The Coordination Toolkit and Coaching Project: Cluster-Randomized Quality Improvement Initiative to Improve Patient Experience of Care Coordination

Background: Given persistent gaps in coordination of care for medically complex primary care patients, efficient strategies are needed to promote better care coordination.

Objective: The Coordination Toolkit and Coaching project compared two toolkit-based strategies of differing intensity to improve care coordination at VA primary care clinics.

Design: Multi-site, cluster-randomized QI initiative.

Participants: Twelve VA primary care clinics matched in 6 pairs.

Interventions: We used a computer-generated allocation sequence to randomize clinics within each pair to two implementation strategies. Active control clinics received an online toolkit with evidence-based tools and QI coaching manual. Intervention clinics received the online toolkit plus weekly assistance from a distance coach for 12 months.

Main measures: We quantified patient experience of general care coordination using the Health Care System Hassles Scale (primary outcome) mailed at baseline and 12-month follow-up to serial cross-sectional patient samples. We measured the difference-in-difference (DiD) in clinic-level-predicted mean counts of hassles between coached and non-coached clinics, adjusting for clustering and patient characteristics using zero-inflated negative binomial regression and bootstrapping to obtain 95% confidence intervals. Other measures included care coordination QI projects attempted, tools adopted, and patient-reported exposure to projects.

Key results: N = 2,484 (49%) patients completed baseline surveys and 2,481 (48%) completed follow-ups. Six coached clinics versus five non-coached clinics attempted QI projects. All coached clinics versus two non-coached clinics attempted more than one project or projects that were multifaceted (i.e., involving multiple components addressing a common goal). Five coached versus three non-coached clinics used 1-2 toolkit tools. Both the coached and non-coached clinics experienced pre-post reductions in hassle counts over the study period (- 0.42 (- 0.76, - 0.08) non-coached; - 0.40 (- 0.75, - 0.06) coached). However, the DiD (0.02 (- 0.47, 0.50)) was not statistically significant; coaching did not improve patient experience of care coordination relative to the toolkit alone.

Conclusion: Although coached clinics attempted more or more complex QI projects and used more tools than non-coached clinics, coaching provided no additional benefit versus the online toolkit alone in patient-reported outcomes.

Trial registration: ClinicalTrials.gov identifier: NCT03063294.