Prognosticating Outcomes and Nudging Decisions With Electronic Records in the ICU Trial

Overview

This is a pragmatic, stepped-wedge, cluster randomized trial testing the real-world effectiveness of two different electronic health record (EHR) behavioral interventions in improving a number of patient- and family-centered processes and outcomes of care among seriously ill hospitalized patients. The investigators hypothesize that outcomes can be improved without raising costs by requiring intensive care unit clinicians to (i) document a prognostic estimate and (ii) provide a justification if they choose not to offer patients the option of comfort-oriented care. To test this hypothesis the investigators will conduct a 33-month trial at 17 intensive care units in 10 hospitals using the same Cerner EHR within Atrium Health System.

The PONDER-ICU trial aims to generate large-scale, experimental evidence regarding the real-world effectiveness of two different electronic health record (EHR) behavioral interventions in improving a number of patient- and family-centered processes and outcomes of care among seriously ill hospitalized patients. The interventions are designed to increase intensive care unit (ICU) physicians' and advanced practitioners' (physician assistants and nurse practitioners) engagement of critically ill patients and caregivers in discussions about alternative treatment options, including care focused on comfort. To achieve this goal, the investigators will conduct a 33-month pragmatic, stepped-wedge cluster randomized clinical trial at 17 ICUs within 10 Atrium Health System hospitals. The investigators hypothesize that outcomes can be improved without raising costs by requiring ICU clinicians to (i) document a prognostic estimate (Intervention A) and (ii) provide a justification if they choose not to offer patients the option of comfort-oriented care (Intervention B). Approximately 4,750 adult patients (1) with chronic life-limiting illness and receiving continuous mechanical ventilation for ≥48 hours will be enrolled. Participating hospitals will be randomized into 5 clusters of 2 hospitals each. Each hospital will first contribute a minimum of 5 months of data collection during usual care in a control phase. Then, using the step-wedge design, all hospitals will implement the two EHR-based interventions. The order in which the interventions will be adopted and the timing of adoption at each hospital will both be determined by random assignment. After 12 months of utilizing Intervention A or Intervention B, each hospital will adopt and implement the second intervention in combination with the other. By the end of the trial, all hospitals will have utilized the combined interventions for at least 4 months. The primary outcome is a composite measure of hospital length of stay and mortality. Secondary outcomes include an array of clinical outcomes, as well as palliative care-related process measures.