B25-22 Hospital-Level Variation in 90-Day Survival Among the Persistently Critically Ill: A Retrospective Cohort Analysis

Rationale Persistent critical illness (PerCI) - defined by intensive care unit (ICU) stays ≥11days due to ongoing clinical instability - is associated with higher morbidity, mortality, and resource utilization. However, recent studies on the persistently critically ill demonstrated greater long-term survival than anticipated. To further explore long-term PerCI survivorship and variables that may impact survival following hospitalization discharge, we aimed to test two hypotheses in a cohort of Veterans Affairs (VA) ICUs: 1) there is hospital-level variation in 90-day PerCI survival, and 2) urban hospitals and hospitals with less fragmented care (characterized by number of outpatient referrals to non-VA centers - i.e., community care referrals [CCR]) are associated with improved 90-day PerCI survival. Methods We retrospectively studied a cohort of Veterans admitted to VA ICUs from 2015-2023. We defined PerCI hospitalizations as ICU stays of ≥ 11 days, based on prior literature. The primary outcome was 90-day post-discharge survival. We used risk- and reliability-adjusted logistic regression models to estimate 90-day survival. We created three multilevel models and quantified variation in 90-day survival across hospitals by using intraclass correlation coefficients (ICC). We compared the ICC for all three models. Model 1 was empty, Model 2 included patient characteristics, and Model 3 included patient and hospital characteristics. We calculated median odds ratios (MOR) from the final model. A MOR of 1.0 implies that the odds of 90-day survival are equivalent across hospitals; the larger the MOR, the more important the hospital-level effects are in driving differences in survival. Results In this retrospective cohort of 105 VA hospitals, we identified 32,110 PerCI ICU hospitalizations, of which 23,617 (73.6%) survived to hospital discharge, with 56.7% (N = 18,222) remaining alive and 47.7% (N = 15,323) living at home at 90-days. After adjusting for patient characteristics, there was some variation in 90-day survival between hospitals, with ICC 0.018 (95% confidence interval [CI]: 0.010-0.026), indicating that 1.8% of variance is attributed to hospitals rather than patient differences. Hospital characteristics (hospital size, teaching status, complexity level, case mix, rurality, and CCR) did not further explain the variance (ICC: 0.018, 95% CI: 0.006-0.030). The MOR for 90-day survival was 1.26 (95% CI: 1.14-1.35). Conclusions More than half of PerCI hospitalizations survive at least 90 days after discharge, with many living at home. While there is hospital variation in 90-day survival, our measured hospital characteristics do not explain this variation. Additional research is needed to further delineate which hospital factors do impact PerCI survival.