Abstract
Background: As previous studies have demonstrated that heart failure (HF) verification is inconsistent, the use of administrative databases to predict heart failure (HF) readmission or all-cause mortality (ACM) requires HF verification. The Framingham Criteria for HF (CHF1 score>4) and European HF guidelines are algorithms that assesses symptoms, physical examination, radiologic findings, echocardiography and brain natriuretic peptide (BNP). We hypothesized that CHF1 score>4 or BNP>100 pg/ml (verifiable HF) will not occur in all patients but if present will predict HF readmission and ACM.
Methods: We queried the electronic medical record from a community-based practice for a HF diagnosis between 2008-2016. We included patients with >3 months follow-up with interpretable Doppler echocardiograms. We recorded demographics, Doppler-echo variables, CHF score>4, BNP, HF readmission, and ACM with follow-up up to 2113 days.
Results: HF phenotypes included preserved ejection fraction (271 patients), mid-range ejection fraction (224 patients), and reduced ejection fraction (144 patients). Verifiable HF varied from 56.9%-60.3%. Patients with verifiable HF had increased ACM for individual phenotypes (HR=2.22 to 3.40, p=0.0009 to <0.0001). Similarly, verifiable HF resulted in increased HF readmission during follow-up for individual phenotypes (HR=1.60-3.50, p=0.0158 to <0.0001). For patients previously readmitted, ACM was increased in patients with HFpEF, HFrEF, and all phenotypes combined despite CHF1 scores<4 and BNP<100 pg/ml.
Conclusions: Verifiable HF predicted ACM and HF readmission in all individual phenotypes. HF readmission also predicted subsequent ACM despite CHF1 scores<4 and BNP<100 pg/ml.
Keywords
Heart failure, Mortality, Readmission, Heart failure phenotypes