The Real-World Effect of Early Screening for Palliative Care Criteria in a Medical Intensive Care Unit: An Instrumental Variable Analysis

Abstract

Rationale: Early identification of intensive care unit (ICU) patients likely to benefit from specialist palliative care could reduce the time such patients spend in the ICU receiving care inconsistent with their goals. Objectives: To evaluate the real-world effects of early screening for palliative care criteria in a medical ICU. Methods: We performed a retrospective cohort study in adults admitted to the ICU using a causal inference approach with instrumental variable analysis. The intervention consisted of screening ICU admissions for palliative care trigger conditions and, if present, offering specialist palliative care consultation, which could be accepted or declined by the ICU. We evaluated specialist palliative care use in pre and postimplementation cohorts from the year before and after screening implementation began (October 2022). In the postimplementation cohort, we compared use of specialist palliative care in those who received early screening versus not. We then estimated the effect of early screening on the primary outcome of days to do-not-resuscitate (DNR) code status or ICU discharge, with death without a DNR order placed at the 99th percentile of the days to DNR or ICU discharge distribution. Secondary outcomes included: DNR order, ICU and hospital lengths of stay, hospice discharge, and mortality metrics. To address unmeasured confounding, we used two-stage least-squares instrumental variables analysis. The instrument, which predicts early screening, comprised weekend versus weekday admission and number of patients meeting palliative care criteria on a patient's ICU Days 1 and 2. Results: Among 1,282 postimplementation admissions, 626 (45%) received early screening, and 398 (28%) received specialty palliative consultation. Early receipt of specialist palliative care was higher in patients who received early screening versus not (17% vs. 1%; P < 0.001), and overall use of specialty palliative care was higher after versus before screening implementation (28% vs. 15%; P < 0.001). In the postimplementation cohort, there were no statistically significant effects of early screening on the primary outcome of days to DNR or ICU discharge (15% relative increase; 95% confidence interval, -11% to +48%) or other secondary outcomes. Conclusions: Despite significantly increased specialty palliative care consultation, there was no evidence that early screening for palliative care criteria affected time to DNR/ICU discharge or other secondary outcomes.

Keywords: critical illness; instrumental variable analysis; palliative care.