Implementation outcomes from a multi-site stepped wedge cluster randomized family caregiver skills training trial.

OBJECTIVE: To assess whether a team collaboration strategy (CONNECT) improves implementation outcomes of a family caregiver skills training program (iHI-FIVES). DATA SOURCES AND STUDY SETTING: iHI-FIVES was delivered to caregivers at eight Veterans Affairs (VA) medical centers. Data sources were electronic health records, staff surveys, and interviews. STUDY DESIGN: In a stepped wedge cluster randomized trial, sites were randomized to a 6-month time interval start date for iHI-FIVES launch. Sites were then randomized 1:1 to either (i) CONNECT, a team collaboration training strategy plus Replicating Effective Programs (REP), brief technical support training for staff, or (ii) REP only (non-CONNECT arm). Implementation outcomes included reach (proportion of eligible caregivers enrolled) and fidelity (proportion of expected trainings delivered). Staff interviews and surveys assessed team function including communication, implementation experience, and their relation to CONNECT and iHI-FIVES implementation outcomes. DATA COLLECTION/EXTRACTION METHODS: The sample for assessing implementation outcomes included 571 Veterans referred to VA home- and community-based services and their family caregivers eligible for iHI-FIVES. Prior to iHI-FIVES launch, staff completed 65 surveys and 62 interviews. After the start of iHI-FIVES, staff completed 52 surveys and 38 interviews. Mixed methods evaluated reach and fidelity by arm. PRINCIPAL FINDINGS: Fidelity was high overall with 88% of expected iHI-FIVES trainings delivered, and higher among REP only (non-CONNECT) compared with CONNECT sites (95% vs. 80%). Reach was 18% (average proportion of reach across eight sites) and higher among non-CONNECT compared with CONNECT sites (22% vs. 14%). Qualitative interviews revealed strong leadership support at high-reach sites. CONNECT did not influence self-reported team function. CONCLUSIONS: A team collaboration strategy (CONNECT), added to REP, required more resources to implement iHI-FIVES than REP only and did not substantially enhance reach or fidelity. Leadership support was a key condition of implementation success and may be an important factor for improving iHI-FIVES reach with national expansion.

Duke Scholars

Author Courtney Harold Van Houtven Population Health Sciences

Author Cynthia Jan Coffman Biostatistics & Bioinformatics, Division of Biostatistics

Author Virginia Wang Population Health Sciences

Author Kelli Dominick Allen Medicine, General Internal Medicine

Author Susan Nicole Hastings Medicine, Geriatrics and Palliative Care

Author George Lee Jackson Population Health Sciences

Author Nina Sperber Population Health Sciences

Author Leah L Zullig Population Health Sciences