Abstract
Dear Editor:
As the field of palliative care grows, it remains unclear whether our services are reaching the patients and caregivers who are most likely to benefit. Identifying this cohort—“the denominator”—is relevant to several aspects of palliative care delivery. At the local level, specialty programs must understand who “the denominator” population is to develop appropriate consultation triggers and prompts, plan for clinical team growth, and evaluate quality of care provided to this population. With a specialty workforce edging closer to its clinical capacity and the need to disseminate palliative care principles and skills to nonspecialty clinicians, identifying patients who would most benefit from either specialist consultations or palliative care-inspired philosophies of care (e.g., advance care planning) remains an important priority.1,2
Parallel to the growth of specialty palliative care services has been the rapid proliferation of electronic medical records (EMRs) in healthcare. In addition to providing mechanisms for billing and documentation of services provided, EMRs contain a trove of data regarding patient, treatment, and utilization characteristics that could identify populations of interest. Combined with clinical alerts or reminders, EMRs have the potential to play an active role in identifying patients who could benefit from palliative care services.
Beginning in 2016, we aimed to realize this potential within the EMR at Cone Health, a community-based health system in central North Carolina. We brought together champions from information technology, primary care, graduate medical education, and palliative care to identify seriously ill Medicare patients who could benefit from advance care planning in the internal medicine residency clinic population as a part of their annual wellness visit. Specifically, we set out to apply three criteria (diagnoses, utilization, and functional limitations) associated with increasing morbidity and mortality in a two-year prospective review of Medicare claims data. 3 Diagnostic codes and hospital admissions or ED visits in the past 12 months were assessed by querying structured data fields related to billing. Among those patients seen in the clinic at least once in the last three years, we also compiled data from their last visit collected by the nurses: the individual's interest in advance care planning, whether or not he/she had advance directives, and reported difficulties with ADLs.
We encountered challenges and learned several lessons in this project, which may inform similar efforts at other institutions. First, our EMR has undergone several updates in the past three years, each of which shifted the location of data and thus required several attempts at finding the needed data for extraction. Second, the accuracy of diagnostic codes is dependent upon the precision of clinician documentation, which sometimes did not reflect the severity of illness if modifiers were missing. Third, our EMR could not capture utilization outside our health system, including among other hospitals and nursing homes. Finally, functional assessments, including ADLs, are not recorded systematically in the EMR or captured by administrative data, and raise concerns about precision and reliability.
Timely and appropriate delivery of palliative care services to those who could benefit most will be aided by closer alignment between data in EMRs and real-time analytics. Challenges still exist, but the potential of leveraging data from multiple sources, including EMRs, to identify patients who could benefit from an extra layer of support is immense. Palliative care should continue to explore how to harness the potential of all data sources to gain insights into our clinical practice. This approach embraces a philosophy that spurs innovation in serious illness care, in ways that are both high value and high tech.