The stratified log-rank χ22 was 15.0 (P < .001) for difference in mortality between groups. HFpEF indicates heart failure with preserved ejection fraction; HFrecEF, heart failure with recovered ejection fraction; and HFrEF, heart failure with reduced ejection fraction.
eFigure. Flowchart of Patient Selection Process Used to Reach the Final Analytic Cohort
eTable 1. Event Rates at 1, 2, and 3 Years of Follow-up
eTable 2. Heart Failure Category and Mortality Risk—Preserved LVEF Defined as >50%
eTable 3. Heart Failure Category and Risk for Death or Hospitalization—Preserved LVEF Defined as >50%
eTable 4. Hospitalization Risk Across Heart Failure Groups—Preserved LVEF Defined as >50%
eTable 5. Baseline Patient Characteristics According to Follow-up Status at Year 1
eTable 6. Heart Failure Category and Risk for Death and Composite End Points After Year 1 (N=1864)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Kalogeropoulos AP, Fonarow GC, Georgiopoulou V, et al. Characteristics and Outcomes of Adult Outpatients With Heart Failure and Improved or Recovered Ejection Fraction. JAMA Cardiol. 2016;1(5):510–518. doi:10.1001/jamacardio.2016.1325
Heart failure (HF) guidelines recognize that a subset of patients with HF and preserved left ventricular ejection fraction (LVEF) previously had reduced LVEF but experienced improvement or recovery in LVEF. However, data on these patients are limited.
To investigate the characteristics and outcomes of adult outpatients with HF and improved or recovered ejection fraction (HFrecEF).
Design, Setting, and Participants
Retrospective cohort study (inception period, January 1, 2012, to April 30, 2012) with 3-year follow-up at cardiology clinics (including HF subspecialty) in an academic institution. The dates of the analysis were May 21, 2015, to August 10, 2015. Participants were all outpatients 18 years or older who received care for a verified diagnosis of HF not attributed to specific cardiomyopathies or other special causes during the inception period.
Type of HF at baseline, classified as HF with reduced ejection fraction (HFrEF) (defined as current LVEF ≤40%), HF with preserved ejection fraction (HFpEF) (defined as current and all previous LVEF reports >40%), and HF with recovered ejection fraction (HFrecEF) (defined as current LVEF >40% but any previously documented LVEF ≤40%).
Main Outcomes and Measures
Mortality, hospitalization rates, and composite end points.
The study cohort comprised 2166 participants. Their median age was 65 years, 41.4% (896 of 2166) were female, 48.7% (1055 of 2166) were white and 45.2% (1368 of 2166) black, and 63.2% (1368 of 2166) had coronary artery disease. Preserved (>40%) LVEF at inception was present in 816 of 2166 (37.7%) patients. Of these patients, 350 of 2166 (16.2%) had previously reduced (≤40%) LVEF and were classified as having HFrecEF, whereas 466 of 2166 (21.5%) had no previous reduced LVEF and were classified as having HFpEF. The remaining 1350 (62.3%) patients were classified as having HFrEF. After 3 years, age and sex–adjusted mortality was 16.3% in patients with HFrEF, 13.2% in patients with HFpEF, and 4.8% in patients with HFrecEF (P < .001 vs HFrEF or HFpEF). Compared with patients with HFpEF and patients with HFrEF, patients with HFrecEF had fewer all-cause (adjusted rate ratio [RR] vs HFpEF, 0.71; 95% CI, 0.55-0.91; P = .007), cardiovascular (RR, 0.50; 95% CI, 0.35-0.71; P < .001), and HF-related (RR, 0.48; 95% CI, 0.30-0.76; P = .002) hospitalizations and were less likely to experience composite end points commonly used in clinical trials (death or cardiovascular hospitalization and death or HF hospitalization).
Conclusions and Relevance
Outpatients with HFrecEF have a different clinical course than patients with HFpEF and HFrEF, with lower mortality, less frequent hospitalizations, and fewer composite end points. These patients may need to be investigated separately in outcomes studies and clinical trials.
Heart failure with preserved ejection fraction (HFpEF) has been reported to account for almost half of all heart failure (HF) cases.1-3 In most studies, the morbidity and mortality of HFpEF were similar or comparable to those of HF with reduced ejection fraction (HFrEF).1,2,4-7 Based on the epidemiological link of HFpEF to aging and age-related risk factors and comorbid conditions, the prevalence of HFpEF is projected to rise.1,7 There is also growing evidence that HFpEF is a diverse entity encompassing a wide range of structural and clinical phenotypes.8-10
It has been recently recognized that a subset of patients with HFpEF previously had reduced left ventricular ejection fraction (LVEF) but had improvement or recovery in LVEF either by natural history or in response to therapy.11 Patients with HF and recovered ejection fraction (HFrecEF) may be clinically distinct from those with persistently preserved or reduced LVEF. However, there are few reports that describe the characteristics and outcomes of patients with HFrecEF,12,13 with these limited data suggesting that such patients may have improved outcomes compared with patients with persistent HFrEF or HFpEF,13 but further research is needed. Most important, data on hospitalizations have not been reported to date.
To enhance our understanding regarding the characteristics and outcomes of patients with HFrecEF, we reviewed the medical records of approximately 2500 outpatients with HF who received care in Emory Healthcare during a prespecified window and collected 3-year follow-up data. Our primary hypothesis was that patients having HFrecEF have lower mortality compared with patients having HFpEF.
Question What are the outcomes of patients having heart failure (HF) with recovered ejection fraction (ie, HF with initially reduced ejection fraction that has partially or fully recovered over time)?
Findings In this cohort study of 2166 adult outpatients with HF, 16.2% of patients were seen with recovered ejection fraction. After 3 years, these patients had significantly lower mortality compared with patients having HF and permanently reduced or preserved ejection fraction, as well as significantly fewer hospitalizations.
Meaning Heart failure with recovered ejection fraction should be treated as a distinct entity for clinical and research purposes.
We reviewed the medical records of 2507 adult (age ≥18 years) outpatients who received care associated with International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 402.X1, 404.X1, 404.X3, and 428.XX14 between January 1, 2012, and April 30, 2012, in Emory Healthcare by cardiologists. The inception time frame was selected to allow for at least 3 years of follow-up. Medical records were reviewed for symptoms, signs, and treatment of HF, as well as the last reported LVEF, previous LVEF documentation, and special causes of HF. The diagnosis of HF was verified in 2302 of 2507 patients (91.8%) based on physician documentation of symptoms, signs, and guideline-based treatment for HF. We excluded the following: (1) patients with specific cardiomyopathies (eg, hypertrophic, stress induced, infiltrative, restrictive, or chemotherapy induced), (2) patients with HF due to adult congenital heart disease, (3) recipients of heart transplant or mechanical circulatory support, (4) patients with primary right-sided disease (eg, right ventricular cardiomyopathy or class I pulmonary arterial hypertension), and (5) patients with primary valvular disease (eFigure in the Supplement). We also excluded 7 patients who had no follow-up encounters and 7 patients who had no quantitative LVEF data. Special causes of HF and all other exclusion criteria were adjudicated on the basis of individual review of medical records. The ICD-9-CM codes were used for only cohort inception. Case verification for HF and all further classification were done by individual medical record review. The final analysis cohort consisted of 2166 patients. The Emory University Institutional Review Board approved the study. Because this was a retrospective study, no informed consent was required. The institutional review board approved the use of data for the purposes of this research.
We classified patients as having (1) HFrEF (defined as current LVEF ≤40% regardless of previous LVEF assessments), (2) HFpEF (defined as current and all previous LVEF reports >40%), or (3) HFrecEF (defined as current LVEF >40% but any previously documented LVEF ≤40%). Patients with current preserved LVEF (>40%) but no previous documentation of LVEF were assigned to HFpEF. We opted for a cutoff of 40% or less for HFrEF on the basis of the American guidelines suggesting that this value is a reasonable definition of HFrEF.11 However, recognizing that there is a subset of patients with borderline LVEF, we classified cases using a greater than 50% cutoff in a secondary analysis. For consistency, all LVEF assessments were based on transthoracic echocardiography. When an LVEF range was reported, the mean value was assigned for analysis purposes (eg, for reported LVEF of 30%-35%, we assigned a value of 32.5%).
Baseline demographic, anthropometric, and vital signs data were collected through the Emory Clinical Data Warehouse from the index clinic visit record. Race was self-reported. For laboratory values and medications, we extracted the last information available up to the index visit date. The presence of baseline chronic conditions was adjudicated according to the latest (2014) algorithm proposed by the Chronic Conditions Data Warehouse,15 which is used to analyze Medicare data.
Outcomes data (death and hospitalizations) were collected through the Emory Clinical Data Warehouse up to April 30, 2015, providing for a minimum of 3 years of follow-up for each alive patient who continued to receive care in Emory Healthcare. For patients who were alive for the last encounter but did not continue to receive care in the system throughout the study period, the last encounter was considered the last date of follow-up for analysis purposes.
We classified hospitalizations based on the primary ICD-9-CM codes using the Clinical Classifications Software, 2015 version, as provided by the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality.16 Hospitalizations with primary ICD-9-CM diagnoses mapping to Clinical Classifications Software category 7 (“Diseases of the circulatory system”) were classified as cardiovascular. Hospitalizations with primary codes 402.X1, 404.X1, 404.X3, and 428.XX were classified as HF related.
We considered (1) time-to-event end points, including mortality and composite end points (death or first hospitalization for any cause, death or first hospitalization for cardiovascular causes, and death or first HF-related hospitalization), and (2) hospitalization rates over the entire follow-up period (all-cause, cardiovascular, and HF-related hospitalizations). We included estimates of composite end points because these values are frequently used in clinical trials.
Based on (1) an interim analysis for distribution of cases in the first 400 eligible patients and (2) 1-year outcomes from our cohort and previous mortality data,13 we calculated that 1900 eligible cases would provide 80% power to detect a difference in mortality between the HFrecEF and HFpEF groups at 2-sided α = .05 (primary hypothesis), allowing for 5% early loss to follow-up. We opted to focus on these 2 groups because of similar LVEF at presentation and the potential of “contamination” of HFpEF studies with HFrecEF cases, which may have a different prognosis. Therefore, we opted to power the study for difference in mortality between these 2 groups. Also, based on our preliminary data, the mortality gradient was smaller between HFpEF and HFrecEF (vs HFrEF and HFrecEF). Therefore, we powered the study to detect the smaller mortality gradient reliably. Because data review was done in calendar month increments, the final sample size exceeded the required sample size. To compare characteristics between groups, we used the Kruskal-Wallis test for continuous variables and the Fisher χ2 test for categorical variables. We used Cox proportional hazards models for mortality and composite time-to-event end points. Because we detected nonproportional mortality between HF groups, we estimated hazard ratios (HRs) separately for year 1 and years 2 to 3 based on an analysis with flexible parametric models showing that a knot at approximately 1 year best represented the time-dependent component. There was no indication of nonproportionality for other time-to-event end points. We used negative binomial regression to estimate rate ratios of hospitalizations between HF groups. The rationale for these models for hospitalization rates has been described previously.17 We considered the following covariates in adjusted models based on associations with the exposure of interest (HF group), associations with outcomes, clinical interpretation, and previous work: age, sex, race (white, black, and other), body mass index, coronary artery disease, comorbid conditions (diabetes, depression, chronic lung disease, chronic kidney disease, and atrial fibrillation), blood pressure, renal function, hemoglobin level, serum sodium and potassium levels, and therapy with β-blockers, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and loop diuretics. To evaluate for nonlinear effects of covariates, we used restricted cubic splines. We performed 2 sensitivity analyses. First, we repeated analyses using a greater than 50% cutoff for preserved LVEF. Second, because outcomes in patients with HFrecEF may reflect lead-time bias (ie, patients with HFrecEF had survived long enough for LVEF to recover),18 we left-truncated follow-up at 1 year and repeated time-to-event analyses. Finally, we investigated for interactions of sex and race with HF groups for outcomes. We used a statistical software program (Stata, version 14.0; StataCorp LP) for analyses.
Among the 2166 eligible patients with HF, 816 (37.7%) were seen with preserved LVEF (>40%) at the index visit. Of 816 patients, 350 (42.9%) were classified as having HFrecEF (16.2%; 95% CI, 14.6%-17.8% of the cohort), and 466 (57.1%) were classified as having HFpEF (21.5%; 95% CI, 19.8%-23.3% of the cohort). The remaining 1350 patients were classified as having HFrEF (62.3%; 95% CI, 60.2%-64.4% of the cohort). The median time from the last reported LVEF was 182 days (25th to 75th percentile, 13-483 days) and did not differ between HF groups (P = .14 for difference). The median LVEFs were 55% (25th to 75th percentile, 50%-60%) among patients with HFpEF, 50% (25th to 75th percentile, 45%-55%) among patients with HFrecEF (P < .001 vs HFpEF), and 25% (25th to 75th percentile, 18%-33%) among patients with HFrEF. The median reduced LVEF among patients with HFrecEF before recovery was 25% (25th to 75th percentile, 18%-33%), and the improvement in LVEF was observed over a median of 976 days (25th to 75th percentile, 377-1797 days). Sixty-eight of 466 (14.6%) HFpEF cases with current preserved LVEF but no previous documentation of LVEF were categorized as HFpEF.
The baseline characteristics of patients are listed in Table 1. Patients with HFrecEF were (1) younger, with male predominance compared with patients with HFpEF and with demographics closer to patients with HFrEF; (2) less likely to have coronary artery disease, diabetes, and kidney disease and had lower systolic blood pressure compared with patients with HFpEF; (3) more likely to be receiving an angiotensin-converting enzyme inhibitor or angiotensin receptor blocker regimen and less likely to be taking loop diuretics, aspirin, or digoxin compared with the other 2 groups; and (4) less likely to be recipients of implantable cardioverter-defibrillator or cardiac resynchronization therapy.
The median follow-up was 3.0 years (25th to 75th percentile, 1.9-3.2 years), with a total of 5367 patient-years. Death occurred in 13.3% (288 of 2166) of the cohort. Age and sex–adjusted mortality at 1, 2, and 3 years was 5.0%, 9.4%, and 14.0%, respectively. Age and sex–adjusted 3-year mortality was 4.8% in patients with HFrecEF vs 13.2% in patients with HFpEF (log-rank χ21 = 5.7, P = .02) and 16.3% in patients with HFrEF (log-rank χ21 = 13.3, P < .001) (Figure, A). Age and sex–adjusted mortality did not differ significantly between the HFrEF and HFpEF groups.
Mortality in patients with HFrecEF decelerated after approximately 1 year. Table 2 lists HRs separately for year 1 and years 2 to 3, with patients with HFrEF as the reference. During year 1, mortality was not different between the 3 groups, but patients with HFrecEF had lower mortality in years 2 to 3 compared with patients with HFrEF or HFpEF. This finding persisted in fully adjusted models. The mean HR for mortality over the 3-year period in the HFrecEF group was 0.56 (95% CI, 0.34-0.92; P = .02) vs HFpEF and 0.55 (95% CI, 0.35-0.87; P = .01) vs HFrEF in adjusted models. Mortality did not differ in patients with HFpEF vs HFrEF (HR, 0.98; 95% CI, 0.71-1.35; P = .90) in adjusted models. There were no significant interactions of HF group with sex or race for mortality.
At 3 years, 55.2%, 41.1%, and 33.6% of patients met the composite end points of death or all-cause hospitalization, death or cardiovascular hospitalization, and death or HF hospitalization, respectively. Patients with HFrecEF were less likely to meet a composite end point compared with patients with either HFrEF or HFpEF, and patients with HFpEF were less likely to meet a composite end point compared with patients with HFrEF (Figure, B-D). These differences persisted in fully adjusted models (Table 3). We list 1-year, 2-year, and 3-year estimates for these end points in eTable 1 in the Supplement.
Patients having HFrecEF were less likely to experience death or hospitalization compared with patients having HFpEF (HR, 0.70; 95% CI, 0.57-0.87; P = .001), but this difference was not significant in fully adjusted models (HR, 0.91; 95% CI, 0.72-1.13; P = .38). However, death or cardiovascular hospitalization (HR, 0.58; 95% CI, 0.43-0.78; P < .001) and death or HF hospitalization (HR, 0.52; 95% CI, 0.37-0.74; P < .001) were less likely in patients with HFrecEF vs HFpEF in adjusted models. There were no significant interactions of HF group with sex or race for these end points.
There were a total of 3813 hospitalizations (73.2 per 100 patient-years). Of this total, 1996 (52.3%) were for cardiovascular causes inclusive of HF (38.3 per 100 patient-years), and 1501 (39.4%) had specifically HF as the primary diagnosis (28.8 per 100 patient-years). Compared with patients with HFrEF, hospitalizations for cardiovascular causes and HF were less frequent in patients with either HFrecEF or HFpEF, and this difference persisted in adjusted models (Table 4). However, all-cause hospitalization rates did not differ significantly between patients with HFpEF and patients with HFrEF.
Patients having HFrecEF had less frequent hospitalizations compared with patients having HFpEF. In fully adjusted models, the rate ratios for hospitalization of patients with HFrecEF vs HFpEF were 0.71 (95% CI, 0.55-0.91; P = .007) for all-cause hospitalizations, 0.50 (95% CI, 0.35-0.71; P < .001) for cardiovascular hospitalizations, and 0.48 (95% CI, 0.30-0.76; P = .002) for HF-related hospitalizations. There were no significant interactions of HF group with sex or race for hospitalization rates.
When LVEF exceeding 50% was used to define preserved LVEF, 486 of 2166 patients (22.4%) had preserved LVEF at the index visit. Of 486 patients, 153 (31.5%) were classified as having HFrecEF (7.1%; 95% CI, 6.0%-8.2% of the cohort) and 333 (68.5%) as having HFpEF (15.4%; 95% CI, 13.9%-17.0% of the cohort). The remaining 1680 patients (77.6%; 95% CI, 75.7%-79.3% of the cohort) were classified as having HFrEF.
The mortality pattern across groups was similar to that observed with the original definition (eTable 2 in the Supplement). Patients with HFrecEF demonstrated lower rates of composite end points and hospitalizations compared with the other 2 groups (eTable 3 and eTable 4 in the Supplement). The relative risk between patients with HFrEF and patients with HFpEF for composite end points was not significant in adjusted models. Hospitalization rates for HF remained significantly lower in patients with HFpEF vs HFrEF.
The HFrecEF outcomes may be differentially affected by lead-time bias (ie, these patients survive long enough for LVEF to recover before entering the study). Therefore, we left-truncated follow-up at 1 year and repeated time-to-event analyses.
The cohort now consisted of 1864 patients after excluding those who died in year 1 (n = 109) or had no follow-up after year 1 (n = 193). Excluded patients (n = 302) were older and more likely to be white and smokers, had lower blood pressure and body mass index, and were more likely to have coronary artery disease and a worse comorbidity profile compared with those with more than 1 year of follow-up (eTable 5 in the Supplement). Patients with HFrecEF still demonstrated lower mortality compared with patients with HFpEF or HFrEF, lower rates of death or hospitalization (all-cause, cardiovascular, and HF-related hospitalizations) compared with patients with HFrEF, and lower rates of death or hospitalization (cardiovascular and HF-related hospitalizations) compared with patients with HFpEF in adjusted models (eTable 6 in the Supplement).
In this retrospective, single-center cohort study from an academic medical center, 42.9% (350 of 816) of outpatients with HF who received cardiology care and had preserved LVEF (>40%) at presentation were actually cases with recovered ejection fraction (HFrecEF). Patients with HFrecEF were younger, predominantly male, and had lower prevalences of coronary artery disease, hypertension, diabetes, chronic lung disease, and atrial fibrillation compared with patients with HFpEF. These patients with HFrecEF had distinctly better 3-year outcomes compared with patients with HFpEF or HFrEF, including lower mortality and fewer all-cause, cardiovascular, and HF-related hospitalizations.
Use of β-blockers, medications targeting the renin-angiotensin-aldosterone axis, and cardiac resynchronization therapy led to improvement in LVEF in a substantial proportion of patients with HFrEF.19-22 In the Improve the Use of Evidence-Based Heart Failure Therapies in the Outpatient Setting22 registry, a performance improvement program aiming to increase use of guideline-directed medical therapy among HF outpatients, the mean LVEF among approximately 4000 patients increased from 25.8% at baseline to 32.3% at 24 months, and 28.6% of patients had a greater than 10% improvement in ejection fraction. These data indicate that LVEF recovery is possible with guideline-directed treatment in a substantial proportion of patients with HFrEF. A previous study12 reported that two-thirds of referral patients seen with preserved LVEF were actually HFrecEF cases, associated with a more benign clinical profile. In another referral cohort, 3 out of 5 patients seen with LVEF of at least 50% were actually HFrecEF cases.13 Because of its referral nature, that cohort was heavily skewed toward HFrEF (84% of patients). However, despite concerns for survival bias in referral cohorts,18 the pattern of event rates between HF groups reported in that study is similar to ours.13 Despite a more inclusive approach in our study (all cardiology clinics vs HF specialist referrals),13 the academic medical center setting still introduces bias toward younger patients with more severe HF. However, in both studies, the mortality reported for HFrecEF is comparable to that of patients without HF or LVEF impairment.23 We extend those observations by providing relative risk estimates for mortality, hospitalizations, and composite end points commonly used in clinical trials in the largest cohort of patients with HFrecEF reported to date. Notwithstanding the differences between these cohorts, taken together these data indicate that patients with HFrecEF should be distinguished from those with HFpEF.
There are several hypotheses regarding the more benign course of HFrecEF, despite the ongoing presence of symptoms and signs of HF. One potential explanation is partial reverse remodeling with a more favorable neurohormonal profile.13 In fact, some patients without scar may have achieved near-normal left ventricular systolic function. In this aspect, it is notable that our patients with HFrecEF had lower prevalence of coronary artery disease, consistent with previous work.13,22 For example, the absence of prior myocardial infarction and nonischemic HF etiology were both associated with a greater than 10% improvement in LVEF in the large Improve Heart Failure Therapies in the Outpatient Setting (IMPROVE HF) registry.22 Other comorbid conditions are also less frequent in HFrecEF compared with HFrEF. In turn, better response to guideline-based treatment might be the result of a more favorable structural profile early in the disease process or genetic variability. Both have been shown to be the case in recipients of β-blockers,24,25 renin-angiotensin-aldosterone axis inhibitors,26,27 and cardiac resynchronization therapy.28,29 Lead-time bias might partially explain the more favorable outcomes in HFrecEF because these patients may have survived long enough for LVEF to recover before entering the study. In our study, the difference in outcomes was still evident after excluding patients who died in year 1 or had no follow-up after year 1.
However, the retrospective design of our study did not allow us to provide insights into potential predictors of LVEF recovery, including clinical and genetic characteristics. That would require a different, longitudinal study design because our present cross-sectional approach of HF group assignment does not allow us to causally link characteristics to LVEF recovery. In addition, lead-time, referral, and ascertainment bias may have exaggerated the favorable outcomes among patients with HFrecEF. An ideal study on the topic would prospectively enroll patients with newly diagnosed HFrEF within a prespecified window of onset and regularly follow up these patients with echocardiography for recovery after guideline-directed medical and device therapy. A long-term extension of that study for outcomes among groups defined, for example, by LVEF recovery after 6 or 12 months would inform us more definitively on the natural history and outcomes of LVEF response groups.
Our study has several limitations. Classification of HF cases was based on clinically available echocardiographic data; therefore, misclassification may have occurred. A number of HFpEF cases did not have previous LVEF assessment. We opted to include these patients and classify them as HFpEF in the belief that a patient is unlikely to have previous stage C HFrEF but no documentation of LVEF. Although such “latent” HFrecEF cases are theoretically possible, they should be uncommon in practice. Similarly, not all patients with reduced LVEF had repeat LVEF assessments in prespecified time points. However, the median time from the last LVEF assessment to cohort inception was similar between the groups, indicating that more recent echocardiographic assessments are a less likely explanation for ascertainment bias toward HFrecEF. Referral bias toward more advanced HF (and therefore HFrEF) is likely because of the tertiary nature of the institution. Considering also the single-center nature of our study, the findings may not be generalizable to other settings. Hospitalizations were ascertained from the Emory Clinical Data Warehouse. Therefore, hospitalization rates may have been underestimated because outside hospitalizations may have occurred. However, we did not observe any differential changes in hospitalization rates across groups over time (eg, fewer hospitalizations in patients with HFrecEF because patients improve), implying that the relative risk for hospitalization should be valid. We were not able to comprehensively adjudicate causes of death; therefore, we could not provide cause-specific data for mortality. The Emory Clinical Data Warehouse incorporates data on vital status from various sources (in-hospital death records, patient communication, financial services, etc). Considering that we did not have institutional review board approval to contact patients and families and obtain death certificates and other information on mode of death, adjudication of cause of death would be possible for only the fraction of deaths that occurred in the hospital. Also, we did not measure neurohormonal markers; therefore, we could not provide these pathophysiological insights. Residual measured and unmeasured confounding may account for some of the findings. We did not study patients with previous transition from preserved to reduced LVEF status as a separate entity.30 All patients who had reduced LVEF at the time of inception were classified as having HFrEF regardless of previous (reduced or preserved) LVEF. We opted for this simple classification because pathophysiology of HF converges once LVEF is reduced regardless of previous LVEF status. Also, there is no current evidence that these patients should be treated differently depending on LVEF trajectory. Finally, to facilitate communication, we opted not to create a “borderline” LVEF category. Instead, we provide a secondary analysis using LVEF greater than 50% as the cutoff for HFpEF. Moreover, we recognize that small improvements in LVEF that still qualify as HFrecEF (eg, a change in LVEF from 35% to 45%) have probably different prognostic implications than larger improvements.30 However, to quantify this effect, a prospective, serial echocardiographic study to ensure consistent timing of LVEF measurements (because rate of improvement could be an important factor) would be needed.
In an outpatient HF cohort receiving cardiology care in an academic medical center, 42.9% (350 of 816) of patients seen with LVEF exceeding 40% were patients with recovered rather than persistently preserved LVEF. These patients had more favorable 3-year outcomes, including all-cause mortality, compared with patients with persistent HFpEF or HFrEF, implying that patients with HFrecEF should be investigated separately in future outcomes studies and clinical trials.
Accepted for Publication: April 11, 2016.
Corresponding Author: Andreas P. Kalogeropoulos, MD, MPH, PhD, Division of Cardiology, Department of Medicine, Emory University School of Medicine, 1462 Clifton Rd NE, Ste 535B, Atlanta, GA 30322 (email@example.com).
Published Online: July 6, 2016. doi:10.1001/jamacardio.2016.1325.
Author Contributions: Dr Kalogeropoulos had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Kalogeropoulos, Fonarow, Georgiopoulou, Siwamogsatham, Papadimitriou, Butler.
Acquisition, analysis, or interpretation of data: Kalogeropoulos, Burkman, Patel, Li, Papadimitriou, Butler.
Drafting of the manuscript: Kalogeropoulos, Georgiopoulou, Burkman, Papadimitriou, Butler.
Critical revision of the manuscript for important intellectual content: Kalogeropoulos, Fonarow, Georgiopoulou, Burkman, Siwamogsatham, Patel, Li, Butler.
Statistical analysis: Kalogeropoulos.
Administrative, technical, or material support: Butler.
Study supervision: Kalogeropoulos, Butler.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and none were reported.
Disclaimer: Dr Fonarow is the Associate Editor for Health Care Quality and Guidelines, JAMA Cardiology, but was not involved in the editorial review or decision to accept the manuscript for publications.