Background: Heart failure (HF) epidemiologic research is problematic because of a lack of universally applied, highly specific diagnostic criteria. We compared 2 sets of diagnostic criteria: one based on standardized clinical and radiographic data (Framingham - FHS) and another based on treating physician’s diagnosis and use of specific HF therapy (Women’s Health Initiative - WHI).
Methods: In the WHI Hormone Therapy (HT) trials of 27,347 post-menopausal women age 50 –79 years, potential first time HF events were identified via participant self-report during semiannual contact or via adjudication of another primary outcome. All events were reviewed and adjudicated by 3 cardiologists using WHI and FHS criteria. Participants were categorized as no HF (HF0; n=25,554) if none of the criteria were met (no WHI/no FHS), HF by one criterion (HF1; n=275) if only WHI (n=259) or FHS (n=8) criteria were met (yes WHI/no FHS or no WHI/yes FHS), and HF by both criteria (HF2; n=331) if both criteria were met (yes WHI/yes FHS). Multivariable Cox models were used to compare cardiovascular disease (CVD) and all-cause mortality between groups with different criteria for HF.
Results: Incidence rates for HF were 1.3 per 1,000 person-years for the HF1 and 1.6 per 1000 person-years for the HF2 groups. The CVD mortality rates were 1.7, 15.2 and 23.4 per 1000 person-years for HF0, HF1 and HF2 groups, respectively (p<0.01 from log-rank test). After adjustment for age, race, blood pressure, body mass index, smoking, alcohol use, hypertension, diabetes, high cholesterol, interim myocardial infarction and HT randomization assignment, the HR for CVD mortality was 5.9 (95% CI 4.4 –7.9) for HF2 and 4.8 (95% CI 3.4 – 6.9) for HF1 (p-value 0.1 for HF1 vs. HF2) when compared to the HF0 group. The risk of all-cause mortality was similar for HF2 (HR 3.0, 95% CI 2.5–3.8) and HF1 (HR 3.3, 95% CI 2.6 – 4.2) compared to the HF0 group.
Conclusions: Women with HF diagnosed by a combination of standardized and clinical HF criteria have somewhat higher CVD mortality rates than when only one set of criteria is used alone. The higher CVD mortality supports a greater specificity of the combined approach. Expanded criteria for HF may provide additional prognostic information in epidemiologic studies.