with HF score 7–13 had an intermediatesurvival rate, while patients with HF score <7 hadthe highest survival rate.

Discussion:This study describes the derivation and testing ofa new scoring system for the prognostic stratifica-tion of outpatients with chronic systolic HF basedon the combination of clinical, echo Doppler,biological findings, loop diuretic dose, and the

1174

Fontanive, et al.

inability to tolerate ACE inhibitors and b-blockers.This multiparametric HF score was purposelydesigned to include the separate contribution ofprognostic markers that are currently evaluatedin clinical practice in an ambulatory setting.

HF is a common clinical syndrome associatedwith a mortality rate that varies according to theseverity of disease and the administration ofappropriate medical therapy. As more than 60prognostic markers in HF have been identified,an integration of the different parameters in therisk stratification is useful. Although it is nowrecognized that risk scores are important to riskstratify patients with chronic HF, for the cost-benefit analysis and implementation of thera-peutic strategies, the models that have beensuggested have limitations that might preventtheir widespread clinical use. In particular, theyinfrequently include echo Doppler variables andnatriuretic peptide assessment.

Some of the scoring systems that have beenproposed are difficult to utilize in an ambulatorysetting especially because they have been devel-oped in patients with advanced HF or becausethey include parameters such as VO2 max thatare not always available in patients with chronicHF.9,20–22 Other models include a wide set of clin-ical, biochemical, and pharmacologic variables,but they are derived from cohorts of patientsfrom large randomized trials that scarcely reflectthe patient population of the real world.8,9 Therecognition of the difficulties in clinical applicabil-ity has brought the introduction of scoring sys-tems that have been derived from ambulatorypatients and that incorporate echo Doppler vari-ables, such as LVEF21 and left atrial size,23 and theevaluation of the circulating levels of natriureticpeptides. However, a major potential flaw ofthese scores is that they were tested in the samepopulation in which they were developed.23

TABLE II

Clinical Characteristics of the Derivation and Testing Groups

Variable All Derivation (n = 250) Testing (n = 388) P-Value

Age (years) 68 � 12 68 � 12 69 � 12 NSMale (%) 78 78 77 NSHeart rate (beats/min) 75 � 15 74 � 15 76 � 14 NSIschemic CM (%) 52 53 52 NSeGFR <60 mL/m2 (%) 27 29 27 NSDM (%) 21 20 21 NSHistory of hypertension 41 37 44 NSAtrial fibrillation (%) 20 18 21 NSLoop diuretics (%) 86 87 86 NSACE inhibitors (%) 79 75 81 NSb-blockers (%) 72 71 78 NSNYHA class > II (%) 43 41 44 NSFramingham criteria* 1.0 (0.5–2.0) 1.0 (0.5–2.0) 1.0 (0.5–2.0) NSCardiac resynchronization therapy (%) 7 8 7 NSInternal cardioverter defibrillator (%) 6 7 6 NSSystolic BP (mmHg) 124 � 17 123 � 16 125 � 17 NSDiastolic BP (mmHg) 74 � 10 73 � 11 74 � 10 NSLV EDVi (mL/m2) 113 � 35 114 � 34 112 � 35 NSLV ESVi (mL/m2) 78 � 31 80 � 30 77 � 32 NSLVEF (%) 32 � 8 31 � 8 32 � 9 NSLV mass index (g/m2) 160 � 38 161 � 35 159 � 39 NSLeft atrial size (mm) 49 � 7 49 � 6 49 � 7 NSModerate-to-severe MR† (%) 27 26 27 NSE/A ratio 1.6 � 1.5 1.7 � 1.6 1.5 � 1.5 NSEDT (msec) 160 � 52 159 � 52 162 � 53 NSTAPSE (mm) 18 � 5 18 � 5 18 � 5 NSPASP (mmHg) 41 � 12 40 � 12 41 � 11 NSNT-proBNP (pg/mL)* 1464 (613–3401) 1443 (616–3173) 1506 (606–3775) NS

ACE = angiotensin converting enzyme; BP = blood pressure; CM = cardiomyopathy; DM = diabetes mellitus; EDT = E-wavedeceleration time; EDVi = end-diastolic volume index; ESVi = end-systolic volume index; eGFR = estimated glomerularfiltration rate; HF = heart failure; LV = left ventricular; LVEF = left ventricular ejection fraction; MR = mitral regurgitation;NT-proBNP = aminoterminal pro-type B natriuretic peptide; NYHA = New York Heart Association; PASP = pulmonary arterysystolic blood pressure; TAPSE = tricuspid annular plane systolic excursion.*Median and interquartile range.†According to the vena contracta method.

1175

Risk Stratification by Heart Failure Score

The 3CHF score is a model that has beenrecently proposed and that was derived from astudy population that included a total of 6274patients diagnosed with chronic HF.24 This scoreis highly accurate in the outcome prediction, asdemonstrated by an AUC of 82%; it containsvariables that are commonly obtainable in ambu-latory patients and considers the presence ofimportant comorbidities. The major drawback ofthis score is that it does not take into accountessential echo Doppler variables, such as indicesof LV remodeling, right ventricular function ormitral parameters, and that it does not include

the evaluation of natriuretic peptides, which arenow considered a cornerstone in the risk stratifi-cation of patients with chronic HF.25 The avail-ability, accuracy, and cost-effectiveness of theecho Doppler technique have made it an indis-pensable tool for the evaluation and prognosticstratification of patients with HF. Particularly, it isimportant not only to acquire information onLVEF but also to evaluate parameters of LVremodeling, LV filling pressure, right ventricularfunction, and presence and severity of functionalMR. The importance of a comprehensive echoDoppler evaluation of patients with HF has beenalso indicated by the recent European Society ofCardiology Guidelines on Heart Failure.1 We haverecently demonstrated the usefulness of combin-ing blood flow and tissue Doppler imaging withcirculating biomarkers to risk stratify patientswith HF.26–28

In this study, it was our aim to design anddevelop a multiparametric HF risk score that canpossibly include important clinical, biochemical,echo Doppler, and therapeutic predictors, whichare routinely evaluated in ambulatory patients toobtain data on cardiac structure and functionas well as neuroendocrine activation and co-morbidities. Increasing HF score was associatedwith an increased risk of mortality in outpatientswith chronic systolic HF. An interesting finding is

TABLE III

Univariable and Multivariable Predictors of All-Cause Mortality in the Derivation Cohort

VariableUnivariableOR (95%CI) P-Value

MultivariableOR (95%CI) P-Value

Age >75 years versus �75 years 2.9 (1.7, 5.2) 0.0002 2.7 (1.5, 4.8) 0.0009Male gender 1. (0.5, 1.9) 0.90Heart rate �80 beats per min versus <80 beats per min 1.6 (0.9, 2.7) 0.10Ischemic versus nonischemic CM 1.6 (0.9, 2.8) 0.10eGFR <60 mL/m2 versus �60 mL/m2 2.5 (1.4, 4.4) 0.003 2.1 (1.2, 3.8) 0.01Diabetes 2.1 (1.1, 4.1) 0.003History of hypertension 1.3 (0.7, 2.2) 0.40Atrial fibrillation 1.4 (0.7, 2.8) 0.40Furosemide >25 mg versus �25 mg daily 4.5 (2.5, 7.9) <0.0001 4.0 (2.2, 7.2) <0.0001No ACE inhibitors 2.5 (1.4, 4.6) 0.003 2.0 (1.1, 3.9) 0.02No b-blockers 1.9 (1.0, 3.3) 0.04NYHA class>II versus � II 2.7 (1.5, 4.6) 0.0006 2.4 (1.4, 4.3) 0.0009Framingham HF Criteria �2.0 versus <2.0 2.9 (1.6, 5.1) 0.0005LV EDVi >96 mL/m2 versus �96 mL/m2 3.6 (1.7, 7.6 0.0007 2.9 (1.4, 6.7) 0.010LV ESVi >42 mL/m2 versus �42 mL/m2 1.1 (0.3, 4.2) 0.80LVEF <30% versus �30% 1.9 (1.1, 3.2) 0.03Moderate-to severe MR 3.3 (1.8, 5.9) <0.0001 2.3 (1.2, 4.5) 0.010EDT <140 msec versus �140 msec 2.3 (1.3, 4.1) 0.003TAPSE �14 mm versus >14 mm 3.1 (1.7, 5.7) 0.0003 2.6 (1.3, 5.1) 0.004PASP >40 mmHg versus �40 mmHg 2.8 (1.6, 4.9) 0.0004NT-pro-BNP >median versus �median 2.2 (1.3, 3.9) 0.0002 2.6 (1.5, 4.6) 0.0009

OR = odds ratio; CI = confidence intervals; TPSE = tricuspid plane systolic excursion; PASP = pulmonary artery systolic blood pres-sure; NYHA = New York Heart Association; MR = mitral regurgitation; HF = heart failure; eGFR = estimated glomerular filtrationrate; ACE = angiotensin converting enzyme; LVEF = left ventricular ejection fraction. For the other abbreviations see Table 1.

Figure 1. Association between the event rate and the riskscore in the entire study population.

1176

Fontanive, et al.

that compromised echo Doppler parametersoutperformed NT-proBNP levels in the predictionof patients’ outcome at follow-up, as it is appar-ent from the incremental analysis. An increasedLVEDV, a reduced TAPSE, and the presence ofa mild-to-moderate MR appear to reflect theseverity of the disease with the inherent patho-physiologic and prognostic implications.

The HF risk scores available in literature resultfrom multivariate analyses based on a high num-ber of predictive factors. The excessive numberof variables increases the P-values and reducesthe odds ratio, thus producing a distorting effecton scores. In this study, we tried to restrain thisbias by using 4 different multivariate analysesbased on homogeneous groups of variables toobtain a global HF score comprising the contri-butions of clinical, instrumental, biochemical andpharmacological prognostic factors.

A

B

C

Figure 2. Receiver operator characteristic (ROC) for the pre-diction of all-cause mortality in the overall study populationobtained using different risk scores. A. The multiparametricclinical and echocardiographic score: 68.5% sensitivity,77.0% specificity, area under the curve [AUC] 77.9%. B. TheMUSIC score: 50.8% sensitivity, 79.2% specificity, AUC71.5%. C. The Congestive Heart Failure (3CHF) score: 48.9%sensitivity, 85.5% specificity, AUC 72.4%.

A

B

Figure 3. Kaplan–Meier plots showing survival for patientsof the derivation A. and in the testing B. cohorts strati-fied according to tertiles of the multiparametric risk score.

1177

Risk Stratification by Heart Failure Score

Limitations:Among the study limitations, the relatively smallnumber of patients and the single-center recruit-ment should be considered. The high mortalityrate reported in this study (28%) may be likelyexplained by the derivation of our patients’population from the real world rather than fromclinical trials, where some of the other HF scoreshave been developed.29 We should also take intoaccount the inclusion of a subgroup of ambula-tory NYHA class 4 patients with persistent severesymptomatic HF. Finally, another reason thatmay contribute to elucidate for the high mortal-ity rate is that quite few patients had beentreated with an internal cardioverter defibrillatorat baseline.

Another limitation is that the score was derivedand tested only in patients with systolic HF, whilepatients with HF and preserved LVEF were notincluded in the study cohorts. Whether tissueDoppler imaging techniques27 or new ultrasoundtechnologies30 could be valuable in designingnewer HF risk scores is an issue that can possiblybe addressed in the forthcoming future.

Conclusion:As there is now evidence that echo Dopplerparameters and circulating natriuretic peptidesare important to risk stratify ambulatory patientswith chronic HF, this HF risk score was designedto encompass echo Doppler indexes andNT-proBNP plasma levels assessed at the time ofthe index echocardiogram. By referring to aHF risk score derived from a number of echoDoppler parameters that are not always accessi-ble to general practitioners, this approach maygive effective information above and beyondLVEF; it could also overcome the jargon of echo-cardiographists and facilitate communicationamong physicians.

References1. McMurray JJ, Adamopoulos S, Anker SD, et al: ESC

Committee for Practice Guidelines. Eur Heart J 2012;33:1787–1847.

2. Campana C, Gavazzi A, Berzuini C, et al: Predictors ofprognosis in patients awaiting heart transplantation.J Heart Lung Transplant 1993;12:756–765.

3. Pocock SJ, Wang D, Pfeffer MA, et al: Predictors of mor-tality and morbidity in patients with chronic heart failure.Eur Heart J 2006;27:65–75.

4. Barlera S, Tavazzi L, Franzosi MG, et al: Predictors of mor-tality in 6975 patients with chronic heart failure in theGruppo Italiano per lo Studio della Streptochinasi nell’Inf-arto Miocardico-Heart Failure trial: Proposal for a nomo-gram. Circ Heart Fail 2013;6:31–39.

5. Senni M, Rodeheffer RJ, Tribouilloy CM, et al: Use ofechocardiography in the management of congestiveheart failure in the community. J Am Coll Cardiol1999;33:164–170.

6. Meta-analysis Research Group in Echocardiography(MeRGE) Heart Failure Collaborators: Independence of

restrictive filling pattern and LV ejection fraction withmortality in heart failure: An individual patient meta-analysis. Eur J Heart Fail 2008;10:786–792.

7. Ballo P, Guarini G, Simioniuc A, et al: Prognostic value ofpulsed tissue Doppler imaging for the assessment of leftventricular systolic function in patients with nonischemicdilated cardiomyopathy. Echocardiography 2012;29:291–297.

8. Ghio S, Temporelli PL, Klersy C, et al: Prognostic rele-vance of a non-invasive evaluation of right ventricularfunction and pulmonary artery pressare in patients withchronic heart failure. Eur J Heart Fail 2013;15:408–414.

9. Aaronson KD, Schwartz JS, Chen TM, et al: Developmentand prospective validation of a clinical index to predictsurvival in ambulatory patients referred for cardiac trans-plant evaluation. Circulation 1997;95:2660–2667.

10. Levy WC, Mozaffarian D, Linker DT, et al: The Seattleheart failure model: Prediction of survival in heart failure.Circulation 2006;113:1424–1433.

11. Zugck C, Haunstetter A, Kr€uger C, et al: Impact of beta-blocker treatment on the prognostic value of currentlyused risk predictors in congestive heart failure. J Am CollCardiol 2002;39:1615–1622.

12. Pfisterer M, Buser P, Rickli H, et al: BNP-guided vs symp-tom-guided heart failure therapy: The Trial of Intensifiedvs Standard Medical Therapy in Elderly Patients WithCongestive Heart Failure (TIME-CHF) randomized trial.JAMA 2009;301:383–392.

13. Lang RM, Bierig M, Devereux RB, et al: Recommenda-tions for chamber quantification: A report from theAmerican Society of Echocardiography’s Guidelines andStandards Committee and the Chamber QuantificationWriting Group, developed in conjunction with the Euro-pean Association of Echocardiography, a branch of theEuropean Society of Cardiology. J Am Soc Echocardiogr2005;18:1440–1463.

14. Dini FL, Demmer RT, Simioniuc A, et al: Right ventriculardysfunction is associated with chronic kidney disease andpredicts survival in patients with chronic systolic heartfailure. Eur J Heart Fail 2012;14:287–294.

15. Nagueh S, Appleton CP, Gillebert TC, et al: Recommen-dations for the evaluation of left ventricular diastolic func-tion by echocardiography. J Am Soc Echocardiogr2009;22:107–133.

16. Hall SA, Brickner ME, Willett DL, et al: Assessment ofmitral regurgitation severity by Doppler color flow map-ping of the vena contracta. Circulation 1997;95:636–642.

17. Stein JH, Neumann A, Preston LM, et al: Echocardiogra-phy for hemodynamic assessment of patients withadvanced heart failure and potential heart transplantrecipients. J Am Coll Cardiol 1997;30:1765–1772.

18. Troughton RW, Frampton CM, Yandle TG, et al: Treat-ment of heart failure guided by plasma aminoterminalbrain natriuretic peptide (N-BNP) concentrations. Lancet2000;355:1126–1130.

19. Cockcroft DW, Gault MH: Prediction of creatinine clear-ance from serum creatinine. Nephron 1976;16:31–41.

20. Lee DS, Austin PC, Rouleau JL, et al: Predicting mortalityamong patients hospitalized for heart failure: Derivationand validation of a clinical model. JAMA 2003;290:2581–2587.

21. O’Connor CM, Whellan DJ, Wojdyla D, et al: Factorsrelated to morbidity and mortality in patients withchronic heart failure with systolic dysfunction. The HF-ACTION predictive risk score model. Circ Heart Fail2012;5:63–71.

22. Agostoni P, Corr�a U, Cattadori G, et al: Metabolic exer-cise test data combined with cardiac and kidney indexes,the MECKI score: A multiparametric approach to heartfailure prognosis. Int J Cardiol 2012 Jul 13 [Epub ahead ofprint].

1178

Fontanive, et al.

23. Vazquez R, Bayes-Genis A, Cygankiewicz I, et al: TheMUSIC Risk score: A simple method for predicting mor-tality in ambulatory patients with chronic heart failure.Eur Heart J 2009;30:1088–1096.

24. Senni M, Parrella P, De Maria R, et al: Predicting heartfailure outcome from cardiac and comorbid conditions:The 3C-HF score. Int J Cardiol 2013;163:206–211.

25. Januzzi JL Jr, Rehman SU, Mohammed AA, et al: Use ofamino-terminal pro-B-type natriuretic peptide to guideoutpatient therapy of patients with chronic left ventricu-lar systolic dysfunction. J Am Coll Cardiol 2011;58:1881–1889.

26. Dini FL, Conti U, Fontanive P, et al: Prognostic value ofN-terminal pro-type-B natriuretic peptide and Dopplerleft ventricular diastolic variables in patients with chronicsystolic heart failure stabilized by therapy. Am J Cardiol2008;102:463–468.

27. Dini FL, Rosa GM, Fontanive P, et al: Combining bloodflow and tissue Doppler imaging with N-terminal pro-type B natriuretic peptide for risk stratification of clinicallystable patients with systolic heart failure. Eur J Echocardi-ogr 2010;11:333–340.

28. Bajraktari G, Miccoli M, Buralli S, et al: Plasma metallo-proteinase-9 and restrictive filling pattern as majorpredictors of outcome in patients with ischemic cardio-myopathy. Eur J Intern Med 2012;23:616–620.

29. Niederseer D, Thaler CW, Niederseer M, et al: Mismatchbetween heart failure patients in clinical trials and the realworld. Int J Cardiol 2013 Jan 24 [Epub ahead of print].

30. Lilli A, Baratto MT, Meglio JD, et al: Left ventricularrotation and twist assessed by four-dimensional speckletracking echocardiography in healthy subjects andpathological remodeling: A single center experience.Echocardiography 2013;30:171–179.

1179

Risk Stratification by Heart Failure Score