Back

Development and operationalization of an electronic palliative care registry in a large integrated health system

Sudat, S. E.; Blanton, K.

2021-06-01 palliative medicine
10.1101/2021.05.29.21257904 medRxiv
Show abstract

BackgroundPalliative care teams generally lack prospective tools to identify individuals who could benefit from specialty palliative care, which hinders their ability to find and treat patients early in their illness trajectories. Health systems are also limited in their ability to assess how well their palliative care services reach the population in need, which in turn makes it much more difficult to determine the quality, value, and effectiveness of those services. This study describes the creation, validation, and operationalization of an electronic registry of patients potentially eligible for palliative care at Sutter Health, a large integrated health system in northern California, US. ResultsThe electronic palliative care eligibility algorithm performed well within the chart review validation sample, with an area under the receiver operating characteristic curve (AUROC) of 0.903 and area under the precision-recall curve (AUPRC) of 0.545. Within the 2.4 million individuals who contacted the SH electronic health record (EHR) during 2017, the final algorithm identified 1.1%-1.8% of patients (26,773-42,847) as potentially eligible for palliative care services. This included approximately 12.9%-17.7% of inpatients (16,392-22,507 out of 126,916), and 1.2%-1.9% of ambulatory care patients (20,140-32,212 out of 1.7 million). ConclusionsA palliative care electronic patient registry was successfully created and operationalized. Performance based on an extensive chart review sample of Sutter Health patients indicated good capture of the palliative care-appropriate population, and this is further supported by successful identification of a majority of enrollees in Sutter Healths palliative care programs during 2017.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.