Author + information
- Received December 18, 2012
- Revision received January 29, 2013
- Accepted February 12, 2013
- Published online June 1, 2013.
- Maria Cecilia Bahit, MD∗,
- Renato D. Lopes, MD, PhD†∗ (, )
- Robert M. Clare, MS†,
- L. Kristin Newby, MD, MHS†,
- Karen S. Pieper, MS†,
- Frans Van de Werf, MD, PhD‡,
- Paul W. Armstrong, MD§,
- Kenneth W. Mahaffey, MD†,
- Robert A. Harrington, MD‖,
- Rafael Diaz, MD¶,
- E. Magnus Ohman, MD†,
- Harvey D. White, DSc#,
- Stefan James, MD, PhD∗∗ and
- Christopher B. Granger, MD†
- ↵∗Reprint requests and correspondence:
Dr. Renato D. Lopes, Duke Clinical Research Institute, Duke University Medical Center, Box 3850, 2400 Pratt Street, Room 0311 Terrace Level, Durham, North Carolina 27705.
Objectives This study sought to describe the occurrence and timing of heart failure (HF), associated clinical factors, and 30-day outcomes in patients with non–ST-segment elevation acute coronary syndromes (NSTE-ACS).
Background Little is known about HF-complicating NSTE-ACS.
Methods Using pooled patient-level data from 7 clinical trials from 1994 to 2008, we describe the occurrence and timing of HF, associated clinical factors, and 30-day outcomes in NSTE-ACS patients. HF at presentation was defined as Killip classes II to III; patients with Killip class IV or cardiogenic shock were excluded. New in-hospital cases of HF included new pulmonary edema. After adjusting for baseline variables, we created logistic regression models to identify clinical factors associated with HF at presentation and to determine the association between HF and 30-day mortality.
Results Of 46,519 NSTE-ACS patients, 4,910 (10.6%) had HF at presentation. Of the 41,609 with no HF at presentation, 1,194 (2.9%) developed HF during hospitalization. A total of 40,415 (86.9%) had no HF at any time. Patients presenting with or developing HF during hospitalization were older, more often female, and had a higher risk of death at 30 days than patients without HF (adjusted odds ratio [OR]: 1.74; 95% confidence interval: 1.35 to 2.26). Older age, higher presenting heart rate, diabetes, prior myocardial infarction (MI), and enrolling MI were significantly associated with HF during hospitalization.
Conclusions In this large cohort of NSTE-ACS patients, presenting with or developing HF during hospitalization was associated with an increased risk of 30-day mortality. Research targeting new strategies to prevent and manage HF in this high-risk population is needed.
Heart failure (HF) is a frequent complication of acute coronary syndrome (ACS) and is associated with poor prognosis (1,2). However, most previous studies have included only patients with ST-segment elevation myocardial infarction (STEMI) (3–5). Less is known about the incidence of or prognosis associated with the occurrence of HF as a complication in non–ST-segment elevation (NSTE)-ACS. A report from the GRACE (Global Registry of Acute Coronary Events) study showed that among >14,000 ACS patients, the incidence of HF was similar in STEMI (15.6%) and non-STEMI (14.7%) patients but was 50% lower among patients with unstable angina (6).
Because the incidence of non-STEMI as a proportion of all MIs is increasing (7), the relevance of assessing determinants and the impact of HF in NSTE-ACS are important. In the present study, we used a pooled dataset from 7 randomized clinical trials that enrolled patients with NSTE-ACS to describe the occurrence and timing of HF complicating NSTE-ACS, to identify factors associated with HF at different times during the index ACS hospitalization, and to evaluate the association of HF with 30-day clinical outcomes.
We analyzed information pooled at the patient level from the databases of 7 large clinical trials that enrolled patients with NSTE-ACS. These trials included GUSTO IIb (Global Use of Strategies to Open Occluded Coronary Arteries IIb) (8), PURSUIT (Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy) (9), PARAGON A (Platelet IIb/IIIa Antagonism for the Reduction of Acute Coronary Syndrome Events in a Global Organization Network-A) (10), PARAGON B (11), ESSENCE (Efficacy and Safety of Subcutaneous Enoxaparin in Non-Q-Wave Coronary Events) (12), SYNERGY (Superior Yield of the New Strategy of Enoxaparin, Revascularization and Glycoprotein IIb/IIIa Inhibitors) (13), and EARLY ACS (Early Glycoprotein IIb/IIIa Inhibition in Non–ST-Segment Elevation Acute Coronary Syndrome) (14). A synopsis of the characteristics of these trials is provided in Table 1. All patients gave written informed consent to participate in the trials, and the protocol for each trial was approved by the ethics committees or institutional review boards of participating hospitals. The current analysis was approved and declared exempt by the institutional review board of Duke University Medical Center.
Heart failure definition
HF was defined at 2 time points. HF at presentation included patients who presented with Killip classes II to III. HF during hospitalization included patients without HF at presentation and an in-hospital complication of HF or pulmonary edema noted on the study case report form. In-hospital complications of HF or of pulmonary edema were collected as prespecified events in the case report forms. HF classification was used as recorded on the case report form; no attempt was made to adjudicate these classifications. Patients with cardiogenic shock were excluded from this analysis because this group of patients has a well-known high rate of 90-day death (approximately 50%) and could potentially have driven our results (2).
Myocardial infarction definition
Myocardial infarction (MI) was defined by the presence of symptoms suggestive of ischemia and new electrocardiographic changes or elevation of creatine kinase (CK) or creatine kinase-myocardial band (CK-MB) levels to 2 times the upper limit of normal.
Baseline characteristics and medications are described using medians with 25th and 75th percentiles for continuous variables and frequencies and proportions for categorical variables.
Multivariable logistic regression modeling with a stepwise selection procedure was used to identify variables associated with HF at presentation and during hospitalization, death from hospital discharge to 30 days postdischarge, and death or MI from hospital discharge to 30 days postdischarge. Candidate variables for these models were selected based on clinical importance, previously developed models, and a review of related literature.
The linearity assumption for continuous variables was evaluated using restricted cubic spline transformations. If the nonlinear transformation significantly improved the model chi-square, then a plot of the logit of the outcome versus the factor of interest was used to determine the most appropriate transformation for modeling. In those cases, a linear spline was used as the final transformation for modeling.
Clinical trial was included as a stratification factor in presenting the association between covariates and HF in the two HF models. We ran an alternate version of each model with trial included as a covariate (rather than as a stratifying factor) to calculate the corresponding C index, recognizing that this would conservatively estimate the statistic.
In the postdischarge efficacy models, HF experience through hospital discharge (defined as none, baseline, or new onset) as well as trial were considered among other candidate variables for stepwise selection.
A total of 53,287 patients were enrolled in the 7 trials. After we excluded patients with cardiogenic shock or Killip class IV at presentation (n = 1,120) and patients with missing HF information on admission (n = 5,648), 46,519 remained for examination in the present study. Among these patients, 4,910 (10.6%) had HF (Killip class II/III or pulmonary edema) at presentation. Of the 41,609 with no HF at presentation, 1,194 (2.9%) developed HF during hospitalization (Fig. 1).
Patients with HF were older and less often male or current smokers (Table 2) than those without HF. Multiple comorbidities and cardiac risk factors were more common among patients with HF. In particular, prior HF was 4 times more common among patients who had HF at presentation or developed HF during hospitalization. More patients with HF had an enrolling MI and ST-segment changes on electrocardiography (ECG) than those with no HF. Presenting heart rate was higher and blood pressure was lower in patients with HF than in those without HF.
Concomitant medications and in-hospital procedures
Treatment with aspirin was similar among patients both with and without HF. Beta-blockers were more commonly prescribed for patients without HF at presentation. Use of angiotensin-converting enzyme inhibitors was more common among patients with HF (Table 3). Patients with HF were less likely to undergo cardiac catheterization (62.3% vs. 70.3%, respectively) and percutaneous coronary interventions (23.4% vs. 33.1%, respectively) than patients without HF.
Predictors of HF at presentation and in-hospital HF
In multivariable logistic regression modeling, older age, female sex, race, current smoker status, higher heart rate, diabetes, prior MI, hypertension, peripheral vascular disease, enrolling MI, and ST-segment changes on the presenting ECG were significantly associated with HF at presentation (Table 4). Among patients with no HF at presentation, the baseline factors associated with HF during hospitalization are shown in Table 5 and were similar to those observed for patients with HF at presentation.
Patients who presented with HF had an approximately 4-fold higher 30-day mortality rate (5.2% vs. 1.3%, p < 0.001, respectively) and were also more likely to suffer an MI (11.1% vs. 8.4%, p < 0.001, respectively) than patients without HF at presentation. The increased risk in patients with HF is seen in-hospital and also between discharge and 30 days.
Patients with HF at presentation or during hospitalization had a higher risk of death at 30 days than patients without HF (adjusted odds ratios [ORs]: 1.74 [95% confidence interval (CI): 1.35 to 2.26] and 2.34 [95% CI: 1.58 to 3.49], respectively). Patients with HF at presentation or in-hospital had a significantly higher risk of death or MI at 30 days than those patients without HF (adjusted OR: 1.76 [95% CI: 1.47 to 2.11] vs. 1.46 [95% CI: 1.04 to 2.05], respectively). There were no differences in prognostic impact between HF at presentation and HF during hospitalization (Fig. 2).
Of the overall population, 32.9% (n = 15,320) had quantitative ejection fraction. The median (25th, 75th quartiles) ejection fraction was lower in patients with baseline HF (49% [35%, 60%]) than in patients without baseline HF (55% [45%, 65%]). Likewise, ejection fraction was lower among patients with in-hospital HF (45% [32%, 55%]) than among those without in-hospital HF (55% [45%, 65%]).
In this study of more than 46,000 patients with NSTE-ACS, we found that patients with HF at presentation or occurring during hospitalization were older and had more comorbidities than patients who did not have HF complicating their NSTE-ACS presentation. Furthermore, patients with HF either at presentation or during hospitalization had substantially worse 30-day outcomes than patients with no HF. This study highlights the fact that even in patients with unstable angina or non-STEMI, HF at any time during hospitalization is independently associated with poor outcomes.
Our study raises the issue of the generalizability of data from clinical trial databases. Even from large multicenter trials, patients who can provide consent and have fewer comorbidities are more often selected for participation. Balancing the shortcomings of patient selection is the much higher quality of data collected in clinical trials, in particular claims data. As we analyzed HF as a postrandomization variable, there is little selection based on the occurrence of HF. Our study draws from several trials that were conducted more than 10 years ago, and it is possible that the occurrence of HF and the association with other outcomes is changing, although this seems unlikely.
HF in ACS
Previous research of HF complicating ACS has been conducted in cases of STEMI (4,5,15–18). Killip classification has been extensively studied in this setting and has been associated with high-risk baseline characteristics including older age, diabetes, prior MI, and prior HF (16,19,20). There are only 2 published studies that assessed prognosis of HF in NSTE-ACS. Mehta et al. (21) examined the incidence, predictors, and clinical outcomes of patients who developed HF after ACS in the OASIS-2 (Organization to Assess Strategies for Ischemic Syndromes) trial (21). The cumulative incidence rates of HF were reported as 2.7% at 24 h, 4.9% at day 7, and 6.3% at 6 months, rates substantially lower than those reported in the GRACE registry (22). Patients with HF in the OASIS-2 trial had a significantly higher risk of death or MI than those with no HF. Patients with HF in that study share similar high-risk characteristics with patients with HF in our study and with STEMI patients with HF (i.e., they were older, more likely female, and more frequently had diabetes, hypertension, and prior MI).
A report from the GRACE registry (6) evaluated the factors associated with HF in the setting of NSTE-ACS and its relationship with outcomes. It is important to mention that the patient population also included patients with STEMI. The authors reported that the overall incidence of HF was 13% and was similar in patients with STEMI (15.6%) and non-STEMI (14.7%) but half as frequent in patients with unstable angina (8.2%). As in our study, patients with HF were older than patients with no HF and were more often women and less likely to smoke. History of diabetes and hypertension, prior MI or TIA/stroke, and prior HF were more common in patients with HF than in those without HF. The high-risk baseline features of the HF population were similar to those of our HF patients. While we found in the subset of patients from whom ejection fraction was collected that ejection fraction was lower in the HF population, a large proportion had ejection fraction in the normal or mildly reduced range.
Treatment of patients with HF
In our study, consistent with the studies by Mehta et al. (21) and the GRACE registry (6), patients with HF were less likely to undergo cardiac catheterization and percutaneous coronary interventions than patients without HF. These findings are consistent with the results of another report from the GRACE registry (23) describing the use of revascularization in NSTE-ACS patients with HF. In that study, only one-fifth of the patients with HF were revascularized compared with 35% of the patients without HF. In addition, among patients with HF, revascularized patients had lower-risk baseline clinical features than those who were not revascularized. Steg and colleagues (23) also showed a consistent reduction in postdischarge death in patients with HF who underwent revascularization, suggesting that a broader application of revascularization in this high-risk group may be beneficial. It is important to mention that among the 7 studies included in our analysis, there were only 2 that by inclusion criteria, required that patients be assigned an early invasive strategy (13,14).
Predictors of HF complicating ACS
Predictors of HF were consistent between our study and the Mehta and GRACE analyses (6,21). In OASIS-2 (21), independent predictors of HF included older age, history of diabetes, prior MI, and admission diagnosis of non-STEMI. In the GRACE registry (6), older age, history of diabetes and hypertension, prior MI, higher pulse rate, STEMI and non-STEMI on admission, and prior use of diuretic agents were among the strongest independent predictors of HF. Similarly, we found that older age, history of diabetes and hypertension, and prior MI were key factors associated with HF at presentation as were MI at enrollment and ST segment changes. Interestingly, some of these factors associated with higher risk of HF following non-STEMI vary over the course of treatment, such as elevated heart rate and lower blood pressure. Thus, close attention to patients with those baseline characteristics in this setting might prompt treatment, such as earlier inhibitors of the renin angiotensin aldosterone system, which might help prevent the development of HF and improve patient outcomes as a result.
Strengths and limitations
This is one of the largest studies to examine the relationship between HF at presentation or during hospitalization and subsequent clinical outcomes in patients with NSTE-ACS. However, this is an observational study and cause and effect relationships between HF and clinical outcomes cannot be concluded. Furthermore, as this was a population of patients enrolled in clinical trials, the results may not be generalizable to the broader population of patients with ACS treated in clinical practice. Additionally, the definition of HF was not standardized or consistently adjudicated. Even though there is no standard definition for HF outside clinical trials (24), we believe the physician-reported data reflect classification in typical clinical practice; thus, our results should be relevant. Second, the time period for patient inclusion in the trials whose databases were included in this analysis dated from 1994 and extended to 2008. As a result, there is wide variation in the treatment of NSTE-ACS in this patient population. Finally, we lack specific data regarding timing of HF prior to hospitalization and timing of HF occurrence during hospitalization. Therefore, we are uncertain whether HF during hospitalization is new onset HF or a consequence of a previous condition or whether other in-hospital complications may have occurred before or after the development of in-hospital HF.
In this large database, HF at presentation was a frequent complication of NSTE-ACS and was associated with higher-risk baseline features and more adverse outcomes. Readily available baseline characteristics identified patients with increased risk of developing HF, which was associated with subsequent mortality. Patients with NSTE-ACS and HF were also undertreated with evidence-based therapies and interventions, suggesting an opportunity to reduce adverse outcomes through adherence to guidelines recommendations. Further prospective studies are needed to assess the best treatment strategies for this high-risk group of patients.
This work was supported internally by the Duke Clinical Research Institute. Dr. Lopes has a financial relationship with Bristol-Myers Squibb. Dr. Newby is a consultant to AstraZeneca, Biosite Inc., Biosite Diagnostics, CV Therapeutics, Daiichi Sankyo, Mosby, Novartis Pharmaceutical Company, Proctor and Gamble, Roche Diagnostic Corp., and Scios; and has received grant support from Adolor Corporation, AstraZeneca, BG Medicine, Bisoite Inc., Biosite Diagnostics, Bristol-Myers Squibb, Medicure, Sanofi-Aventis, and Schering Plough Corporation. Dr. Mahaffey has financial relationships with Amylin, AstraZeneca, Baxter, Cordis, Eli Lilly, GlaxoSmithKline, Ikaria, Johnson & Johnson, Luipold, Medtronic, Merck, Novartis, Pozen, Regado, Roche Diagnostic, Schering-Plough, Abbott Vascular, Amgen, Boehringer Ingleheim, Cubist PharmaCeuticals, Daiichi Sankyo, Edwards Lifesciences, Guidant, Portola, Novartis, Sanofi, The Medicines Company, Adolor, Bayer Healthcare, Bristol-Myers Squibb, Exeter Group, Genentech, Orexigen, Ortho McNeill, WebMD, Elsevier, Biotronik, Forest, Gilead Science, Pfizer, Polymedix, St. Jude Medical, Dialouges, Haemonetics, Johns Hopkins University, South East Area Health Education Center, and Sun Pharma. Dr. Ohman is a consultant for AstraZeneca, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead Sciences, Janssen Pharmaceuticals, Liposcience, Merck, Pozen Inc., Roche, Sanofi-Aventis, The Medicines Co., and WebMD; and has received research grants from Daiichi Sankyo, Eli Lilly & Co., and Gilead Sciences. Dr. White has financial relationships with Aventis, Eli Lilly, The Medicines Company, U.S. National Institutes of Health, Pfizer, Roche, Johnson & Johnson, Schering-Plough, Merck Sharpe & Dohme, Astra Zeneca, GlaxoSmithKline, Daiichi Sankyo Pharma Development, and Bristol-Myers Squibb. Dr. Granger has financial relationships with Boehringer Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Hoffmann-La Roche, Lilly, Pfizer, Sanofi-Aventis, Takeda, The Medicines Company, and AstraZeneca. Anthony N. DeMaria, MD, acted as Guest Editor for this article.
- Abbreviations and Acronyms
- acute coronary syndrome
- confidence interval
- heart failure
- myocardial infarction
- non–ST-segment elevation
- odds ratio
- ST-segment elevation myocardial infarction
- Received December 18, 2012.
- Revision received January 29, 2013.
- Accepted February 12, 2013.
- American College of Cardiology Foundation
- Gheorghiade M.,
- Bonow R.O.
- Spencer F.A.,
- Meyer T.E.,
- Gore J.M.,
- Goldberg R.J.
- Wu A.H.,
- Parsons L.,
- Every N.R.,
- Bates E.R.
- Steg P.G.,
- Dabbous O.H.,
- Feldman L.J.,
- et al.
- The PARAGON Investigators
- ↵(2002) Randomized, placebo-controlled trial of titrated intravenous lamifiban for acute coronary syndromes. Circulation 105:316–321.
- Ferguson J.J.,
- Califf R.M.,
- Antman E.M.,
- et al.
- Rott D.,
- Behar S.,
- Gottlieb S.,
- Boyko V.,
- Hod H.
- Steg P.G.,
- Kerner A.,
- Van de Werf F.,
- et al.
- Zannad F.,
- Stough W.G.,
- Pitt B.,
- et al.