Moderate aortic stenosis identified during intensive care admission for a non-cardiac primary diagnosis and 90-day mortality: a retrospective cohort study using the MIMIC-IV echocardiography database

Abstract

Abstract Background Moderate aortic stenosis is associated with reduced survival in ambulatory cardiology populations, but its prognostic significance when identified during an intensive care admission for a non-cardiac primary diagnosis is unknown. We examined its association with 90-day mortality in this setting. Methods Retrospective cohort using the Medical Information Mart for Intensive Care IV (MIMIC-IV) version 3.1 and its echocardiogram-matched subset (MIMIC-IV-ECHO). We included each adult’s first intensive care admission for a non-cardiac primary diagnosis with an interpretable baseline transthoracic echocardiogram recording peak aortic jet velocity. Stenosis was graded from the peak velocity (no stenosis < 2.0, sclerosis or mild 2.0–2.9, moderate 3.0–3.9, severe ≥ 4.0 m/s). The echocardiogram defined time zero; the primary outcome was all-cause mortality within a fixed 90-day window. Associations were estimated with multivariable Cox models adjusted for acute-illness severity and comorbidity, with propensity-score matching, velocity–gradient concordance grading, pre-specified ejection-fraction subgroups, and an E-value. Results Of 12,878 patients (10,511 no stenosis, 1,797 sclerosis, 411 moderate, 159 severe), 3,605 died within 90 days. Crude 90-day mortality rose across grades (26.1%, 34.4%, 40.4%, 46.5%). After adjustment, the hazard increased monotonically: sclerosis 1.17 (95% CI 1.07–1.28), moderate 1.33 (1.13–1.56), severe 1.61 (1.27–2.03). The moderate-stenosis association was modest but consistent, attenuating slightly in the most conservative specifications (propensity-score matched 1.20 [1.02–1.40]; parsimonious 1.22 [1.05–1.43]); the E-value was 1.73 (lower limit 1.41). It was numerically concentrated in the first two weeks (1.58 [1.30–1.93], days 0–14) and numerically larger with reduced (1.47 [1.10–1.97]) than preserved ejection fraction (1.10 [0.87–1.39]); neither the proportional-hazards nor the grade-by-ejection-fraction interaction test was significant (interaction P = 0.21), and both are exploratory numerical patterns, not demonstrated effect modification. Conclusions In non-cardiac critical illness, velocity-defined moderate aortic stenosis identified during the intensive care admission was independently associated with higher 90-day mortality, with exploratory numerical patterns of greater early risk and of greater risk with reduced ejection fraction. These data identify a higher-risk phenotype; they do not establish a causal adverse effect of moderate valvular obstruction itself. It is best regarded as a prognostic marker warranting closer surveillance rather than, on current evidence, a target for valve intervention.