Heart failure with preserved ejection fraction (HFpEF) is recognized as a major cause of cardiovascular morbidity and mortality. An ability to identify patients with HFpEF who are at increased risk for adverse outcomes can facilitate their more careful management. We studied the patients having heart failure (HF) using data from the Heart Failure Adherence and Retention Trial (HART). HART enrolled 902 patients in the New York Heart Association (NYHA) class II or III who had been recently hospitalized for HF to study the impact of self-management counseling on the primary outcome of death or HF hospitalization. In HART, 208 patients had HFpEF and 692 had HF with reduced ejection fraction (HFrEF) and were followed for a median of 1,080 days. Two final multivariate models were developed. In patients having HFpEF, predictors of primary outcome were male gender (odds ratio [OR] 3.45, p = 0.004), NYHA class III (OR 3.05, p = 0.008), distance covered on a 6-minute walk test (6-MWT) of <620 feet (OR 2.81, p = 0.013), and <80% adherence to prescribed medications (OR 2.61, p = 0.018). In patients having HFrEF, the predictors were being on diuretics (OR 3.06, p = 0.001), having ≥3 co-morbidities (OR 2.11, p = 0.0001), distance covered on a 6-MWT of <620 feet (OR 1.94, p = 0.001), NYHA class III (OR 1.90, p = 0.001), and age >65 years (OR 1.63, p = 0.01). In conclusion, indicators of functional status (6-MWT and NYHA class) were common to both patients with HFpEF and those with HFrEF, whereas gender and adherence to prescribed therapy were unique to patients having HFpEF in predicting death or HF hospitalization.
Given the increasing recognition of heart failure with preserved ejection fraction (HFpEF) as a growing and difficult-to-treat clinical problem, the identification of predictors of adverse outcomes can help to identify those patients who are at the highest risk and who would benefit from more personalized and aggressive management. To be useful, such predictors should be easy to identify in routine clinical practice, thereby making them potentially valuable in personalizing the approach to patient care, monitoring disease progression, and evaluating therapeutic effectiveness. Comparing these predictors among patients having HFpEF and those having heart failure with reduced ejection fraction (HFrEF) can further our understanding of the differences in the 2 subtypes of heart failure (HF).
Methods
We analyzed data from the Heart failure Adherence and Retention Trial (HART). HART was a single-center, multiple-hospital, partially blinded, randomized controlled behavioral trial that was based in the Chicago metropolitan area. HART was designed to assess the impact of self-management counseling versus education alone on the composite primary outcome of death or HF hospitalizations in patients with HFrEF or those with HFpEF. HART enrolled a total of 902 patients. Of the patients who could be classified, 692 had HFrEF and 208 had HFpEF. Details on patient enrollment, data collection, and follow-up within HART have been reported elsewhere. Briefly, patients having HF were recruited through inpatient and outpatient screening and through referrals from cardiologists and internists. The recruitment continued from October 2001 to October 2004. The follow-up was completed in May 2007. All patients were receiving some form of active HF treatment, including diuretics, for the previous 3 months. HFrEF was defined as an ejection fraction of ≤40% by echocardiography, radiographic ventriculography, or radionuclide ventriculography. HFpEF was defined as an ejection fraction of >40% by 1 of the 3 previously listed methods and ≥1 previous hospitalizations for HF.
Baseline data were collected on demographics, medications, co-morbidities, and adherence to medications. The median follow-up period was 1,080 days. Primary end points were ascertained through blind adjudication by a designated team of cardiologists. All patients, or in the case of death, their family members, were contacted every 3 months by telephone to ascertain occurrence of a death or hospitalization. Reports of death were confirmed by medical record, death certificate, emergency medical services record, or queries from the Social Security Death Index. HF admissions were adjudicated by the presence of shortness of breath, peripheral edema, or chest radiographic evidence of pulmonary edema without evidence of another disease process accounting for symptoms or signs. HF admissions were confirmed if the patient responded to HF therapy or had a documented decrease in left ventricular function.
Medication adherence was tested using electronic pill cap monitoring. The patient was asked to place a month’s supply of an angiotensin-converting enzyme inhibitor (angiotensin receptor blocker, β blocker, or diuretics, if the patient was not taking an angiotensin-converting enzyme inhibitor) into a Medication Event Monitoring System electronic pill cap container (MEMS V TrackCap; AARDEX, Zug, Switzerland). They were taught to use it for the ensuing month. Adherence to drug therapy was defined by way of the percentage of pills taken relative to pills prescribed, with a cut-off point of <80% indicating nonadherence. New York Heart Association (NYHA) class was assessed by the treating physicians at the time of enrollment and during follow-up. Six-minute walk test (6-MWT) was performed by measuring the distance that patients could walk during a period of 6 minutes. For analysis, distance covered on a 6-MWT was dichotomized at the lowest tertile. Glomerular filtration rate was calculated using the Cockroft-Gault equation. Diabetes was self-reported at the time of enrollment and during follow-up. Other co-morbid conditions that were assessed included previous myocardial infarction, hypertension, cancer, stroke, arthritis, lung disease, liver disease, asthma, sleep apnea, and Parkinson’s disease. Depression was assessed using Geriatric Depression Scale with a score of >10 having high sensitivity and specificity for diagnosing depression. Other psychosocial factors that were assessed using standardized questionnaires included quality of life, purpose in life, and social support.
Statistical analyses began with a description of the baseline characteristics in overall population of 900 patients and then a comparison of patients with HFpEF and those with HFrEF. To identify predictors of the primary outcome (death or HF hospitalization), univariate unadjusted odds ratios reflecting risk for the primary end point were computed for each of the baseline factors separately in patients with HFpEF and those with HFrEF. Next, and again separately, for subgroups of patients having HFpEF and those having HFrEF, a backward stepwise multivariate elimination strategy was used with all factors, with an unadjusted odds ratio marginally different from 1 (i.e., p <0.40) included in a saturated model and then iteratively assessed for elimination. The criterion for remaining in the model was p ≤0.15. Both likelihood ratio and Hosmer-Lemeshow tests were used to assess model fit. Unadjusted Kaplan-Meier curves were used to evaluate the time to event during the follow-up period for patients having HFpEF.
Results
Table 1 lists the baseline patient characteristics of the total cohort and the differences in baseline characteristics between patients with HFpEF and those with HFrEF. The average age of all the patients in the trial was 63.6 years, 47% were women, and 60% were Caucasian. A total of 74 (36%) of the 208 patients having HFpEF had a primary event. Of these, 48 (29%) were hospitalized for HF and 39 (20%) died. Of the 48 patients who were hospitalized for HF, 27 (56%) were hospitalized once and 21 (44%) were hospitalized more than once during the follow-up. A total of 259 (37%) of the 692 patients having HFrEF had a primary event. Of these, 162 (23%) were hospitalized for HF and 148 (21%) died. Of the 162 who were hospitalized, 86 (53%) were hospitalized once and 76 (47%) were hospitalized more than once during follow-up. These numbers do not add up to the total with primary event because those who were hospitalized and later died were counted only once. Table 2 lists the univariate odds of primary event by baseline characteristic in patients having HFpEF and those having HFrEF.
| Characteristic | All Patients, n = 900 (%) | HFpEF, n = 208 (%) | HFrEF, n = 692 (%) |
|---|---|---|---|
| Age (yrs), mean ± SD* | 63.6 ± 13.5 | 67.3 ± 13.0 | 62.4 ± 13.4 |
| Women* | 426 (47) | 136 (65) | 290 (42) |
| Minority race or ethnicity | 361 (40) | 79 (38) | 282 (41) |
| Lesser than high school education | 393 (44) | 82 (39) | 311 (45) |
| Annual family income of <$30,000 | 426 (52) | 106 (56) | 320 (50) |
| Married or living with someone else as if unmarried* | 502 (56) | 93 (45) | 409 (60) |
| In treatment arm | 450 (50) | 107 (51) | 343 (50) |
| NYHA class III | 284 (32) | 64 (31) | 220 (32) |
| 6-MWT distance (feet), mean ± SD* | 821 ± 465 | 718 ± 449 | 852 ± 465 |
| Hypertension** | 675 (75) | 168 (81) | 507 (74) |
| Diabetes mellitus | 361 (40) | 89 (43) | 272 (39) |
| Co-morbid conditions, mean ± SD*** | 3.2 ± 1.7 | 3.5 ± 1.6 | 3.1 ± 1.7 |
| Total number of medications, mean ± SD | 6.8 ± 3.0 | 6.7 ± 2.9 | 6.8 ± 3.0 |
| ACE inhibitor or ARB use* | 772 (86) | 163 (78) | 609 (88) |
| β-Blocker use* | 635 (71) | 111 (53) | 524 (76) |
| Major depressive symptoms | 264 (29) | 66 (32) | 198 (29) |
| Social support-emotional, mean ± SD | 75.2 ± 22.2 | 75.6 ± 22.5 | 75.0 ± 22.1 |
| Purpose in life, mean ± SD | 4.5 ± 0.8 | 4.4 ± 0.8 | 4.5 ± 0.8 |
| Quality of life, mean ± SD | |||
| SF-36 | |||
| Physical function* | 48.2 ± 24.9 | 43.2 ± 22.7 | 49.7 ± 25.4 |
| Energy and vitality, | 46.5 ± 23.7 | 44.4 ± 23.8 | 47.1 ± 23.6 |
| Quality of Life Index—Cardiac | |||
| Satisfaction with health and function | 4.3 ± 1.0 | 4.2 ± 1.1 | 4.3 ± 1.0 |
| Satisfaction with psychological or spiritual function | 4.7 ± 1.1 | 4.7 ± 1.1 | 4.8 ± 1.0 |
| Nonadherence to drug therapy | 273 (37) | 60 (36) | 213 (37) |
| Sodium intake, median (IQR), (mg/day) | 3,332 (2,647–4,269) | 3,129 (2,615–4,095) | 3,416 (2,655–4,290) |
| Current smoker | 85 (9.5) | 20 (9.6) | 65 (9.4) |
| Body mass index (kg/m 2 ), mean ± SD* | 31.0 ± 7.7 | 32.9 ± 8.4 | 30.5 ± 7.4 |
| Self-efficacy at self-management, mean ± SD | 7.6 ± 1.7 | 7.4 ± 1.8 | 7.7 ± 1.7 |
| Risk Factor | HFpEF | HFrEF |
|---|---|---|
| Odds Ratio (95% CI) | Odds Ratio (95% CI) | |
| NYHA class III symptoms | 3.47 (1.82–6.63)* | 2.43 (1.72–3.42)* |
| Men | 2.63 (1.40–4.76)** | 0.85 (0.62–1.17) |
| Distance covered on 6-MWT of <620 (feet) | 2.22 (1.20–4.11)*** | 2.88 (2.00–4.15)* |
| Diabetes mellitus | 2.03 (1.12–3.68)*** | 1.94 (1.40–2.69)* |
| Medication adherence of <80% | 2.15 (1.08–4.28)*** | 1.81 (1.25–2.60)* |
| ≥3 Co-morbidities | 1.90 (0.94–3.84) | 2.63 (1.87–3.71)*** |
| Self-management treatment arm | 1.67 (0.93–3.02) | 0.82 (0.59–1.12) |
| Hypertension | 1.74 (0.78–3.87) | 1.28 (0.89–1.85) |
| On diuretics | 2.64 (0.55–12.80) | 3.30 (1.83–5.94)* |
| Coronary artery disease | 1.45 (0.80–2.63) | 1.37 (0.98–1.91) |
| On ACE-I or ARB | 0.68 (0.33–1.39) | 0.48 (0.29–0.78)** |
| Glomerular filtration rate of ≤30 (ml/min/m 2 ) | 1.55 (0.60–4.0) | 0.53 (0.30–0.95)*** |
| Geriatric Depression Scale score of ≥10 | 1.32 (0.71–2.46) | 1.79 (1.26–2.54)* |
| Age >65 (yrs) | 1.28 (0.68–2.39) | 2.00 (1.45–2.76)* |
| On β blocker | 0.80 (0.45–1.43) | 0.61 (0.42–0.88)** |
| Education less than high school | 1.24 (0.68–2.25) | 1.39 (1.01–1.91)*** |
| Atrial fibrillation | 1.24 (0.68–2.27) | 1.38 (0.98–1.94) |
| Sleep apnea | 1.28 (0.63–2.57) | 1.30 (0.84–2.03) |
| Ever smoker | 0.84 (0.46–1.51) | 0.89 (0.64–1.25) |
| Income of <$30,000.00/yr | 1.17 (0.63–2.15) | 1.19 (0.85–1.66) |
| Diabetes and blood pressure at goal | 0.87 (0.46–1.64) | 0.74 (0.50–1.09) |
| Asthma | 0.90 (0.44–1.84) | 1.29 (0.80–2.09) |
| Minority race or ethnicity | 0.95 (0.52–1.74) | 1.21 (0.88–1.68) |
| Kidney disease and blood pressure at goal | 0.97 (0.53–1.79) | 0.66 (0.46–0.94)*** |
Table 3 lists the results of the multivariate model aimed at identifying independent predictors of death or HF hospitalization in patients with HFpEF and those with HFrEF. The likelihood ratio test was significant (p <0.0001) in models for both groups, whereas the Hosmer-Lemeshow test was insignificant (p = 0.4211 for patients with HFpEF and p = 0.0533 for those with HFrEF), indicating a good model fit. Predictors common to both subgroups included indicators of functional status, that is, NYHA class and distance covered on a 6-MWT. Unique predictors for patients having HFpEF were male gender and medication nonadherence and for those having HFrEF were being on diuretics, having ≥3 co-morbidities, and age >65 years. Figure 1 presents unadjusted time-to-event Kaplan-Meier curves portraying these associations in the subgroup of patients having HFpEF.
| Characteristics | HFpEF | HFrEF |
|---|---|---|
| OR (95% CI) | OR (95% CI) | |
| Men vs Women | 3.45 (1.47–8.07)* | |
| NYHA class III vs II | 3.05 (1.33–6.98)* | 1.90 (1.28–2.82)* |
| Baseline medication adherence of <80% vs ≥80% | 2.61 (1.18–5.76)** | |
| Distance covered on 6-MWT of <620 vs ≥620 (feet) | 2.81 (1.24–6.40)** | 1.94 (1.30–2.90)* |
| Self-management vs enhanced education arm | 2.16 (0.98–4.76) | |
| ≥3 Co-morbidities | 2.11 (1.44–3.10)*** | |
| On diuretics | 3.06 (1.57–5.97)*** | |
| Age >65 yrs | 1.63 (1.11–2.40)* | |
| Minority race or ethnicity | 1.34 (0.91–1.97) | |
| CKD with BP ≤130/80 mm Hg; no CKD with BP ≤140/90 mm Hg | 0.69 (0.46–1.05) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
