Promoting early goals of care conversations in the CICU with a surprise question-based EHR workflow

The Surprise Question (SQ) - Would you be surprised if this patient died within the next 6 months? - is a validated tool for mortality prediction. The Mount Sinai Cardiac Intensive Care Unit (CICU) incorporated the SQ into a novel EHR workflow to identify patients who would benefit from early initiation of Palliative Care (PC).

Implementation of the SQ proceeded in two steps. During the feasibility pilot (December 2021-March 2022), providers answered the SQ using an EXCEL spreadsheet for all CICU patients, without changing other workflows. In April 2022, the CICU launched a new workflow-column built into the Epic patient-list dashboard with the SQ as the backbone. For patients with SQ answers of "NO," providers were prompted to facilitate and document a goals of care (GOC) conversation. We conducted a retrospective, observational, quasi-experimental study of all admissions to the CICU with SQ = NO between December 2021-September 2022. Clinical data was obtained via EHR query and chart review. We compared the frequency and timing of GOC conversations and the likelihood of redirected GOC (defined as code status change and/or hospice discharge) during the 3-month pilot versus the 6-month implementation period.

195 admissions were included: median [IQR] age 72.0 [61.0, 84.0] years; LOS > 5 days 43.6%; CICU mortality 17.9%. These clinical characteristics were comparable between the pilot (N = 57) and implementation (N = 138) periods. However, ICU interventions (i.e. mechanical ventilation, renal replacement therapy) were more common among the pilot cohort (52.6% vs. 33.3%, p = .015). For the primary outcomes, compared to the pilot period, there was a significantly higher frequency of GOC conversations (61.4% vs. 81.2%, p = .004) and GOC conversations < 2 days from CICU admission (40.4% vs. 61.6%, p = .007) in the intervention period. There was no difference in the likelihood of redirected GOC towards comfort or no escalation (28.1% vs. 21.0%, p = .288).

We facilitated earlier GOC conversations directed to critically ill patients with high mortality risk by integrating the SQ into the EHR.