Prospectively assessing risk of acute care utilization for patients with an active cancer diagnosis.

Abstract

208 Background: Acute care utilization (ACU)--emergency department visits or hospitalizations--is common in patients with cancer. As many as 83% of all patients with cancer visit the emergency department annually; nearly three quarters of patients with advanced cancer are hospitalized in the year after diagnosis. Much of this ACU may be preventable. Identifying patients at risk for ACU using model-based approaches has shown potential for risk stratifying certain patient subgroups. However, a model applicable to any patient with an active cancer diagnosis is needed. We developed a real time clinical prediction model to assess risk for acute care utilization in patients with an active cancer diagnosis. Methods: We completed a retrospective cohort analysis of patients with an active cancer diagnosis (defined as at least one medical oncology encounter in a 12 month period) at one health system. Clinical factors with potential to impact disease progression and ACU were identified through a clinical review. Significant variables were defined by multivariate logistic regression. Risk of ACU was further characterized through the development of a point scoring system to define the upper decile of patients at highest risk. Results: We included 8,246 patient records in the analysis. Seven variables were determined to be statistically significant: An emergency department visit in the last 90 days, chronic obstructive pulmonary disease, congestive heart failure, chronic kidney disease, low hemoglobin, low albumin, and low absolute neutrophil count. The model produced an overall C-statistic of 0.726 Each significant variable was assigned a score of 0 or 1 (with the exception of ED visits, which were given one point for each visit, with three points maximum). Each patient received a total score, resulting from the summation of the individual variable scores. An evaluation of the distribution of points determined that 10% of the patients achieved a score of 2 or higher and contributed to 46% of ACU in the last 90 days. Patients receiving 0 points were defined as low risk (73% of patients contributing to 30% of ED/admissions). Patients receiving 1 point were deemed intermediate risk (17% of patients contributing to 24% of ED/admissions). Conclusions: Risk of acute care utilization for patients with an active cancer diagnosis can be prospectively assessed. This tool is currently integrated into our clinical practice and is updated every 14 days, or any time the chart is accessed. Assessment of efficacy is ongoing.