Prognostic Value of Heart RateVariability in Chronic Congestive HeartFailure (Veterans Affairs’ Survival Trialof Antiarrhythmic Therapy in Congestive

Heart Failure)Kenneth C. Bilchick, MD, Barry Fetics, MSE, Ronnie Djoukeng, Susan Gross Fisher, PhD,

Ross D. Fletcher, MD, Steven N. Singh, MD, Erez Nevo, MD, DSC, andRonald D. Berger, MD, PhD

Although the value of heart rate variability (HRV) for riskstratification after acute myocardial infarction has beendemonstrated, the value of low HRV as a predictor ofsudden cardiac death in patients with ischemic cardio-myopathy has not been shown convincingly to date. Weretrospectively analyzed electrocardiographic data from179 patients in the Veterans Affairs’ Survival Trial ofAntiarrhythmic Therapy in Congestive Heart Failure todetermine if HRV (expressed as the SD of the normal-to-normal RR intervals [SDNN]) would be useful as a pre-dictor of overall mortality and sudden death. Becauseour goal was to identify high-risk patients, we comparedpatients in the lowest quartile of HRV with the remaining

patients. Among the 127 patients meeting inclusion cri-teria, SDNN <65.3 ms (the lowest quartile) was the soleindependent factor predictive of survival in a multivari-ate model (p � 0.0001). A Cox proportional-hazardsmodel revealed that each increase of 10 ms in SDNNconferred a 20% decrease in risk of mortality (p �0.0001). Furthermore, patients with SDNN <65.3 mshad a significantly increased risk of sudden death (p �0.016). Thus, HRV was the sole independent predictor ofoverall mortality and was significantly associated withsudden death in this population. �2002 by ExcerptaMedica, Inc.

(Am J Cardiol 2002;90:24–28)

Previous data addressing the relation of heart ratevariability (HRV) and sudden cardiac death in

patients with heart failure have yielded mixed results.The goal of the present study was to clarify whetherHRV, as measured by the SD of the normal-to-normalRR intervals (SDNN), could predict sudden death inpatients with heart failure and investigate how well itsuse as a predictor of mortality compares with otherrisk stratifiers such as left ventricular ejection fraction(LVEF). To answer this question, we used electrocar-diographic data from the Veterans Affairs’ SurvivalTrial of Antiarrhythmic Therapy in Congestive HeartFailure (CHF-STAT), which studied the effectivenessof amiodarone in reducing sudden death and mortalityin a population of 674 patients with EF �40% andprimarily New York Heart Association (NYHA) classII and III heart failure.1

METHODSCHF-STAT study design: The design of the CHF-

STAT study is described elsewhere.1,2 Briefly, patientswith a documented history of congestive heart failure(EF �40%), whether ischemic or nonischemic in origin,and at least 10 ventricular premature beats/hour, unac-companied by symptoms, were eligible for the study.Exclusion criteria included myocardial infarction withinthe 3 months before enrollment, symptomatic ventriculararrhythmia, a history of aborted sudden cardiac arrest orsustained ventricular tachycardia, uncontrolled thyroiddisease, women of childbearing age, the need for antiar-rhythmic therapy, prolongation of the QRS or QT inter-val, a serious disease other than heart disease likely to befatal within 3 years, symptomatic hypertension, and sys-tolic blood pressure �90 mm Hg. The diagnosis ofischemic cardiomyopathy was assessed by a clinicianand required 1 of the following: at least 1 coronarystenosis �50%, chest pain with reversible electrocardio-graphic changes, or a positive stress isotope study.

Six hundred seventy-four patients were randomizedto either amiodarone (800 mg once daily for 14 days,then 400 mg once daily for 50 weeks, and then 300 mgonce daily until the end of the study) or placebo through-out the trial. A 24-hour continuous electrocardiogramwas obtained at baseline, 2 weeks, 1, 3, 6, 9, and 12months, and then every 6 months. Patients were recruitedfor the study over a period of 3.5 years and followed for1 additional year after the enrollment period. The studyend points were overall mortality and sudden death.

From the Department of Medicine, Johns Hopkins University School ofMedicine, Baltimore, Maryland; and the Department of Veterans Af-fairs, Cooperative Studies Program, Hines, Illinois. This work wassupported by Grants P50 HL52307 and R44 HL61903 from theNational Heart, Lung, and Blood Institute, Bethesda, Maryland. TheCHF-STAT study was supported by the Cooperative Studies Program ofthe U.S. Department of Veterans Affairs Research and Development Ser-vice, Sanofi Winthrop Recherche, Paris, France; and Wyeth-Ayerst Lab-oratories, Philadelphia, Pennsylvania. Manuscript received December4, 2001; revised manuscript received and accepted February 26,2002.

Address for reprints: Ronald D. Berger, MD, PhD, Carnegie 592,Johns Hopkins Hospital, 600 North Wolfe Street, Baltimore, Maryland21287. E-mail: rberger@jhmi.edu.

24 ©2002 by Excerpta Medica, Inc. All rights reserved. 0002-9149/02/$–see front matterThe American Journal of Cardiology Vol. 90 July 1, 2002 PII S0002-9149(02)02380-9

Holter analysis: The earliest available Holter (seeResults section) from a random sample of 179 CHF-STAT patients was digitized using a Del-Mar Holterrecorder (Irvine, California) and analyzed using a cus-tomized computer program at the Johns Hopkins Hos-pital. For each 24-hour Holter, the SDNN was calcu-lated by investigators blinded to patients’ outcomes.Data segments with excessive ectopy (�10%), atrialfibrillation, pacing, or problems with recording qualitywere excluded.

Statistical analysis: It was determined a priori that theindependent factor of interest would be SDNN dichoto-mized by the lowest quartile versus the remainder of thestudy sample. Baseline demographic and disease char-acteristics between SDNN groups were compared toidentify imbalances that might necessitate inclusion in a

multivariate analysis. The Student’s ttest or the chi-square test were used forcontinuous and categorical variables,respectively. Continuous variables in-cluded heart rate, age, frequency ofventricular premature contractions,and systolic blood pressure. Categori-cal variables that were examined weregender, presence of ischemic cardio-myopathy, LVEF (lowest quartile vsremainder), NYHA class, and treat-ment group. Study outcomes were de-fined as overall mortality and mortalityrelated to sudden death. Sudden deathwas defined prospectively as any deathoccurring �1 hour from the onset ofsymptoms (in the absence of severeleft ventricular failure or shock), wit-nessed resuscitated or unresuscitatedventricular fibrillation, or unwitnesseddeath in an apparently stable patient.

Kaplan-Meier survival techniques and the log-rank testwere used in the analysis. Survival was calculated fromthe date of study randomization to the date of the out-come (death or sudden death); patients not reaching anoutcome were censored at the last follow-up date or thedate of nonsudden death, as appropriate. Variables en-tered into the multivariate Cox model included factorsthat differed between the SDNN groups (p �0.10) aswell as patient age, treatment group, and the presence ofischemic cardiomyopathy. In all analyses, a 2-sided �level of 0.05 was considered statistically significant.

RESULTSDemographics: Of the 179 patients in our sample,

52 were excluded from our analysis. Reasons forexclusion were excessive ectopy (n � 21), atrial fi-

TABLE 1 Baseline Characteristics of Study Population (n � 127)

Overall(n � 127)†

SDNN �65.3(n � 32)†

SDNN �65.3(n � 95)† p Value*

Age (yrs) 65 � 9 63 � 9 65 � 9 NSLVEF (%) 26 � 9 23 � 9 27 � 8 0.0068Systolic BP (mm Hg) 129 � 19 123 � 18 131 � 19 0.054HR (beats/min) 79 � 13 86 � 11 77 � 13 0.0016VPC frequency (/h) 150 � 229 163 � 247 146 � 221 NSMen 127 (100%) 32 (100%) 95 (100%)Ischemic cardiomyopathy 97 (76%) 23 (72%) 74 (78%)NYHA class II† 69 (56%) 15 (48%) 54 (58%)NYHA class III† 51 (41%) 14 (45%) 37 (40%)NYHA class IV† 4 (3%) 2 (7%) 2 (2%)On � blocker 8 (6%) 0 (0%) 8 (8%)On ACE inhibitor 116 (91%) 28 (88%) 88 (93%)On amiodarone 72 (57%) 14 (44%) 58 (61%)

*The p values refer to comparison between SDNN groups.†NYHA class unavailable for 3 patients. For NYHA class, n � 124 for Overall, n � 31 for SDNN

�65.3, and n � 93 for SDNN �65.3.ACE � angiotensin-converting enzyme inhibitor; BP � blood pressure; HR � heart rate; VPC �

ventricular premature contraction.

FIGURE 1. Overall mortality ofpatients in the lowest quartile ofSDNN versus the remainder.

HEART FAILURE/HRV AND CONGESTIVE HEART FAILURE 25

brillation (n � 14), pacing (n � 12), and problemswith the quality of the Holter recording (n � 5).

The study analysis included 127 patients (mean age64.7 years). Table 1 lists the demographics for thesepatients. Seventy-two were assigned to amiodaroneand 55 to placebo. Most patients were in NYHA classII and III with a mean LVEF of 26%; slightly morethan 3/4 had ischemic cardiomyopathy. The sampledid not differ significantly from the overall CHF-STAT study population with respect to the presence ofischemic cardiomyopathy, gender, NYHA class, useof angiotensin-converting enzyme inhibitors, use of �blockers, age, systolic blood pressure, LVEF, andheart rate. There was a significant difference in ven-tricular premature contractions per hour between ourstudy group (150 � 229) and the entire CHF-STATstudy population (293 � 401; p � 0.0001), mostlikely due to our exclusion of patients with a highdegree of ectopy.

The SDNN for the study sample was 90.0 � 36.1ms. Based on a cutoff of 65.3 ms (first quartile),subjects were categorized for analysis of low (n � 32)

and high (n � 95) SDNN. Statisticaldifferences (p �0.10) betweenSDNN groups (lowest quartile vs allothers) were identified. Subjects withlower SDNN had lower LVEF, lowersystolic blood pressure, and highermean heart rate. These variables,with the addition of age, the presenceof ischemic cardiomyopathy, andrandomization to amiodarone, wereentered into a multivariate Coxmodel.

Prerandomization tapes wereavailable for 90 patients. For the re-maining 37 patients included in ouranalysis, the median time between

randomization and the Holter recording studied was30 days.

Overall survival: Of the 127 patients in our studypopulation, there were a total of 59 deaths during a meanfollow-up of 34 months. Patients with SDNN in thelowest quartile (SDNN �65.3 ms) were at significantlygreater risk for all-cause mortality than those with SDNNin the upper 3 quartiles (p � 0.0001) (Figure 1). A Coxproportional-hazards model revealed that each increaseof 10 ms conferred a 20% decrease in risk of mortality (p� 0.0001). Figure 2 provides the survival distribution foreach specific quartile of SDNN. This strong associationbetween low SDNN and mortality was observed withineach treatment group (amiodarone, p � 0.0001; placebo,p � 0.004).

Multivariate analysis demonstrated that SDNN wasthe only independent factor predictive of survival inthis study sample (risk ratio 3.72; p � 0.0001) (Table2). Interestingly, LVEF, previously identified in theVeterans Administration cohort as the most importantpredictor of survival, did not significantly predict sur-vival in this multivariate model that included SDNN.

FIGURE 2. Overall mortality byquartile of SDNN.

TABLE 2 Multivariate Analysis for Overall Survival

Factor

Overall Mortality Sudden Death

RR p Value RR p Value

SDNN �65.3 ms 3.72 0.0001 2.40 0.088*Ejection fraction �20% 1.03 0.939 1.67 0.330Systolic BP 1.00/mm Hg 0.735 1.01/mm Hg 0.579HR 1.06/beats/min 0.425 1.01/beat/min 0.583Age 1.02/yr 0.210 1.00/yr 0.875On amiodarone 1.16 0.596 1.24 0.635Ischemic cardiomyopathy 1.23 0.55 3.77 0.096

*The p value for SDNN in a univariate analysis was 0.016.Abbreviations as in Table 1.

26 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 90 JULY 1, 2002

In fact, when considered in a univariate analysis,LVEF �20% was associated with a 76% increase inrisk of death compared with the greater than threefoldincrease in risk observed with low SDNN.

Sudden death: In this study population, 21 patientsexperienced sudden death. Figure 3 shows that SDNNwas significantly associated with sudden death in theoverall patient population, with those in the lowestquartile at increased risk (p � 0.016). In a multivariateanalysis, patients with SDNN �65.3 ms had a riskratio of 2.40 for sudden death, although only a trendwas demonstrated (p � 0.088). The presence of isch-emic cardiomyopathy was the only other factor asso-ciated with an increased risk for sudden death in themultivariate analysis.

DISCUSSIONThe key finding in this study was that SDNN, a

measure of HRV, was the sole predictor of overallmortality and was significantly associated with suddendeath in this chronic heart failure cohort. Bigger etal3,4 established the prognostic value of SDNN mea-sured shortly after acute myocardial infarction. Ourfindings extend these results to the chronic heart fail-ure period. Our study population represents a crosssection of patients with heart failure, predominantlyNYHA class II and III. Approximately 3/4 of thesepatients had cardiomyopathy of ischemic origin, andthe remaining 1/4 had nonischemic dilated cardiomy-opathy; most had mild to moderate ventricular ectopy.

There was a strong association between SDNN andmortality in these patients with chronic heart failure,even after adjustment for LVEF—a controversial find-ing based on current published reports discussed be-low. LVEF measured soon after a myocardial infarc-tion was shown to predict mortality in the early1980s.5 Recent postinfarction studies of both HRVand LVEF have confirmed this finding, but also dem-

onstrated the prognostic value of HRV in this popu-lation.6–15 For example, the Autonomic Tone andReflexes After Myocardial Infarction trial investiga-tors prospectively studied the prognostic values ofSDNN and LVEF in 1,284 postinfarction patients andfound that depressed SDNN carried a similar mortalityrelative risk to that of low LVEF: 3.2 (p � 0.005) and3.9 (p � 0.0003), respectively.16

Our current understanding of the prognostic valueof SDNN in chronic heart failure is based mostly onthe 4 prospective studies shown in Table 3. Three ofthe studies enrolled patients with mostly ischemiccardiomyopathy, whereas 1 study enrolled patientswith idiopathic dilated cardiomyopathy. The degree ofheart failure varied considerably among these studies.The patient populations studied by Brouwer et al17

and Szabo et al18 were most similar to ours in theproportion of patients with ischemic disease and theseverity of heart failure. Brouwer et al17 found noassociation between SDNN and mortality, whereasSzabo et al18 reported that SDNN predicted cardiacdeath. Nolan et al19 found that SDNN independentlypredicted mortality in a patient population with lesssevere ischemic and nonischemic cardiomyopathy.Furthermore, Szabo et al18 found that LVEF indepen-dently predicted mortality, whereas the other inves-tigators17–19 did not. Given these conflicting re-ports, our finding that SDNN, but not LVEF, indepen-dently predicted mortality increases our understandingof the relative roles of SDNN and LVEF for prognosisin patients with moderate to severe heart failure ofpredominantly ischemic origin.

Perhaps the most important finding of the presentstudy was the statistically significant association be-tween sudden cardiac death and depressed SDNN inpatients with chronic heart failure. Although this as-sociation reached statistical significance only with

FIGURE 3. Sudden death in patientsin the lowest quartile of SDNN ver-sus the remainder.

HEART FAILURE/HRV AND CONGESTIVE HEART FAILURE 27

univariate analysis, multivariate analysis demon-strated a trend with a relative risk of 2.4 for patientswith SDNN in the lowest quartile. The 3 studies ofpatients with predominantly ischemic cardiomyopathyfailed to show any predictive value of SDNN forsudden death,17–19 although Brouwer et al17 did find asignificant association between Poincare plots andsudden death. Although a significant association be-tween SDNN and sudden death was found by Fauchieret al,20 their study enrolled only patients with idio-pathic dilated cardiomyopathy. Several studies of pa-tients in the early postinfarction period have foundHRV to be a useful predictor of arrhythmic events post-infarction.4,7,8,11,14

In summary, our findings demonstrate that SDNNhas a strong and independent association with mortal-ity in a patient population with moderate to severeheart failure of predominantly ischemic origin, as wellas a significant association with sudden death.

Acknowledgment: We thank Steven Schulman,MD, and David Kass, MD, for their many suggestionsin the writing of this manuscript.

1. Singh SN, Fletcher RD, Fisher SG, Singh BN, Lewis HD, Deedwania PC,Massie BM, Colling C, Lazzeri D. Amiodarone in patients with congestive heartfailure and asymptomatic ventricular arrhythmia. Survival Trial of Antiarrhyth-mic Therapy in Congestive Heart Failure. N Engl J Med 1995;333:77–82.2. Singh S, Fletcher RD, Fisher S, Deedwania P, Lewis D, Massie B, Singh B,Colling CL. Congestive heart failure: survival trial of antiarrhythmic therapy(CHF STAT). The CHF STAT Investigators. Control Clin Trials 1992;13:339–350.3. Bigger JT Jr, Fleiss JL, Rolnitzky LM, Steinman RC. Frequency domainmeasures of heart period variability to assess risk late after myocardial infarction.J Am Coll Cardiol 1993;21:729–736.4. Bigger JT Jr, Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger RE, Rottman JN.Frequency domain measures of heart period variability and mortality after myo-cardial infarction. Circulation 1992;85:164–171.5. Risk stratification and survival after myocardial infarction. N Engl J Med1983;309:331–336.6. Kleiger RE, Miller JP, Bigger JT Jr, Moss AJ. Decreased heart rate variabilityand its association with increased mortality after acute myocardial infarction.Am J Cardiol 1987;59:256–262.7. Farrell TG, Bashir Y, Cripps T, Malik M, Poloniecki J, Bennett ED, Ward DE,

Camm AJ. Risk stratification for arrhythmic events in postinfarction patientsbased on heart rate variability, ambulatory electrocardiographic variables and thesignal-averaged electrocardiogram. J Am Coll Cardiol 1991;18:687–697.8. Odemuyiwa O, Malik M, Farrell T, Bashir Y, Poloniecki J, Camm J. Com-parison of the predictive characteristics of heart rate variability index and leftventricular ejection fraction for all-cause mortality, arrhythmic events and suddendeath after acute myocardial infarction. Am J Cardiol 1991;68:434–439.9. Copie X, Hnatkova K, Staunton A, Fei L, Camm AJ, Malik M. Predictivepower of increased heart rate versus depressed left ventricular ejection fractionand heart rate variability for risk stratification after myocardial infarction. Resultsof a two-year follow-up study. J Am Coll Cardiol 1996;27:270–276.10. Vaishnav S, Stevenson R, Marchant B, Lagi K, Ranjadayalan K, Timmis AD.Relation between heart rate variability early after acute myocardial infarction andlong-term mortality. Am J Cardiol 1994;73:653–657.11. Cripps TR, Malik M, Farrell TG, Camm AJ. Prognostic value of reducedheart rate variability after myocardial infarction: clinical evaluation of a newanalysis method. Br Heart J 1991;65:14–19.12. Zuanetti G, Neilson JM, Latini R, Santoro E, Maggioni AP, Ewing DJ.Prognostic significance of heart rate variability in post-myocardial infarctionpatients in the fibrinolytic era. The GISSI-2 results. Gruppo Italiano per lo Studiodella Sopravvivenza nell’ Infarto Miocardico. Circulation 1996;94:432–436.13. Quintana M, Storck N, Lindblad LE, Lindvall K, Ericson M. Heart ratevariability as a means of assessing prognosis after acute myocardial infarction. A3-year follow-up study. Eur Heart J 1997;18:789–797.14. Lanza GA, Guido V, Galeazzi MM, Mustilli M, Natali R, Ierardi C, Milici C,Burzotta F, Pasceri V, Tomassini F, Lupi A, Maseri A. Prognostic role of heartrate variability in patients with a recent acute myocardial infarction. Am J Cardiol1998;82:1323–1328.15. Odemuyiwa O, Poloniecki J, Malik M, Farrell T, Xia R, Staunton A,Kulakowski P, Ward D, Camm J. Temporal influences on the prediction ofpostinfarction mortality by heart rate variability: a comparison with the leftventricular ejection fraction. Br Heart J 1994;71:521–527.16. La Rovere MT, Bigger JT Jr, Marcus FI, Mortara A, Schwartz PJ. Baroreflexsensitivity and heart-rate variability in prediction of total cardiac mortality aftermyocardial infarction. ATRAMI (Autonomic Tone and Reflexes After Myocar-dial Infarction) Investigators. Lancet 1998;351:478–484.17. Brouwer J, van Veldhuisen DJ, Man in ‘t Veld AJ, Haaksma J, Dijk WA,Visser KR, Boomsma F, Dunselman PH. Prognostic value of heart rate variabilityduring long-term follow-up in patients with mild to moderate heart failure. TheDutch Ibopamine Multicenter Trial Study Group. J Am Coll Cardiol 1996;28:1183–1189.18. Szabo BM, van Veldhuisen DJ, van der Veer N, Brouwer J, De Graeff PA,Crijns HJ. Prognostic value of heart rate variability in chronic congestive heartfailure secondary to idiopathic or ischemic dilated cardiomyopathy. Am J Cardiol1997;79:978–980.19. Nolan J, Batin PD, Andrews R, Lindsay SJ, Brooksby P, Mullen M, Baig W,Flapan AD, Cowley A, Prescott RJ, Neilson JM, Fox KA. Prospective study ofheart rate variability and mortality in chronic heart failure: results of the UnitedKingdom heart failure evaluation and assessment of risk trial (UK-heart). Circu-lation 1998;98:1510–1516.20. Fauchier L, Babuty D, Cosnay P, Fauchier JP. Prognostic value of heart ratevariability for sudden death and major arrhythmic events in patients with idio-pathic dilated cardiomyopathy. J Am Coll Cardiol 1999;33:1203–1207.

TABLE 3 Previous Studies Investigating Predictive Value of SDNN in Heart Failure

Authors PatientsSDNN:

MortalitySDNN:

Sudden DeathLVEF: Mortality

or Sudden Death

Brouwer et al17 (1996) EF 0.29 � 0.0975% ischemic CM(n � 95)

No* No* No

Szabo et al18 (1997) EF 0.27 (SD omitted)75% ischemic CM(n � 129)

Yes (cardiacdeath)

No Yes (cardiacdeath)

Nolan et al19 (UK-Heart) (1998) EF 0.41 � 0.1776% ischemic CM(n � 433)

Yes No No

Fauchier et al20 (1999) EF 0.34 � 0.12100% idiopathicdilated CM(n � 116)

Yes Yes No

*Poincare plots were predictive of mortality and sudden death in Brouwer et al.CM � cardiomyopathy.

28 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 90 JULY 1, 2002