Poster 054: CRITICAL RESOURCES USE IN PATIENTS WITH ACUTE DECOMPENSATED HEART FAILURE WITHOUT CARDIOGENIC SHOCK ADMITTED TO CARDIAC INTENSIVE CARE UNIT

Background: Admissions for ADHF in the absence of cardiogenic shock (CS) has been the predominant indication for admission to the cardiac intensive care unit (CICU). However, a significant proportion of these patients do not need CICU-restricted therapies. We developed and validated a prediction tool to identify patients that need CICU admission.

Methods: We included consecutive patients admitted to CICU with ADHF in the absence of cardiogenic shock (CS) from 2014-2025 at University Health Network. Baseline characteristics were compared using Shapiro–Wilk and Kruskal-Wallis test, multivariable logistic regression with 1,000 bootstrapping was used to evaluate the independent association of each variable with outcome of interest. Kaplan-Meier curves are used to report 1-year and Cox-proportional hazard ratio with time variant used to assess for covariates association with 1 year mortality. To generate CICU resource use prediction model, the dataset was split into a derivation cohort (2014–2021) and a validation cohort (2022–2025). The final model variables were selected form a pool of forty candidate predictor variables. Variable selection was guided by prior literature, clinical relevance, and availability at the time of clinical presentation. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test, with a P value of less than 0.05 indicating poor calibration. Calibration was further assessed visually by plotting observed versus predicted probabilities across deciles of predicted risk.

Outcome: We included 890 patients that were admitted to CICU with ADHF in the absence of CS. A large proportion of patients, 344 (39%) did not use CICU restricted resource during their CICU admission. The most frequently used resources were vasopressors (30.5%) and central lines (21.8%) (Figure 1). Those with CICU resources use, had lower mean arterial pressure and sodium, elevated creatinine and longer median hospital and CICU length of stay (Table 1). Thirty-day and 90-day readmission rates did not differ significantly between those that used CICU restricted resource and those that did not. Although in-hospital (OR 1.89, 95% CI 1.01-3.45) is higher among CICU resource users, one-year mortality (HR 1.31, 95% CI 0.93-1.84) is not. Multivariable logistic regression demonstrates that odds of mortality increase with age, elevated creatinine and lower MAP (Figure 2). CICU resource use prediction model was developed based on readily available clinical variables demonstrated an AUC of 0.746 and calibration (Hosmer–Lemeshow P>0.05) (Figure 3).

Conclusion: We identified key characteristics associated with poor clinical outcome and CICU resource use amongst patients with ADHF in absence of CS. We developed and validated a risk score that accurately identifies patients likely to use CICU resource and may serve as important clinical tool particularly in resource limited settings.