Patient-reported performance status as a predictive tool for mortality and supportive care needs in patients with advanced cancer.

Abstract

1569 Background: Performance status is a key prognostic factor among patients with advanced cancer and can inform the need for supportive care services. However, assessment and documentation of performance status is inconsistently integrated into standard oncology care. Patient-reported performance status (PRPS) is a surrogate for clinician-assessed performance status and has the potential to address this challenge. We piloted a decision support tool in our electronic health record (EHR) that uses PRPS to prompt clinician assessment of performance status and suggest referrals to supportive care services. Methods: We identified patients with advanced cancer using a pre-built and validated registry driven by ICD codes in our EHR (EPIC). For patients with advanced cancer seen at one of our clinics, a multiple-choice question with options derived from the Eastern Cooperative Oncology Group (ECOG) scale in patient-friendly language was added to the pre-visit questionnaire sent electronically prior to oncology visits. For patients with a poor PRPS (2 or greater), a clinical decision support tool alerted providers of the patient’s self-reported performance status, encouraged input of the clinician’s ECOG assessment, and suggested referral orders to supportive care services such as palliative care, nutrition, and social work. We assessed the relationship between PRPS and ECOG with mortality using logistic regression and by calculating median time to death stratified by patients’ highest (poorest) score. Results: Between 3/11/24 and 3/11/25, 3,376 PRPS surveys were completed by 1,343 oncology patients. Seventeen percent of patients reported poor PRPS (≥2). Worsening PRPS was strongly associated with mortality (3% for PRPS 0 vs. 44% for PRPS 4) and with shorter median time-to-death (163 days for PRPS 0 vs. 19 days for PRPS 4). PRPS correlated with clinician-assessed ECOG (ρ=0.3, p<0.001) and predicted mortality (ρ=0.8) comparably with ECOG (ρ=0.7). By contrast, ECOG documentation by clinicians was incomplete in 39% of patients. Conclusions: Results suggest feasibility of a PRPS-driven decision support tool to flag patients with advanced cancer and declining function at greatest risk of mortality. PRPS can help allocate limited supportive care resources and prompt advance care planning discussions for patients with greatest need.