U-Shaped Association Between Left Ventricular Dimension and Mortality in Patients With Heart Failure

Importance: Left ventricular structural assessment is fundamental in heart failure (HF). However, the independent prognostic value of left ventricular size beyond its association with ejection fraction remains poorly defined in real-world, large-scale populations. Objective: To investigate the association between left ventricular dimension and mortality risk within a large, nationwide cohort with HF. Design, Setting, and Participants: This was a nationwide cohort study using data from the Chinese Cardiovascular Association Database-Heart Failure Center Registry. Patients were enrolled from January 1, 2018, to May 31, 2022. The multicenter study involved 723 centers across 31 provincial-level administrative regions in mainland China. The study included patients hospitalized with HF. Patients were categorized into groups with a small, normal, or large left ventricle (LV) according to American Society of Echocardiography criteria for LV end-diastolic diameter (LVEDD). Data analysis was conducted from March to June 2025. Exposure: LVEDD measured by echocardiography. Main Outcomes and Measures: The primary and secondary end points were all-cause mortality and cardiovascular mortality, respectively. Results: A total of 273 921 patients (median [IQR] age, 71.0 [62.0-79.0] years; 161 589 male [59.0%]) hospitalized with HF were included in this study. A significant U-shaped association was found between LVEDD and both all-cause and cardiovascular mortality (P for nonlinearity <.001). Both a small LV (adjusted HR [aHR], 1.32; 95% CI, 1.28-1.37; P < .001) and a large LV (aHR, 1.38; 95% CI, 1.35-1.40; P < .001) were independently associated with elevated all-cause mortality. Sex-specific optimal LVEDD thresholds were identified (47 mm for male patients, 43 mm for female patients), with each 1-mm deviation associated with a significant increase in mortality risk. These findings were consistent across prespecified subgroups and were further corroborated by analysis of LVEDD indexed to body surface area and by extensive sensitivity analyses, including competing risk models and complete-case analyses. Conclusions and Relevance: This large-scale study established that a U-shaped association exists between LVEDD and mortality in HF, suggesting that both abnormally small and abnormally large ventricles signify high risk, likely through distinct mechanisms. These findings support the integration of LV size assessment into routine risk stratification to guide personalized management.