Comparison of Two Methods for Implementing Comfort Care Order Sets in the Inpatient Setting: a Cluster Randomized Trial

Background There is an ongoing need for interventions to improve quality of end-of-life care for patients in inpatient settings. Objective To compare two methods for implementing a Comfort Care Education Intervention for Palliative Care Consultation Teams (PCCT) in Veterans Affairs Medical Centers (VAMCs). Design Cluster randomized implementation trial conducted March 2015–April 2019. PCCTs were assigned to a traditional implementation approach using a teleconference or to an in-person, train-the-champion workshop to prepare PCCTs to be clinical champions at their home sites. Participants One hundred thirty-two providers from PCCTs at 47 VAMCs. Interventions Both training modalities involved review of educational materials, instruction on using an electronic Comfort Care Order Set, and coaching to deliver the intervention to other providers. Main Measurements Several processes of care were identified a priori as quality endpoints for end-of-life care (last 7 days) and abstracted from medical records of veterans who died within 9 months before or after implementation ( n = 6,491). The primary endpoint was the presence of an active order for opioid medication at time of death. Secondary endpoints were orders/administration of antipsychotics, benzodiazepines, and scopolamine, do-not-resuscitate orders, advance directives, locations of death, palliative care consultations, nasogastric tubes, intravenous lines, physical restraints, pastoral care visits, and family presence at/near time of death. Generalized estimating equations were conducted adjusting for potential covariates. Key Results Eighty-eight providers from 23 VAMCs received teleconference training; 44 providers from 23 VAMCs received in-person workshop training. Analyses found no significant differences between intervention groups in any process-of-care endpoints (primary endpoint AOR (CI) = 1.18 (0.74, 1.89). Furthermore, pre-post changes were not significant for any endpoints (primary endpoint AOR (CI) = 1.16 (0.92, 1.46). Analyses may have been limited by high baseline values on key endpoints with little room for improvement. Conclusion Findings suggest the clinical effectiveness of palliative care educational intervention was not dependent on which of the two implementation methods was used. Trial Registration ClinicalTrials.gov identifier: NCT02383173