Management and outcomes following an acute coronary event in patients with chronic heart failure 1999–2007

Study population

For this analysis we limited our study to the Australian and New Zealand population enrolled in the GRACE registry between 1999 and 2007 (n = 6540, 11 sites). Full details of GRACE methodology have been published elsewhere.8 GRACE is designed to reflect an unbiased population of patients with ACS. Patients enrolled in the GRACE registry were at least 18 years of age, and presented with symptoms suggestive of coronary ischaemia. In addition to symptoms, patients were required to have either electrocardiographic (ECG) changes consistent with ACS, elevation of serum cardiac biomarkers of myocardial necrosis, or documented coronary artery disease. ACS precipitated by non-cardiovascular co-morbidity such as anaemia or trauma, was excluded. To enrol an unselected population, the first 10–20 consecutive eligible patients were recruited from each site per month. Data were collected by trained co-ordinators using standardized case report forms. Demographic characteristics, medical history, presenting symptoms, biochemical and electrocardiographic findings, treatment practices, and a variety of hospital outcome data were collected. Selected study outcomes were assessed at 6 months post-discharge. Standardized definitions for all patient-related variables and clinical diagnoses were used.

We included all patients with a diagnosis of CHF recorded in their medical record at the time of hospitalization (irrespective of whether they had clinical signs and symptoms of heart failure at presentation) and a confirmed discharge diagnosis of an ACS [ST elevation myocardial infarction (STEMI), non-ST elevation myocardial infarction (NSTEMI), and unstable angina pectoris (UAP)] based on prior published criteria.8 We included patients who presented directly to hospital as well as patients transferred from an outlying hospital. Patients with ACS but without CHF were used for comparison.

We evaluated rates of in-hospital investigations and procedures [including coronary angiography (CA), percutaneous coronary intervention (PCI), coronary artery bypass grafring (CABG), and echocardiography] as well as guideline-based in-hospital medications [including aspirin, thienopyridine, unfractionated heparin, low molecular weight heparin, glycoprotein IIb/IIIa inhibitors, beta-blockers, statins, angiotensin-converting enzyme (ACE) inhibitors, and angiotensin II receptor blockers (ARBs)], and referral to cardiac rehabilitation on discharge. Additionally we assessed in-hospital adverse events (mortality, acute decompensated cardiac failure, stroke, major bleeding, and renal failure) and 6 month outcomes (readmission for cardiac disease and mortality).

Our study cohort was derived from 11 metropolitan centres, of which 8 (73%) were teaching hospitals with onsite cardiac catheterization facilities. Five of 11 (45%) centres had onsite cardiac surgical facilities. At non-interventional and non-surgical sites it would be routine practice to transfer patients for angiography and revascularization (PCI or CABG). To account for these centres, our rates of CA, PCI, and CABG included patients if they were referred to another hospital for the procedure or had the procedure as an outpatient provided these were booked from the index admission. For all other investigations and medications, patients were only considered if they had the investigation or medication during the index admission.

Statistical analysis

Data are summarized as frequencies and percentages for categorical variables. Continuous variables are presented as mean ± standard deviation. The χ2 statistic and logistic regression analysis was used to compare treatments, in-hospital outcomes, and 6 month readmission rates between those with and without CHF. To adjust for potential confounders, all regression models were adjusted for baseline demographic and clinical characteristics including age, sex, cardiac history and risk factors, co-morbidities, GRACE risk score at presentation, and discharge diagnosis.

The initial years of GRACE enrolment in Australia and New Zealand excluded patients who died within 24 h of admission, therefore early (1999–2001) in-hospital mortality data are less robust and are not included in the analysis. In those who survived until discharge, 6 month mortality was compared between those with and without CHF. For comparison of 6 month mortality, the logistic regression model was adjusted for additional covariates including in-hospital receipt of guideline-advocated medications (aspirin, beta-blocker, ACE inhibitor, statin, and ARB), receipt of revascularization (PCI and CABG), and in-hospital complications. For the purpose of the regression analysis, continuous variables were divided into quartiles and modelled as nominal or ordinal variables as appropriate.

For comparative evaluation of temporal trends, we divided the period of enrolment into three yearly groups (1999–2001, 2002–2004, and 2005–2007). We further evaluated the temporal trend (increase or decrease) in the outcome variable across year groups for patients with CHF vs. those without. The double-sided Cochran–Armitage test for trend or logistic regression was used to evaluate temporal trends. A P-value of <0.05 was used as a cut-off for statistical significance. The analysis was performed using SAS software version 9.1 (SAS Institute Inc., Cary, NC, USA)

Baseline characteristics

A total of 5556 patients with a discharge diagnosis of ACS (1723 STEMI, 2098 NSTEMI, and 1794 UAP) were enrolled in the Australia–New Zealand population of GRACE between 1999 and 2007. Of these patients, 609 had CHF at the time of admission and 4947 had no documentation of CHF. Patients with CHF were on average 10 years older, were more likely to be female, had greater co-morbidities and cardiac risk factors, and were more likely to have a prior history of angina, myocardial infarction, and revascularization by CABG (see Table 1). At presentation, CHF patients were less likely to have initial positive cardiac biomarkers or have ST changes on ECG, but were much more likely to have signs of heart failure at presentation (Killip class >1 in 54.6% vs. 15.6% in those without CHF, P < 0.01). As expected, the mean GRACE risk score at presentation was significantly higher for those with CHF (157.8 vs. 124.4, P < 0.01). Patients with CHF were more likely to be receiving cardiac medication at presentation (all medications P <0.05), and to have a diagnosis of NSTEMI rather than STEMI compared with those without CHF.

Temporal analysis of baseline characteristics (Table 2) shows that the relative prevalence of CHF patients amongst those presenting with an ACS event remained relatively unchanged during the 9 year enrolment period (P = 0.08). Similarly, presenting age, gender, cardiac risk factors, the overall presenting risk (as calculated by the mean GRACE risk score at presentation), and the discharge diagnosis remain largely unchanged. However, the Killip class at presentation improved with time, with a reduction in the proportion of patients with Killip class > 1 (P for trend = 0.05). The proportion of patients with myocardial infarction and prior revascularization (PCI or CABG) increased with time. Additionally, a temporal increase in the absolute percentage of pre-hospital (chronic) use of evidence-based medication was noted for thienopyridines (19.8%), beta-blockers (23.0%), statins (24.8%), and ARBs (23.4%).

In-hospital medications (Tables 3 and 4)

The unadjusted prescription rates of all medications were lower in those with CHF, with the exception of ACE inhibitors and diuretics. When adjusted for selected baseline covariates, in-hospital prescription of aspirin, statin, and low molecular weight heparin were the only agents that were significantly lower in those with CHF. There was no overall difference in adjusted rate of beta-blockers, thienopyridines, and ARBs between those with and without CHF (Table 3).

During the conduct of the study, there was a substantial increase in the prescription of thienopyridines and statins in both patients with and without CHF. Conversely the increase in beta-blockers (20.1%) and ARBs (16.4%) favoured those with CHF, with only a modest increase noted in those without CHF (5.3% and 10%, respectively), while a temporal increase in low molecular weight heparin (13.6%) was only seen in those with CHF.

Investigations, procedures, and referral to cardiac rehabilitation (Tables 3 and 4)

When adjusted for baseline covariates, overall rates of CA, PCI, CABG, and referral for cardiac rehabilitation were substantially lower in those with CHF. There was no difference in performance of echocardiography.

Temporal data indicate an increase in echocardiography in patients with and without CHF. While a temporal increase in CA favoured those with CHF(16.1% vs. 9.2%, P < 0.01), this failed to translate into an increase in PCI. Conversely, in those without CHF, a 10.9% increase in PCI was noted during the same time period (P for trend < 0.01). There was no change in CABG in either group. Increased referral for cardiac rehabilitation was seen in both groups and was most pronounced in those with CHF, with an absolute increase in referral rates of 31.8% compared with a 13.0% increase in those without CHF.

In-hospital outcomes (Table 5)

For patients surviving discharge, follow-up was complete for 97.1% of patients with CHF and for 96.8% of patients without CHF. Crude in-hospital mortality, decompensated heart failure, stroke, and renal failure were higher in those with CHF, but, when adjusted for covariates, only renal failure differed between those with and without CHF [adjusted odds ratio (OR) 1.76, 95% confidnece interval (CI) 1.15–2.71 for patients with CHF].

A statistically significant declining trend in mortality occurred in those with CHF (P for trend = 0.02) but not in those without CHF (P for trend = 0.06). There was a significant temporal decline in rates of acute heart failure in both groups (both trend P < 0.01), with an absolute decline in acute heart failure of 23.1% in those with CHF and of 6.5% in those without CHF. There was no significant temporal trend in in-hospital rates of stroke, major bleeding, or renal failure in either group.

Readmission at 6 months

Overall crude and adjusted readmission rates for heart disease were significantly higher in those with CHF (adjusted OR 1.47, 95% CI 1.17–1.90). There was a consistent and significant temporal decline in the readmission rate at 6 months in those with CHF (absolute decline 16.8%, P for trend = 0.01) and in those without CHF (absolute decline 7.0%, P for trend = 0.01).

Post-discharge (6 month) mortality

The overall 6 month all-cause mortality rate of CHF patients was significantly higher than that of those without CHF (adjusted OR 2.25, 95% CI 1.55–3.27). While the cause of death was not recorded for all patients with CHF, in those for whom a cause of death was recorded (64/86), the majority (56/64, 87.5%) were due to cardiac causes with the rest reported to be due to pulmonary emboli (8/64, 12.5%). When all-cause post-discharge mortality was stratified by the year group, a significant temporal decline in mortality was observed in those with CHF (14% absolute decline in mortality, P for trend < 0.01). There was no temporal decline in mortality observed in patients without CHF (P for trend = 0.64). In the multivariate analysis, revascularization (receipt of PCI or CABG) and in-hospital and use of evidence-based medication (aspirin, beta-blocker, clopidogrel, statin, ACE, or ARB) were not significantly associated with a reduction in mortality (all P > 0.05, data not shown)

Discussion

Patients with CHF who are hospitalized with an ACS event are older, have more co-morbidities, and carry a greater burden of pre-existing cardiac disease. Nearly 55% had evidence of heart failure at presentation, as indicated by a Killip class >1, and had substantially higher risk of in-hospital death as measured by the GRACE risk score. This is reflected in a substantially higher crude rate of adverse events both in-hospital and at 6 months. However, once adjusted for covariates, adverse events in hospital were comparable between those with and without CHF (with the exception of renal failure), suggesting that increasing age, burden of cardiac disease, and co-morbidities amongst those with CHF rather than heart failure per se contributed to these adverse events. In contrast, CHF was independently associated with a substantial increase in readmission and doubling of adjusted post-discharge mortality at 6 months. The only other contemporary study to evaluate the impact of an ACS event in patients with existing heart failure originates from the CHARM study which reported a 30 day crude mortality of 30% amongst CHF patients following a myocardial infarction (3.6% following UAP).9 This exceeds the mortality observed in our study, which is most probably due to this study selectively enrolling patients with symptomatic heart failure (New York Heart Association class II–IV).9 In comparison, our observational cohort represents a much more unselected and heterogeneous heart failure population reflective of actual clinical practice and is similar to reported observational studies of heart failure populations undergoing PCI.10

Despite the adverse outcomes amongst those with CHF, there were several encouraging observations from our study. Firstly, during the during the 9 year enrolment period, outcomes for patients with CHF have improved dramatically. Amongst patients with CHF, the in-hospital incidence of acute decompensated heart failure declined nearly 70% accompanied by a 52% decline in in-hospital mortality. Furthermore, both readmissions (16.8% absolute decline) and 6 month mortality (14% absolute decline) showed marked improvement in patients with CHF. The improvement in outcomes for patients with CHF and ACS in our population parallels the observed decline in rate of hospitalization for heart failure and the declining mortality from heart failure admissions reported in epidemiological studies.11^,12 Furthermore, in the post-myocardial infarction setting contemporary studies have consistently shown a 30–50% decline in the incidence of acute heart failure and sudden cardiac death,13 a fact which has been attributed to improved specific medical therapies as well as a decrease in the prevalence of risk factors such as hypertension.14 Interestingly, our study showed no improvement in mortality (in-hospital or at 6 months) in those without CHF. While the risk of a lethal event is still disproportionately carried by patients with CHF, our study suggests that most of the improvement in post-ACS mortality over the last decade has occurred in patients with CHF.

The improved outcomes for CHF patients were temporally related to an increase in the pre-hospital and in-hospital use of evidence-based therapies. We observed increasing rates of pre-hospital use of statins, beta-blockers, ARBs, and thienopyridines, and a higher likelihood of prior revascularization (CABG and PCI) among patients with CHF at presentation, suggesting better primary and/or secondary prevention strategies in the community. In contrast, in-hospital use of evidence-based treatments amongst CHF patients was lower for most medications. Our patients with both coronary disease and heart failure are older than the general ACS population and have multiple cardiac and non-cardiac co-morbidities. These factors are well known to be associated with increasing ‘frailty’, polypharmacy, functional and cognitive impairment, and disability,8 and are negative predictors of evidence-based treatment in patients with heart failure.7 Furthermore, elderly patients and those with co-morbidities (who are often those affected by CHF) are frequently excluded from trials of heart failure therapy.15 This may result in considerable uncertainty regarding the benefits of these medications in patients with CHF, which may explain the lower use of prognostically beneficial medications. Similar concerns may explain the lower rates of invasive procedures amongst patients with CHF despite the acceptable mortality and high long-term survival offered by revascularization.16 In contrast, use of non-invasive echocardiography [the most common mode of measurement of left ventricular (LV) function] was low but comparable between the groups. While patients may have had LV function assessed prior to hospitalization, assessment of LV function post-ACS event is a well recognized quality measure17,18 and is associated with greater concordance with guideline recommendations.19

Promisingly, however, there was a temporal improvement in in-hospital use of low molecular weight heparin, statins, thienopyridines, beta-blockers, and ARBs, all of which are prognostically beneficial in ACS care. Furthermore, the magnitude of observed increase in beta-blockers and ARBs was greater in those with CHF. Similarly the magnitude of increase in CA rates over the 9 year period was greater in those with CHF. The improvement in secondary prevention was also supported by a marked 31.8% absolute increase in the referral for cardiac rehabilitation amongst patients. While there are no large-scale randomized controlled trials indicating improved survival for CHF patients with cardiac rehabilitation, exercise training reduces heart failure-related hospitalizations and can improve health-related quality of life.20 These temporal changes suggest better recognition of CHF patients as high risk patients, better targeting of therapies toward those with CHF, and greater acceptance by clinicians to generalize the benefit of these medications from the clinical trial setting to those with CHF.

Study limitations

Our identification of prior heart failure was based on a clinical definition and may have been subject to bias. However, 76% of patients were on chronic treatment with diuretics pre-hospital, which is supportive of a population with heart failure. In those with prior heart failure in whom LV function was assessed in hospital, almost 77% had some degree of LV impairment. In a large European survey of heart failure,21 which utilized a clinical definition of heart failure, only 46% of evaluated patients had LV systolic dysfunction despite 83% having a clinical diagnosis of heart failure and 78% receiving diuretic therapy. Thus the clinical definition of heart failure used in the present study is consistent with previous studies reflecting actual clinical practice. Furthermore, our study did not distinguish between systolic heart failure vs. heart failure with normal ejection fraction. The benefit of guideline-recommended medications may be limited to those with systolic dysfunction.

The earlier identification of relatively mild cases of heart failure (due to improved recognition and better diagnostic techniques) leading to an apparent increase in survival (lead time bias) is a well known issue with observational studies.22 However our enrolment criteria for CHF and the relative prevalence of CHF patients amongst our population remained the same throughout the study. Additionally age, gender, discharge diagnosis, and, importantly, the mean GRACE risk score at presentation remained largely unchanged during the enrolment period, suggesting that lead time bias is unlikely to explain the improvement in outcomes seen in our study.

Our analysis did not show a statistically significant contribution of increasing use of evidence-based therapies or revascularization during hospitalization to outcomes of CHF patients, and this may have reflected lack of power in this observational study. In addition, this ACS-focused registry did not consistently collect data on key therapies (e.g. type of beta-blocker, aldosterone antagonist vs. loop diuretic) and interventions of proven benefit in heart failure.23 The receipt of device therapies (implanted cardiac defibrillators, cardiac resynchronization therapy, or left ventricular assist devices) and outpatient heart failure management (e.g. specialist heart failure clinics, heart failure nurses, home-based heart failure interventions, etc.) were not recorded and may have contributed to the improvement in outcomes observed in our cohort.