Patient-reported symptom profiles: A tool to direct palliative care referral.

Abstract

93 Background: Both routine PRO monitoring and specialty palliative care are associated with a survival benefit in advanced cancer patients, yet the integration of these efforts has not been described. As PRO collection becomes standard of care, there is a need to understand how PROs may be leveraged to identify patients who could benefit from palliative care referral. Methods: We applied latent profile analysis (LPA) to PROs from a national palliative care registry collected from 2008 – 2017 on 745 solid tumor patients at the time of initial outpatient palliative care visit. LPA is an established method to identify unobserved groups from variables of interest. We used patients’ responses to 11 questions across 4 palliative care domains (9 ESAS items and 2 items for social and existential distress) to generate “PRO profiles” and examined their clinical and demographic correlates using multinomial logistic regression. Results: Four PRO profiles were identified using 9 ESAS items (Table). Young age, metastatic disease, and tumor type (including breast, GYN, GI) differed across PRO profiles compared to lung. Patients with PPS < 70% were 4.5 times more likely to be in the High vs. Low symptom profile compared to those with PPS > 70%. Subgroup analyses showed correlation of social and existential distress PROs with Mood and High profiles. Conclusions: Cancer patients referred to outpatient palliative care can be differentiated into clinically meaningful PRO profiles using brief, routinely collected data in the real-world setting. PRO profiles provide richer data on patient needs compared to prognosis estimates or cancer stage, and synthesize big data for use in clinical practice and epidemiologic research. PRO profiles should include comprehensive palliative domains and be tested as screening thresholds for palliative care referral. [Table: see text]