Influence of Nonfatal Hospitalization for Heart Failure on Subsequent Mortality in Patients With Chronic Heart Failure
Background— Patients with chronic heart failure (HF) are at increased risk of both fatal and nonfatal major adverse cardiovascular events. We used data from the Candesartan in Heart failure: Assessment of Reduction in Mortality and morbidity (CHARM) trials to assess the influence of nonfatal hospitalizations for HF on subsequent mortality rates in a broad spectrum of HF patients.
Methods and Results— In the present study, 7599 patients with New York Heart Association class II to IV HF and reduced or preserved left ventricular ejection fraction were randomized to placebo or candesartan. We assessed the risk of death after discharge from a first hospitalization for HF using time-updated Cox proportional-hazards models on 7572 patients for whom discharge data were available. Of 7572 patients, 1455 (19%) had at least 1 HF hospitalization, and 586 of 1819 deaths occurred after discharge from an HF hospitalization. The mortality rate was increased after HF hospitalizations, even after adjustment for baseline predictors of death (hazard ratio, 3.15; 95% confidence interval, 2.83 to 3.50). Longer duration of HF hospitalization enhanced the risk of dying, as did repeat HF hospitalizations. Moreover, risk of death was highest within a month of discharge and then declined progressively over time, particularly for death resulting from HF progression and for sudden cardiac death. We observed a similar pattern of risk associated with all-cause hospitalization, although the magnitude was less than that with HF hospitalization.
Conclusions— In patients with chronic HF, the risk of death is greatest in the early period after discharge after a hospitalization for HF and is directly related to the duration and frequency of HF hospitalizations. These findings suggest a role for increased surveillance in the early postdischarge period of greatest vulnerability after an HF admission.
Received February 15, 2007; accepted July 10, 2007.
Patients with chronic heart failure (HF) are at increased risk of both fatal and nonfatal major adverse cardiovascular events.1,2 Nonfatal events, including hospitalization for HF, myocardial infarction, and stroke, contribute to increased cost3 and reduced quality of life4 and portend increased risk of both additional nonfatal events and death.5,6 Although common, distressing, and expensive, symptomatic worsening of HF leading to hospital admission usually is not associated with a high inpatient mortality rate, but it is associated with an increased risk of subsequent death, although this relationship has not been examined in detail. A better understanding of the relationship between HF hospitalization and postdischarge death in patients with HF is important for a number of reasons. First, there are implications for clinical care and postdischarge surveillance. Second, a better understanding of this relationship may allow improvement of prognostic models. These, in turn, may aid targeting of expensive monitoring technologies and treatments. Clarification of the prognostic importance of HF hospitalization also is important from the perspective of clinical trials in which the use of this nonfatal outcome as a component of composite mortality-morbidity end points has been controversial.7
Clinical Perspective p 1487
The Candesartan in Heart failure: Reduction in Mortality and morbidity (CHARM) Program studied the effect of candesartan compared with placebo in a broad spectrum of patients with symptomatic HF and either reduced or preserved left ventricular ejection fraction (LVEF).8–11 We used data from the CHARM trials to assess the influence of a nonfatal hospitalization for HF on subsequent death in this broad spectrum of HF patients.
Patients: The CHARM Program
The CHARM program consisted of 3 independent, concurrently performed trials in which 7599 patients with New York Heart Association class II to IV chronic HF of ≥4 weeks’ duration were randomized to candesartan (target dose, 32 mg once daily) or matching placebo added to conventional background treatments. Patients were enrolled in the individual CHARM trials according to LVEF and baseline treatment with an angiotensin-converting enzyme (ACE) inhibitor. Patients with an LVEF ≤40% who were not receiving an ACE inhibitor because of intolerance were enrolled in CHARM-Alternative (n=2028). Patients who had an LVEF ≤40% and who were taking an ACE inhibitor were enrolled in CHARM-Added (n=2548). Patients with an LVEF >40% (with or without concomitant ACE inhibitor use) were randomized into CHARM-Preserved (n=3023). Overall, follow-up ranged from a median of 41 months in CHARM-Added and 37 months in CHARM-Preserved to 34 months in CHARM-Alternative (38 months in the overall CHARM Program). The primary outcome in the individual component trials of the CHARM program was the composite of cardiovascular death or hospital admission with worsening HF analyzed by time to first event, and the primary outcome of the overall CHARM Program was all-cause mortality. The design, baseline findings, and primary results of the CHARM program have been reported in detail.8–12
End Point Assessment
All first nonfatal major cardiovascular end points, including admission for HF, myocardial infarction, resuscitated sudden death, or stroke, were adjudicated, and all deaths were reviewed by an independent clinical events committee. HF hospitalization required that a patient had typical symptoms and signs, treatment with an intravenous diuretic, and at least an overnight hospital stay. All deaths were classified as cardiovascular or noncardiovascular; cardiovascular deaths were further subclassified as secondary to sudden death, progressive HF death, myocardial infarction, or other.
First hospitalizations for HF were related to subsequent risk of death by the use of time-updated Cox proportional-hazards models.13 The time-updated variable for HF hospitalization comes into play at hospital discharge. Thus, HF hospitalizations that led directly to death (ie, the patient never left the hospital) were not attributed to the subsequent mortality risk. The 27 patients for whom the discharge date was missing were excluded, leaving 7572 patients for the present analysis. Only a first hospitalization for HF was considered for the initial analysis. Both unadjusted and covariate-adjusted hazard ratios are reported, with adjustment for those baseline predictors of all-cause mortality identified in previous analyses.6 We further explored the relationship between duration of hospital stay and risk of death after discharge, the relationship between hospitalization for HF and specific causes of death, and the effect of multiple HF hospitalizations on subsequent mortality rates. For the analysis of multiple hospitalizations, only those hospitalizations that investigators had identified as resulting from worsening HF were included, and the sample size was reduced to 7549 patients because of missing data. Death rates by both length of hospital stay and time since HF hospitalization were obtained with a Poisson survival model with covariate adjustment. To determine the prognostic significance of any hospitalization in this population, we also assessed the relationship between all first hospitalizations for any cause and subsequent mortality rates.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Of the 7572 patients included in this analysis, 1455 had at least 1 hospitalization for HF and were subsequently discharged from hospital. Of these, 869 patients survived to the end of the trials, and 586 patients died during the follow-up period. Regardless of whether patients had an LVEF ≤40% or >40%, patients with a hospitalization for HF were older, were more likely to be diabetic, had worse baseline New York Heart Association class, were more likely to have been hospitalized for HF in the 6 months before randomization, and were less likely to have been randomized to candesartan (Table 1). Of the patients who were hospitalized for HF, those who subsequently died during follow-up were older and were more likely to have cardiomegaly, to be underweight, and to have been hospitalized for HF in the 6 months before randomization, also regardless of whether LVEF was ≤40% or >40%.
Of the 1819 overall deaths in the trial, 586 occurred after hospital discharge for a first HF hospitalization. The estimated crude hazard for all-cause mortality after discharge following a first hospitalization for HF (Table 2) was 4.55 times that of patients never hospitalized for HF or yet to be hospitalized and discharged (95% confidence interval [CI], 4.11 to 5.03) and remained elevated after adjustment for baseline predictors of mortality (hazard ratio, 3.15; 95% CI, 2.83 to 3.50).
The mortality risk after a hospitalization for HF subsequently declined over time (Figure 1), with an estimated 6-fold excess risk in the first month after discharge falling to a doubling of risk after 2 years after discharge compared with those not hospitalized for HF. The risk of dying also was related to the length of HF hospitalization, with long HF hospitalizations (≥22 days) carrying more than double the mortality risk of short HF hospitalizations (≤7 days) (Figure 2). Both time from HF hospitalization and duration of HF hospitalization were predictive of death independently of each other (Figure 3).
Patients who died after a hospitalization for HF were at the greatest increased risk of dying of progressive HF, followed by sudden death (Figure 4). For each cause of death, the risk of death decreased with time from discharge, an effect that was most marked for death resulting from progressive HF. The relationships between time from discharge and subsequent mortality outcomes and between duration of HF hospitalization and mortality outcomes were similar in patients with LVEF >40% and ≤40% and in patients from different geographic regions, with no interaction observed on the basis of LVEF >40% or ≤40% or geography.
The risk of death increased with each additional HF hospitalization, with a nearly 30% cumulative incremental risk associated with discharge from a second or third HF hospitalization (Figure 5). However, ≥4 HF hospitalizations carried no additional risk. Duration of longest HF hospitalization, time from most recent discharge, and number of HF hospitalizations all contributed substantively to a combined model of mortality based on HF hospitalizations. Hospitalization for any cause similarly increased the risk of subsequent death, with a similar declining risk after hospital discharge, although the magnitude of the risk was less compared with hospitalization for HF (Table 3).
The CHARM program, which enrolled a broad group of symptomatic HF patients across the full spectrum of LVEF, provides a unique opportunity to assess the influence of an HF hospitalization on subsequent risk of dying after discharge. We found that hospitalization for HF conferred a significantly increased risk of subsequent death. Moreover, this risk was inversely related to the time since discharge, was increased after longer hospital stay, and was further increased after additional HF hospitalizations.
We observed that a number of factors affect the influence of HF hospitalization on subsequent death after discharge, including duration of hospital stay, time from HF hospitalization, and number of HF hospitalizations. Peripheral edema, duration of treatment with intravenous diuretic, renal impairment, concurrent respiratory problems, and factors requiring social intervention have been associated with increased length of stay.14 The thresholds for the decision to hospitalize and its duration can be influenced by HF severity and comorbidity, as well as local practice patterns. We adjusted for the former by including those baseline covariates that most contributed to death in our overall multivariable mortality model.6 That we did not observe any heterogeneity based on geography suggests that these findings are valid even accounting for local practice pattern differences.
The finding that the risk of death was highest in the immediate postdischarge period and declined steeply and exponentially thereafter is in contrast to the relatively stable overall mortality risk over time observed in HF trials in general, including CHARM, and is reminiscent of the initially high mortality risk observed in the post–myocardial infarction setting.15 Indeed, a discrete HF hospitalization event appears to place a patient in a much higher risk category. This was true for patients in CHARM whether they had reduced or preserved LVEF.
That risk is highest in the immediate postdischarge period has potentially important clinical implications. The initially high and subsequently declining risk indicates that patients are particularly vulnerable in the early postdischarge period and require heightened surveillance, especially if the hospital admission has been prolonged or if the patient has had prior admissions for worsening HF. We also observed that the risk of sudden unexpected death was highest in the immediate postdischarge period, a finding that also has been seen in the early post–myocardial infarction period.16 Early intervention at clinic or home visits might help to detect and treat HF patients before decompensation, especially because we found that the greatest increased risk and most likely cause of death after discharge was progressive HF. Similarly, careful monitoring of electrolytes may be important in ameliorating the risk of sudden death, particularly because these patients have generally experienced recent changes in diuretic dose and the addition of other treatments that influence blood chemistry. Indeed, patients identified as being at particularly high risk using these simple measures might represent a selected population in which new but relatively expensive technologies permitting remote monitoring (using either external or internal devices) that have shown promise in clinical trials might play a role.17 Although our findings suggest a relative concordance of hospitalization for HF and mortality risk in the immediate postdischarge period, we cannot say that this would invalidate HF hospitalization as an independent end point in all populations, especially because some therapies might reduce hospitalization without reducing long-term mortality rates.
A number of limitations of this analysis should be noted. Although we had data on length of HF hospitalization and baseline data from randomization on comorbidity, we could not easily incorporate time-updated comorbidity data into our models. It is likely that additional factors beyond HF hospitalization duration and number such as change in New York Heart Association class or weight loss may have contributed to risk and that more detailed knowledge of patient characteristics at the time of HF hospitalization may add to a time-updated predictive model. In addition, we were not able to determine to what extent patients were discharged with the anticipation that they would die soon. Because a number of patients with advanced HF are discharged to hospice or a chronic care facility with the expectation that death is imminent, the increased risk of death observed in the early postdischarge period may be elevated secondary to an overrepresentation of these types of patients. Finally, we cannot exclude potential bias in categorizing cause of death when death occurred shortly after a HF hospitalization.
In summary, we have found that HF hospitalization, its duration, and its frequency are important predictors of increased subsequent death after discharge in symptomatic chronic HF patients with reduced or preserved LVEF. This risk appears to be greatest in the early postdischarge period and declines over time. These findings may help to identify HF patients at greatest risk of dying and suggest a role for increased surveillance in the early postdischarge period.
Source of Funding
The CHARM program was funded by AstraZeneca.
Drs Solomon, Dobson, Pocock, McMurray, Granger, Yusuf, Swedberg, Young, and Pfeffer have received research funding from AstraZeneca. Drs Pocock, McMurray, Granger, Yusuf, Swedberg, Young, and Pfeffer have consulted for or have received honoraria from AstraZeneca. Dr Michelson is an employee of AstraZeneca. Dr Skali reports no disclosures.
Solomon SD, Anavekar N, Skali H, McMurray JJ, Swedberg K, Yusuf S, Granger CB, Michelson EL, Wang D, Pocock S, Pfeffer MA, for the Candesartan in Heart failure Reduction in Mortality and morbidity (CHARM) Investigators. Influence of ejection fraction on cardiovascular outcomes in a broad spectrum of heart failure patients. Circulation. 2005; 112: 3738–3744.
McMurray JJ, Andersson FL, Stewart S, Svensson K, Solal AC, Dietz R, Vanhaecke J, van Veldhuisen DJ, Ostergren J, Granger CB, Yusuf S, Pfeffer MA, Swedberg K. Resource utilization and costs in the Candesartan in Heart failure: Assessment of Reduction in Mortality and morbidity (CHARM) programme. Eur Heart J. 2006; 27: 1447–1458.
Lewis EF, Lamas GA, O’ Meara E, Granger CB, Dunlap ME, McKelvie RS, Probstfield JL, Young JB, Michelson EL, Halling K, Carlsson J, Olofsson B, McMurray JJ, Yusuf S, Swedberg K, Pfeffer MA, for the CHARM Investigators. Characterization of health-related quality of life in heart failure patients with preserved versus low ejection fraction in CHARM. Eur J Heart Fail. 2006; 9: 83–91.
Solomon SD, Wang D, Finn P, Skali H, Zornoff L, McMurray JJ, Swedberg K, Yusuf S, Granger CB, Michelson EL, Pocock S, Pfeffer MA. Effect of candesartan on cause-specific mortality in heart failure patients: the Candesartan in Heart failure: Assessment of Reduction in Mortality and morbidity (CHARM) program. Circulation. 2004; 110: 2180–2183.
Pocock SJ, Wang D, Pfeffer MA, Yusuf S, McMurray JJ, Swedberg KB, Ostergren J, Michelson EL, Pieper KS, Granger CB. Predictors of mortality and morbidity in patients with chronic heart failure. Eur Heart J. 2006; 27: 65–75.
Skali H, Pfeffer MA, Lubsen J, Solomon SD. Variable impact of combining fatal and nonfatal end points in heart failure trials. Circulation. 2006; 114: 2298–2303.
Pfeffer MA, Swedberg K, Granger CB, Held P, McMurray JJ, Michelson EL, Olofsson B, Ostergren J, Yusuf S, Pocock S, for the CHARM Investigators and Committees. Effects of candesartan on mortality and morbidity in patients with chronic heart failure: the CHARM-Overall Programme. Lancet. 2003; 362: 759–766.
Granger CB, McMurray JJ, Yusuf S, Held P, Michelson EL, Olofsson B, Ostergren J, Pfeffer MA, Swedberg K, for the CHARM Investigators and Committees. Effects of candesartan in patients with chronic heart failure and reduced left-ventricular systolic function intolerant to angiotensin-converting-enzyme inhibitors: the CHARM-Alternative Trial. Lancet. 2003; 362: 772–776.
McMurray JJ, Ostergren J, Swedberg K, Granger CB, Held P, Michelson EL, Olofsson B, Yusuf S, Pfeffer MA, for the CHARM Investigators and Committees. Effects of candesartan in patients with chronic heart failure and reduced left-ventricular systolic function taking angiotensin-converting-enzyme inhibitors: the CHARM-Added Trial. Lancet. 2003; 362: 767–771.
Yusuf S, Pfeffer MA, Swedberg K, Granger CB, Held P, McMurray JJ, Michelson EL, Olofsson B, Ostergren J, for the CHARM Investigators and Committees. Effects of candesartan in patients with chronic heart failure and preserved left-ventricular ejection fraction: the CHARM-Preserved Trial. Lancet. 2003; 362: 777–781.
Pfeffer MA, McMurray JJ, Velazquez EJ, Rouleau JL, Kober L, Maggioni AP, Solomon SD, Swedberg K, Van de Werf F, White H, Leimberger JD, Henis M, Edwards S, Zelenkofske S, Sellers MA, Califf RM, for the Valsartan in Acute Myocardial Infarction Trial Investigators. Valsartan, captopril, or both in myocardial infarction complicated by heart failure, left ventricular dysfunction, or both. N Engl J Med. 2003; 349: 1893–1906.
Solomon SD, Zelenkofske S, McMurray JJ, Finn PV, Velazquez E, Ertl G, Harsanyi A, Rouleau JL, Maggioni A, Kober L, White H, Van de Werf F, Pieper K, Califf RM, Pfeffer MA, for the Valsartan in Acute Myocardial Infarction Trial (VALIANT) Investigators. Sudden death in patients with myocardial infarction and left ventricular dysfunction, heart failure, or both. N Engl J Med. 2005; 352: 2581–2588.
Cleland JG, Louis AA, Rigby AS, Janssens U, Balk AH, for the TEN-HMS Investigators. Noninvasive home telemonitoring for patients with heart failure at high risk of recurrent admission and death: the Trans-European Network-Home-Care Management System (TEN-HMS) study. J Am Coll Cardiol. 2005; 45: 1654–1664.
Patients with chronic heart failure are at increased risk for both fatal and nonfatal cardiovascular events, including hospitalization for recurrent episodes of heart failure. We used data from the Candesartan Heart failure: Reduction in Mortality and morbidity (CHARM) program to elucidate the subsequent risk of death associated with a heart failure hospitalization. We found that the mortality rate was highest early after a heart failure hospitalization and declined rapidly thereafter. Moreover, patients who were hospitalized longer or more frequently had a higher subsequent mortality risk. These data suggest that chronic heart failure patients are most vulnerable in the immediate aftermath of a hospital admission and suggest a potential role for increased surveillance during this period.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
Guest Editor for this article was Shigetake Sasayama, MD.