Predicting mortality in patients with advanced heart failure

Cardiology Review® Online, February 2004, Volume 21, Issue 2

From the Department of Cardiopulmonary Transplantation, Texas Heart Institute at St. Luke’s Episcopal Hospital, Houston, Texas

Heart failure is becoming increasingly common and is associated with high morbidity and mortality. Improvements in medical therapy have greatly improved survival, but many patients with heart failure still become progressively sicker. For these patients, the only widely available and definitive treatment is cardiac transplantation. The survival benefit from transplantation, however, is conferred only on the sickest patients and is limited by the shortage of available donor organs.1

Newer therapeutic options, such as biventricular pacemakers, implantable cardioverter-defibrillators (ICDs), and left ventricular assist devices, have entered clinical use with the aim of improving outcomes.2 Such therapies are expensive, however, and their widespread application in

inappropriate patients with heart failure will likely contribute to spiral-

ing health care costs. Better ways of properly identifying those patients

who would benefit the most from heart transplantation or device therapy are needed.

Clinical risk stratification and decision making in heart failure remain relatively difficult and highlight the need for newer, widely available measures for risk stratification. Several risk-stratification models have been proposed, but many incorporate measures that have predictive value only when used alone, not when used in combination with other variables.3-5 Of the few multivariate models investigated so far, the Heart Failure Survival Score has been most useful for predicting the outcome of patients on transplant waiting lists.1

Corrected QT (QTc) interval duration on the surface electrocardiogram (ECG) has been shown to predict mortality in various clinical settings,6 although its prognostic value in heart failure is not clear. Plasma B-type natriuretic peptide (BNP) level may predict outcome in patients with heart failure, and it may be cost-effective as well. Increased BNP levels are associated with functional deterioration, mortality, and sudden death in heart failure patients.7 We therefore examined the combined prognostic utility of prolonged QTc interval and elevated BNP levels in predicting survival in patients with advanced heart failure.

Patients and methods

We reviewed the records of patients referred to our heart failure clinic from May 1, 2001 to December 1, 2001. We selected patients with BNP levels exceeding 400 pg/mL who had been classified as having New York Heart Association (NYHA) functional class III or IV heart failure for at least 2 months before referral. We excluded patients with pacemakers or ICDs and patients taking class III antiarrhythmic drugs. At evaluation, plasma BNP levels were measured, and standard 12-lead ECGs were recorded. QT interval duration was determined by averaging three consecutive beats recorded through leads II and V4 and corrected using the Bazett formula. After 6 months of follow-up, the rates of all-cause death (primary end point) and cardiac death, sudden cardiac death, pump failure death, and left ventricular assist device implantation (secondary end points) were determined.

Differences between the groups were analyzed by 1-factor analysis of variance, and categorical variables were compared using a chi-square test. Univariate and multivariate stepwise Cox proportional hazard regression analyses were done to identify independent predictors of the primary and secondary end points. A P value of .05 was considered significant. Survival for patients with prolonged versus normal QTc intervals was compared using the Kaplan-Meier method.


Of 371 eligible patients, 130 were excluded. We were able to determine baseline characteristics for the remaining 241 patients (table 1). Forty-six of these patients died, nine underwent heart transplantation, and 17 underwent implantation of a left ventricular assist device. Of the 46 deaths, 24 were attributed to pump failure, 18 to sudden cardiac death, and four to noncardiac causes.

A prolonged QTc interval was defined as one longer than 440 milliseconds. The mean QTc interval duration was significantly longer

in patients who died of pump fail-ure (491 ± 31 milliseconds; P < .001) or sudden cardiac death (480 ± 42 milliseconds; P < .001) than it was in survivors (438 ± 43 milliseconds). The QTc interval was prolonged

in 122 patients (51%), and it was

normal in 119 (49%). BNP levels were similar in patients with prolonged versus normal QTc inter-

vals (786 ± 321 pg/mL versus 733 ± 274 pg/mL).

At 6 months, all-cause mortality, our primary end point, was significantly higher (P < .001) for patients with prolonged versus normal QTc intervals. The same was true of pump failure death (P < .001) and sudden death (P < .001). In left ventricular assist device recipients, the QTc in-terval before device implantation was prolonged in 14 patients (82%) and normal in three (8%; P < .004).

The multivariate analysis revealed that the combination of prolonged QTc interval and elevated BNP level predicted all-cause, cardiac, and pump failure death; but only prolonged QTc interval predicted sudden cardiac death and the need for left ventricular assist device implantation (table 2). Survival was three times higher in patients who had normal compared with prolonged QTc intervals despite similar BNP levels (P < .001; figure).


Our results suggest that, by itself, prolonged QTc interval in heart failure patients is a strong predictor of pump failure death and sudden cardiac death. In combination with elevated BNP levels (> 400 pg/mL), it is an adverse prognostic sign.

The QTc interval is affected by cardiac and noncardiac factors. On the one hand, the QTc interval on

the surface ECG, which reflects the time between the initial fast depolarization and repolarization of the left ventricle, is highly dependent

on T-wave morphology.8 In patients with heart disease, ventricular re-polarization becomes increasingly heterogeneous and may lead to prolonged QTc intervals.9 On the other hand, QTc interval is also affected by noncardiac stimuli, such as inflammation and the changes in autonomic nervous tone that are often seen in advanced heart failure.10

Although these factors may limit the value of the QTc interval when analyzing the electrophysiological properties of the ventricular myocardium, it also suggests that the QTc interval may be combined with other measures in assessing heart failure.11 For instance, peak exercise oxygen uptake (VO2max), which predicts survival in patients with advanced heart failure, also correlates inversely with QTc interval duration in patients awaiting transplantation.12,13 Therefore, a prolonged QTc interval may be a marker not only of ventricular repolarization instability but also of advanced disease in heart failure patients.

Our results also suggest that plasma BNP elevation is a strong in-dependent predictor of all-cause, cardiac, and pump failure death,

but not of sudden cardiac death, in patients with heart failure. This is corroborated by other recent studies. Harrison and colleagues found a correlation between high BNP lev-els (> 480 pg/mL) and increased

risk of heart failure admission or death and worsening prognosis.14 Anand and colleagues found that changes in BNP levels over time corresponded with changes in mortality and morbidity.15

Conversely, a recent study of heart failure patients with a left ventricular ejection fraction of less than 35% by Berger and colleagues suggested that high BNP levels (> 130 pg/mL) do increase the risk of sudden death.16 Although this finding contradicts ours, the discrepancy may be explained by the fact that our patient cohort included only patients with BNP levels above 400 pg/mL. Thus, the patients in our study would have had a mean BNP level much greater than the cutoff point used in the study by Berger and colleagues.


QTc interval prolongation was the only independent predictor of the need for left ventricular assist device implantation in our patient cohort. QTc duration appears to decrease after left ventricular assist device implantation.17 Therefore, the combination of prolonged QTc in-terval and elevated BNP levels in patients with heart failure may be

a useful tool for identifying those patients who would benefit most from left ventricular assist devices and other advanced device-based therapies. In this relatively simple way, the clinician should be able to use limited health care resources more effectively while providing necessary and proven therapies to those who need them most.