Background: Serious Illness Conversations (SICs) lead to a better understanding of patient values and preferences, improved patient and family satisfaction, and other benefits. Prognostic awareness and early identification of patients who may benefit from an SIC remains a challenge. Epic’s Risk of Unplanned Readmission (Readmission Risk Score), a composite score (0-100) that includes clinical factors such as diagnoses, laboratory values, medications, orders, and encounters to predict 30-day readmission, is widely available across health systems that utilize Epic’s electronic health record (EHR). The aims of this study are to assess if Epic’s Readmission Risk Score calculated on admission is a useful predictor of six-month mortality, and if it and other established factors are associated with SIC documentation during hospitalization.
Methods: We conducted retrospective analyses of general medicine patient encounters at a community teaching hospital affiliated with an academic medical center from October 2018 to August 2019. All data were obtained from the enterprise data warehouse (EDW), including data from our EHR and other administrative sources. We identified all encounters associated with death within six months following hospitalization. A receiver operating characteristic (ROC) curve was used to evaluate whether Epic Readmission Risk Score on admission was a useful predictor of six-month mortality. Hospital encounters with a documented SIC were matched to encounters without a documented SIC (controls) in a ratio of 1:3, matched by gender and age (±5 years) to facilitate full exploration of factors associated with SIC documentation. A multivariable, paired logistic regression model was used to assess the strength of association of Epic Readmission Risk Score on admission with documentation of an SIC while hospitalized. A secondary analysis using Van Walraven / Elixhauser comorbidity score was generated as a comparative model.
Results: We identified 6,853 encounters (5,143 patients) during the study period: 246 (3.6%) encounters were associated with death within six months. The area under the ROC curve (AUC) for Epic’s Readmission Risk score to predict 6-month mortality was 0.71 (Figure 1). In 59 encounters (0.86%), an SIC was documented during hospitalization, and these were paired to 177 controls without a documented SIC. The first model (Table 1) demonstrated that the odds of an SIC increased by 40% (OR 1.40, 95% CI [1.22 – 1.60], p=0.01) for every 10-percentage point increase in the Epic Readmission Risk Score. Our secondary model showed that Van Walraven / Elixhauser comorbidity score and any hospitalization within the past year were significantly associated with a documented SIC during hospitalization.
Conclusions: Epic’s Readmission Risk Score when calculated on admission was predictive of six-month mortality with acceptable discrimination and was significantly associated with SIC documentation during hospitalization. This result was corroborated by a secondary model using Van Walraven / Elixhauser comborbidity score. Because the Readmission Risk Score is widely available at most institutions with Epic’s EHR, our analysis provides practical utility for identifying patients who may benefit from early SICs and subsequent downstream actions (e.g., palliative care consult, hospice referral). Because our study is retrospective and conducted at a single site, further prospective evaluation would be necessary to validate these findings at similar institutions.
