A Validated EMR-Based Algorithm to Detect and Quantify Palliative Care Needs in Seriously Ill Hospitalized Patients

• Published in: Journal of pain and symptom management 2024-05, Vol.67 (5), p.e628-e629
• Main Authors: Sharpless, Joanna, MD, Goto, Yuika, MD, O'Riordan, David L., PhD, Pantilat, Steven, MD, Schoenherr, Laura A., MD
• Format: Article
• Language: English
• Subjects: Anesthesia; Pain Medicine
• ISSN: 0885-3924
• DOI: 10.1016/j.jpainsymman.2024.02.062

Outcomes1. Participants will describe how population-based methods to systematically identify palliative care need make access to palliative care services more equitable by mitigating clinician bias, ensure limited resources are provided to those most likely to benefit, and allow programs to quantify population need to argue for additional resources. 2. Participants will analyze the criteria included in our validated algorithm, identify those criteria that can be directly applied in their own clinical setting, and propose changes to criteria that require tailoring to maximize their utility in their local environment. Key MessageThis validated algorithm could fundamentally change the approach to care for people with serious illness from one in which patients are “lucky” to receive palliative care to one in which their needs are reliably and proactively identified and addressed. It could reduce disparities by mitigating clinician bias in referral and help programs quantify population need to advocate for additional resources. IntroductionPopulation-based methods to identify patients with palliative care (PC) needs are necessary to provide equitable and efficient access to PC services. ObjectivesCreate a validated algorithm embedded in the electronic medical record (EMR) to identify PC needs among hospitalized patients with serious illness. MethodsInterdisciplinary workgroups generated criteria to identify PC needs from either the EMR or clinician interview using literature review and clinical experience. Criteria were validated using a modified Delphi process and used to screen all hospitalized patients in one week in 2021. Workgroups used this survey's results to identify the EMR-based criteria that correlated best with PC need and developed a weighting system to generate a composite needs score. This score was validated against a “gold standard” of clinician assessment of unmet PC need from our 2021 survey. ResultsOf 76 criteria initially generated, 56 emerged from our validation process and were used to screen 224 of 268 patients (84%) identified as having serious illness. Of these, 28 criteria in four domains (psychological, physical, and spiritual distress; ACP needs) were found to correlate best with PC need. When compared with clinician assessment of unmet PC need, our composite need score resulted in an acceptable Area Under the Receiver Operating Characteristics Curve of 0.70. Furthermore, we found that patients referred for PC had a significantly h