OUP user menu

Impact of hospitalization for acute coronary events on subsequent mortality in patients with chronic heart failure

Putte Abrahamsson, Joanna Dobson, Christopher B. Granger, John J.V. McMurray, Eric L. Michelson, Marc Pfeffer, Stuart Pocock, Scott D. Solomon, Salim Yusuf, Karl Swedberg
DOI: http://dx.doi.org/10.1093/eurheartj/ehn503 338-345 First published online: 11 November 2008


Aims We explored the impact of having a hospital admission for an acute coronary syndrome (ACS) on the subsequent prognosis among patients with chronic heart failure (CHF).

Methods and results A total of 7599 patients with CHF, New York Heart Association Classes II–IV, were randomly assigned to candesartan or placebo. We assessed the risk of death after a first ACS using time-updated Cox proportional hazard models adjusted for baseline predictors. During a mean follow-up of 3.3 years, 1174 patients experienced at least one ACS. Myocardial infarction (MI) was the first ACS in 442 subjects and unstable angina (UA) in 732. After these events, 219 (49.5%) and 167 (22.8%) patients died during follow-up. The early risk of death was more pronounced after MI: 30.2% died within 30 days compared with 3.6% after UA. After an ACS event, the risk of death declined steadily over time, although 18 months after an MI the risk was still twice that of patients without an ACS.

Conclusion Patients with CHF, who develop an ACS, have markedly increased subsequent mortality, particularly in the early phase after an MI.

  • Chronic heart failure
  • Acute myocardial infarction
  • Unstable angina pectoris
  • Prognosis


Heart failure is a major cause of death and hospitalizations in the ageing population of developed countries and is present in 1–2% of the adult population and in 6–10% of the elderly.1,2 Heart failure is the most common cause for hospital admission among elderly patients, and accounts for ∼5% of medicine and geriatric admissions.3,4 The major causes of heart failure in developed countries are hypertension and ischaemic heart disease.5 Patients with heart failure have a poor long-term survival and a high risk of hospitalization which is often recurring. Previous community studies report that 30–40% of patients with chronic heart failure (CHF) die within 1 year from diagnosis and 60–70% within 5 years.2 Hospitalization for worsening of heart failure is associated with a greatly increased risk of subsequent death.6 The risk is highest early after discharge, but diminishes with time. Readmissions or hospitalizations with long duration increase the risk of death even further.7 A similar pattern is observed for patients who have had a myocardial infarction (MI) complicated by heart failure, left ventricular systolic dysfunction or both. The hazard for death is highest in the first 30 days and remains elevated for 6 months after MI. In the longer-term mortality rates varying from 2.5 to 10% per year have been reported.810 In case of a post-MI re-hospitalization of any cause, the subsequent risk of death is increased even further, in particular, early after hospitalization and diminishing with time.11 Patients with CHF deteriorate mainly because of worsening heart failure or sudden cardiac death. However, acute coronary syndromes (ACSs) also contribute to a poorer prognosis in CHF patients, though there has been little evidence to date regarding the relative impact of unstable angina (UA) and acute MI to deterioration of patients with CHF. The present analysis uses the large dataset from the CHARM programme to explore the impact of having an ACS event, i.e. hospitalization for acute MI or UA pectoris, on the subsequent risk of death among patients with CHF, with or without impaired left ventricular systolic function.


The Candesartan in Heart failure: Assessment of Reduction in Mortality and morbidity (CHARM) programme consisted of three independent but related studies in which 7599 patients with New York Heart Association (NYHA) Classes II–IV CHF, were randomly assigned to double blind treatment with candesartan, target dose 32 mg once daily, or matching placebo in addition to conventional heart failure treatment, between March 1999 and March 2001.12 Patients were enrolled in either of the three component studies according to left ventricular ejection fraction (LVEF) and ACE inhibitor treatment at baseline. CHARM-added consisted of patients with LVEF ≤40% who were on treatment with an ACE inhibitor, CHARM-alternative of patients with LVEF ≤40% who were intolerant to ACE inhibitors and CHARM-preserved of patients with LVEF >40%. Major exclusion criteria were serum-creatinine ≥265 mmol/L (≥3 mg/dL), serum potassium ≥5.5 mmol/L (≥5.5 mEq/L), known bilateral renal artery stenosis, symptomatic hypotension, women of childbearing potential, critical aortic or mitral valve stenosis, stroke, MI or open heart surgery within the last 4 weeks.

After the initial titration phase, patients were followed at scheduled visits every 4 months and the follow-up time ranged from a median of 41 months in CHARM-added and 37 months in CHARM-preserved to 34 months in CHARM-alternative. The programme was completed as planned with follow-up concluded in March 2003. The primary endpoint in the individual CHARM-studies was the composite of cardiovascular death or hospital admission for worsening heart failure analysed by time to first event, and the primary endpoint of the overall CHARM programme was all-cause mortality.

In the present analysis of the overall CHARM programme, we examined the impact of experiencing an ACS event, i.e. hospitalization for MI or UA pectoris, on the subsequent risk of death from any cause.

Hospitalization with an ACS was defined as either (i) an admission with MI or UA as the primary cause or (ii) one or other these events occurring during an admission for another cause. Investigator reported diagnoses were used and only the first hospitalization with an ACS after randomization was analysed. Patients with an adjudicated MI who died on the day of the event with no record of hospitalization, were also counted as an ACS hospitalization. MI was defined by the following: creatine kinase or creatine kinase MB concentrations more than twice the upper limit of normal or troponin I or T levels twice the upper limit of normal or three times the upper limit of normal within 24 h after an percutaneous coronary angioplasty, or five times the upper limit of normal within 24 h after coronary artery bypass surgery. In addition to the bio-markers above, diagnosis of an MI also required either electrocardiographic changes in two or more contiguous leads with new Q-waves, ischaemic ST-T changes or left bundle branch block, or typical clinical presentation of MI consisting of characteristic chest pain lasting for >20 min, pulmonary oedema or cardiogenic shock. The definition of UA was not specified in the study protocol and for this type of ACS we used the diagnosis reported by the investigator. The primary endpoint in this analysis was death from any cause.


First hospitalizations for ACS were related to subsequent risk of mortality using time-updated Cox proportional hazards models adjusted for baseline predictors.13 The time-updated variables for MI and UA hospitalizations were initiated on the MI date or hospital admission date respectively. Thus, both hospitalizations that led directly to death (i.e. patient never left the hospital) and those where the patient was eventually discharged from hospital were attributed to the subsequent mortality risk. Explicitly, the Cox models included a time-updated binary covariate for each specified time interval after the MI or UA event (0–1, 2–7, 8–30 days, 31 days–6 months, 6–18 months, and 18+ months). If both MI and UA occurred in the same hospitalization, the event was counted as an MI. Only the first ACS hospitalization was considered for the initial analysis. Both crude death rates and covariate adjusted hazard ratios were reported, with adjustment for the strongest of those baseline predictors of all-cause mortality identified in previous analyses.14 The baseline predictors included were age, ejection fraction, diabetes, body mass index, gender, NYHA class, smoking, cardiomegaly, prior CHF hospitalization, and randomized treatment allocation. We then explored the impact of all ACS hospitalizations and hospitalizations due to worsening heart failure on the subsequent risk of death. For these analyses, if MI, UA or worsening heart failure occurred in the same hospitalization, the hospitalization was attributed to the most severe of these events where MI and UA were considered the most and least severe, respectively. All significance levels were two-sided. Analyses were performed using the statistical software Stata 10.0 (StataCorp, College Station, TX, USA).


Of the 7599 patients in the CHARM programme, 1174 (15.4%) experienced at least one ACS hospitalization during a mean 3.3 years follow-up. The rate of ACS hospitalization was found to be significantly lower in the second 6 months than in the first 6 months after randomization (rate ratio: 0.53, 95% CI 0.47, 0.60, P < 0.001). Beyond 6 months the event rate was linear. A first hospitalization for such an acute event was MI in 442 subjects and UA in 732. After these MI and UA events 219 (49.5%) and 167 (22.8%), respectively, subsequently died during follow-up. Of these deaths after MI and UA, 125 (57.1%) and 18 (10.8%), respectively, occurred during the hospitalization due to the event. Twenty-one percent of all deaths in the CHARM programme were preceded by an ACS event. Table 1 shows the associations of these first acute events to certain key baseline variables, previously identified to be strong predictors of mortality.14 Overall, the acute coronary event rate was 5.9 events per 100 patient years. The event rate increased with increasing age, worse NYHA class and was higher for diabetics compared with non-diabetics, particularly insulin-treated. Event rates were also higher for patients with a prior heart failure hospitalization and for those with a prior MI. The event rate was similar for those with LVEF ≤40% as compared with >40%. As age increased, a greater proportion of the first acute coronary events were MIs. Similarly, the first event was more likely to be an MI for diabetics compared with non-diabetics and for males compared with females. The proportion of patients experiencing an MI, rather than UA, as their first event was greater for current smokers, patients with lower body mass index and patients with LVEF ≤40%.

View this table:
Table 1

First acute coronary events by baseline variables

Baseline variablesNo. of patientsNo. (%) without acute coronary eventNo. (%) with acute coronary eventEvent rate per 100 patient yearsNo. (%) of events which were MI
All patients7599642511745.90442 (38)
Age (years)
 <6021851922 (30)263 (22)4.2665 (25)
 60–6923952015 (31)380 (32)5.97138 (36)
 70–7924091999 (31)410 (35)6.88183 (45)
 80+610489 (8)121 (10)8.6156 (46)
LVEF (%)
 ≤4048044085 (64)719 (61)5.77290 (40)
 >4027952340 (36)455 (39)6.12152 (33)
 Insulin-treated707530 (8)177 (15)10.8481 (46)
 Other14561216 (19)240 (20)6.5596 (40)
 No54364679 (73)757 (64)5.18265 (35)
Body mass index (kg/m2)
 <22.5889749 (12)140 (12)6.5155 (39)
 22.5–24.912721082 (17)190 (16)5.8188 (46)
 25–29.930682580 (40)488 (42)6.07188 (39)
 30–34.915791338 (21)241 (21)5.5783 (34)
 35+791676 (11)115 (10)5.4328 (24)
 Male51994399 (68)800 (68)5.86317 (40)
 Female24002026 (32)374 (32)5.98125 (33)
NYHA class
 II34162966 (46)450 (38)4.83167 (37)
 III39853294 (51)691 (59)6.78266 (38)
 IV198165 (3)33 (3)8.199 (27)
Current smoker
 Yes1114969 (15)145 (12)4.9068 (47)
 No64855456 (85)1029 (88)6.07374 (36)
 Yes16671406 (22)261 (22)6.3096 (37)
 No59325019 (78)913 (78)5.79346 (38)
Previous HF hospitalization
 Within 6 months27952329 (36)466 (40)6.60171 (37)
 >6 months26312260 (35)371 (32)5.30150 (40)
 No21731836 (29)337 (29)5.76121 (36)
Allocated treatment
 Candesartan38033226 (50)577 (49)5.70217 (38)
 Placebo37963199 (50)597 (51)6.10225 (38)
Previous MI
 Yes40043168 (49)836 (71)8.41318 (38)
 No35953257 (51)338 (29)3.39124 (37)
Previous angina pectorisa
 Yes43523424 (53)928 (79)8.51314 (34)
 No32473001 (47)246 (21)2.73128 (52)
  • aPast or present angina pectoris.

Hospitalization for acute coronory syndrome: unadjusted mortality analysis

Figure 1 shows the subsequent mortality risk over time following MI and UA events. The much higher immediate death rate within the first week after MI was clear. For both MI and UA, the death rate steadily declined over time. At all time intervals, the death rate after MI was substantially higher than the death rates after UA.

Figure 1

Mortality incidence following myocardial infarction and unstable angina events.

Even over 18 months after MI, the death rate was still double that of patients who had not experienced an ACS event. However, 18 months after UA, the mortality rate only slightly (non-significantly) exceeded that for event-free patients.

Hospitalization for acute coronory syndrome: adjusted mortality analysis

Since some baseline variables such as age affect both risk of acute coronary events and risk of death it was relevant to undertake analyses of mortality risk after acute coronary events which adjusted for baseline variables. Thus, Table 2 and Figure 2 show the results from Cox models with MI or UA events as time-updated covariates, and baseline factors as fixed covariates. These adjusted analyses show broadly similar patterns of increased risk to the unadjusted analyses in Figure 1. Note the baseline variables all remain highly significant predictors of mortality. The slightly reduced mortality hazard on candesartan vs. placebo was similar to that previously published.12

Figure 2

Hazard ratios for mortality (with 95% confidential interval) by (a) time since myocardial infarction and (b) time since unstable angina adjusted for baseline predictors.

View this table:
Table 2

Mortality risk by (a) time since myocardial infarction and (b) time since unstable angina using Cox models with time-updated covariates for the event and fixed covariates for baseline predictors

VariablesHazard ratio (95% CI)P-value
Time-updated variables
 Never had MI or UA event or before first MI or UA event1.00
After first MI or UA event
  (a) MI
  0–1 days309.60 (239.37, 400.43)<0.001
  2–7 days63.95 (45.72, 89.45)<0.001
  8–30 days16.39 (11.39, 23.58)<0.001
  31 days–6 months2.61 (1.78, 3.83)<0.001
  6–18 months2.32 (1.69, 3.20)<0.001
  18+ months1.96 (1.26, 3.06)0.003
 (b) UA
  0–1 days13.28 (4.97, 35.51)<0.001
  2–7 days9.87 (5.12, 19.03)<0.001
  8–30 days3.78 (2.19, 6.53)<0.001
  31 days–6 months2.43 (1.83, 3.24)<0.001
  6–18 months1.36 (1.04, 1.80)0.027
  18+ months1.21 (0.88, 1.68)0.241
Baseline variables
 Age (per 10 years over age 60)1.73 (1.62, 1.84)<0.001
 Ejection fraction (per 5% decrease below 45)1.17 (1.14, 1.20)<0.001
 Diabetes: insulin-treated1.80 (1.56, 2.08)<0.001
 Diabetes: other1.49 (1.33, 1.67)<0.001
 BMI (per 1 kg/m2 decrease below 27.5)1.07 (1.06, 1.09)<0.001
 Female0.76 (0.69, 0.85)<0.001
 NYHA Class III1.45 (1.31, 1.61)<0.001
 NYHA Class IV2.52 (2.03, 3.12)<0.001
 Current smoker1.31 (1.15, 1.49)<0.001
 Cardiomegaly1.36 (1.23, 1.51)<0.001
 Prior CHF hosp within 6 months1.44 (1.27, 1.63)<0.001
 Prior CHF hosp but not within 6 months1.17 (1.03, 1.33)0.017
 Candesartan0.92 (0.84, 1.01)0.093

Hospitalization for heart failure or acute coronory syndrome

We also examined any hospitalizations for an ACS or worsening heart failure and how those, collectively, influenced the risk of death. The number of hospitalizations for MI, UA, and heart failure were 554, 999, and 3839 in 478, 654, and 1917 patients, respectively. Thus, hospitalizations for heart failure were much more common than for UA or MI. Of the 478 and 654 patients with at least one hospitalization for MI or UA, 157 (33%) and 203 (31%) also had at least one hospitalization for heart failure.

Table 3 models how mortality risk was affected by each hospitalization for MI, UA, or worsening heart failure and also by baseline predictors. After an ACS event, we found that the most recent event predicted prognosis with no additional risk after repeated ACS events. However, after hospitalization for worsening heart failure, additional hospitalizations accumulated additional mortality risk. With a second or third re-hospitalization for worsening heart failure, risks were increased by 46 and 94%, respectively. The risk of death after hospitalization for heart failure was much higher during the first 30 days, which then declined but remained significantly higher compared with those patients not hospitalized. Table 3 also demonstrates that the subsequent mortality risk after hospitalization for MI was much higher than for admission for UA, and notably, also for worsening heart failure.

View this table:
Table 3

Time-updated Cox model for mortality hazard after multiple hospitalizations for myocardial infarction, unstable angina or worsening heart failure, adjusted for baseline predictors

Hospitalization for myocardial infarctionHazard ratio95% Confidence interval
Time after most recent MI event
 0–1 days261.07204.20, 333.76
 2–7 days56.4541.13, 77.48
 8–30 days11.468.04, 16.33
 31 days–6 months1.681.16, 2.44
 6–18 months1.601.14, 2.23
 18+ months1.290.79, 2.13
Hospitalization for unstable angina
 Time after most recent unstable angina event
  0–1 days9.703.98, 23.63
  2–7 days5.032.38, 10.61
  8–30 days1.850.96, 3.56
  31 days–6 months1.210.88, 1.66
  6–18 months0.860.64, 1.17
  18+ months0.880.59, 1.32
Hospitalization for worsening heart failure
 Time after most recent heart failure event
  0–1 days14.149.27, 21.59
  2–7 days19.8715.82, 24.96
  8–30 days12.2710.33, 14.56
  31 days–6 months3.513.00, 4.10
  6–18 months1.951.64, 2.33
  18+ months1.721.34, 2.21
  • Extra multiplier on hazard if this is a second hospitalization for worsening heart failure: 1.46, this is a third (or more) hospitalization for worsening heart failure: 1.94.


The present analysis showed that hospitalization for an ACS event in patients with CHF resulted in a very high risk of subsequent death. This added risk was striking, even after adjustment for key baseline predictors of risk. The risk was highest in the early phase but declined with time. The early risk was most pronounced for patients who had an MI, with a 300-fold risk-increase during the first day, a 64-fold increase during the first week and a 16-fold increase during the first month. Even though risk declined with time, the patients who had an MI were still at twice the risk of death after 18 months. In comparison, for patients with UA, there was a 13-fold risk-increase during the first day, declining to the risk of the overall CHF population within 18 months.

The CHARM programme comprised a broad spectrum of patients with CHF, with or without left ventricular systolic dysfunction, with an average age of 66 years, two-thirds of whom were male and mainly with NYHA Classes II–III.12 The most common causes of heart failure in developed countries are ischaemic heart disease and hypertension5 which was also reflected in the CHARM programme where 52.6% of the patients had a previous MI, 23.8% had current angina pectoris and 55% had a history of hypertension.12 Therefore, the CHARM population reflects a CHF population in developed countries and thus gives clinically valuable and relevant information for routine heart failure care.

We have previously reported that patients in the CHARM programme, who were hospitalized for worsening of heart failure, had a substantially increased mortality risk.7 Acute coronary events accounted for a minor part of all hospitalizations in the CHARM trials.12 As the CHARM programme included a broad range of patients with CHF, we found that the risk of developing ACS events was similar for patients with LVEF ≤40 and >40%. We have previously reported that the risk of death declined with increasing LVEF until an ejection fraction of ∼45%. This trend was also observed for cardiovascular death and the components of cardiovascular death, with the exception of fatal stroke.15

The EuroHeart Failure Survey II explored the characteristics of hospitalized acute heart failure patients. ACS was reported to be the precipitating factor in 23% of patients with acute decompensated CHF.16 However, previous reports on the impact of acute coronary events on prognosis among patients with CHF are scarce. One possible reason is that non-fatal MI in CHF is not a common outcome. In CHARM, ∼6.3% of randomized patients had a subsequent fatal or non-fatal MI during follow-up which suggests that a very large study is needed to analyse the risk of this added complication. In CHARM, 1174 patients were hospitalized for a first ACS event with 386 subsequent deaths that allow a more robust analysis of this complication. The present findings are in agreement with the SOLVD trials in patients with CHF or left ventricular dysfunction ∼10 years prior to the CHARM programme.17 In SOLVD, the 650 and 1094 patients who developed an MI and an episode of UA had a much higher early subsequent risk of death compared with patients who did not develop such events (1 year mortality 55.4 vs. 7.3% and 14.6 vs. 9.0%, respectively). Twenty-five percent of all deaths in the SOLVD trials were preceded by an MI and 21% of all deaths in the CHARM programme were preceded by hospitalization for an ACS event.

Heart failure complicating acute MI has also been reported to increase the risk. From the GRACE registry, it was concluded that patients with an acute MI, who presented with heart failure on admission, were at a three- to four-fold increased risk of death during the first 6 months as compared with MI patients without heart failure. The risk was even higher for patients who developed heart failure during hospitalization.18 Thus, a similar pattern of increased early hazard after an ACS event with heart failure was seen in both the present study and in the GRACE registry; however, the mortality rate was much higher in the present study. The cumulative mortality after 18 months was ∼50% for MI patients and ∼20% patients with UA in our study as compared with ∼20% for MI patients complicated with heart failure in the GRACE registry.


The first ACS event, either as the primary cause of hospitalization or as a major event during hospitalization as reported by the investigator, was considered as the index event in the present analysis. An adjudication committee adjudicated MIs leading to death or hospitalization. Thus, only a minority of all ACS events in the present analysis was adjudicated and hospitalizations for UA, in particular, were included as investigator reported.

The length of hospital stay may be of importance for prognosis but was not analysed.

Even though the CHARM population seems to reflect the overall heart failure population in developed countries, a selection bias cannot be ruled out, and the relevance to routine heart failure care may therefore be questioned.


Worsening of heart failure is the most common cause for hospitalization among patients with CHF and is associated with a subsequent increased risk of death. However, in those patients with CHF who are hospitalized for an MI or an episode of UA, the present study shows that the subsequent mortality is markedly increased, in particular, in the early phase and after an MI. Mortality risk declines for the most part within 6 months. The risk among these patients is much higher than reported from previous studies focused on MI. Thus, serious therapeutic re-assessment should be done early for patients with CHF who have an acute coronary event.


The CHARM programme was sponsored by AstraZeneca R&D, Mölndal, Sweden. Drs M.P., K.S., C.B.G., J.J.V.M., S.P., and S.Y. have served as consultants to or received research grants and honoraria from AstraZeneca and/or other major pharmaceutical companies. Dr E.L.M. is an employee of AstraZeneca. Drs P.A. and J.D. have no relationships to disclose related to this manuscript.

Conflict of interest: none declared.


View Abstract