This case-control study evaluates the performance of the H2FPEF and HFA-PEFF algorithms to diagnose heart failure with preserved ejection fraction among ambulatory patients with dyspnea.
Key Points
Question
What is the performance of the H2FPEF and HFA-PEFF algorithms to diagnose heart failure with preserved ejection fraction (HFpEF) compared with the invasive gold standard of an elevated pulmonary capillary wedge pressure (PCWP) during exercise?
Findings
In this case-control study of 736 patients, H2FPEF and HFA-PEFF scores provided discriminatory information to identify HFpEF, but there was superior discrimination for the H2FPEF score. The use of an alternative criterion, the PCWP/cardiac output slope, led to potential misclassification in 20% of patients.
Meaning
The findings suggest that the H2FPEF and HFA-PEFF algorithms can discriminate patients with HFpEF from control individuals among ambulatory patients with dyspnea and that the H2FPEF score provides superior diagnostic performance despite fewer input variables.
Abstract
Importance
Diagnosis of heart failure with preserved ejection fraction (HFpEF) among dyspneic patients without overt congestion is challenging. Multiple diagnostic approaches have been proposed but are not well validated against the independent gold standard for HFpEF diagnosis of an elevated pulmonary capillary wedge pressure (PCWP) during exercise.
Objective
To evaluate H2FPEF and HFA-PEFF scores and a PCWP/cardiac output (CO) slope of more than 2 mm Hg/L/min to diagnose HFpEF.
Design, Setting, and Participants
This retrospective case-control study included patients with unexplained dyspnea from 6 centers in the US, the Netherlands, Denmark, and Australia from March 2016 to October 2020. Diagnosis of HFpEF (cases) was definitively ascertained by the presence of elevated PCWP during exertion; control individuals were those with normal rest and exercise hemodynamics.
Main Outcomes and Measures
Logistic regression was used to evaluate the accuracy of HFA-PEFF and H2FPEF scores to discriminate patients with HFpEF from controls.
Results
Among 736 patients, 563 (76%) were diagnosed with HFpEF (mean [SD] age, 69 [11] years; 334 [59%] female) and 173 (24%) represented controls (mean [SD] age, 60 [15] years; 109 [63%] female). H2FPEF and HFA-PEFF scores discriminated patients with HFpEF from controls, but the H2FPEF score had greater area under the curve (0.845; 95% CI, 0.810-0.875) compared with the HFA-PEFF score (0.710; 95% CI, 0.659-0.756) (difference, −0.134; 95% CI, –0.177 to −0.094; P < .001). Specificity was robust for both scores, but sensitivity was poorer for HFA-PEFF, with a false-negative rate of 55% for low-probability scores compared with 25% using the H2FPEF score. Use of the PCWP/CO slope to redefine HFpEF rather than exercise PCWP reclassified 20% (117 of 583) of patients, but patients reclassified from HFpEF to control by this metric had clinical, echocardiographic, and hemodynamic features typical of HFpEF, including elevated resting PCWP in 66% (46 of 70) of reclassified patients.
Conclusions and Relevance
In this case-control study, despite requiring fewer data, the H2FPEF score had superior diagnostic performance compared with the HFA-PEFF score and PCWP/CO slope in the evaluation of unexplained dyspnea and HFpEF in the outpatient setting.
Introduction
Exertional dyspnea is one of the most common symptoms reported in clinical practice, particularly among older adults and adults with obesity.1 Heart failure (HF) with preserved ejection fraction (HFpEF) is a common etiology that is important to identify to guide proper treatment.2,3,4 Diagnosis of HFpEF is straightforward among patients with overt congestion but is often challenging in ambulatory patients with exertional dyspnea, in whom symptoms and hemodynamic abnormalities often manifest only during exertional activities.5,6,7,8,9,10,11 Hemodynamic exercise testing has thus emerged as the gold standard to definitively establish or refute the diagnosis among these patients,7,9,12 but estimates from community-based studies still show that many patients with dyspnea due to HFpEF remain undiagnosed.13
Clinical scoring models are widely available in cardiovascular medicine to enhance diagnosis, risk stratification, and medical decision-making, but few are routinely used in clinical practice because of complexity, time requirements for application, and lack of external validation reducing confidence in their veracity.14,15,16 Two scoring models were developed to enhance diagnostic evaluation in HFpEF: the H2FPEF and HFA-PEFF algorithms.7,10 These models have been validated using the reference standards of diagnosis established by clinical impression,17,18,19,20 trial eligibility,21,22,23 and previous HF hospitalization,24,25 but to our knowledge, no multicenter study has yet validated these approaches using the invasive gold standard.7 In this study, we sought to validate HFA-PEFF and H2FPEF scores in patients with unexplained dyspnea from a large, multicenter, international cohort, with case-control status ascertained definitively using the gold standard—an elevated pulmonary capillary wedge pressure (PCWP) during exercise.
Methods
This international, multicenter, case-control study included patients with unexplained dyspnea and normal EF (≥50%) who underwent evaluation for possible HFpEF in 3 centers in the US (Mayo Clinic, Rochester, Minnesota; Johns Hopkins Hospital, Baltimore, Maryland; and Medical University of South Carolina, Charleston) and 3 international centers (Amsterdam University Medical Centre, Amsterdam, the Netherlands; Aarhus University Hospital, Aarhus, Denmark; and Alfred Hospital, Melbourne, Victoria, Australia) from March 2016 to October 2020. Because HFA-PEFF and H2FPEF scores require echocardiographic variables, only patients with echocardiographic data available were included. The H2FPEF score was originally derived and then validated from a cohort of patients who underwent invasive exercise testing for the evaluation of unexplained dyspnea at 1 center between 2006 and 2016.10 Because of the potential for overinflation of diagnostic performance, patients from this original derivation and validation cohort were not included. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. The study was approved by the Mayo Clinic institutional review board, which waived informed consent because of the use of deidentified patient data.
Case Definitions and Ascertainment
Across all centers, patients were studied in the supine position at rest and during supine cycle ergometry exercise to exhaustion.9,10 For the primary analysis, positive diagnosis of HFpEF (cases) was defined as elevated PCWP at rest (≥15 mm Hg) or during exercise (≥25 mm Hg) at cardiac catheterization based on current guidelines.7 Control individuals were defined as those with normal rest and exercise hemodynamics across all centers, with symptoms deemed primarily related to deconditioning. More details of exclusion criteria and evaluation are given in the eMethods in the Supplement.
H2FPEF and HFA-PEFF Scores
The European Society of Cardiology HFA-PEFF score is calculated from echocardiography and natriuretic peptide data (eFigure 1 in the Supplement).7 Scores of 0 to 1 are considered to represent low probability of HFpEF (rule out), and scores of 5 to 6 represent high probability (rule in).7 The H2FPEF score includes echocardiographic and clinical variables and ranges from 0 to 9 (Figure).10 H2FPEF scores of 0 to 1 are associated with a low probability of HFpEF (<25%; rule out), and scores of 6 to 9 are associated with a high probability of HFpEF (>90%; rule in).10 Patients with scores in the intermediate category for both scores require additional exercise testing for diagnosis.
Figure. Comparison of H2FPEF and HFA-PEFF Scores for Diagnosis of Heart Failure With Preserved Ejection Fraction (HFpEF).
H2FPEF and HFA-PEFF scores of 0 to 1 indicated a low probability of HFpEF; H2FPEF scores of 6 to 9 and HFA-PEFF scores of 5 to 6 indicated a high probability of HFpEF. Intermediate probability of HFpEF required additional exercise testing for diagnosis. C, The P value is for the difference in area under the curve between scores. ROC indicates receiver operating characteristic.
An additional sensitivity analysis was performed in which HFpEF case status was defined using a recently proposed alternative hemodynamic criterion: the slope of increase in PCWP compared with the increase in cardiac output (CO) with exercise of more than 2 mm Hg/L/min rather than the absolute PCWP at rest and exercise.26
Statistical Analysis
Logistic regression with receiver operating characteristic curves was used to assess the strength of association and discriminatory performance of the variables included in the HFA-PEFF and H2FPEF scores. Because these scores are used as both rule-out and rule-in tests, sensitivity of the suggested low-probability HFpEF cut points (≤1 for both scores) and specificity of the suggested high-probability cut points (≥6 for the H2FPEF score and ≥5 for the HFA-PEFF score) were tested by logistic regression compared with the gold standard diagnosis from invasive exercise testing. Positive predictive value (PPV) and negative predictive value (NPV) were plotted as a function of disease prevalence based on sensitivity and specificity, according to formulas derived from Bayes theorem. Full details are included in the eMethods in the Supplement. Two-sided P < .05 was considered significant. We analyzed data using JMP, version 14.1.0 (JMP Statistical Discovery LLC) and BlueSky Statistics, version 7.40 (BlueSky Statistics LLC).
Results
Among 736 patients with unexplained dyspnea who had data necessary to calculate the HFA-PEFF or H2FPEF score across all centers, 563 (76%) were diagnosed with HFpEF (mean [SD] age, 69 [11] years; 334 [59%] female) and 173 (24%) represented controls (mean [SD] age, 60 [15] years; 109 [63%] female). The distribution of scores, patients from participating centers, and the clinical characteristics of patients are presented in eTables 1 to 3 and eFigure 2 in the Supplement. Overall, compared with controls, patients with HFpEF were older, were more likely to have obesity, and had a higher prevalence of hypertension, atrial fibrillation (AF), and kidney dysfunction with higher natriuretic peptide levels (Table 1).
Table 1. Baseline Characteristics of Study Patientsa.
| Characteristic | Control individuals (n = 173) | Patients with HFpEF (n = 563) | P value |
|---|---|---|---|
| Age, y | 60 (15) | 69 (11) | <.001 |
| Sex, No. (%) | |||
| Female | 109 (63) | 334 (59) | .42 |
| Male | 64 (37) | 229 (41) | |
| Body mass indexb | 27.9 (5.8) | 32.4 (7.2) | <.001 |
| Comorbidities, No. (%) | |||
| Hypertension | 111 (64) | 463 (82) | <.001 |
| Diabetes | 25 (14) | 108 (19) | .15 |
| Atrial fibrillation | |||
| Any | 14 (8) | 212 (38) | <.001 |
| Paroxysmal | 10 (6) | 126 (22) | <.001 |
| Permanent | 4 (2) | 86 (15) | <.001 |
| Hemoglobin level, g/dL | 13.1 (1.6) | 13.0 (1.6) | .20 |
| NTproBNP level, mean (95% CI), pg/mL | 83 (50-198) | 273 (99-868) | <.001 |
| Creatinine level, mg/dL | 0.98 (0.32) | 1.12 (0.62) | .007 |
| Echocardiography | |||
| LV end diastolic dimension, mm | 47 (5) | 48 (7) | .09 |
| LV mass index, g/m2 | 78 (25) | 86 (27) | <.001 |
| Relative wall thickness | 0.39 (0.10) | 0.41 (0.10) | .001 |
| Septal wall thickness, mm | 9.4 (2.3) | 10.3 (2.3) | <.001 |
| LV ejection fraction, % | 62 (6) | 61 (6) | .03 |
| LA volume index, mL/m2 | 32 (12) | 37 (14) | <.001 |
| Septal E/e′ ratio | 9.3 (3.8) | 12.9 (6.5) | <.001 |
| Septal e′, cm/s | 8 (3) | 7 (2) | <.001 |
| TR velocity, m/s | 2.45 (0.45) | 2.79 (0.58) | <.001 |
| PASP, mm Hg | 30 (11) | 39 (15) | <.001 |
| Invasive hemodynamics | |||
| Rest PCWP, mm Hg | 10 (3) | 16 (6) | <.001 |
| Rest mean PAP, mm Hg | 18 (5) | 28 (10) | <.001 |
| Rest CO, L/minc | 5.5 (1.6) | 5.3 (1.9) | .36 |
| Peak PCWP, mm Hg | 18 (5) | 32 (6) | <.001 |
| Peak mean PAP, mm Hg | 31 (8) | 45 (11) | <.001 |
| Peak CO, L/minc | 10.6 (3.3) | 9.1 (3.4) | <.001 |
| PCWP/CO slope, mean (95% CI), mm Hg/L/minc | 1.6 (1.0-2.7) | 4.2 (2.6-7.5) | <.001 |
| PCWP/CO slope >2 mm Hg/L/min, No. (%)c | 47 (32) | 367 (84) | <.001 |
Abbreviations: CO, cardiac output; e′, septal mitral annulus tissue relaxation velocity in early diastole; E/e′, ratio of early diastolic mitral inflow velocity to e′; HFpEF, heart failure with preserved ejection fraction; LA, left atrium; LV, left ventricle; NA, not applicable; NTproBNP, N-terminal pro brain natriuretic peptide; PAP, pulmonary artery pressure; PASP, pulmonary artery systolic pressure; PCWP, pulmonary capillary wedge pressure; TR, tricuspid regurgitation.
SI conversion factors: To convert BNP to ng/L, multiply by 1.0; creatinine to μmol/L, multiply by 88.4; hemoglobin to g/L, multiply by 10.0.
Data are presented as mean (SD) values unless otherwise indicated.
Calculated as weight in kilograms divided by height in meters squared.
PCWP/CO slope was available for 146 controls and 437 patients with HFpEF.
On echocardiography, patients with HFpEF were more likely to have diastolic dysfunction, with higher noninvasive estimates of filling pressure (higher early diastolic mitral inflow velocity to septal mitral annulus tissue relaxation velocity [E/e′] ratio), lower e′, and greater estimated right ventricular systolic pressure than controls (Table 1). Ejection fraction was slightly lower in the group with HFpEF. Left atrial volume, left ventricular mass, relative wall thickness, and septal hypertrophy were greater in the group with HFpEF.
Individual Components of the Scores
All components of the H2FPEF score and nearly all components of the HFA-PEFF score demonstrated greater prevalence or severity in patients with HFpEF compared with controls (Table 2). All components of the H2FPEF score discriminated patients with HFpEF from controls individually, and most of the HFA-PEFF score components were also predictive (Table 3 and eFigure 3 in the Supplement).
Table 2. Prevalence of Components of H2FPEF and HFA-PEFF Scores.
| Score component | No. (%) | P value | |
|---|---|---|---|
| Control individuals (n = 173) | Patients with HFpEF (n = 563) | ||
| H2FPEF score | |||
| Age >60 y | 97 (56) | 461 (82) | <.001 |
| Obesitya | 50 (29) | 335 (59) | <.001 |
| Treated with ≥2 antihypertensives | 64 (37) | 375 (67) | <.001 |
| Any AF | 14 (8) | 212 (38) | <.001 |
| E/e′>9 | 66 (39) | 381 (71) | <.001 |
| PASP>35 mm Hg | 25 (17) | 249 (49) | <.001 |
| HFA-PEFF score | |||
| Functional domain | |||
| Septal e′<7 (age <75 y) or <5 (age >75 y) | 45 (27) | 187 (35) | .05 |
| Septal e′<7 | 58 (34) | 258 (48) | .002 |
| E/e′≥15 | 16 (10) | 152 (28) | <.001 |
| TR velocity >2.8 | 15 (13) | 171 (40) | <.001 |
| E/e′ of 9-14 | 50 (30) | 221 (41) | .006 |
| Functional domain scoreb | |||
| 2 | 63 (36) | 373 (66) | <.001 |
| 1 | 19 (11) | 80 (14) | |
| 0 | 91 (53) | 110 (20) | |
| Morphological domain | |||
| LAVI>34 (sinus) or >40 (AF) | 58 (36) | 237 (48) | .008 |
| LAVI of 29-34 (sinus) or 34-40 (AF) | 34 (21) | 90 (18) | .42 |
| RWT>0.42 and LVMI≥149 (male) or ≥122 (female) | 2 (1) | 17 (3) | .27 |
| RWT>0.42 | 45 (26) | 233 (43) | <.001 |
| LV wall thickness ≥12 | 18 (11) | 151 (28) | <.001 |
| LVMI≥115 (male) or ≥95 (female) | 21 (13) | 113 (21) | .01 |
| Morphological domain scorec | |||
| 2 | 60 (35) | 243 (43) | .003 |
| 1 | 49 (28) | 198 (35) | |
| 0 | 64 (37) | 118 (21) | |
| Natriuretic peptide domain | |||
| NTproBNP level >220 (sinus) or >660 (AF)d | 31 (24) | 243 (52) | <.001 |
| NTproBNP level of 125-220 (sinus) or 375-660 (AF)e | 24 (18) | 75 (16) | .54 |
| Natriuretic peptide scoref | |||
| 2 | 31 (24) | 243 (52) | <.001 |
| 1 | 24 (18) | 75 (16) | |
| 0 | 76 (58) | 148 (32) | |
Abbreviations: AF, atrial fibrillation; e′, septal mitral annulus tissue relaxation velocity in early diastole; E/e′, ratio of early diastolic mitral inflow velocity to e′; HFpEF, heart failure with preserved ejection fraction; LAVI, left atrial volume index; LV, left ventricle; LVMI, LV mass index; NTproBNP, N-terminal pro brain natriuretic peptide; PASP, pulmonary artery systolic pressure; RWT, relative wall thickness; TR, tricuspid regurgitation.
Defined as a body mass index greater than 30 (calculated as weight in kilograms divided by height in meters squared).
Scores range from 0 to 2, with higher scores indicating higher functional domain score component of HFA-PEFF score.
Scores range from 0 to 2, with higher scores indicating higher morphological domain score component of HFA-PEFF score.
Includes 37 patients with BNP measurements from 1 center who had BNP greater than 80 (sinus) or greater than 240 (AF).
Includes 37 patients with BNP measurements from 1 center who had BNP of 35 to 80 (sinus) or 105 to 240 (AF).
Scores range from 0 to 2, with higher scores indicating higher natriuretic peptide domain score component of HFA-PEFF score.
Table 3. Operating Characteristics of Individual Score Components in Isolation.
| Score component | OR (95% CI) | AUC (95% CI) | Sensitivity | Specificity | P value |
|---|---|---|---|---|---|
| H2FPEF score | |||||
| Age >60 y | 3.55 (2.45-5.13) | 0.629 (0.589-0.669) | 82 | 44 | <.001 |
| Obesitya | 3.60 (2.49-5.21) | 0.652 (0.614-0.693) | 59 | 71 | <.001 |
| ≥2 Antihypertensives | 3.38 (2.37-4.82) | 0.648 (0.606-0.688) | 67 | 63 | <.001 |
| Any AF | 6.84 (3.86-12.12) | 0.648 (0.619-0.676) | 38 | 92 | <.001 |
| E/e′>9 | 3.91 (2.72-5.62) | 0.662 (0.621-0.704) | 72 | 61 | <.001 |
| PASP>35 mm Hg | 4.69 (2.95-7.46) | 0.660 (0.622-0.697) | 49 | 83 | <.001 |
| HFA-PEFF score | |||||
| Functional domain | |||||
| Septal e′<7 (age <75 y) or <5 (age >75 y) | 1.47 (1.00-2.16) | 0.541 (0.502-0.580) | 35 | 73 | .048 |
| Septal e′<7 | 1.77 (1.23-2.53) | 0.568 (0.527-0.610) | 48 | 66 | .002 |
| E/e′>15 | 3.82 (2.21-6.60) | 0.595 (0.566-0.625) | 29 | 90 | <.001 |
| TR velocity >2.8 m/s | 4.53 (2.54-8.05) | 0.637 (0.598-0.676) | 40 | 87 | <.001 |
| E/e′>9 | 3.91 (2.72-5.62) | 0.662 (0.621-0.704) | 72 | 61 | <.001 |
| Functional domain score | 4.88 (3.30-7.21) | 0.676 (0.634-0.720) | NA | NA | <.001 |
| Morphological domain | |||||
| LAVI>34 mL/m2 (SR) or >40 mL/m2 (AF) | 1.65 (1.14-2.38) | 0.561 (0.517-0.604) | 48 | 64 | .008 |
| LAVI>29 mL/m2 (SR) or >34 mL/m2 (AF) | 0.83 (0.54-1.30) | 0.514 (0.478-0.551) | NA | NA | .42 |
| RWT>0.42 g/m2 and LVMI≥149 (male) or ≥122 g/m2 (female) | 2.75 (0.63-12.0) | 0.510 (0.499-0.521) | NA | NA | .18 |
| RWT>0.42 | 2.11 (1.44-3.09) | 0.583 (0.544-0.623) | 43 | 74 | <.001 |
| LV wall thickness ≥12 mm | 3.23 (1.91-5.45) | 0.585 (0.556-0.615) | 28 | 89 | <.001 |
| LVMI≥115 (male) or ≥95 g/m2 (female) | 1.86 (1.12-3.07) | 0.542 (0.512-0.573) | 21 | 87 | .02 |
| Morphological domain score | 1.47 (1.19-1.82) | 0.580 (0.532-0.627) | 79 | 37 | <.001 |
| Natriuretic peptide domain | |||||
| NTproBNP level >220 (SR) or >660 (AF)b | 3.52 (2.26-5.47) | 0.642 (0.600-0.685) | 52 | 76 | <.001 |
| NTproBNP level >125 (SR) or >375 (AF)c | 0.86 (0.52-1.42) | 0.511 (0.474-0.548) | NA | NA | .55 |
| Natriuretic peptide score | 1.98 (1.58-2.49) | 0.660 (0.612-0.708) | NA | NA | <.001 |
Abbreviations: AF, atrial fibrillation; AUC, area under the curve; e′, septal mitral annulus tissue relaxation velocity in early diastole; E/e′, ratio of early diastolic mitral inflow velocity to e′; LAVI, left atrial volume index; LV, left ventricle; LVMI, LV mass index; NA, not applicable; NTproBNP, N-terminal pro brain natriuretic peptide; OR, odds ratio; PASP, pulmonary artery systolic pressure; RWT, relative wall thickness; SR, sinus rhythm; TR, tricuspid regurgitation.
Defined as a body mass index greater than 30 (calculated as weight in kilograms divided by height in meters squared).
Includes 37 patients with BNP measurements from 1 center who had BNP greater than 80 (sinus) or greater than 240 (AF).
Includes 37 patients with BNP measurements from 1 center who had BNP of 35 to 80 (sinus) or 105 to 240 (AF).
Validation of Diagnostic Utility
The H2FPEF and HFA-PEFF scores discriminated cases from controls overall among the 627 patients for whom the H2FPEF score could be calculated and the 594 patients for whom the HFA-PEFF score could be calculated. Among the 485 patients with all necessary data available to calculate both scores, the H2FPEF score had a higher area under the curve (AUC) (0.845; 95% CI, 0.810-0.875; P < .001) compared with the HFA-PEFF score (0.710; 95% CI, 0.659-0.756; P < .001) (difference, −0.134; 95% CI, –0.177 to −0.094; P < .001) (Table 4 and Figure). The false-negative rate was higher with the HFA-PEFF score than with the H2FPEF score; a total of 47% (20 of 43) of patients with a HFA-PEFF score of 0 and 61% (33 of 54) of patients with a HFA-PEFF score of 1 were found to have HFpEF based on invasive testing (Figure). Overall, there was a 55% false-negative rate with HFA-PEFF estimation of low probability compared with 25% using the H2FPEF score. In contrast, for the H2FPEF score, 7% (2 of 27) of patients with a score of 0 and 36% (15 of 42) of patients with a score of 1 were found to have HFpEF. False-positive rates were low for both algorithms in the rule-in range. A total of 279 of 298 patients (94%) with either an HFA-PEFF or H2FPEF score in the rule-in range had HFpEF, whereas 98% (117 of 120) of patients with both scores in the rule-in range had HFpEF (eFigure 4 and eTable 4 in the Supplement).
Table 4. Diagnostic Performance of the H2FPEF and HFA-PEFF Algorithms.
| Overall score | OR per unit change (95% CI) | AUC (95% CI) | P value | AUC comparisona | P value |
|---|---|---|---|---|---|
| H2FPEF (n = 627) | 2.18 (1.89 to 2.52) | 0.845 (0.810 to 0.875) | <.001 | 1 [Reference] | NA |
| HFA-PEFF (n = 594) | 1.53 (1.37 to 1.72) | 0.710 (0.659 to 0.756) | <.001 | –0.134 (–0.177 to –0.094) | <.001 |
Abbreviations: AUC, area under the curve; NA, not applicable; OR, odds ratio.
Positive predictive values and NPVs varied with disease prevalence. eFigure 5 in the Supplement shows the PPV and NPV for the rule-in and rule-out thresholds for the H2FPEF and HFA-PEFF scores plotted as a function of disease prevalence. In cohorts in which disease prevalence was high, differences in PPV between scores were not clinically significant, but the 2 scores differed more by NPV, which was greater for the rule-out H2FPEF score. Conversely, in populations with low disease prevalence, modeling predicted greater discrimination with the H2FPEF compared with the HFA-PEFF, with higher PPV for rule-in values (eFigure 5 in the Supplement).
Concordance and Discordant Group Differences
Most patients had concordant estimations of probability using H2FPEF and HFA-PEFF scores (62.5% [303 of 485]), with the remaining patients (37.5% [182 of 485]) having discordant estimations of probability (eFigure 4 in the Supplement). Discrepancies in HFpEF probability using H2FPEF and HFA-PEFF scores ranging between 28% and 41% have been reported by other researchers,13,18 similar to the 38% discordant result observed in the present study. For patients with a low H2FPEF score but an intermediate or a high HFA-PEFF score, the prevalence of true HFpEF was 30% (6 of 20). Among patients with a low HFA-PEFF score but an intermediate or a high H2FPEF probability, the prevalence of true HFpEF was 78% (31 of 40).
Among the discrepantly categorized groups, body mass index, hypertension history, invasively verified HFpEF, and exertional pulmonary vascular pressures were higher in the H2FPEF high-probability and HFA-PEFF low- and intermediate-probability groups, whereas left atrial volume and N-terminal pro brain natriuretic peptide (NTproBNP) levels were higher in the HFA-PEFF high-probability and H2FPEF low- and intermediate-probability groups (eTable 5 in the Supplement).
Subgroup and Sensitivity Analyses
Discrimination for the H2FPEF and HFA-PEFF scores across multiple subgroups of interest is given in eTable 6 in the Supplement. Among patients with a low NTproBNP level and HFpEF, an incorrect low-probability score among patients with true HFpEF was more common with the HFA-PEFF score (31.3% [46 of 147]) compared with the H2FPEF score (4.2% [5 of 120]) (P < .001) (eFigure 6 in the Supplement). Patients with HFpEF and elevated NTproBNP levels were more likely to be classified as having high probability of HFpEF using the HFA-PEFF score (75.1% [181 of 241]) compared with the H2FPEF score (57.4% [120 of 209]), but the H2FPEF score remained robust even in the subset of patients with high NTproBNP levels and among patients with AF (eTable 6 in the Supplement).
Use of a PCWP/CO slope more than 2 mm Hg/L/min rather than peak exercise PCWP of 25 mm Hg or higher to define HFpEF case status led to reclassification of 20% (117 of 583) of patients, with 70 HFpEF cases reclassified from case to control and 47 controls reclassified as cases (eTable 7 in the Supplement). The H2FPEF and HFA-PEFF scores poorly discriminated patients with HFpEF from controls using this alternative hemodynamic definition to define HFpEF case status (AUC, 0.673-0.690) (eTable 8 in the Supplement), but reclassified patients had clinical, echocardiographic, and hemodynamic findings that challenged the accuracy of reclassification.
Patients reclassified from controls to cases by PCWP/CO slope had findings less typical of HFpEF, including lower body mass index, left ventricular mass, E/e′ ratio, and prevalence of AF. In contrast, patients reclassified from cases to controls by PCWP/CO slope retained features typical of HFpEF in the community, with a high prevalence of obesity and AF and elevated pulmonary vascular pressures that predispose to lung congestion at rest and during exercise. Two-thirds (66% [46 of 70]) of patients reclassified from cases to controls by the PCWP/CO slope had an elevated resting PCWP, even without exercise testing.
Discussion
In this multicenter, international, case-control study, we evaluated a large series of patients with unexplained exertional dyspnea for suspected HFpEF. Case status was ascertained definitively using directly measured PCWP during invasive exercise testing. This study’s data validated both scoring systems against the reference standard defined by the Heart Failure Association of the European Society of Cardiology for diagnosis of HFpEF as an elevation in PCWP during exercise,7 and greater sensitivity and overall diagnostic accuracy was found for the H2FPEF score. The use of an elevated PCWP/CO slope as an alternative diagnostic criteria for HFpEF, as recently proposed,26,27 reclassified 20% of patients and was associated with poorer accuracy for the HFA-PEFF and H2FPEF scores, with clinical, echocardiographic, and hemodynamic features of reclassified patients being less typical of HFpEF seen in the community.
Dyspnea is a commonly encountered patient concern in clinical practice. Diagnosis of HFpEF is often challenging in this setting9 but is important to provide insight into symptom causality, guide treatment decisions, and preclude unnecessary testing for pulmonary, neuromuscular, and other disorders. Despite increasing awareness in general practice, recent data suggest that HFpEF remains underrecognized. Selvaraj and colleagues13 found that patients with self-reported dyspnea, no documented HFpEF, and an elevated H2FPEF or HFA-PEFF score had HF event rates that were similar to those among patients with clinically diagnosed HFpEF. This finding suggests that many of the patients had undiagnosed HFpEF, and this cohort constituted one-third of the total population with HFpEF in that community-based study, underlining the potential magnitude of the problem. Although it is not feasible or warranted to perform invasive exercise testing for all patients at risk of HFpEF, underdiagnosis may deprive afflicted patients of effective treatments.2,3,4
Clinical risk scores offer a solution that may boost awareness and help inform medical decision-making regarding further evaluation, particularly when instruments are externally validated and not overly complex.14,15,16 The H2FPEF score is an evidence-based algorithm that combines echocardiographic and clinical variables and was previously derived and validated in populations other than those in the present study.10 Discrimination of patients with HFpEF from controls in this study was similar to that in the previous derivation study; the AUC to distinguish patients with HFpEF from controls with the H2FPEF score was 0.845 in this study and was 0.841 in the previous study,10 indicating robust reproducibility and external validation.
The HFA-PEFF score is a somewhat more complex algorithm using data from echocardiography and natriuretic peptide testing.7 The reasons for the greater diagnostic accuracy observed in the present study with the H2FPEF score compared with the HFA-PEFF score may be associated with the primary differences among score components, including obesity, AF, and natriuretic peptides (eFigure 1 in the Supplement). Echocardiographic and natriuretic peptide abnormalities are informative when present but have been shown to have poor sensitivity for HFpEF.9,28 HFpEF is known to be associated with comorbidities, including obesity, hypertension, and AF,5,6,8 and the inclusion of clinical factors provides orthogonal information to echocardiography to estimate pretest probability, potentially explaining the superior diagnostic performance of the H2FPEF score.
A history of AF is associated with HFpEF.29,30,31,32 The HFA-PEFF score requires higher NTproBNP levels and left atrial sizes to support the diagnosis of HFpEF in AF, essentially raising the diagnostic threshold for HFpEF in the presence of AF rather than using the presence of AF as a biomarker of potential HFpEF. Obesity is one of the most common factors associated with HFpEF, and natriuretic peptide levels are known to be lower (or even normal) in individuals with HFpEF and obesity despite equal or high filling pressures,28,33,34 which may conceal the presence of HFpEF in patients with obesity.
In a sensitivity analysis comparing the discrepantly categorized patients (eTable 5 in the Supplement), those with a low-probability HFA-PEFF score and higher H2FPEF score were more likely to have hemodynamic evidence of HFpEF and other characteristics typical of HFpEF. Both scores include exercise testing among patients with intermediate probability, and in these cases, the true diagnosis would be ascertained definitively through exercise evaluation regardless of which score was applied.7,8,10 Therefore, for clinical purposes, the important differences reside in the rule-in and rule-out performances. Of note, 54 of 97 patients (56%) with low HFA-PEFF scores of 0 or 1 (deemed exclusionary for HFpEF)7 were found to have HFpEF, and the diagnosis would have been missed in these patients. The validation of both scores with the use of the invasive standard across multiple centers in our study suggests that the scoring systems may be useful to enrich clinical trials for genuine HFpEF, particularly because both scores have been shown to also predict high event rates.22,25
Previous studies17,18,19,20,21,22,23,24 validated these scores using less robust reference standards and were variably limited by a lack of appropriate dyspneic controls, use of variable case definitions, and of note, the absence of definitive case or control ascertainment. These previous validation studies used HFpEF case status defined by previous HF hospitalization, clinical impressions, or trial criteria and controls who were asymptomatic or had no cardiovascular history.
Previous HF hospitalization is pathognomonic for HFpEF, and the diagnosis of HFpEF is not considered to be challenging in this cohort.5 Other studies have used asymptomatic controls, but a control group that also seeks care for dyspnea and in whom disease is definitively excluded should be included to accurately evaluate diagnostic accuracy.
To our knowledge, only 1 other study has used an invasive approach to evaluate discrimination using HFA-PEFF and H2FPEF scores. Churchill et al26 reported similar accuracy for the 2 scores (AUC of 0.73 for HFA-PEFF and 0.74 for H2FPEF). Accuracy for the HFA-PEFF score in the present study was similar to the accuracy in that study, but discrimination using the H2FPEF score was greater in the present study. There are several potential reasons for this difference. One explanation may be patient selection. The study by Churchill et al26 included data from a younger (mean age, 59 years) and selective cohort of 156 patients identified over 12 years at a single center (approximately 13 patients per year),26 limiting external validity, compared with the present, multicenter study, which included 736 patients.
An important difference may be the definition of HFpEF used as the gold standard. Churchill et al.26 used a lower cutoff value for exercise PCWP to define case status (≥15 mm Hg) if there was an increased PCWP/CO slope, whereas the present study used a more restrictive reference standard of a higher exercise PCWP (≥25 mm Hg) without adjustment for CO.7 These definitions agreed for most patients (80%), but there were differences in the remaining 20%.26 Patients reclassified from controls to cases based on the PCWP/CO slope in the present study had features that are not typical of HFpEF, including young age, low comorbidity burden, and normal PCWP and pulmonary artery pressures, raising questions of whether these patients should be diagnosed with HFpEF. Conversely, patients reclassified from cases to controls based on the PCWP/CO slope had features typical of HFpEF, including high comorbidity burden, obesity, diastolic dysfunction, and elevated pulmonary vascular pressures that cause dyspnea (eTable 6 in the Supplement). Therefore, the lower diagnostic performance of the 2 scores in the study by Churchill et al26 may be associated with limitations of the PCWP/CO slope in defining HFpEF case status, and these limitations may explain the poorer discrimination observed using the PCWP/CO slope with both scores in the present study.
Limitations
The study has limitations. All patients were referred to catheterization for diagnosis, but invasive assessment has become routine practice at many centers and is necessary to definitively ascertain case or control status. Referral patterns for invasive cardiopulmonary exercise testing likely differed across the institutions studied, introducing bias. Although our modeling estimates suggested similar test characteristics at lower disease prevalence, the overall high prevalence of HFpEF in the study sample may impact generalizability, and further study in a broader population is needed. Pressure waveforms were interpreted individually at each center rather than through a central core laboratory. Although individual interpretations may introduce greater variability, this is reflective of real-world clinical practice, enhancing generalizability. Whereas this multicenter study included patients evaluated across 3 continents, there was no representation from Africa, South America, or Asia,20 where prevalence of obesity and other comorbidities differ; thus, this study’s results may not apply to these regions.
Conclusions
In this case-control study, both the H2FPEF and the HFA-PEFF algorithms discriminated patients with HFpEF from controls with high specificity among patients presenting with unexplained dyspnea, but the H2FPEF score provided superior sensitivity and overall diagnostic accuracy despite the requirement of fewer input variables.
eMethods.
eTable 1. Characteristics by Center Including All Participants
eTable 2. HFpEF Characteristics by Participating Center
eTable 3. Control Characteristics by Participating Center
eTable 4. Diagnostic Performance of H2FPEF and HFA-PEFF Algorithms
eTable 5. Characteristics of Discordant H2FPEF and HFA-PEFF Groups
eTable 6. Diagnostic Performance of the H2FPEF and HFA-PEFF Algorithms in Ambulatory HFpEF Subgroups of Interest
eTable 7. Patients Classified Discordantly by Exercise PCWP≥25 mm Hg and PCWP/CO Slope >2
eTable 8. Comparative Diagnostic Performance of the H2FPEF and HFA-PEFF Algorithms Using PCWP/CO Slope >2 mm Hg/L/min to Define HFpEF
eFigure 1. Categorization of HFpEF Probability by the H2FPEF and HFA-PEFF Scores
eFigure 2. Distribution of H2FPEF and HFA-PEFF Scores
eFigure 3. Distribution of HFA-PEFF Domain Scores in Invasively Confirmed HFpEF Patients Vs Controls
eFigure 4. 3 × 3 Confusion Matrix Demonstrating Concordance and Reclassification Among Low, Intermediate, and High Probability by the H2FPEF and HFA-PEFF Scores (Total, n = 485)
eFigure 5. Positive and Negative Predictive Values for the H2FPEF Score (Red/Pink) and the HFA-PEFF Score (Black/Gray) Plotted As a Function of HFpEF Prevalence. The Prevalence of HFpEF in the Present Study Is Shown As a Green Dashed Line
eFigure 6. Distribution of HFpEF Probabilities by the H2FPEF and HFA-PEFF Score in Patients With Obese HFpEF As Well As Those With Low BNP HFpEF
References
- 1.Ramalho SHR, Santos M, Claggett B, et al. Association of undifferentiated dyspnea in late life with cardiovascular and noncardiovascular dysfunction: a cross-sectional analysis from the ARIC study. JAMA Netw Open. 2019;2(6):e195321. doi: 10.1001/jamanetworkopen.2019.5321 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Anker SD, Butler J, Filippatos G, et al. ; EMPEROR-Preserved Trial Investigators . Empagliflozin in heart failure with a preserved ejection fraction. N Engl J Med. 2021;385(16):1451-1461. doi: 10.1056/NEJMoa2107038 [DOI] [PubMed] [Google Scholar]
- 3.Nassif ME, Windsor SL, Borlaug BA, et al. The SGLT2 inhibitor dapagliflozin in heart failure with preserved ejection fraction: a multicenter randomized trial. Nat Med. 2021;27(11):1954-1960. doi: 10.1038/s41591-021-01536-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Solomon SD, McMurray JJV, Anand IS, et al. ; PARAGON-HF Investigators and Committees . Angiotensin-neprilysin inhibition in heart failure with preserved ejection fraction. N Engl J Med. 2019;381(17):1609-1620. doi: 10.1056/NEJMoa1908655 [DOI] [PubMed] [Google Scholar]
- 5.Redfield MM. Heart failure with preserved ejection fraction. N Engl J Med. 2017;376(9):897. [DOI] [PubMed] [Google Scholar]
- 6.Shah SJ, Borlaug BA, Kitzman DW, et al. Research priorities for heart failure with preserved ejection fraction: National Heart, Lung, and Blood Institute Working Group summary. Circulation. 2020;141(12):1001-1026. doi: 10.1161/CIRCULATIONAHA.119.041886 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pieske B, Tschöpe C, de Boer RA, et al. How to diagnose heart failure with preserved ejection fraction: the HFA-PEFF diagnostic algorithm: a consensus recommendation from the Heart Failure Association (HFA) of the European Society of Cardiology (ESC). Eur Heart J. 2019;40(40):3297-3317. doi: 10.1093/eurheartj/ehz641 [DOI] [PubMed] [Google Scholar]
- 8.Borlaug BA. Evaluation and management of heart failure with preserved ejection fraction. Nat Rev Cardiol. 2020;17(9):559-573. doi: 10.1038/s41569-020-0363-2 [DOI] [PubMed] [Google Scholar]
- 9.Obokata M, Kane GC, Reddy YN, Olson TP, Melenovsky V, Borlaug BA. Role of diastolic stress testing in the evaluation for heart failure with preserved ejection fraction: a simultaneous invasive-echocardiographic study. Circulation. 2017;135(9):825-838. doi: 10.1161/CIRCULATIONAHA.116.024822 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Reddy YNV, Carter RE, Obokata M, Redfield MM, Borlaug BAA. A simple, evidence-based approach to help guide diagnosis of heart failure with preserved ejection fraction. Circulation. 2018;138(9):861-870. doi: 10.1161/CIRCULATIONAHA.118.034646 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Obokata M, Reddy YNV, Borlaug BA. Diastolic dysfunction and heart failure with preserved ejection fraction: understanding mechanisms by using noninvasive methods. JACC Cardiovasc Imaging. 2020;13(1 Pt 2):245-257. doi: 10.1016/j.jcmg.2018.12.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Borlaug BA, Nishimura RA, Sorajja P, Lam CS, Redfield MM. Exercise hemodynamics enhance diagnosis of early heart failure with preserved ejection fraction. Circ Heart Fail. 2010;3(5):588-595. doi: 10.1161/CIRCHEARTFAILURE.109.930701 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Selvaraj S, Myhre PL, Vaduganathan M, et al. Application of diagnostic algorithms for heart failure with preserved ejection fraction to the community. JACC Heart Fail. 2020;8(8):640-653. doi: 10.1016/j.jchf.2020.03.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Damen JA, Hooft L, Schuit E, et al. Prediction models for cardiovascular disease risk in the general population: systematic review. BMJ. 2016;353:i2416. doi: 10.1136/bmj.i2416 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Steyerberg EW, Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation. Eur Heart J. 2014;35(29):1925-1931. doi: 10.1093/eurheartj/ehu207 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Aakre CA, Dziadzko MA, Herasevich V. Towards automated calculation of evidence-based clinical scores. World J Methodol. 2017;7(1):16-24. doi: 10.5662/wjm.v7.i1.16 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Barandiarán Aizpurua A, Sanders-van Wijk S, Brunner-La Rocca HP, et al. Validation of the HFA-PEFF score for the diagnosis of heart failure with preserved ejection fraction. Eur J Heart Fail. 2020;22(3):413-421. doi: 10.1002/ejhf.1614 [DOI] [PubMed] [Google Scholar]
- 18.Sanders-van Wijk S, Barandiarán Aizpurua A, Brunner-La Rocca HP, et al. The HFA-PEFF and H2 FPEF scores largely disagree in classifying patients with suspected heart failure with preserved ejection fraction. Eur J Heart Fail. 2021;23(5):838-840. doi: 10.1002/ejhf.2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sepehrvand N, Alemayehu W, Dyck GJB, et al. External validation of the H2F-PEF model in diagnosing patients with heart failure and preserved ejection fraction. Circulation. 2019;139(20):2377-2379. doi: 10.1161/CIRCULATIONAHA.118.038594 [DOI] [PubMed] [Google Scholar]
- 20.Ouwerkerk W, Tromp J, Jin X, et al. Heart failure with preserved ejection fraction diagnostic scores in an Asian population. Eur J Heart Fail. 2020;22(9):1737-1739. doi: 10.1002/ejhf.1851 [DOI] [PubMed] [Google Scholar]
- 21.Segar MW, Patel KV, Berry JD, Grodin JL, Pandey A. generalizability and implications of the H2FPEF score in a cohort of patients with heart failure with preserved ejection fraction. Circulation. 2019;139(15):1851-1853. doi: 10.1161/CIRCULATIONAHA.118.039051 [DOI] [PubMed] [Google Scholar]
- 22.Myhre PL, Vaduganathan M, Claggett BL, et al. Application of the H2 FPEF score to a global clinical trial of patients with heart failure with preserved ejection fraction: the TOPCAT trial. Eur J Heart Fail. 2019;21(10):1288-1291. doi: 10.1002/ejhf.1542 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Parcha V, Malla G, Kalra R, et al. Diagnostic and prognostic implications of heart failure with preserved ejection fraction scoring systems. ESC Heart Fail. 2021;8(3):2089-2102. doi: 10.1002/ehf2.13288 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tada A, Nagai T, Omote K, et al. Performance of the H2FPEF and the HFA-PEFF scores for the diagnosis of heart failure with preserved ejection fraction in Japanese patients: a report from the Japanese multicenter registry. Int J Cardiol. 2021;342:43-48. doi: 10.1016/j.ijcard.2021.08.001 [DOI] [PubMed] [Google Scholar]
- 25.Verbrugge FH, Reddy YNV, Sorimachi H, Omote K, Carter RE, Borlaug BA. Diagnostic scores predict morbidity and mortality in patients hospitalized for heart failure with preserved ejection fraction. Eur J Heart Fail. 2021;23(6):954-963. doi: 10.1002/ejhf.2142 [DOI] [PubMed] [Google Scholar]
- 26.Churchill TW, Li SX, Curreri L, et al. Evaluation of 2 existing diagnostic scores for heart failure with preserved ejection fraction against a comprehensively phenotyped cohort. Circulation. 2021;143(3):289-291. doi: 10.1161/CIRCULATIONAHA.120.050757 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Eisman AS, Shah RV, Dhakal BP, et al. Pulmonary capillary wedge pressure patterns during exercise predict exercise capacity and incident heart failure. Circ Heart Fail. 2018;11(5):e004750. doi: 10.1161/CIRCHEARTFAILURE.117.004750 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Anjan VY, Loftus TM, Burke MA, et al. Prevalence, clinical phenotype, and outcomes associated with normal B-type natriuretic peptide levels in heart failure with preserved ejection fraction. Am J Cardiol. 2012;110(6):870-876. doi: 10.1016/j.amjcard.2012.05.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Reddy YNV, Obokata M, Gersh BJ, Borlaug BA. High prevalence of occult heart failure with preserved ejection fraction among patients with atrial fibrillation and dyspnea. Circulation. 2018;137(5):534-535. doi: 10.1161/CIRCULATIONAHA.117.030093 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Reddy YNV, Obokata M, Verbrugge FH, Lin G, Borlaug BA. Atrial dysfunction in patients with heart failure with preserved ejection fraction and atrial fibrillation. J Am Coll Cardiol. 2020;76(9):1051-1064. doi: 10.1016/j.jacc.2020.07.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Reddy YNV, Obokata M, Egbe A, et al. Left atrial strain and compliance in the diagnostic evaluation of heart failure with preserved ejection fraction. Eur J Heart Fail. 2019;21(7):891-900. doi: 10.1002/ejhf.1464 [DOI] [PubMed] [Google Scholar]
- 32.Patel RB, Shah SJ. Therapeutic targeting of left atrial myopathy in atrial fibrillation and heart failure with preserved ejection fraction. JAMA Cardiol. 2020;5(5):497-499. doi: 10.1001/jamacardio.2020.0136 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Obokata M, Reddy YNV, Pislaru SV, Melenovsky V, Borlaug BA. Evidence supporting the existence of a distinct obese phenotype of heart failure with preserved ejection fraction. Circulation. 2017;136(1):6-19. doi: 10.1161/CIRCULATIONAHA.116.026807 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Reddy YNV, Lewis GD, Shah SJ, et al. Characterization of the obese phenotype of heart failure with preserved ejection fraction: a RELAX trial ancillary study. Mayo Clin Proc. 2019;94(7):1199-1209. doi: 10.1016/j.mayocp.2018.11.037 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods.
eTable 1. Characteristics by Center Including All Participants
eTable 2. HFpEF Characteristics by Participating Center
eTable 3. Control Characteristics by Participating Center
eTable 4. Diagnostic Performance of H2FPEF and HFA-PEFF Algorithms
eTable 5. Characteristics of Discordant H2FPEF and HFA-PEFF Groups
eTable 6. Diagnostic Performance of the H2FPEF and HFA-PEFF Algorithms in Ambulatory HFpEF Subgroups of Interest
eTable 7. Patients Classified Discordantly by Exercise PCWP≥25 mm Hg and PCWP/CO Slope >2
eTable 8. Comparative Diagnostic Performance of the H2FPEF and HFA-PEFF Algorithms Using PCWP/CO Slope >2 mm Hg/L/min to Define HFpEF
eFigure 1. Categorization of HFpEF Probability by the H2FPEF and HFA-PEFF Scores
eFigure 2. Distribution of H2FPEF and HFA-PEFF Scores
eFigure 3. Distribution of HFA-PEFF Domain Scores in Invasively Confirmed HFpEF Patients Vs Controls
eFigure 4. 3 × 3 Confusion Matrix Demonstrating Concordance and Reclassification Among Low, Intermediate, and High Probability by the H2FPEF and HFA-PEFF Scores (Total, n = 485)
eFigure 5. Positive and Negative Predictive Values for the H2FPEF Score (Red/Pink) and the HFA-PEFF Score (Black/Gray) Plotted As a Function of HFpEF Prevalence. The Prevalence of HFpEF in the Present Study Is Shown As a Green Dashed Line
eFigure 6. Distribution of HFpEF Probabilities by the H2FPEF and HFA-PEFF Score in Patients With Obese HFpEF As Well As Those With Low BNP HFpEF