Article Title

Development of an Algorithm to Prospectively Identify Palliative Care-Eligible Patients From the Electronic Health Record

Publication Date



information technology, palliative care, dissemination and implementation of innovations


Background: For referral-based health care programs, enrollment is usually triggered by a negative health event. This can mean that referral occurs late in the course of illness, even if patients would have benefited from earlier enrollment. In addition, a referral-based model gives little information about the total number of patients in need and provides no opportunity for outreach to better serve potentially eligible patients. The ability to prospectively identify appropriate patients using data from the electronic health record (EHR) presents a possible answer to these challenges. We present an evaluation of an EHR-based prospective identification method focused on palliative care-eligible patients.

Methods: Sutter Health is a health system in Northern California with 24 hospitals, more than 5,000 physicians and an embedded home health and hospice organization. Using Medicare fee-for-service (FFS) claims data from our decedents from 2009 to 2014, we created a preliminary algorithm to identify patients who were eligible for a palliative care program. We then validated this identification system against a cohort of enrolled patients. Work is currently underway to generalize this algorithm using data from Sutter Health’s EHR (Epic) to prospectively identify palliative care-appropriate patients. We plan to update our current results and perform a more rigorous analysis with access to both referral and enrollment data. We also are incorporating clinician chart reviews to identify false positives.

Results: Use of this algorithmic approach on a retrospective cohort of deceased Medicare FFS palliative care enrollees resulted in identification of 93% of enrolled patients, though it was not possible to determine the number of patients identified who were not appropriate. The expanded EHR-based analysis will supplement this gap and also provide a more comprehensive view of algorithm performance.

Conclusion: Prospective identification of palliative care-eligible patients from the wealth of EHR data presents an important opportunity to use data to better serve our patients. An effective identification system also could be an important tool for clinicians and managers in identifying underserved populations, allocating program resources and planning for program growth.




June 28th, 2017


August 10th, 2017