Following the Breadcrumbs of Palliative Care Financial Sustainability to Big Data

Abstract

Dear Editor:

The success and future of palliative care (PC) is jeopardized by its deficit funding, which limits its availability and sustainability. However, data science research can translate into improved clinical practice, policy, and payment reform to support high-value services such as PC into the future. The Center to Advance PC (CAPC) reports that PC billable collections on average cover only 30% of service expenses in a traditional fee-for-service payment model given the significant time and coordination of care involved in consultations for seriously ill patients.¹ This large gap in reimbursement creates an ongoing inherent challenge to find and lobby for sustained funding primarily from revenue generated by the entire health system. Real-world data, such as from electronic health records (EHRs) and claims data, are necessary to generate economic evidence to support the new and evolving discipline of PC. Evidence-based research from large data repositories will allow the PC field to advance clinical care, develop payment mechanisms to support growth and sustainability through improved quality outcomes at reduced costs.

Evidence demonstrates that the delivery of early PC is associated with improved patient satisfaction, improved symptom control and quality of life, reduced health care utilization, and improved survival in a population of lung cancer patients.² Given our unique population, traditional PC research retrospectively examines the quality of care provided during the last 6 months of life in decedent populations (Table 1). To identify critical gaps in the availability of PC data, our team conducted a gap analysis by comparing the ideal data elements with those currently captured in a unique and robust large statewide data repository—the OneFlorida Clinical Research Consortium Data Trust.³ The OneFlorida Data Trust is a large repository that contains robust longitudinal patient-level data of ∼15 million Floridians (>60% of the population in the third largest state in the United States), including cancer registry, vital statistics, along with the EHR data from its 13 heterogenous public and private clinical partners. The Data Trust is one of nine Patient-Centered Outcomes Research Institute (PCORI)-funded clinical data research networks, part of the national Patient-Centered Outcomes Research Network (PCORnet).⁴

Table 1.

Palliative Care Research Data Elements: Gap Analysis of Data Availability to Support Research at the Level of a Single Institution or Across Institutions

PC data element	Availability in health system datasets	Availability in the OneFlorida Data Trust	Gap analysis: (barrier to implementation)
Decedents mortality data	No, unless the patient died inside hospital	Yes, tracked by third party (Datavant)	N/A
Hospitalizations for decedents in last 6 months of life: • Emergency department visits • Medical and intensive care units admissions • Length of stay and timing of admissions • Hospitalization within 30 days of death • 30-day readmissions	Yes	Yes	N/A
Hospital intensity (e.g., revenue codes for intensive care unit days)	Yes	No, but can be calculated using proxy data	Claims data considered proprietary. Public CMS claims data available through ResDAC-linked access to OneFlorida
Professional fee (carrier) data including ICD-10 V66.7/Z51.5/PC codes	Yes	No, but can be calculated using proxy data	Claims data considered proprietary. Public CMS claims data available through ResDAC-linked access to OneFlorida PC ICD codes not sensitive or specific
Ability to identify individual PC providers' professional fee billing by specialty or NPI, including use and timing of PC inpatient, outpatient visits, and home visits	Yes	Yes, but with limitations	PC providers may practice two disciplines making tracking PC specific data difficult. OneFlorida will work toward linking NPI board certification specialties
Revenue codes for prehospice/PC (069x)	No	No	Limited use
Use and timing of hospice services	No	No	Outside private vendor data
Fiscal outcomes—costs, contribution margin, and net margin (i.e., charges and payments)	Yes	No, but can be calculated using proxy data	Claims data considered proprietary. Public CMS claims data available through linked access to OneFlorida

CMS, Center for Medicare and Medicaid Services; N/A, not applicable; NPI, National Provider Identifier; ResDAC, Research Data Assistance Center; PC, palliative care.

Our assessment of the data elements available in the OneFlorida Data Trust revealed robust data far exceeding data available for traditional single institution research. Gap analysis of current and ideal data elements necessary to perform high-quality PC health economic research are listed in Table 1. This table contrasts data elements currently available with those needed for high-quality PC research based on literature review. In addition, data allow for future development of a quantifiable PC phenotype and application of artificial intelligence methods for data analysis. This brief commentary summarizing a data repository gap analysis of PC variables is a call to action for leaders, health systems, EHR vendors, and data repositories to translate knowledge into action to fill this critical gap. Doing so will allow the PC field to grow the body of evidence to support services over time as medicine evolves, especially during this critical tipping point of “value versus volume” health care delivery.⁵

References

Center to Advance Palliative Care: Financing a Palliative Care Program. www.capc.org/toolkits/funding-a-palliative-care-program/. (Last accessed November 1, 2019).

Rabow

, Kvale

, Barbour

, et al.: Moving upstream: A review of the evidence of the impact of outpatient palliative care. J Palliat Med, 2013; 16:1540–1549.

University of Florida Clinical and Translational Science Institute: OneFlorida Clinical Research Consortium. www.ctsi.ufl.edu/ctsa-consortium-projects/oneflorida. (Last accessed November 1, 2019).

PCORnet—the National Patient-Centered Clinical Research Network: The PCORnet Common Data Model. https://pcornet.org/data-driven-common-model/. (Last accessed November 1, 2019).

Graham

, Logan

, Harrison

, et al.: Lost in knowledge translation: Time for a map?. J Contin Educ Health Prof, 2006; 26:13–24.