ABSTRACT: ECHO PROJECT
Patients undergoing complex surgeries are most vulnerable during the immediate postoperative period; thus,
handoffs from the OR (operating room) to ICU (intensive care unit) require seamless communication and
coordination between surgical, anesthesia, and critical care teams. Postoperative handoffs are a threat to
patient safety, causing ~35% of medical errors in the US. To mitigate these errors, the National Patient Safety
Goal (2E) necessitated the “standardization” of handoff process and content, which resulted in adoption of
information transfer checklists, handoff process-based protocols, or both. Although such strategies have
improved handoff quality, our meta-analysis found that such improvements were temporary and had limited
sustainability, due to the structured formats imposing “rigid” standardization with limited flexibility and support
for interactive and personalized communication. Our central hypothesis is that a flexible standardization
approach will lead to not only improvements in information sharing, but also improvements in shared
understanding of patient risks, handoff interactivity, and handoff duration. Towards this end, we propose to
develop the INTERACT (Intelligent interactive care continuity) handoff bundle, a flexible, standardized, EHR-
integrated, and resilient sociotechnical intervention comprised of a: (1) telemedicine-augmented handoff
process (i.e., the social component) supported by a (2) machine learning (ML)-augmented handoff report (i.e.,
the technical component). INTERACT underscores the importance of using a perioperative telemedicine suite
as a safety net to support resilience to errors in OR-ICU handoff process and content. The ML-augmented
handoff report supports personalized communication of core (i.e., standardized) and tailored (flexible) content
based on predicted patient risks for postoperative complications. Aim 1 will focus on updating our current ML
models for predicting risks associated with postoperative complications, based on state-of-the-art imputation
and feature engineering techniques. We will enhance our model-agnostic explanation framework to support
postoperative handoffs and decision-making, which will also be validated with a summative user evaluation
study. Aim 2 will follow a user-centered design approach to iteratively develop and test the INTERACT bundle
including handoff report design ideation, and usability testing, and lastly, the INTERACT bundle in-situ
simulations. Aim 3 will adopt a Hybrid Type 1 trial design and the Care Transitions Framework to evaluate the
effectiveness and implementation-potential of the INTERACT bundle. Our primary outcome is information
sharing score (i.e., a measure of information completeness), while secondary outcomes include information
inaccuracies, realized errors and adverse events, and ICU length of stay. With an integrated multidisciplinary
approach to improving perioperative care transitions, the proposed INTERACT bundle will address the stated
AHRQ FOA goals of “defragmenting information, improving communication, and assuring care team access to
reliable and complete health information; and empowering care teams to improve health outcomes.”