Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Jun 1.
Published in final edited form as: Circ Heart Fail. 2016 Jun;9(6):10.1161/CIRCHEARTFAILURE.115.003116 e003116. doi: 10.1161/CIRCHEARTFAILURE.115.003116

Predicting Heart Failure With Preserved and Reduced Ejection Fraction: The International Collaboration on Heart Failure Subtypes

Jennifer E Ho 1,*, Danielle Enserro 1,*, Frank P Brouwers 1,*, Jorge R Kizer 1,*, Sanjiv J Shah 1, Bruce M Psaty 1, Traci M Bartz 1, Rajalakshmi Santhanakrishnan 1, Douglas S Lee 1, Cheeling Chan 1, Kiang Liu 1, Michael J Blaha 1, Hans L Hillege 1, Pim van der Harst 1, Wiek H van Gilst 1, Willem J Kop 1, Ron T Gansevoort 1, Ramachandran S Vasan 1, Julius M Gardin 1, Daniel Levy 1, John S Gottdiener 1,*, Rudolf A de Boer 1,*, Martin G Larson 1,*
PMCID: PMC4902276  NIHMSID: NIHMS788516  PMID: 27266854

Abstract

Background

Heart failure (HF) is a prevalent and deadly disease, and preventive strategies focused on at-risk individuals are needed. Current HF prediction models have not examined HF subtypes. We sought to develop and validate risk prediction models for HF with preserved and reduced ejection fraction (HFpEF, HFrEF).

Methods and Results

Of 28,820 participants from four community-based cohorts, 982 developed incident HFpEF and 909 HFrEF during a median follow-up of 12 years. Three cohorts were combined and a 2:1 random split used for derivation and internal validation, with the fourth cohort as external validation. Models accounted for multiple competing risks (death, other HF subtype, unclassified HF). The HFpEF-specific model included age, sex, systolic blood pressure, body mass index, antihypertensive treatment, and prior myocardial infarction; it had good discrimination in derivation (c-statistic 0.80, 95% CI 0.78–0.82) and validation samples (internal 0.79, 95% CI 0.77–0.82; external 0.76, 95% CI 0.71–0.80). The HFrEF-specific model additionally included smoking, left ventricular hypertrophy (LVH), left bundle branch block (LBBB), and diabetes; it had good discrimination in derivation (c-statistic 0.82, 95% CI 0.80–0.84) and validation samples (internal 0.80, 95% CI 0.78–0.83; external 0.76, 95% CI 0.71–0.80). Age was more strongly associated with HFpEF, and male sex, LVH, LBBB, prior myocardial infarction, and smoking with HFrEF (P for each comparison ≤ 0.02).

Conclusions

We describe and validate risk prediction models for HF subtypes, and show good discrimination in a large sample. Some risk factors differed between HFpEF and HFrEF, supporting the notion of pathogenetic differences among HF subtypes.

Keywords: heart failure, epidemiology, risk factor, ejection fraction

Heart failure (HF) is a major growing public health burden worldwide. One in five men and women will develop HF in their lifetime.1 In the United States alone, it is estimated that over 8 million people will be living with HF by the year 2030, with projected direct medical costs of HF doubling in the next 20 years to $53 billion, and similar trends are projected worldwide.2,3 The need for strategies to prevent HF has sharpened the focus on identifying and treating high-risk asymptomatic individuals, classified as American Heart Association and American College of Cardiology (ACC/AHA) stage A or B HF.2,4 Several recent initiatives have advocated for primary prevention and aggressive treatment of at-risk individuals,5,6 yet this group of individuals remains ill-defined with respect to their exact risk for HF. This emphasis on disease prevention highlights the potential importance of risk prediction of future HF.

Prior population-based studies on HF risk prediction models have often lacked external validation, and more importantly, none have taken into account HF subtypes.7 Current recommended therapies for HF with preserved ejection fraction (HFpEF) differ considerably from therapies instituted for HF with reduced ejection fraction (HFrEF).4,8 It is also thought that HFpEF and HFrEF are two distinct HF phenotypes with different etiologic factors.9 Accordingly, we hypothesized that risk predictors of HFpEF would be distinct from those preceding HFrEF. Previous studies have revealed some differences in risk factor profiles among incident HFpEF vs HFrEF.10–12 Findings were restricted to single cohorts, however, and risk prediction was limited by the challenge of handling multiple competing risks when examining HF subtypes.

To examine HF subtype-specific risk profiles, we assembled an international consortium of four longitudinal community-based cohorts, each of which classified incident HF cases as HFpEF or HFrEF. We developed and validated separate risk prediction models for HFpEF and HFrEF. Our findings may help delineate and phenotype individuals at risk for specific HF subtypes, and may be useful in targeting future preventive strategies.

METHODS

Study Sample

Four prospective, observational community-based cohorts with adjudicated incident HF outcomes were included: the Framingham Heart Study (FHS) original and offspring cohorts, the Cardiovascular Health Study (CHS), the Prevention of Renal and Vascular Endstage Disease (PREVEND), and the Multi-Ethnic Study of Atherosclerosis (MESA).13–17 Participants from the following baseline examinations were included: FHS original cohort exam 16 (1979–1982) or 24 (1995–1998), FHS offspring cohort exam 2 (1979–1983) or 6 (1995–1998), CHS exam 1 (1989–1990; 1992–1993 for supplemental African-American cohort), PREVEND exam 1 (1997–1998), or MESA exam 1 (2000–2002). Individuals with prevalent HF (n=472), age <30 years at baseline examination (n=379), or those with missing covariates (n=2177) were excluded, leaving 28,820 individuals for analysis. Participants were monitored for the first HF event occurring up to 15 years after the baseline examination. Written informed consent was obtained, and institutional review board approval was obtained at all participating institutions. For cohort-specific details, see Supplemental Methods in the Supplement.

Clinical Assessment

All participants had detailed medical history, physical examination, fasting laboratory assessment, and electrocardiography at the baseline examination. All potential risk factors were evaluated and harmonized across cohorts whenever possible. Blood pressure was taken as the average of two seated measurements. Body mass index was calculated as weight divided by height2 and expressed as kg/m2. Diabetes mellitus was defined as a fasting glucose ≥126 mg/dL, random glucose ≥200 mg/dL, or the use of hypoglycemic medications. Modest alcohol use was defined as ≥1 drink per day in both men and women. Electrocardiographic left ventricular (LV) hypertrophy was defined based on accepted voltage and ST-segment criteria (Supplemental Methods).

Definition of Incident Heart Failure Subtypes

Individuals were followed prospectively for the occurrence of incident HF or death. Outcomes were adjudicated using established protocols by study investigators within each cohort after review of all available outpatient and hospital records. HF was defined using a combination of signs and symptoms (Supplemental Methods). Records were reviewed for LV function assessment at or around the time of the first HF presentation. Each incident HF event was categorized as HFpEF (LVEF >45%), HFrEF (LVEF ≤45%), or unclassified (no LV function assessment available). LV function was ascertained by echocardiography in over 85% of cases in all 4 cohorts.

Statistical Analysis

Baseline clinical characteristics were summarized by cohort. Individual-level data were pooled for FHS, CHS, and PREVEND after harmonizing definitions of clinical variables. Internal derivation and validation sets were created using a random 2:1 split in the pooled sample.

Cumulative incidence rates of HFpEF and HFrEF were estimated using a Kaplan-Meier-like method accounting for competing risks (death, other HF subtype, unclassified HF).18 Fine-Gray proportional sub-distribution hazards models were fitted for overall HF, and separately for HFpEF and HFrEF.19 Given that incident all-cause mortality was not negligible among cohorts studied, death was treated as a competing risk event in order to avoid informative censoring. In addition to accounting for competing risk of death, analyses also accounted for competing risks of other HF subtype and unclassified HF in PSHREG (SAS).20 First, age- and sex-adjusted models were fitted for each clinical covariate. Given sex differences in HDL cholesterol, sex-specific centered HDL was used in models. Covariates associated with HF at a P≤0.10 were entered into a stepwise selection model, forcing in age and sex, and using a Bonferroni-corrected P-value threshold for retention (P=0.05/number of covariates considered for entry). A strata statement was included to specify study cohorts within the pooled analysis.

We assessed validation using the final multivariable HF subtype-specific models and cohort-specific null hazards. To assess model discrimination, the c-statistic was calculated using predicted event probabilities and times to events.21 We created model-based risk deciles to assess calibration. Overall calibration was assessed using the Hosmer-Lemeshow approach. Calibration was also assessed visually by plotting model-based versus non-parametric estimates of 10-year cumulative incidence (Supplemental Methods). External validation was performed in MESA using the same approach, fitting the model from derivation sample with the null sub-distribution hazard derived from the FHS sample. MESA was selected as the validation cohort after CHS, PREVEND, and FHS data had already been merged, and was deemed a reasonable validation cohort given a comparable age and sex distribution with the other three cohorts.

In secondary analyses, we examined whether variables were associated differentially with risk of HFpEF vs HFrEF. We took all covariates from HFpEF and HFrEF models, then we compared subtype-specific coefficients using the Lunn-McNeil method.22 Given the lack of consensus around which LVEF cut-point is used to define HF subtypes, we repeated primary analyses after re-classification of HFpEF and HFrEF cases using and LVEF of 50%. In exploratory analyses, sex*covariate interactions were tested by adding interaction terms to HF subtype-specific models. Stratified age- and sex-adjusted models by cohort and derivation/validation sets were constructed to compare effects among subgroups. All statistical analyses were conducted with SAS version 9.4 for Windows (Cary, NC).

RESULTS

In the primary cohorts, there were 22,142 participants: 9496 (43%) from FHS, 5277 (24%) from CHS, and 7369 (33%) from PREVEND. Mean ages of participants were 58 ± 14 years in FHS, 73 ± 6 years in CHS, and 49 ± 12 years in PREVEND; over half of participants were women. Baseline clinical characteristics by cohort are detailed in Table 1.

Table 1.

Baseline characteristics by cohort

CHS
(n=5277)		 PREVEND
(n=7369)		 FHS
(n=9496)		 MESA
(n=6678)
Demographics
Age, years 73 (6) 49 (12) 58 (14) 62 (10)
Women, n (%) 3038 (57) 3702 (50) 5207 (55) 3521 (53)
Race
  White, n (%) 4470 (85) 7001 (96) 9496 (100) 2561 (38)
  Black, n (%) 778 (15) 65 (1) 0 (0) 1838 (28)
  Asian, n (%) 3 (0.1) 159 (2) 0 (0) 0 (0)
  American Indian, n (%) 11 (0.2) 0 (0) 0 (0) 0 (0)
  Chinese American, n (%) 0 (0) 0 (0) 0 (0) 798 (12)
  Hispanics, n (%) 0 (0) 0 (0) 0 (0) 1481 (22)
  Other, n (%) 15 (0.3) 89 (1) 0 (0) 0 (0)
Clinical characteristics
Systolic blood pressure, mm Hg 136 (21) 129 (20) 130 (20) 127 (21)
Diastolic blood pressure, mm Hg 71 (11) 74 (10) 77 (10) 72 (10)
Heart rate, bpm 68 (11) 69 (10) 67 (11) 63 (10)
Body mass index, kg/m2 26.7 (4.7) 26.1 (4.2) 26.7 (4.8) 28.3 (5.5)
Antihypertensive treatment, n (%) 2396 (45) 1001 (14) 2430 (26) 2479 (37)
Diabetes mellitus, n (%) 819 (16) 272 (4) 611 (6) 841 (13)
Current smoker, n (%) 626 (12) 2518 (34) 2245 (24) 872 (13)
Modest alcohol use, n (%) 751 (14) 1895 (26) 3052 (32) 3191 (48)
Previous myocardial infarction, n (%) 418 (8) 405 (6) 295 (3) 0 (0)
Previous coronary heart disease, n (%) 909 (17) 305 (4) 724 (8) 0 (0)
Previous stroke, n (%) 197 (4) 69 (1) 235 (3) 0 (0)
Laboratory and ECG characteristics
Total cholesterol, mg/dL 212 (39) 218 (44) 212 (41) 194 (36)
HDL cholesterol, mg/dL 54 (16) 51 (15) 50 (16) 51 (15)
ECG LV hypertrophy, n (%) 227 (4) 174 (2) 189 (2) 242 (4)
Left bundle branch block, n (%) 83 (2) 30 (0.4) 108 (1) 23 (0.3)
Right bundle branch block, n (%) 233 (4) 86 (1) 242 (3) 158 (2)
Incident HF events
HFpEF, n 386 113 296 114
HFrEF, n 338 193 340 111
Open in a new tab

Data are mean (SD) unless otherwise noted. To convert cholesterol to mmol/L, multiply values by 0.0259.

Cumulative Incidence of HFpEF and HFrEF

Over a mean follow-up of 13.2 ± 3.6 years, there were 715 incident HF events in FHS, of which 636 (89%) were classified according to HF subtype. Similarly, over 11.4 ± 4.3 years in CHS, there were 1304 HF events, of which 724 (56%) were classified. In PREVEND, there were 306 incident HF events over 11.5 ± 2.9 years of follow-up, all of which were classified.

In total across the three cohorts, there were 1,666 classified HF events: 795 (48%) individuals were classified as HFpEF and 871 (52%) as HFrEF. Among classified HF, frequencies of HF subtypes varied by cohort: HFpEF was more common in CHS among classified events (53%) whereas HFrEF was more common in FHS (54%) and PREVEND (63%). Cohort-specific cumulative incidence rates of HFpEF and HFrEF are presented in the Figure.

Figure. Cumulative incidence of HF in CHS, FHS, and PREVEND over 12 years.

Figure

Figure

Open in a new tab

The cumulative incidence of HFpEF is displayed in panel A, and that of HFrEF in panel B. Cumulative incidence estimates accounted for competing risk of death, other HF subtype, and unclassified HF.

Derivation of HFpEF and HFrEF-specific Risk Prediction Models

Derivation and validation samples were created from pooled FHS, CHS, and PREVEND cohorts using a random 2:1 split. Baseline characteristics were similar across derivation (n=14,759) and validation (n=7383) samples (Supplemental Table 1).

HF subtype-specific predictors were first evaluated in the derivation cohort in age- and sex-adjusted analyses. Significant predictors of incident HFpEF included age, systolic blood pressure, body mass index, HDL cholesterol, antihypertensive treatment, diabetes, and previous MI when using a Bonferroni-corrected P-value threshold (Table 2). Predictors of incident HFrEF included age, sex, systolic and diastolic blood pressure, HDL cholesterol, body mass index, smoking status, antihypertensive treatment, ECG LV hypertrophy, left bundle branch block, diabetes mellitus, previous MI, and previous stroke in age- and sex-adjusted models (Table 2).

Table 2.

Risk factors for HFpEF and HFrEF in derivation set; individual factors age and sex-adjusted

HFpEF (n=545) HFrEF (n=584)
sHR* (95% CI) P sHR* (95% CI) P
Age, per 10 years 2.00 (1.86–2.16) <0.0001 1.88 (1.75–2.01) <0.0001
Male sex 0.92 (0.77–1.09) 0.31 2.00 (1.69–2.37) <0.0001
Systolic BP, per 20 mm Hg 1.19 (1.10–1.29) <0.0001 1.27 (1.18–1.37) <0.0001
Diastolic BP, per 10 mm Hg 0.98 (0.91–1.07) 0.69 1.12 (1.04–1.21) 0.004
Heart rate, per 10 bpm 0.98 (0.91–1.06) 0.64 1.08 (1.01–1.17) 0.03
Body mass index, per 4 kg/m2 1.33 (1.25–1.41) <0.0001 1.26 (1.18–1.35) <0.0001
Antihypertensive treatment 1.71 (1.43–2.05) <0.0001 1.86 (1.56–2.21) <0.0001
Diabetes mellitus 1.76 (1.39–2.21) <0.0001 2.34 (1.90–2.89) <0.0001
Current smoker 0.87 (0.67–1.12) 0.27 1.28 (1.04–1.58) 0.02
Modest alcohol use 0.74 (0.59–0.94) 0.01 0.80 (0.65–0.98) 0.03
Previous myocardial infarction 1.59 (1.20–2.11) 0.001 2.95 (2.37–3.68) <0.0001
Previous stroke 1.17 (0.78–1.76) 0.45 1.77 (1.27–2.48) 0.0008
Total cholesterol, per 40 mg/dL 1.02 (0.93–1.10) 0.73 1.05 (0.96–1.14) 0.31
HDL cholesterol, per 15 mg/dL 0.83 (0.75–0.91) <0.0001 0.77 (0.70–0.86) <0.0001
ECG LV hypertrophy 1.29 (0.88–1.89) 0.20 2.60 (1.92–3.52) <0.0001
Left bundle branch block 1.53 (0.92–2.54) 0.10 3.68 (2.51–5.40) <0.0001
Right bundle branch block 1.18 (0.80–1.76) 0.40 0.79 (0.51–1.30) 0.30
Open in a new tab

BP, blood pressure; CI, confidence interval; sHR, sub-distribution hazard ratio;

*

Hazard ratio is expressed per increase in continuous variables as specified in the table, and for presence versus absence of dichotomous variables. To convert cholesterol to mmol/L, multiply values by 0.0259.

Multivariable models were then developed using a stepwise approach to predict HFpEF and HFrEF separately, with age and sex forced in. The final HFpEF-specific risk model included age, sex, systolic blood pressure, body mass index, antihypertensive treatment, and previous MI. Specifically, the relative risk of HFpEF increased 90% per 10 years of age (HR 1.90, 95% CI 1.74–2.07), 14% per 20 mmHg systolic blood pressure (HR 1.14, 95% CI 1.05–1.24), 28% per 4 kg/m2 body mass index (HR 1.28, 95% CI 1.21–1.37), 42% if taking antihypertensive treatment (HR 1.42, 95% CI 1.18–1.71), and 48% with previous MI (HR 1.48, 95% CI 1.12–1.96, Table 3). Sex was forced into the model, and did not predict HFpEF in multivariable analyses (P=0.43).

Table 3.

Final Risk Prediction Models for HFpEF and HFrEF in the derivation set; multivariable models included all variables shown

HFpEF sHR* (95% CI) P
Age, per 10 years 1.90 (1.74–2.07) <0.0001
Male sex 0.93 (0.78–1.11) 0.43
Systolic BP, per 20 mmHg 1.14 (1.05–1.24) 0.003
Body mass index, per 4 kg/m2 1.28 (1.21–1.37) <0.0001
Antihypertensive treatment 1.42 (1.18–1.71) 0.0002
Previous myocardial infarction 1.48 (1.12–1.96) 0.006
HFrEF sHR* (95% CI) P
Age, per 10 years 1.66 (1.52–1.80) <0.0001
Male sex 1.84 (1.55–2.19) <0.0001
Systolic BP, per 20 mmHg 1.20 (1.10–1.30) <0.0001
Body mass index, per 4 kg/m2 1.19 (1.11–1.28) <0.0001
Antihypertensive treatment 1.35 (1.13–1.63) 0.001
Diabetes mellitus 1.83 (1.48–2.26) <0.0001
Current smoker 1.41 (1.14–1.75) 0.0015
Previous myocardial infarction 2.60 (2.08–3.25) <0.0001
ECG LV hypertrophy 2.12 (1.55–2.90) <0.0001
Left bundle branch block 3.17 (2.11–4.78) <0.0001
Open in a new tab

BP, blood pressure; CI, confidence interval; sHR, sub-distribution hazard ratio;

*

Hazard ratio is expressed per increase in continuous variables as specified in the table, and for presence versus absence of dichotomous variables.

The HFrEF-specific multivariable risk model included age, sex, systolic blood pressure, body mass index, smoking status, antihypertensive treatment, LV hypertrophy, left bundle branch block, diabetes, and previous MI. Specifically, the relative risk of HFrEF increased 66% per 10-years of age (HR 1.66, 95% CI 1.52–1.80), 84% for men (HR 1.84, 95% CI 1.55–2.19), 20% per 20 mm Hg systolic blood pressure (HR 1.20, 95% CI 1.10–1.30), 19% per 4 kg/m2 body mass index (HR 1.19, 95% CI 1.11–1.28), 41% in current smokers (HR 1.41, 95% CI 1.14–1.75), 35% if taking antihypertensive treatment (HR 1.35, 95% CI 1.13–1.63), 112% in presence of ECG LV hypertrophy (HR 2.12, 95% CI 1.55–2.90), 217% in presence of left bundle branch block (HR 3.17, 95% CI 2.11–4.78), 83% with diabetes mellitus (HR 1.83, 95% CI 1.48–2.26), and 160% with previous MI (HR 2.60, 95% CI 2.08–3.25, Table 3). Final subtype-specific risk prediction models and examples of 10-year estimated HF risk calculations are provided in Supplemental Table 2.

Performance Metrics and Validation of HF Subtype-Specific Risk Models

The final HFpEF risk prediction model had a c-statistic of 0.80 (95% CI 0.78–0.82) in the derivation sample. When the HFpEF model was applied to the validation sample, the c-statistic was 0.79 (95% CI 0.77–0.82). The model was well calibrated in both derivation and validation sets (X2 statistic 5.29, P=0.73 and X2 statistic 9.02, P=0.34, respectively).

The c-statistic for the HFrEF-specific prediction model was 0.82 (95% CI 0.80–0.84) in the derivation sample. In the validation sample, the c-statistic was 0.80 (95% CI 0.78–0.83). Calibration was reasonable in derivation and validation sets (X2 statistic 13.35, P=0.10 and X2 statistic 14.19, P=0.08, respectively).

External Validation of HF Subtype-Specific Risk Models

External validation was performed among 6,678 MESA participants, with a mean age of 62 ± 10 years and 53% women (Table 1). Over a mean follow-up of 10.4 ± 2.6 years, there were 254 incident HF events, of which 114 were classified as HFpEF, and 111 as HFrEF. The HFpEF-specific model had good discrimination with a c-statistic of 0.76 (95% CI 0.71–0.80), as did the HFrEF-specific model (c-statistic 0.76, 95% CI 0.71–0.80). Both models had good calibration in MESA (X2 statistic 4.54, P=0.81 for HFpEF, X2 statistic 7.56, P=0.48 for HFrEF).

Differential Effects of Predictors on HFpEF vs HFrEF

We tested whether clinical covariates had identical effects on HFpEF vs HFrEF (pooled FHS, CHS, and PREVEND participants) using the Lunn-McNeil method (Table 4).22 Men had higher risk than women for HFrEF but not HFpEF (P for comparison<0.0001). Left bundle branch block and previous MI increased risk more strongly for HFrEF than for HFpEF (P for comparison ≤0.0008 for both). Additionally, age appeared to have a greater risk associated with HFpEF than HFrEF; smoking status and LV hypertrophy were more strongly associated with HFrEF than HFpEF (P for comparison ≤0.02 for all).

Table 4.

Differential effects of risk factors on HFpEF vs HFrEF in multivariable analysis

HFpEF HFrEF P for equality
sHR* (95% CI) sHR* (95% CI)
Age, per 10 years 1.91 (1.78–2.06) 1.69 (1.59–1.81) 0.02
Male sex 0.91 (0.79–1.05) 1.87 (1.63–2.16) <0.0001
Systolic BP, per 20 mmHg 1.13 (1.05–1.21) 1.20 (1.12–1.28) 0.24
Body mass index, per 4 kg/m2 1.28 (1.22–1.36) 1.18 (1.11–1.25) 0.05
Antihypertensive treatment 1.41 (1.21–1.65) 1.33 (1.14–1.54) 0.59
Diabetes mellitus 1.42 (1.17–1.72) 1.58 (1.32–1.90) 0.44
Current smoker 1.04 (0.85–1.28) 1.44 (1.21–1.72) 0.02
Previous myocardial infarction 1.30 (1.02–1.67) 2.70 (2.25–3.24) <0.0001
Left ventricular hypertrophy 1.16 (0.84–1.60) 2.08 (1.60–2.69) 0.009
Left bundle branch block 1.30 (0.81–2.09) 3.65 (2.62–5.09) 0.0008
Open in a new tab

BP, blood pressure; CI, confidence interval; sHR, sub-distribution hazard ratio; bold face indicates significant P-value using a Bonferroni-corrected threshold of 0.005 for number of variables tested.

*

Hazard ratio is for the presence versus absence of dichotomous predictors, and per increase in continuous predictors as specified in the table with all covariates shown in the model simultaneously.

Secondary Analyses and Interactions

After re-classifying HF subtypes using an LVEF 50% as the cut-point, we found that among a total of 1,891 classified HF events across the 4 cohorts, 105 of 909 individuals originally deemed HFpEF were re-classified as HFrEF (5.6% of all classified HF). Age- and sex-adjusted analyses demonstrated minor differences, and the final risk prediction models were also similar after re-classification, with similar directionality and magnitude of effects and slightly less significant p-values in the HFpEF model owing to smaller number of events (Supplemental Tables 3 and 4). Model discrimination remained similar for the HFrEF model, with slightly lower c-statistics noted for the HFpEF model in the validation cohorts (Supplemental Table 5).

In exploratory analyses, we found no significant sex*covariate interactions in the final HF subtype-specific models using the derivation cohort (Supplemental Table 6). We also examined age- and sex-adjusted hazard ratios of all predictors in relation to HF subtypes in the following subgroups: derivation and validation samples, and three primary cohorts. Results were similar across subgroups (Supplemental Tables 7 and 8). Secondary analyses examined a risk prediction model for overall HF (Supplemental Table 9).

DISCUSSION

We developed and validated separate risk prediction models for HFpEF and HFrEF among four longitudinal community-based cohorts with over 1800 incident HF events classified by subtype. These models accounted for multiple competing risks, including death and other HF subtype. Specifically, age, blood pressure, body mass index, and previous MI predicted HFpEF, whereas these risk factors in addition to sex, smoking status, LV hypertrophy, left bundle branch block, and diabetes mellitus predicted incident HFrEF. HF subtype-specific risk models appeared robust upon internal and external validation.

Given the substantial morbidity, mortality, and rising costs associated with HF,2,23 preventive strategies are urgently needed.6 Fundamental elements for developing a disease prevention strategy are a clear understanding of disease risk factors, and the ability to define an at-risk population. Prior studies have focused on HF risk prediction models among community-based samples,24–30 individuals captured at the insurance interface,31 and specific populations enrolled in clinical trials.32,33 Across most studies including ours, age, male sex, hypertension, obesity, diabetes, and prior MI are predictors of risk for incident HF, as summarized in a recent systematic review.7 However, the clinical applicability of existing HF risk prediction models remains unclear, as generalizability has not been tested rigorously. One strength of our study is both internal and external validation of our risk prediction models with robust discrimination and calibration in over 25,000 individuals. We chose LVEF 45% as the cut-off to define HF subtypes, consistent with previously published manuscripts.34 In addition, Solomon et al previously demonstrated that the lower the LVEF, the higher the risk of adverse events.35 However, once elevated to above 45%, LVEF no longer contributes to cardiovascular risk.35 In addition, our secondary analyses after re-classification of HF subtypes using a cut-off of LVEF 50% demonstrated similar results.

Importantly, no previous HF risk prediction studies examined HF subtype-specific prediction models. While the clinical phenotype of HF is largely similar, prior evidence suggests that HFpEF and HFrEF might be separate entities within the spectrum of HF, with distinct etiologies as evidenced by unique patterns of cardiac and cellular remodeling and responses to therapy.9,36 We hypothesized that HF risk prediction may similarly be distinct when examining HFpEF vs HFrEF. Indeed, prior studies have indicated different clusters of risk factors preceding HFpEF vs HFrEF.10–12 However, given the challenge of small sample sizes and multiple competing risks, risk prediction was not previously reported separately for these HF subtypes. We now extend these findings across 4 richly phenotyped community-based cohorts, and have developed and validated HF subtype-specific risk prediction models using readily available clinical measures. While both models had excellent discrimination, the HFrEF model had a higher c-statistic when compared with the HFpEF model in both the derivation and internal validation samples.

Our findings show substantial overlap in modifiable and non-modifiable risk factors leading to both subtypes of HF (age, obesity, hypertension, previous MI), but also support some differences in predictors of HFpEF vs HFrEF. Specifically, male sex, ECG LV hypertrophy, left bundle branch block, previous MI, and smoking status were more strongly associated with HFrEF compared with HFpEF, and age was a stronger predictor of HFpEF compared with HFrEF. These differences were also reflected in the stepwise selection process of unique covariates included in the final HF subtype specific models, and they have been substantiated by prior studies. For example, while ischemic heart disease can lead to both HFpEF and HFrEF, prior MI was more frequently associated with HFrEF in the ADHERE registry,37 and thus it is not surprising that risk factors for coronary artery disease (previous MI, male sex, smoking) are more strongly associated with HFrEF in our study. Similarly, left bundle branch block is known to lead to deterioration of LV systolic function and incite progression to HF via electrical and mechanical remodeling,38 and supports our observed association with HFrEF. The finding that electrocardiographic LV hypertrophy is more strongly associated with HFrEF vs HFpEF is notable. LV hypertrophy can progress to both HFpEF and HFrEF in hypertensive heart disease.39 Differential effects of LV hypertrophy on HFpEF and HFrEF may be due to poor sensitivity of ECG LVH for detecting LV hypertrophy on echo, particularly in women and in the setting of obesity.40

Although we found that discrimination of the HFpEF risk prediction model was not as good as that of the HFrEF prediction model, our data provide novel insight into this phenotype as well. For instance, although it has been recognized that a large portion of patients with HFpEF suffer from coronary artery disease, this is generally not viewed as a predictor for HFpEF. Herein, we demonstrate that a history of MI is predictive for the development of HFpEF, although less strong than it is for HFrEF. Likewise, it has been postulated that HFpEF is a disease of “elderly females”, but the value of sex for prediction of HFpEF was not as strong as been put forward before. Clearly, HFpEF risk prediction remains more difficult, and additional parameters may improve our models.

Several limitations deserve mention. First, our models focused only on readily available clinical characteristics, similar to 10-year cardiovascular disease risk prediction models.41 We acknowledge that the use of a Bonferroni-corrected p-value threshold is a conservative approach, and may result in exclusion of predictors with small or modest effects from the final risk prediction models. Prior studies have demonstrated the potential utility of imaging and biomarkers, including natriuretic peptides and high-sensitivity troponin, in refining clinical HF risk prediction models.29,42–45 Whether the addition of biomarkers and other noncardiac comorbidities such as pulmonary disease, anemia, sleep apnea, and kidney disease may improve discrimination and differentially predict HF subtype will need to be examined in future studies. Next, HFpEF and HFrEF by definition included only individuals who underwent LV function assessment at or around the time of HF presentation, leaving 27% of cases as unclassified HF. This may have led to differential bias and potential cohort-based differences given that a greater proportion of HF cases in CHS were unclassified. However, most covariates had similar effects across cohorts, and given widely varying proportions of unclassified HF among cohorts, this may be reassuring that unclassified HF may not have differentially affected HF subtypes. Definitions of HF and LVH varied slightly by cohort, and may underlie cohort-specific differences observed. We did not account for relatedness among Framingham original cohort and offspring samples, which may have influenced FHS-specific results. External validation in the ethnically diverse MESA cohort demonstrated robust risk models, although we did not have adequate power to perform race-specific HF subtype analyses.

In summary, we developed and validated separate risk prediction models for HFpEF and HFrEF by leveraging data from four longitudinal community-based cohorts spanning a broad range of ages. We found substantial overlap in risk factors for incident HFpEF and HFrEF, although some risk factors displayed differential effects on HF subtypes. While therapies to prevent mortality in patients with symptomatic HF are predicated on the distinction between HFpEF and HFrEF,4 current preventive strategies in the preclinical stages of HF do not distinguish between HF subtype. Recent data support the notion that a screening strategy targeting at-risk patients in primary care clinics may prevent HF-related outcomes,46 although community-wide screening using biomarkers or echocardiography is not supported.47 Further studies are needed to examine the clinical utility of HF subtype risk prediction, with the ultimate goal of targeted preventive strategies.

Supplementary Material

Supplemental Material

Clinical Perspective.

Heart failure accounts for a substantial burden of total health care costs world wide, and about half of individuals presenting with heart failure have heart failure with preserved as opposed to reduced ejection fraction (HFpEF and HFrEF, respectively). Risk prediction specific to heart failure subtype may be able to improve upon existing risk prediction tools and guide future preventive strategies. We developed and validated separate risk prediction models for HFpEF and HFrEF by leveraging data from four longitudinal community-based cohorts spanning a broad range of ages. We found substantial overlap in risk factors for incident HFpEF and HFrEF, although some risk factors displayed differential effects on HF subtypes. Recent data support the notion that a screening strategy targeting at-risk patients in primary care clinics may prevent heart failure related outcomes. Future studies are needed to examine the clinical utility of heart failure subtype risk prediction, with the ultimate goal of targeted preventive strategies.

Acknowledgments

Sources of Funding

This work was partially supported by the National Heart, Lung and Blood Institute (Framingham Heart Study, contract N01-HC25195 and HHSN268201500001I; Cardiovascular Health Study, contracts HHSN268201200036C, HHSN268200800007C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and grant U01HL080295; Multi-Ethnic Study of Atherosclerosis contract N01-HC95159 through N01-HC95166 and N01-HC95169; N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168 and N01-HC-95169), the National Center for Research Resources (Multi-Ethnic Study of Atherosclerosis UL1-TR-000040 and UL1-TR-001079). The Cardiovascular Health Study received additional contribution from the National Institute of Neurological Disorders and Stroke, and R01AG023629 from the National Institute on Aging. A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Dr. Ho is supported by K23-HL116780. Dr. Lee is supported by a clinician-scientist award from the Canadian Institutes of Health Research. Dr. Ho 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.

Disclosures: Dr. Psaty serves on a DSMB for a clinical trial funded by the manufacturer (Zoll LifeCor) and on the Steering Committee of the Yale Open Data Access Project funded by Johnson & Johnson

REFERENCES

  • 1.Lloyd-Jones DM, Larson MG, Leip EP, Beiser A, D’Agostino RB, Kannel WB, Murabito JM, Vasan RS, Benjamin EJ, Levy D Framingham Heart Study. Lifetime risk for developing congestive heart failure: the Framingham Heart Study. Circulation. 2002;106:3068–3072. doi: 10.1161/01.cir.0000039105.49749.6f. [DOI] [PubMed] [Google Scholar]
  • 2.Heidenreich PA, Albert NM, Allen LA, Bluemke DA, Butler J, Fonarow GC, Ikonomidis JS, Khavjou O, Konstam MA, Maddox TM, Nichol G, Pham M, Piña IL, Trogdon JG. Forecasting the impact of heart failure in the United States: a policy statement from the American Heart Association. Circ Heart Fail. 2013;6:606–619. doi: 10.1161/HHF.0b013e318291329a. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.McMurray JJ, Petrie MC, Murdoch DR, Davie AP. Clinical epidemiology of heart failure: public and private health burden. Eur Heart J. 1998;(19 Suppl P):P9–P16. [PubMed] [Google Scholar]
  • 4.Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE, Drazner MH, Fonarow GC, Geraci SA, Horwich T, Januzzi JL, Johnson MR, Kasper EK, Levy WC, Masoudi FA, McBride PE, McMurray JJV, Mitchell JE, Peterson PN, Riegel B, Sam F, Stevenson LW, Tang WHW, Tsai EJ, Wilkoff BL. 2013 ACCF/AHA Guideline for the Management of Heart Failure. Circulation. 2013;128 doi: 10.1161/CIR.0b013e31829e8776. e240-327-e327. [DOI] [PubMed] [Google Scholar]
  • 5.Lloyd-Jones DM, Hong Y, Labarthe D, Mozaffarian D, Appel LJ, Van Horn L, Greenlund K, Daniels S, Nichol G, Tomaselli GF, Arnett DK, Fonarow GC, Ho PM, Lauer MS, Masoudi FA, Robertson RM, Roger V, Schwamm LH, Sorlie P, Yancy CW, Rosamond WD. Defining and setting national goals for cardiovascular health promotion and disease reduction: the American Heart Association’s strategic Impact Goal through 2020 and beyond. 2010;121:586–613. doi: 10.1161/CIRCULATIONAHA.109.192703. [DOI] [PubMed] [Google Scholar]
  • 6.Schocken DD, Benjamin EJ, Fonarow GC, Krumholz HM, Levy D, Mensah GA, Narula J, Shor ES, Young JB, Hong Y. Prevention of heart failure: a scientific statement from the American Heart Association Councils on Epidemiology and Prevention, Clinical Cardiology, Cardiovascular Nursing, and High Blood Pressure Research; Quality of Care and Outcomes Research Interdisciplinary Working Group; and Functional Genomics and Translational Biology Interdisciplinary Working Group. Circulation. 2008;117:2544–2565. doi: 10.1161/CIRCULATIONAHA.107.188965. [DOI] [PubMed] [Google Scholar]
  • 7.Echouffo-Tcheugui JB, Greene SJ, Papadimitriou L, Zannad F, Yancy CW, Gheorghiade M, Butler J. Population Risk Prediction Models for Incident Heart Failure: A Systematic Review. Circ Heart Fail. 2015;8:438–447. doi: 10.1161/CIRCHEARTFAILURE.114.001896. [DOI] [PubMed] [Google Scholar]
  • 8.McMurray JJV, Adamopoulos S, Anker SD, Auricchio A, Böhm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Gomez-Sanchez MA, Jaarsma T, Køber L, Lip GYH, Maggioni AP, Parkhomenko A, Pieske BM, Popescu BA, Rønnevik PK, Rutten FH, Schwitter J, Seferovic P, Stepinska J, Trindade PT, Voors AA, Zannad F, Zeiher A. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012. Eur Heart J. 2012;33:1787–1847. doi: 10.1093/eurheartj/ehs104. [DOI] [PubMed] [Google Scholar]
  • 9.Borlaug BA, Redfield MM. Diastolic and systolic heart failure are distinct phenotypes within the heart failure spectrum. Circulation. 2011;123:2006–2013. doi: 10.1161/CIRCULATIONAHA.110.954388. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Ho JE, Gona P, Pencina MJ, Tu JV, Austin PC, Vasan RS, Kannel WB, D’Agostino RB, Lee DS, Levy D. Discriminating clinical features of heart failure with preserved vs. reduced ejection fraction in the community. Eur Heart J. 2012;33:1734–1741. doi: 10.1093/eurheartj/ehs070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ho JE, Lyass A, Lee DS, Vasan RS, Kannel WB, Larson MG, Levy D. Predictors of new-onset heart failure: differences in preserved versus reduced ejection fraction. Circ Heart Fail. 2013;6:279–286. doi: 10.1161/CIRCHEARTFAILURE.112.972828. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Brouwers FP, de Boer RA, van der Harst P, Voors AA, Gansevoort RT, Bakker SJ, Hillege HL, van Veldhuisen DJ, van Gilst WH. Incidence and epidemiology of new onset heart failure with preserved vs. reduced ejection fraction in a community-based cohort: 11-year follow-up of PREVEND. Eur Heart J. 2013;34:1424–1431. doi: 10.1093/eurheartj/eht066. [DOI] [PubMed] [Google Scholar]
  • 13.Dawber TR, Kannel WB, Lyell LP. An approach to longitudinal studies in a community: the Framingham Study. Ann N Y Acad Sci. 1963;107:539–556. doi: 10.1111/j.1749-6632.1963.tb13299.x. [DOI] [PubMed] [Google Scholar]
  • 14.Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979;110:281–290. doi: 10.1093/oxfordjournals.aje.a112813. [DOI] [PubMed] [Google Scholar]
  • 15.Psaty BM, Kuller LH, Bild D, Burke GL, Kittner SJ, Mittelmark M, Price TR, Rautaharju PM, Robbins J. Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol. 1995;5:270–277. doi: 10.1016/1047-2797(94)00092-8. [DOI] [PubMed] [Google Scholar]
  • 16.Diercks GF, Janssen WM, van Boven AJ, Bak AA, de Jong PE, Crijns HJ, van Gilst WH. Rationale, design, and baseline characteristics of a trial of prevention of cardiovascular and renal disease with fosinopril and pravastatin in nonhypertensive, nonhypercholesterolemic subjects with microalbuminuria (the Prevention of REnal and Vascular ENdstage Disease Intervention Trial [PREVEND IT]) Am J Cardiol. 2000;86:635–638. doi: 10.1016/s0002-9149(00)01042-0. [DOI] [PubMed] [Google Scholar]
  • 17.Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, Greenland P, Jacob DR, Kronmal R, Liu K, Nelson JC, O’Leary D, Saad MF, Shea S, Szklo M, Tracy RP. Multi-ethnic study of atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–881. doi: 10.1093/aje/kwf113. [DOI] [PubMed] [Google Scholar]
  • 18.Gaynor JJ, Feuer EJ, Tan CC, Wu DH. On the use of cause-specific failure and conditional failure probabilities: examples from clinical oncology data. J Am Stat Assoc. 1993;86:770–778. [Google Scholar]
  • 19.Fine JP, Gray RJ. A Proportional Hazards Model for the Subdistribution of a Competing Risk. J Am Stat Assoc. 1999;94:496–509. [Google Scholar]
  • 20.Kohl M, Plischke M, Leffondré K, Heinze G. PSHREG: a SAS macro for proportional and nonproportional subdistribution hazards regression. Comput Methods Programs Biomed. 2015;118:218–233. doi: 10.1016/j.cmpb.2014.11.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Pencina MJ, D’Agostino RB, Song L. Quantifying discrimination of Framingham risk functions with different survival C statistics. Stat Med. 2012;31:1543–1553. doi: 10.1002/sim.4508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Lunn M, McNeil D. Applying Cox regression to competing risks. Biometrics. 1995;51:524–532. [PubMed] [Google Scholar]
  • 23.Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Blaha MJ, Dai S, Ford ES, Fox CS, Franco S, Fullerton HJ, Gillespie C, Hailpern SM, Heit JA, Howard VJ, Huffman MD, Judd SE, Kissela BM, Kittner SJ, Lackland DT, Lichtman JH, Lisabeth LD, Mackey RH, Magid DJ, Marcus GM, Marelli A, Matchar DB, McGuire DK, Mohler ER, Moy CS, Mussolino ME, Neumar RW, Nichol G, Pandey DK, Paynter NP, Reeves MJ, Sorlie PD, Stein J, Towfighi A, Turan TN, Virani SS, Wong ND, Woo D, Turner MB. Heart disease and stroke statistics--2014 update: a report from the American Heart Association. Circulation. 2014;129:e28–e292. doi: 10.1161/01.cir.0000441139.02102.80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Chahal H, Bluemke DA, Wu CO, McClelland R, Liu K, Shea SJ, Burke G, Balfour P, Herrington D, Shi P, Post W, Olson J, Watson KE, Folsom AR, Lima JAC. Heart failure risk prediction in the Multi-Ethnic Study of Atherosclerosis. Heart. 2015;101:58–64. doi: 10.1136/heartjnl-2014-305697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Agarwal SK, Chambless LE, Ballantyne CM, Astor B, Bertoni AG, Chang PP, Folsom AR, He M, Hoogeveen RC, Ni H, Quibrera M, Rosamond WD, Russell SD, Shahar E, Heiss G. Prediction of Incident Heart Failure in General Practice: The ARIC Study. Circ Heart Fail. 2012 doi: 10.1161/CIRCHEARTFAILURE.111.964841. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Kalogeropoulos A, Psaty BM, Vasan RS, Georgiopoulou V, Smith AL, Smith NL, Kritchevsky SB, Wilson PWF, Newman AB, Harris TB, Butler J. Cardiovascular Health Study. Validation of the health ABC heart failure model for incident heart failure risk prediction: the Cardiovascular Health Study. Circ Heart Fail. 2010;3:495–502. doi: 10.1161/CIRCHEARTFAILURE.109.904300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Djoussé L, Driver JA, Gaziano JM. Relation between modifiable lifestyle factors and lifetime risk of heart failure. JAMA. 2009;302:394–400. doi: 10.1001/jama.2009.1062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Butler J, Kalogeropoulos A, Georgiopoulou V, Belue R, Rodondi N, Garcia M, Bauer DC, Satterfield S, Smith AL, Vaccarino V, Newman AB, Harris TB, Wilson PWF. Kritchevsky SB, Health ABC Study. Incident heart failure prediction in the elderly: the health ABC heart failure score. Circ Heart Fail. 2008;1:125–133. doi: 10.1161/CIRCHEARTFAILURE.108.768457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.deFilippi CR, de Lemos JA, Christenson RH, Gottdiener JS, Kop WJ, Zhan M, Seliger SL. Association of serial measures of cardiac troponin T using a sensitive assay with incident heart failure and cardiovascular mortality in older adults. JAMA. 2010;304:2494–2502. doi: 10.1001/jama.2010.1708. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Kannel WB, D’Agostino RB, Silbershatz H, Belanger AJ, Wilson PW, Levy D. Profile for estimating risk of heart failure. Arch Intern Med. 1999;159:1197–1204. doi: 10.1001/archinte.159.11.1197. [DOI] [PubMed] [Google Scholar]
  • 31.Goyal A, Norton CR, Thomas TN, Davis RL, Butler J, Ashok V, Zhao L, Vaccarino V, Wilson PWF. Predictors of incident heart failure in a large insured population: a one million person-year follow-up study. Circ Heart Fail. 2010;3:698–705. doi: 10.1161/CIRCHEARTFAILURE.110.938175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Wong YW, Thomas L, Sun JL, McMurray JJV, Krum H, Hernandez AF, Rutten GEHM, Leiter LA, Standl E, Haffner SM, Mazzone T, Martinez FA, Tognoni G, Giles T, Califf RM. Predictors of incident heart failure hospitalizations among patients with impaired glucose tolerance: insight from the Nateglinide And Valsartan in Impaired Glucose Tolerance Outcomes Research study. Circ Heart Fail. 2013;6:203–210. doi: 10.1161/CIRCHEARTFAILURE.112.000086. [DOI] [PubMed] [Google Scholar]
  • 33.Pfister R, Cairns R, Erdmann E, Schneider CA. A clinical risk score for heart failure in patients with type 2 diabetes and macrovascular disease: an analysis of the PROactive study. Int J Cardiol. 2013;162:112–116. doi: 10.1016/j.ijcard.2011.05.056. [DOI] [PubMed] [Google Scholar]
  • 34.Lee DS, Gona P, Vasan RS, Larson MG, Benjamin EJ, Wang TJ, Tu JV, Levy D. Relation of disease pathogenesis and risk factors to heart failure with preserved or reduced ejection fraction. Circulation. 2009;119:3070–3077. doi: 10.1161/CIRCULATIONAHA.108.815944. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Solomon SD, Anavekar N, Skali H, McMurray JJV, Swedberg K, Yusuf S, Granger CB, Michelson EL, Wang D, Pocock S, Pfeffer MA. Influence of ejection fraction on cardiovascular outcomes in a broad spectrum of heart failure patients. Circulation. 2005;112:3738–3744. doi: 10.1161/CIRCULATIONAHA.105.561423. [DOI] [PubMed] [Google Scholar]
  • 36.Paulus WJ, Tschöpe C. A novel paradigm for heart failure with preserved ejection fraction: comorbidities drive myocardial dysfunction and remodeling through coronary microvascular endothelial inflammation. J Am Coll Cardiol. 2013;62:263–271. doi: 10.1016/j.jacc.2013.02.092. [DOI] [PubMed] [Google Scholar]
  • 37.Yancy CW, Lopatin M, Stevenson LW, De Marco T, Fonarow GC. ADHERE Scientific Advisory Committee and Investigators. Clinical presentation, management, and in-hospital outcomes of patients admitted with acute decompensated heart failure with preserved systolic function: a report from the Acute Decompensated Heart Failure National Registry (ADHERE) Database. J Am Coll Cardiol. 2006;47:76–84. doi: 10.1016/j.jacc.2005.09.022. [DOI] [PubMed] [Google Scholar]
  • 38.Zannad F, Huvelle E, Dickstein K, van Veldhuisen DJ, Stellbrink C, Køber L, Cazeau S, Ritter P, Maggioni AP, Ferrari R, Lechat P. Left bundle branch block as a risk factor for progression to heart failure. Eur J Heart Fail. 2007;9:7–14. doi: 10.1016/j.ejheart.2006.04.011. [DOI] [PubMed] [Google Scholar]
  • 39.Drazner MH. The progression of hypertensive heart disease. Circulation. 2011;123:327–334. doi: 10.1161/CIRCULATIONAHA.108.845792. [DOI] [PubMed] [Google Scholar]
  • 40.Levy D, Labib SB, Anderson KM, Christiansen JC. Determinants of sensitivity and specificity of electrocardiographic criteria for left ventricular hypertrophy. Circulation. 1990;81:815–820. doi: 10.1161/01.cir.81.3.815. [DOI] [PubMed] [Google Scholar]
  • 41.Goff DC, Lloyd-Jones DM, Bennett G, Coady S, D’Agostino RB, Gibbons R, Greenland P, Lackland DT, Levy D, O’Donnell CJ, Robinson JG, Schwartz JS, Shero ST, Smith SC, Sorlie P, Stone NJ, Wilson PWF, Jordan HS, Nevo L, Wnek J, Anderson JL, Halperin JL, Albert NM, Bozkurt B, Brindis RG, Curtis LH, Demets D, Hochman JS, Kovacs RJ, Ohman EM, Pressler SJ, Sellke FW, Shen W-K, Smith SC, Tomaselli GF. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129:S49–S73. doi: 10.1161/01.cir.0000437741.48606.98. [DOI] [PubMed] [Google Scholar]
  • 42.Brouwers FP, van Gilst WH, Damman K, Van den Berg MP, Gansevoort RT, Bakker SJL, Hillege HL, van Veldhuisen DJ, van der Harst P, de Boer RA. Clinical risk stratification optimizes value of biomarkers to predict new-onset heart failure in a community-based cohort. Circ Heart Fail. 2014;7:723–731. doi: 10.1161/CIRCHEARTFAILURE.114.001185. [DOI] [PubMed] [Google Scholar]
  • 43.Wannamethee SG, Welsh P, Whincup PH, Lennon L, Papacosta O, Sattar N. N-terminal pro brain natriuretic peptide but not copeptin improves prediction of heart failure over other routine clinical risk parameters in older men with and without cardiovascular disease: population-based study. Eur J Heart Fail. 2014;16:25–32. doi: 10.1093/eurjhf/hft124. [DOI] [PubMed] [Google Scholar]
  • 44.Nambi V, Liu X, Chambless LE, de Lemos JA, Virani SS, Agarwal S, Boerwinkle E, Hoogeveen RC, Aguilar D, Astor BC, Srinivas PR, Deswal A, Mosley TH, Coresh J, Folsom AR, Heiss G, Ballantyne CM. Troponin T and N-terminal pro-B-type natriuretic peptide: a biomarker approach to predict heart failure risk--the atherosclerosis risk in communities study. Clin Chem. 2013;59:1802–1810. doi: 10.1373/clinchem.2013.203638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Smith JG, Newton-Cheh C, Almgren P, Struck J, Morgenthaler NG, Bergmann A, Platonov PG, Hedblad B, Engström G, Wang TJ, Melander O. Assessment of conventional cardiovascular risk factors and multiple biomarkers for the prediction of incident heart failure and atrial fibrillation. J Am Coll Cardiol. 2010;56:1712–1719. doi: 10.1016/j.jacc.2010.05.049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Ledwidge M, Gallagher J, Conlon C, Tallon E, O’Connell E, Dawkins I, Watson C, O’Hanlon R, Bermingham M, Patle A, Badabhagni MR, Murtagh G, Voon V, Tilson L, Barry M, McDonald L, Maurer B, McDonald K. Natriuretic peptide-based screening and collaborative care for heart failure: the STOP-HF randomized trial. JAMA. 2013;310:66–74. doi: 10.1001/jama.2013.7588. [DOI] [PubMed] [Google Scholar]
  • 47.Wang TJ, Levy D, Benjamin EJ, Vasan RS. The epidemiology of “asymptomatic” left ventricular systolic dysfunction: implications for screening. Ann Intern Med. 2003;138:907–916. doi: 10.7326/0003-4819-138-11-200306030-00012. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS788516-supplement-Supplemental_Material.pdf (119.3KB, pdf)

RESOURCES