Balanced informational feedback as part of an implementation/deimplementation bundle to support deimplementation...

Effective de-implementation approaches to reduce low-value or harmful care are limited. Prolonged antimicrobial use after skin closure is ineffective and increases harm but is nonetheless common following cardiovascular implantable electronic device (CIED) procedures. The aim of this quasi-experimental Implementation/Effectiveness study was to use an implementation/de-implementation (I/D) bundle of strategies to promote de-implementation of guideline-discordant prolonged antimicrobial use and improve outcomes.

Hybrid III Implementation/Effectiveness study grounded in learning/unlearning theory and guided by the integrated-Promoting Action on Research Implementation in Health Services (i-PARIHS) framework with a non-randomized parallel group trial design (3 intervention and 8 control sites) over a 12-month period with an additional 12-months of follow up to assess sustainment. The I/D bundle included: blended facilitation, local champions, provider education, and access to locally-customized audit-feedback data about facility-level guideline compliance with benchmarking, CIED infections, and antimicrobial harms (acute kidney injury, allergic events, Clostridiodes difficile infections). Implementation outcomes were assessed quantitatively using a difference-in-differences model and qualitatively; clinical outcomes were assessed quantitatively.

2,293 CIED procedures were performed at the 3 intervention sites and 5,273 at the 8 matched control sites. Following I/D bundle intervention, rates of guideline non-compliance fell in the intervention sites but not the controls (OR, 0.16, 95% CI, 0.12-0.21). There was no change in CIED infections following the intervention (OR, 0.96, 95% CI, 0.48-1.93, p = 0.91) and a trend toward a reduction in composite antimicrobial harms (OR = 0.59; 95% CI: 0.34, 1.02, p-value = 0.059). Improvements were sustained during the 12-month period after active implementation and associated with a decrease in composite harm (aOR = 0.57; 95% CI: 0.34-0.91, p = 0.02). Strategies identified as effective for promoting de-implementation included education, locally-customized dashboard audit-feedback data with benchmarking to high-performing sites and feedback about both harm and benefit, blended facilitation, and provider's readiness to change. Review and updating of hospital factors that promote sustainment of ineffective care, such as automated order sets, were also important. Practices learned and introduced from other practice settings were important barriers to change.

The I/D bundle that included audit-feedback about patient-level harms of guideline discordant care was an effective strategy for reducing low-value care and improving quality.

ClinicalTrials.Gov: NCT05020418.