EF indicates ejection fraction; HFmrEF, heart failure with midrange ejection fraction; LVAD, left ventricular assist device; LVEF, left ventricular ejection fraction.
eTable 1. Univariate analysis of clinical variables associated with the risk of all-cause mortality.
eTable 2. Comparison of outcomes for patients with a previous normal LVEF without a clinical diagnosis of heart failure compared with patients with heart failure with preserved ejection fraction.
eTable 3. Heart failure category and risk for death or hospitalization excluding patients with a history of malignancy treated with chemotherapy.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Brann A, Janvanishstaporn S, Greenberg B. Association of Prior Left Ventricular Ejection Fraction With Clinical Outcomes in Patients With Heart Failure With Midrange Ejection Fraction. JAMA Cardiol. 2020;5(9):1027–1035. doi:10.1001/jamacardio.2020.2081
Does change from prior left ventricular ejection fraction (LVEF) affect outcomes in patients with heart failure with midrange ejection fraction?
In this cohort study of 448 patients with LVEF between 40% and 50%, patients whose previous LVEF had deteriorated from greater than 50% to midrange levels were at higher risk of adverse outcomes compared with patients whose prior LVEF had increased from less than 40%.
In patients with heart failure with midrange ejection fraction, directional change in LVEF from past measurement may be an important determinant of future clinical events.
Patients categorized as having heart failure (HF) with left ventricular ejection fraction (LVEF) in the midrange between 40% and 50% (HFmrEF) are known to be at increased risk of future events. Although patients can transition into the midrange through either improvement or deterioration in their LVEF, there is limited information available assessing the association of directional change in LVEF with future events. Understanding the association between change in LVEF and the clinical course of patients with HFmrEF would be of value in guiding management strategies.
To determine whether risk of clinical events experienced by patients with HFmrEF varies according to whether LVEF improved or deteriorated into the range of 40% to 50% from previous measurements.
Design, Setting, and Participants
In this retrospective cohort study, patients were identified from the electronic health records at the UC San Diego Health System who had an LVEF measured between 40% and 50% on transthoracic echocardiography (TTE) performed during the calendar year of 2015 and who also had at least 1 prior TTE for comparison. The clinical course of these patients was then followed from the time of the index TTE through December 2018. Data were analyzed from January to March 2019.
Main Outcomes and Measures
The composite of all-cause mortality and all-cause hospitalization, the composite of cardiovascular mortality and HF hospitalization, and each of the individual components.
Of the 448 patients who were identified with HFmrEF, 278 (62.1%) were male, and the mean (SD) age was 67.4 (9.7) years. Left ventricular ejection fraction improved from less than 40% in 157 patients (35.0%), deteriorated from greater than 50% in 224 patients (50.0%), and remained between 40% and 50% over time in 67 patients (15.0%). Compared with patients whose LVEF improved from less than 40% to midrange levels, patients whose LVEF deteriorated from greater than 50% had higher risk of all-cause mortality and hospitalization (hazard ratio, 1.34; 95% CI, 1.10-1.82; P = .03) and of cardiovascular mortality and HF hospitalization (hazard ratio, 1.71; 95% CI, 1.08-2.50; P = .02), and these differences persisted after multivariable analysis. Outcomes did not differ significantly between patients whose LVEF improved and those in whom it remained stable.
Conclusion and Relevance
In a cohort of patients with HFmrEF from a large academic medical center, the clinical course was strongly influenced by the directional change in LVEF from prior study. Patients whose LVEF deteriorated into midrange levels experienced a significantly higher risk of adverse clinical events than patients whose LVEF had improved. These results suggest that directional change in LVEF from prior measurements should be considered when devising management strategies for patients with HFmrEF.
Contemporary management of heart failure (HF) is guided by knowledge of a patient’s left ventricular ejection fraction (LVEF). Historically, patients have been categorized as having HF with reduced ejection fraction (HFrEF), defined as an LVEF less than 40%, or HF with preserved ejection fraction (HFpEF), based on an LVEF of 50% or greater.1-3 While these definitions clearly distinguish between these 2 groups, they exclude patients with HF with midrange ejection fraction (HFmrEF). Many clinical trials testing therapies for HF have restricted enrollment to patients with either HFrEF or HFpEF. As a consequence, there is less information known about the clinical characteristics of patients with HFmrEF, and there is a paucity of evidence on which to base recommendations for therapy. To address this gray area in HF taxonomy, recent guidelines for the management of HF acknowledged the group of patients with LVEF between 40% and 50% as having “HFpEF, borderline”1 or “HF with mid-range ejection fraction.”2
Transition into the HFmrEF category can occur by either deterioration from a previously normal LVEF or improvement from one that was reduced. While intuitively there is reason to think that these subgroups might differ from each other, there is little information available describing their characteristics and clinical course. Consequently, we categorized patients with HFmrEF according to directional change in LVEF from a prior measurement to determine if clinical characteristics and risk of future events differed between these subgroups.
The University of California, San Diego, Institutional Review Board approved this study and granted a waiver of informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.
For this analysis, patients with an EF from 40% to 50% are designated as having HFmrEF. We identified all patients 18 years or older with a documented LVEF in this range on transthoracic echocardiography (TTE) performed at the University of California, San Diego, in 2015. This time frame allowed at least 2 years of follow-up for clinical events. Echocardiograms acquired in outpatient and inpatient settings were included. Retrospective electronic health record (EHR) review was performed to determine whether the index LVEF had improved from less than 40%, remained between 40% and 50%, or was reduced from greater than 50% on prior TTE that was performed at least 3 months prior to the index study. A 3-month period was used to avoid inclusion of patients who may have been experiencing an acute illness and to allow for sufficient time for patients with HFrEF to achieve recovery of systolic function. Patients who only had studies done within the 3 months of the index TTE or were without a prior TTE and recipients of heart transplant or mechanical circulatory support were excluded. Figure 1 depicts the CONSORT diagram for patients included in the study. Demographic data and clinical information were collected from each patient’s EHR at the closest time point to the index TTE. Race/ethnicity data were collected from the demographic section in the EHR, which is self-reported. Each patient’s medical record was reviewed to determine if they had a documented diagnosis of HF or signs and symptoms consistent with HF. New York Heart Association functional class was recorded as noted in the medical records (365 of 448 patients [81.4%]) or, if not specifically stated, assigned according to the history obtained that was most proximate within a 2-month window to the index TTE. For patients who were alive for the last encounter in the EHR but did not continue to receive care in our system throughout the study period, the last encounter was considered the last date of follow-up for analysis purposes.
Patients were categorized according to whether their LVEF either improved from a prior EF of less than 40% (improved group), was consistently between 40% and 50% on all past TTEs (stable group), or was previously normal (greater than 50%) on all past TTEs (deteriorated group). The term deteriorated was used for patients with previous EF greater than 50% since a clinical diagnosis of HF was not required for inclusion. The terms improved and stable were used for patients with previous EF less than 40% and EF stable between 40% and 50%, respectively, since these patients either had symptomatic HF or evidence of persistent left ventricular dysfunction, consistent with a diagnosis of American College of Cardiology/American Heart Association stage B HF. If a patient had multiple prior TTEs with conflicting LVEFs (less than 40% and greater than 50%), they were first categorized as improved if any prior LVEF was less than 40%, since a diagnosis of a reduced LVEF would likely have affected the medical therapy they were prescribed. The remaining patients were categorized in the deteriorated group if their LVEF had been measured as greater than 50% on at least 1 past TTE or in the stable group if all LVEF measurements were between 40% and 50%.
Clinical events were ascertained using information in the EHR and were based on review of the primary diagnoses documented on each discharge summary during the follow-up period. The events included in this analysis were all-cause mortality, cardiovascular mortality, all-cause hospitalization, first HF hospitalization, and the composites of death or hospitalization from any cause and death from cardiovascular cause or first HF hospitalization.
Characteristics of study participants at enrollment were summarized using standard descriptive statistics; analysis of variance, Kruskal-Wallis tests, and Fisher exact tests were used to compare characteristics between the 3 categories. Clinical outcomes were evaluated through time-to-event analyses. Kaplan-Meier curves were created for each group and compared using the log-rank test. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) for univariate and multivariate associations for each end point. Variables used for adjustment were selected according to a priori assumptions and results of univariate analysis (eTable 1 in the Supplement). These included age, sex, history of coronary artery disease (CAD), atrial fibrillation, chronic kidney disease, history of malignancy, and N-terminal pro–brain natriuretic peptide levels. Additional sensitivity analyses were performed, including adjustments for both diabetes and hypertension, since these variables are known to be significant risk factors for HF. Since not all patients in the deterioration group had a history of HF, an additional analysis was performed comparing patients with and without a diagnosis of HF who were included in the deterioration subgroup. We also repeated analysis of end points excluding patients from the deterioration group who had received chemotherapy for treatment of a malignancy.
Two-sided P values less than .05 were considered statistically significant. All data analysis was conducted in SPSS Statistics version 25.0 (IBM Corp).
Of 928 patients with an LVEF between 40% and 50% on TTE identified at our institution during 2015, 467 patients did not have a prior TTE for comparison and 13 patients had a history of heart transplant (Figure 1). Of the remaining 448 patients with HFmrEF, 278 (62.1%) were male, and the mean (SD) age was 67.4 (9.7) years. Left ventricular ejection fraction improved from less than 40% in 157 patients (35.0%), deteriorated from greater than 50% in 224 patients (50.0%), and remained between 40% and 50% over time in 67 patients (15.0%). The median (interquartile range) time between the index TTE and the prior TTE was 1226 (570-2995) days, 1159 (419-2884) days, and 1422 (524-2233) days for the improved, stable, and deteriorated groups, respectively.
Baseline clinical characteristics for the entire HFmrEF cohort and each subgroup are shown in Table 1. Patients in the deteriorated group were more likely to be female and have a history of receiving chemotherapy. Patients in the improved group were younger and more likely to have a history of chronic kidney disease. A documented HF diagnosis or signs and symptoms of HF were present in the EHR in 142 of 157 patients (90.4%) in the improved group, 54 of 67 (81%) in the stable group, and 116 of 224 (51.8%) in the deteriorated group (P < .001). The improved subgroup also had more severe HF symptoms, with a greater prevalence of patients with New York Heart Association class III or IV HF. Most other clinical variables, including CAD and all laboratory values, were similar in the 3 patient subgroups. Patients in the improved group were more likely to be receiving an angiotensin-converting enzyme inhibitor, angiotensin receptor blocker, β-blocker, mineralocorticoid receptor antagonist, and diuretics and to have received implantable cardioverter-defibrillator or cardiac resynchronization therapy than patients in the other subgroups. There were no differences between the subgroups for location where the index TTE was obtained (inpatient vs outpatient).
The clinical course of the 3 subgroups of the HFmrEF population was observed over a median (interquartile range) time of 2.24 (1.87-2.53) years. As shown in Table 2, patients in the deteriorated group experienced a 1.34-fold increase in risk of combined all-cause mortality and hospitalization compared with patients in the improved group (HR, 1.34; 95% CI, 1.10-1.82; P = .03), a difference that was maintained with multivariable analysis. This difference was due to trends toward increased risk of both all-cause mortality (HR, 1.48; 95% CI, 0.96-2.23; P = .08) and all-cause hospitalization (HR, 1.31; 95% CI, 0.99-1.80; P = .06). Patients in the deteriorated group were also at 1.71-fold higher risk of the composite of cardiovascular mortality and HF hospitalization (HR, 1.71; 95% CI, 1.08-2.50; P = .02). This difference also persisted after multivariable analysis. Increased risk in patients in the deteriorated group for this composite was predominantly due to higher risk of cardiovascular mortality (HR, 1.90; 95% CI, 0.95-3.78; P = .07), as risk of HF hospitalization was not increased. No significant differences in outcomes between patients in the improved and stable groups were seen for any of the end points in either unadjusted or multivariate analysis. Figure 2 shows the Kaplan-Meier curve estimates for both composite end points and for their components. Similar results were found after adjustment for diabetes and hypertension.
Although most patients in the improved and stable groups had been diagnosed with HF at the time of the index TTE, a substantially lower number of patients in the deteriorated group did not have this diagnosis recorded in their EHR. To determine whether increased risk of events in the deteriorated group was related to a diagnosis of HF, patients in the deteriorated group were subdivided according to whether or not they had been diagnosed with HF at the time of their TTE and their outcomes were compared. No significant differences in outcomes between patients with and without a diagnosis of HF were noted (eTable 1 in the Supplement). A significantly higher number of patients in the deteriorated group had a history of malignancy treated with chemotherapy. However, even after exclusion of patients who had received chemotherapy, differences between the deteriorated and improved groups persisted (eTable 2 in the Supplement).
The major findings of this study are that the characteristics and clinical course of patients with HFmrEF vary considerably between subgroups defined by the directional change in LVEF from prior study. Patients whose LVEF deteriorated to the 40% to 50% range from a previously higher level had significantly higher morbidity and mortality than patients whose LVEF had improved from a lower value. The increased risk experienced by patients in the deteriorated group emphasizes the need for careful follow-up of this group and for therapeutic strategies that improve outcomes in this population.
There is less available information about the clinical characteristics, natural history, and effective treatments for HFmrEF than for other categories of HF because patients with HFmrEF have mostly been excluded from clinical trials for HF. From data that are available, 12% to 21% of patients with HF are categorized as having HFmrEF.4-9 Patients with HFmrEF appear to have clinical characteristics that are generally more similar to those with HFpEF except for the prevalence of CAD, which is similar to that seen in patients with HFrEF.4-10 However, our results show that within subgroups of the HFmrEF population defined by directional change in their LVEF, there are also substantial differences. Patients in the deterioration group were older and more likely to be female but less likely to be symptomatic or have experienced an HF hospitalization than patients in the improved group. Coronary artery disease was diagnosed in about half of the patients in our study, but there was no significant difference in CAD prevalence between the subgroups defined by previous LVEF. Not surprisingly, since neurohormonal blocking agents are recommended in HF management guidelines for patients with HFrEF but not HFpEF, patients in the deteriorated group were less likely to be receiving traditional HF therapies, such as angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, β-blockers, mineralocorticoid receptor antagonist, diuretics, and devices, but were more likely to be treated with calcium channel blockers than patients in the improved group. Overall, these differences underscore the considerable heterogeneity between patients in the HFmrEF population.
Prior studies have shown that patients with HFmrEF have similar long-term outcomes as patients with either HFrEF or HFpEF.4,5,10 While that may be the case for patients with HFmrEF as a group, our findings suggest that their clinical course is strongly influenced by the directional change in LVEF that brought them into the midrange. Notably, patients whose LVEF had deteriorated from greater than 50% were at higher risk of mortality and morbidity than patients whose EF had improved from less than 40%. The substantial differences in the number of patients who had died or were hospitalized during the relatively brief period of observation in patients in the deteriorated group (Figure 2) underscores their high risk of future events. While risk of all-cause death and all-cause hospitalization was driven by higher risk of both components of this composite in the deteriorated group, the increased risk of cardiovascular mortality and HF hospitalization in the deteriorated group was due almost exclusively to a higher mortality risk. The divergence between all-cause and HF hospitalizations in the deteriorated group is due to a higher percentage of non–HF-related hospitalizations in this group, a finding that is reminiscent of results from previous studies showing that patients with HFpEF experience a higher risk of non-HF hospitalization than patients with HFrEF. Our findings indicate that this trend continues even when patients from these groups transition into the midrange category.
We explored factors that might be responsible for the higher risk of clinical events in patients in the deteriorated group. Although several baseline characteristics varied between groups, these differences did not seem to have affected outcomes since differences in risk persisted even after adjustment for multiple variables. The use of medical therapy was not standardized during the follow-up period, and it is possible that the more favorable outcomes in the improved group may be explained by the higher usage of guideline-directed medical therapy in this group. It is also worth noting that while the prevalence of CAD was similar in the 3 subgroups, it was more common in patients in the deteriorated group identified in our study than in other HFpEF populations.4-7 This relatively high prevalence of CAD might have influenced outcomes, as a history of CAD and new ischemic events have been associated with a higher risk of mortality and deterioration in LVEF in HFpEF.11,12 Patients with HFpEF and CAD who receive complete revascularization have been reported to experience less deterioration in LVEF and lower mortality compared with patients without complete revascularization.12 While effective medical options for HFpEF are limited, these findings suggest that aggressive control of risk factors and early detection and treatment of CAD could potentially improve outcomes in this population.
We considered whether a diagnosis of HF at the time of the TTE in the deteriorated subgroup might be related to the higher risk of future events, but our results indicate that this was not the case. We also assessed whether the relatively unfavorable clinical course of the deteriorated subgroup might be driven by events in patients who had received chemotherapy, but our results did not support this (eTable 3 in the Supplement). However, the relatively small number of patients in each of the subgroups may not have allowed us to detect an impact of differences in baseline characteristics or medical treatments on outcomes.
There is limited information in the literature describing the association of directional change in LVEF with outcomes in patients with HFmrEF. A registry-based study reported a 73% prevalence of recovered EF within an HFmrEF population,13 a figure much higher than the 35% seen in our patients. However, this prior study evaluated patients with clinical evidence of HF who had been entered into a registry, while our study included consecutive patients identified based on LVEF alone. These differences in ascertainment likely resulted in the differences in prevalence of recovered EF between the studies. Although this registry-based study13 also found patients with LVEF that deteriorated from greater than 50% to be at higher risk of future events than other patients in the registry, this was based on the clinical course of only 29 patients, all of whom were symptomatic with HFpEF. In contrast, our study population included 221 patients in the deteriorated group, and it included patients with and without symptoms of HF. However, analysis of patients with and without symptoms suggests that a diagnosis of HF per se in patients in the deteriorated group does not determine the subsequent clinical course, as patients experienced similar clinical course regardless of their symptomatic status. It is possible that the rational for obtaining the TTE from which the index LVEF in the 40% to 50% range was identified was based on either a suspicion of HF or symptoms that might have suggested cardiac disease, albeit not specifically HF.
Nadruz et al14 have similarly demonstrated that among patients with a midrange LVEF, recovered systolic function is a marker of more favorable prognosis. In a cohort of patients who underwent clinically indicated cardiopulmonary exercise testing, those with HFmrEF who had recovered from a prior reduced EF had lower risk of composite events than those with HFrEF and HFmrEF with no recovered EF and similar prognosis compared with HFpEF. However, this study did not evaluate patients with HFmrEF who had deteriorated from a prior normal LVEF and who were found to have high risk of adverse outcomes, as in our study.14 Gu et al15 have also studied the association of directional change of LVEF with clinical outcomes in HF. In a cohort of patients with HF, those who experienced an increase in LVEF (including from HFrEF to HFmrEF or HFpEF or from HFmrEF to HFpEF) had improved outcomes compared with patients with a stable LVEF or a decline in LVEF (including from HFpEF to HFmrEF or HFrEF or from HFmrEF to HFrEF). These findings suggest that the directional change in LVEF may provide important prognostic value across the spectrum of LVEF, not just within the midrange, as demonstrated in our study.15
Our study has limitations. Our results are based on data derived from retrospective EHR review of patients from a single large academic medical center, and the findings may not be generalizable to other settings. Nearly half of patients with HFmrEF were excluded because they did not have a prior LVEF assessment available in our medical records. Assessment of LVEF was based on a retrospective review of the LVEF entered in the EHR by multiple readers and is subject to both intraobserver and interobserver variability. This may have affected both the initial categorization of LVEF as well as the changes in LVEF between tests. Inclusion criteria were based on LVEF and did not require that patients have a diagnosis of HF. Additionally, the number of patients in the stable group was relatively small, and this may have precluded our ability to discern significant differences between groups.
We found that the characteristics and clinical course of patients with HFmrEF vary substantially between subgroups defined according to directional changes in LVEF from prior study. These findings should alert clinicians to seek information about prior LVEF in patients with HFmrEF and to consider the impact of directional changes in LVEF on patient outcomes when planning management strategies. Within the HFmrEF population, more favorable outcomes were seen in patients with a prior reduced EF who were more likely to be treated with guideline-directed medical therapy. Whether this supports initiation of guideline-directed medical therapy for all patients with a midrange LVEF warrants further investigation. Finally, the observation that patients with deteriorated LVEF are at substantially higher risk than other patients in the HFmrEF population identifies a subgroup of patients for whom clinical trials designed to help improve outcomes are badly needed.
Accepted for Publication: April 9, 2020.
Corresponding Author: Barry Greenberg, MD, Division of Cardiovascular Medicine, University of California, San Diego, 9452 Medical Center Dr, La Jolla, CA 92037-7411 (firstname.lastname@example.org).
Published Online: June 17, 2020. doi:10.1001/jamacardio.2020.2081
Author Contributions: Drs Brann and Greenberg had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Brann, Greenberg.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Brann, Greenberg.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Brann, Greenberg.
Study supervision: Greenberg.
Conflict of Interest Disclosures: Dr Greenberg has received grants from Rocket Pharmaceuticals and personal fees from ACI Clinical, Actelion Pharmaceuticals, Akcea Therapeutics, Amgen, EBR Systems, Ionis Pharmaceuticals, Janssen Pharmaceuticals, Merck, Sanofi, Novartis, Viking Therapeutics, and Zensun. No other disclosures were reported.
Meeting Presentation: This study was presented as a poster at the Heart Failure Congress of the European Society of Cardiology; May 26, 2019; Athens, Greece.