Abstract
Abstract
Background
Multiple regurgitant lesions coexist in critically ill patients, yet cohort-level analyses typically represent each valve separately. We evaluated whether a Multi-Valvular Burden (MVB) score could summarize regurgitant involvement and whether its prognostic information differed from models retaining individual valve grades.
Methods
This retrospective cohort used MIMIC-IV v3.1 linked with MIMIC-IV-ECHO. Adults with an ICU admission and complete ordinal grading of tricuspid, mitral, and aortic regurgitation on an index echocardiogram within 72 h of ICU admission were included. MVB was the arithmetic sum of TR, MR, and AR grades (0–12). Cox models assessed 1-year all-cause mortality. Additional analyses compared MVB with separate valve-grade models, tested a high-burden definition requiring MVB > = 4 and at least two involved valves, used missing-aware AS/MS evidence proxies, and examined LVEF with structured RV-function classification.
Results
The cohort included 5,362 patients; 1,979 (36.9%) died within 1 year. Mortality increased from 24.6% at MVB 0 to 47.4% at MVB > = 4. After clinical adjustment, continuous MVB remained associated with mortality (HR 1.07 per point, 95% CI 1.05–1.09;
P
= 4 was associated with higher mortality (HR 1.35, 95% CI 1.22–1.48;
P
= 4 had at least two affected valves. A model with separate TR, MR, and AR grades showed similar discrimination to the MVB model (C-index 0.7585 vs. 0.7582). Incremental discrimination over the clinical model was small (C-index 0.7543 to 0.7582; delta 0.0039), although likelihood-ratio improvement was statistically significant (chi-square 50.61;
P
Conclusions
MVB is an exploratory summary marker rather than an outcome-derived or mechanistic weighting system. Its near-identical discrimination to a model retaining individual valve grades and very modest incremental discrimination suggest that it is best viewed as a compact descriptive marker of broader cardiovascular burden within systemic critical illness, not as a stand-alone clinical decision tool or evidence of a causal pathway to mortality.
Clinical trial number
Not applicable.
Trial registration
Not applicable.