Background: Identifying patients who may benefit from a Serious Illness Conversation (SIC) in the hospital is an important step in increasing SIC timeliness. Epic’s Readmission Risk Score (RRS) is an electronic health record integrated composite score (0-100%) that includes diagnostic, laboratory, medication, order, and utilization variables to predict unplanned, 30-day readmission and was found by our group to be associated with six-month mortality. The aim of this study is to assess if RRS combined with clinician response to the “surprise question” aids in identification of hospitalized patients who may benefit from SICs.
Methods: In this retrospective study, patient encounters >18 y.o. with an admission to a general medicine service at an academic medical center from January 2019 to October 2021 and an RRS >28% were included and randomly selected electronically. Three clinicians (2 MDs and 1 fourth-year medical student) independently performed chart reviews to answer the following questions: 1) Would you be surprised if the patient died in the next 12 months? 2) Can the patient meaningfully participate in an SIC? 3) If the patient cannot participate in an SIC, why not? Each chart was reviewed by 2 clinicians with disagreements resolved by consensus-based discussion. Patient encounter characteristics and outcomes including standardized documentation of an SIC and post-discharge, 3-month mortality are reported descriptively. Fisher’s exact test was used to assess statistically significant differences.
Results: A total of 202 patient encounter charts were reviewed. Clinicians felt they would not be surprised if 156 patients (77%) died in the next 12 months (i.e., answered “no” to the surprise question). Of these patients, 119 (59%) would be able to meaningfully participate in an SIC (Figure 1). Primary reasons for inability to participate included encephalopathy and dementia. Patients where clinicians answered “no” to the surprise question were older and had a higher Elixhauser/Van Walraven comorbidity score. RRS was similar between groups, although patients where clinicians answered “no” to the surprise question were less likely to have 5 or more hospitalizations in the prior year. Patients where clinicians answered “no” to the surprise question were more likely to have a documented SIC (9% vs 0%, p value 0.04), palliative care consultation (23% vs 2%, p value <.01), change in code status from full to do not resuscitate/ do not intubate (17% vs 0%, p value <.01), hospice enrollment (3% vs 0%, p value 0.58) and post-discharge, 3-month mortality (8% vs 0%, p value 0.04) compared to patients where clinicians answered “yes” to the surprise question.
Conclusions: In a medical record review of patients with an RRS > 28%, clinicians felt they would not be surprised if 77% of patients died in the next 12 months. Most of these patients were determined to be able to meaningfully participate in an SIC. Patients where clinicians answered “no” to the surprise question were more likely to have a documented SIC and post-discharge, 3-month mortality. RRS, coupled with clinician acumen, may be a practical though imperfect means to identify patients who may benefit from an SIC during hospitalization. Encephalopathy and dementia were primary reasons for inability to participate in an SIC, suggesting need to perform SICs earlier in the life-course. Further research is needed to determine if SIC documentation in these patients changes advance care planning outcomes and patient experience.
