Abstract 4144997: Pro-B-Type Natriuretic Peptide Kinetics across Pre-, Index, and Post-Acute COVID-19 in Hospitalized Acute on Chronic Heart Failure Patients: A Learning Health System Science Initiative

Introduction: Myocardial injury in patients hospitalized with acute on chronic heart failure concurrent with index SARS-CoV-2 (CoV-2) infection is well described, though studies incorporating pre- and post-acute COVID-19 (PAC) are lacking. We address this gap by estimating intensity of acutely decompensated heart failure (ADHF) using time-series pro-BNP levels across hospitalizations pre- vs. respectively index and initial readmission (PAC1). Hypothesis: Case time series analysis will reveal association (p<0.05) between changes in baseline (<24h of admission) pro-BNP levels at pre-index (<180d) vs. respectively index and PAC1 (<90d) with hospital length of stay (LOS) / Medicare Severity Diagnosis Related Groups Geometric Length of Stay (GMLOS) ratio (LOS/GMLOS) >1.0. Methods: Electronic health records underwent IRB exempted extraction of administrative, demographic, and clinical-pathophysiologic data. Continuous data summarized with median [IQR] were compared using repeated-measures ANOVA and discrete data with chi-squared testing. Confounders were statistically balanced. Results: Among 1,413 patients hospitalized between March 10, 2020 and February 29, 2024 discharged alive, not to hospice, harboring ADHF and CoV-2 infection, 91 respectively had baseline pro-BNP at Index and a pre-index hospitalization. Twenty-two underwent PAC1. Statistically isomorphic age, sex and race included 76 (68, 84) years, 43% females distributed by White (83%), Black (11%) and other races (6%). Similarly,chronic comorbidities were pooled for summary including hypertension (87%), chronic lung disease (46%), diabetes (41%), renal disease (36%), obesity (35%), and deficiency anemia (35%). Median times pre-DC to index hospitalization (56 [23, 114]) vs. Pre-DC to PAC1 admission (37 [15, 45]) days (p<0.001). Median pro-BNP (pg/mL) at Pre- (1948 [903, 4309]), Index (3081 [1458, 7230]), and PAC1 (2505 [ 1628, 8048]). Threshold increase in baseline pro-BNP associated with increased risk of LOS/GMLOS >1.0 between pre-COVID-19 vs. index and PAC1 respectively was >20% (p=0.013) and >14% (p=0.039). Conclusions: Time series pro-BNP levels obtained at serial hospitalizations can identity ADHF patients likely to exceed MS DRG GMLOS. Moreover, self-matched designs may strengthen patient-oriented learning health system science aiming to understand causal relationships between exposures and identify treatable traits to blunt or obviate duration of ADHF requiring hospitalization.