We designed a retrospective study of consecutive patients admitted to the coronary care unit and the hospitalization ward of 2 Spanish hospitals due to ACS from November 2003 through May 2014 (n=7033). The contemporary cohort was comprehensive and the only exclusion criterion was the absence of available LVEF data or missing values during the index hospitalization (825 patients); the final cohort was based on 6208 patients. The study protocol and the review of the clinical history was approved by the ethics committee of the coordinating hospital.

Patients were classified as having ST-segment elevation myocardial infarction (STEMI) or non—ST-segment elevation ACS that included unstable angina and non—STEMI. The diagnosis of myocardial infarction was made according to the 2012 ESC definition.15 Diagnosis of unstable angina required the presence of suggestive symptoms together with objective evidence of myocardial ischemia on stress testing or detection of a culprit lesion of 50% on coronary angiography, in addition to cardiac biomarkers below the upper normal laboratory limit.15 Both centers are involved in regional STEMI systems of care and primary angioplasty, the strategy of choice within the inclusion period, was performed in> 90% of STEMI patients in both centers.

Risk factors, clinical antecedents, treatments, complementary tests, and main diagnosis at discharge were collected from all patients by trained medical staff. The diagnostic and therapeutic ACS protocols at both centers included blood sample determinations in the emergency room and in fasting state after hospital admission. We identified patients with prior coronary artery disease by searching for those who already had a clinical diagnosis of myocardial infarction, and/or a history of angina or angina-driven coronary revascularization. Prior HF was identified if patients had at least 1 hospitalization in which HF was the main diagnosis or if they had typical signs and symptoms consistent with HF syndrome along with compatible imaging studies (X-ray or echocardiogram).

We defined index HF as the presence of pulmonary rales or the use of intravenous diuretics or intravenous inotropic drugs during ACS admission and described the level of HF severity by Killip class.16 The highest class observed during the first 7 days of hospital stay was used in the present analysis. Two-dimensional transthoracic echocardiography was performed in all participants by a level III-certified echocardiographer as part of routine clinical practice during hospitalization and LVEF was assessed according to the international Simpson method17 at discharge.

The final population included 6208 patients. They were divided into 2 groups depending on the development of HF (Killip=1 or Killip ≥ 2). In each group, the patients were classified according to LVEF following the current HF-ESC guidelines cutoff13 (Figure 1).

Figure 1.

Flowchart of the study patients. ACS, acute coronary syndrome; LVEF, left ventricular ejection fraction.

(0.25MB).

We used a well-established protocol for postdischarge follow-up: after discharge, patients were followed up in a monographic consultation of ischemic heart disease and primary care. The structured follow-up was carried out through the electronic history by reviewing all medical assistance and hospital records and resorting in certain cases to telephone contact.

Primary endpoints were all-cause mortality and HF after hospital discharge. The development of HF on follow-up was identified if the patient had at least 1 hospitalization with HF as the main diagnosis.

Quantitative variables are presented as mean±standard deviation and differences were assessed by the Student t test or ANOVA test. Qualitative variables are presented as percentages and differences were analyzed by the chi-square test. Multicollinearity between LVEF and HF during hospitalization, revascularization and age was rejected since variance inflation factors were low.

All-cause mortality after hospital discharge was assessed by survival analyses. The observed event risk was calculated as a Kaplan-Meier estimate using the log rank test.

All-cause mortality predictors were assessed by Cox regression models, after verification of the proportional risk assumption by the Schoenfeld residuals test, using all variables that obtained P values <.1 in the univariate analysis or could have plausible clinical implication; the results are presented as hazard ratios (HR) and 95% confidence intervals (95%CI). The first model included all the variables and, after identification of a positive interaction between HF and LVEF categories, a second model was carried out stratified by HF during the index admission. The discriminative and calibration ability of survival models were assessed by means of Harrell's C-statistic and the Gronnesby and Borgan test, respectively. The incidence of HF could be affected by patient death and, therefore, the usual techniques for time-to-event analysis would provide biased or uninterpretable results due to the presence of competing risks. To avoid such effects, we applied the model introduced by Fine and Gray18 to test the competing events. The incidence of HF is presented in cumulative incidence function graphs and results of the multivariate analysis as a subhazard ratio (sHR). Harrell's C-statistic test was used to assess the model's discrimination while calibration was tested by the Gronnesby and Borgan test. Patients lost during follow-up were categorized as missing, as well as those who lacked any of the main variables for the analyses, although these were very few.

Statistical difference was accepted at P <.05. All analyses were performed using STATA 14.2 (StataCorp, 2009, Stata Statistical Software, Release 14, College Station, TX, StataCorp LP).

A total of 5064 patients had an ACS without HF and were assigned to 1 of 3 groups according to LVEF: 290 (5.7%) with LVEF <40%, 540 (10.7%) with LVEF 40% to 49%, and 4234 (83.6%) with LVEF ≥ 50%. The characteristics of the patients are shown in Table 1. Age, history of hypertension, coronary artery disease, previous revascularization (surgical or percutaneous), previous HF, renal function, blood pressure, heart rate, STEMI, treatment with angiotensin-converting enzyme inhibitors (ACEIs)-angiotensin receptor blockers, mineralocorticoid receptor antagonists, and diuretics were statistically significantly different between groups.

The remaining 1144 patients developed HF (Killip ≥ 2) during hospitalization: 395 (34.6%) with LVEF <40%, 251 (21.9%) with LVEF 40% to 49%, and 498 (43.5%) with LVEF ≥ 50%. Their characteristics are described in Table 2. The main differences between LVEF groups were found in age, sex, and history of hypertension, previous HF, systolic blood pressure, heart rate, hemoglobin, STEMI presentation and treatment with ACEIs, mineralocorticoid receptor antagonists, and diuretics.

The incidence of HF was increasingly higher in each category of LVEF (Figure 2). In-hospital mortality was 4.0% (95%CI, 3.4-4.5) and was increasingly higher in each LEVF category: 1.75% (LVEF ≥ 50%), 5.44% (LVEF 40%-49%), and 17.60% (LVEF <40%) (P <.01), as well as in patients who developed HF compared with those who did not: 22.5% vs 1.4% (P <.01).

Figure 2.

Distribution of heart failure incidence according to LVEF categories and maximal Killip class. LVEF, left ventricular ejection fraction.

(0.08MB).

Postdischarge follow-up was available in 91.8% of the patients with a median of 4.4 years [interquartile range, 2.2-6.5]. All-cause mortality was 22.5% (95%CI, 21.5-23.5) at was much higher in patients with HF during the ACS hospitalization (50.4% vs 17.6%; P <.001). As shown in Figure 3A, in patients without HF during the ACS hospitalization, the lowest rate was noted in patients LVEF ≥ 50% (15.4%), followed by those with LVEF 40% to 49% (25.4%) and LVEF <40% (29.4%); in contrast, no differences were observed in patients who developed HF during the ACS hospitalization (Figure 3B).

Figure 3.

Kaplan-Meier curves for time free of all-cause mortality according to LVEF categories in the entire cohort (A), patients without HF (B) and patients with HF (C) during hospitalization. ACS, acute coronary syndrome; HF, heart failure; LVEF, left ventricular ejection fraction.

(0.32MB).

We performed a landmark analysis with the assessment of 1-year mortality and the results for all-cause mortality were the same (Figure of the supplementary material).

The incidence of HF after hospital discharge was 13.3% (95%CI, 12.5-14.1) and was higher in patients with HF during the ACS hospitalization (31.5% vs 10.0%; P <.001). As shown in Figure 4A, among patients without HF during the ACS hospitalization, the highest rate was found in those with LVEF <40% (18.2%) or LVEF 40% to 49% (17.2%), and was much lower in patients with LVEF ≥ 50% (8.3%). Nonetheless, no differences were observed according to LVEF among patients with HF during the ACS hospitalization (Figure 4B).

Figure 4.

Cumulative incidence of postdischarge HF function plots according to LVEF categories in the entire cohort (A) patients without HF (B) and patients with HF (C) during hospitalization. ACS, acute coronary syndrome; HF, heart failure; LVEF, left ventricular ejection fraction.

(0.3MB).

Multivariate analysis of all-cause mortality and HF was adjusted by age, sex, diabetes mellitus, previous coronary artery disease, previous HF, complete revascularization, treatment with diuretics, ACEIs, and beta-blockers, and LVEF.

Heart failure during the hospitalization, LVEF <40%, and LVEF 40% to 49% were associated with all-cause mortality and postdischarge HF in a first analysis (Table 3), and a positive interaction was found between HF at admission and LVEF; therefore, the final model was designed with these interactions. The discriminative (Harrell's C-statistic, 0.79; 95%CI, 0.78-0.80) and calibration (Gronnesby and Borgan test P=.73) ability of the survival models were high. Higher LVEF, as well as complete revascularization and female sex, was associated with lower mortality only in patients without HF during the ACS hospitalization. The effect of age on long-term mortality was lower in patients who had HF during hospitalization (P=.01).

Independent predictors of postdischarge HF are presented in Table 4. Left ventricular ejection fraction 40% to 49% was independently associated with lower postdischarge HF only in patients who did not develope HF during hospitalization. The discriminative (Harrell's C-statistic, 0.76; 95%CI, 0.72-0.81) and calibration (Gronnesby and Borgan test P=.73) ability of competing risk regression models were adequate for their intended purpose. As an intern sensitivity analysis, we investigated the possible changes in clinical features of patients throughout the long inclusion period by dividing the cohort into 3 time intervals (2004-2006, 2007-2010, and 2011-2014). No differences were found in age or sex and only a slight increase in the prevalence of hypertension was noted. The percentage of STEMI patients increased significantly during the inclusion period but the mean GRACE score did not. The prevalence of LVEF <40% (15.7%, 15.8%, 18.6%; P=.09) or HF during hospitalization (18.3%, 19.3%, 17.6%; P=.337) did not vary between the 3 time periods.

This study has several strengths; it represents a contemporary cohort with a large sample and statistical power so the results may be generalizable to real-world clinical practice. We included a large number of patients with a very long follow-up. We examined the different prognosis of patients with ACS and LVEF <40%, 40% to 49%, and ≥ 50% and also analyzed the long-term prognostic factors depending on HF development.

The study also has some limitations. First, it is an observational and retrospective analysis and we could not measure LVEF in 12.2% of the study population; previous studies have reported that missing echocardiographic data is not uncommon in clinical practice.33 In addition, 8.2% of patients were lost to follow-up. We used the classification proposed by the ESC-HF guidelines in the ACS setting, which could be a study limitation. As in any observational study, we cannot rule out the effect of residual confounding due to unmeasured variables. In addition, there may be appropriate contraindications to adjunctive pharmacotherapy or invasive angiography that were not collected or known to us. We used the Simpson method to estimate LVEF according to international recommendations with its inherent limitations. We had no data on any biomarker with utility in HF, such as B-type natriuretic peptide, midregional proatrial natriuretic peptide, soluble ST2, or galectin-3. Our study began in 2004, which explains why the use of beta-blockers and ACEIs was lower than estimated. Over the years, the use of these drugs has increased progressively in both institutions. We did not identify treatment adherence or changes in LVEF during follow-up. In addition, the percentage of ACEIs, beta-blockers and mineralocorticoid antagonist that was maintained during admission is unknown. Finally, revascularization and complete revascularization rates were not too high, which merely reflects daily clinical practice in 2 centers with available catheterization laboratories, as previously described.