Advance care planning norms may contribute to hospital variation in end-of-life ICU use: a simulation study

Abstract

Background: There is wide variation in end-of-life (EOL) intensive care unit (ICU) use among academic medical centers (AMCs). Our objective was to develop hypotheses regarding medical decision-making factors underlying this variation.

Methods: This was a high-fidelity simulation experiment involving a critically and terminally ill elder, followed by a survey and debriefing cognitive interview and evaluated using triangulated quantitative-qualitative comparative analysis. The study was conducted in 2 AMCs in the same state and health care system with disparate EOL ICU use. Subjects were hospital-based physicians responsible for ICU admission decisions. Measurements included treatment plan, prognosis, diagnosis, qualitative case perceptions, and clinical reasoning.

Results: Sixty-seven of 111 (60%) eligible physicians agreed to participate; 48 (72%) could be scheduled. There were no significant between-AMC differences in 3-month prognosis or treatment plan, but there were systematic differences in perceptions of the case. Case perceptions at the low-intensity AMC seemed to be influenced by the absence of a do-not-resuscitate order in the context of norms of universal code status discussion and documentation upon admission, whereas case perceptions at the high-intensity AMC seemed to be influenced by the patient's known metastatic gastric cancer in the context of norms of oncologists' avoiding code status discussions.

Conclusions: In this simulation study of 2 AMCs, hospital-based physicians had different perceptions of an identical case. We hypothesize that different advance care planning norms may have influenced their decision-making heuristics.

Keywords: Medicare; cancer; heuristics; intensive care; national health policy; palliative care; physician decision making; qualitative research; simulation; terminal care; variation.

Figure 1. Simulation

This is a screen shot of the simulation, as observed from the control room and recorded on video. The simulation included a hospital room, standardized patients, a medical chart, and bedside vital signs tracings. The case depicted a 78 year-old man with metastatic gastric cancer, accompanied by his caregiver wife. The patient was transferred from a skilled nursing facility with hypoxia most likely attributable to cancer progression. Physician subjects were summoned to the bedside by a nurse, approximately 8 hours after his transfer and initial evaluation, to assess the patient’s gradually-increasing tachypnea, tachycardia, hypotension, and hypoxia.

Figure 2. Subject recruitment

We recruited subjects from staff lists. We ascertained eligibility from departmental staff and contacted physicians. 38 (53%) of 72 eligible physicians from the low-intensity AMC agreed to participate, of whom 26 (68%) could be scheduled, for a total participation rate of 36%. 29 (74%) of 39 eligible physicians from the high-intensity AMC agreed to participate, of whom 22 (76%) could be scheduled, for a total participation rate of 56%.

Figure 3. Hypothesized mechanism underlying variation in end-of-life ICU use

We hypothesize that norms related to referring oncologist use of non-standard therapies and willingness to broach end-of-life issues, patient expectations, and norms of code status assessment and documentation by admitting physicians contribute to differences in ICU admission rates. High-intensity AMC norms favor ICU admission whereas low-intensity AMC norms favor ICU non-admission. The power of these norms was revealed by their apparent influence on subject physicians’ intuitive judgments (e.g., heuristics), diagnoses, and subsequent treatment decisions.