Using Data to Risk Stratify a Palliative Care Population

Topic: Building Palliative Care Into the Organizational DNA

Intermountain HealthCare_Krutka.April 2017_10 18 17.pdf.png

One of the largest barriers for seriously ill patients in need of Palliative Care, is timely identification of palliative care needs. In the acute care setting, many admitting and consulting physicians struggle with identification of a patients in need of expert level Palliative Care. Intermountain Healthcare, in collaboration with Cerner Corporation, developed an electronic palliative care identification algorithm, or trigger, which leverages historical and near real-time data to identify patients that would benefit from palliative care services. During this project, several iterations of the algorithm ran in silent mode in the electronic medical record. After each iteration, chart reviews were completed by the clinical palliative care team, which allowed for confirmation of appropriately identified patients and subsequent adjustments to the algorithm. The goal of this tool is to assist clinicians in determining the appropriate time to consult palliative care, while also increasing access to palliative care services across the continuum of care. With these adjustments, this tool has become more accurate with a current positive predictive value (PPV) of 94%. It has identified nearly 26% of this 310 bed hospital’s adult population as appropriate for palliative care services. The high patient volume combined with the current staffing model of the hospital revealed the need for a risk stratification. Primary palliative care patients are defined as lower risk and may be managed by the attending physician or primary care physician with strengthened educational resources for the medical team. Secondary palliative care is defined as high risk patients who need expert consultation during the current hospital encounter or post hospitalization by ambulatory specialty palliative care team. During the fourth iteration, patients were separated into one of these two groups. 438 unique patient charts were reviewed. 65% were found to need an expert secondary palliative care consult, 28% needed strengthened primary level palliative care, and 7% were found to not be appropriate for palliative care interventions. The next phase of the project will transition the current algorithm into a more sophisticated machine learning data model, where we will use our current algorithm as an identifier and use clinical data to help us define the palliative care population into High, Moderate or Low risk of palliative care intervention.

Author

  • April Krutka, DO
  • Medical Director Palliative Care, McKay Dee Hospital
  • Intermountain Healthcare
  • 4403 Harrison Blvd
  • Ogden, UT 84403
  • (801) 387-7900

Co-authors

  • Emmie Gardner, MSW, LCSW, ACHE
  • Hannah Luetke-Stahlman, MPA

Register Now

Register for the 2018 CAPC National Seminar and Pre-Conference Workshops by September 24 for Early Bird pricing.

Learn More and Register 

If you need assistance, please use the customer support form.