Practice Variation in Vasoactive Medication Use for Cardiogenic Shock Across ICU Subtypes: A Retrospective Cohort Study

BACKGROUNDCardiogenic shock has significant associated morbidity and mortality with a wide range of vasoactive management strategies. However, the extent to which variation in vasoactive and inodilator therapies is explained by patient level variables versus clinical practice variation remains underexplored. METHODSThe cohort included 4,525 patients admitted to ICUs at Michigan Medicine from 01/01/2014-12/31/2023 diagnosed with cardiogenic shock identified by billing codes. Vasoactive medication utilization was compared across various ICUs (cardiovascular, cardiovascular surgical, non-cardiac). Mixed-effect multiple logistic modeling was used to evaluate to what extent variation in inodilator use was associated with fixed patient-level variables (i.e. prior cardiac arrest) compared to ICU location as a random effect. RESULTSPatient encounters were classified as cardiovascular ICU (n = 1,355), cardiovascular surgical ICU (n = 1,405), non-cardiac ICU (n = 723), and multiple ICUs (n = 1,042). Vasoactive and inodilator medication use varied significantly. Inodilators were more frequently used in the cardiac ICU (44.4% [95% CI 40.4-48.3%]), cardiothoracic surgical ICU (64.9% [61.8-68.0%]), and multiple ICU patients (60.8% [57.0-64.6%]) and less frequently in non-cardiac ICU patients (18.8% [12.2-25.4%]). Heart failure was associated with more frequent inodilator use (OR 3.82 [3.06-4.80]), while increased age (OR 0.993 [0.989 - 0.997]), male sex (0.80 [0.70 - 0.91]), and prior cardiac arrest (0.68 [0.58 - 0.81]) were associated with lower use. ICU location attributed to 15.8% of variance in inodilator use, while fixed patient-level variables combined accounted for 6.1% of variance. CONCLUSIONA substantial portion of variation in vasoactive medication utilization was attributed to ICU setting.