AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
A Standardized Hospital Discharge Mobility Assessment Provides Superior Prognostic Fit vs Cardiac Biomarkers and LVEF in Predicting 1-Year Outcomes
By Ramonfaur D, Ryan D, Vest AR et al.
Background: HF risk stratification typically relies on cardiac-specific metrics like LVEF and natriuretic peptides, yet HF is a multisystem disease that often results in impaired physical function, particularly in the context of acute heart failure hospitalization (AHF). The AM-PAC “6-Clicks” basic mobility short form is routinely assessed by inpatient physical therapists at many health systems and scored from 6-24 (higher = greater mobility). In a separate analysis, we demonstrated that the 6-Clicks score was independently associated with 1-year HF outcomes, but its prognostic value relative to established HF-specific markers NT-proBNP and LVEF is unknown. Methods: We identified adults discharged after a HF hospitalization across 12 hospitals in 2 states (OH, FL); patients with cardiac amyloid, LVAD, or heart transplant were excluded. Multivariable Cox models adjusted for sex, race, age group, eGFR, NT-proBNP, atrial fibrillation/flutter, and prior HF admission predicted a composite of all-cause hospitalization and death, and each component separately, over 1 year. Because the candidate models are non-nested (6-Clicks vs. NT-proBNP; 6-Clicks vs. LVEF), they cannot be compared by a standard likelihood ratio test. We therefore used the Vuong test for non-nested models, which evaluates two things: (1) whether the models are statistically distinguishable (the variance test), and, if so, (2) which model fits better (the directional test, based on the difference in pointwise log-likelihoods). The base covariate set was held identical across comparisons, substituting only the marker of interest (discharge 6-Clicks, NT-proBNP, or LVEF). Results: Among 31,181 patients (50.1% female; 67.8% aged 65-89 years; mean discharge 6-Clicks 17.0±5.3). For every outcome, the Vuong variance test confirmed that the competing models were statistically distinguishable (P<0.001 for all). On directional testing, the 6-Clicks model fit significantly better than the NT-proBNP model for the composite, all-cause hospitalization, and death (P<0.001 for all), while the reverse comparison (NT-proBNP favored over 6-Clicks) was non-significant in every case. The same pattern held against LVEF: the 6-Clicks model fit each outcome significantly better (P<0.001 for all), with the reverse comparison non-significant. Conclusions: To our knowledge, this represents the largest analysis of a standardized mobility score in AHF. Across non-nested model comparisons, the discharge 6-Clicks score provided superior prognostic fit compared with NT-proBNP and LVEF in predicting 1-year all-cause readmission, death, and their composite, suggesting physical function may have underrecognized and readily assessable prognostic value in patients hospitalized with HF.