Organizers and Facilitators: Katie Raffel, Jeff Schnipper, Peter Barish, Andy Auerbach

Background

The April 11, 2025 HOMERuN Collaborative Session focused on refining and discussing three potential programs for improving diagnostic accuracy in inpatient general medicine, as part of a broader initiative under the Achieving Diagnostic Excellence through Prevention and Teamwork (ADEPT) study. 

The meeting featured updates from various ADEPT intervention groups, with breakout sessions for in-depth discussions on diagnostic timeouts, second opinion programs, and patient engagement strategies.

Diagnostic Timeouts

Diagnostic timeouts encourage clinicians to use a structured tool to review common gaps in diagnostic processes. Key reflections on diagnostic timeouts included:

• Implementing diagnostic timeouts on the day of admission or within 48-72 hours could yield high benefits in identifying diagnostic errors early. Current triggers such as rapid response (RRT) calls and intensive care unit (ICU) transfers might be too late in the patient's trajectory.
• Recommendation to involve bedside nursing during diagnostic timeouts, possibly during interdisciplinary rounds.
• Utilizing artificial intelligence (AI) to identify triggers for diagnostic timeouts and involving bedside nursing during these timeouts could enhance the feasibility and impact of this intervention.

"I think if you were using RRT as a trigger, it would be fairly easy to add in a diagnostic timeout and document your decision making because that's already part of an existing structure. If you were trying to broaden it, I worry it would be very difficult to get."

Second Opinion Programs

Second opinion programs usually involve a physician separate from the one providing care providing a second view (or cross-check) of a clinical picture — this can be initiated by the second opinion physician or the primary physician.

• Establishing structured approaches (like antibiotic stewardship has) and possibly billing models for second opinion consultations can help integrate this intervention into existing workflows.
• Challenges include recognizing the need for help and potential biases in how cases are presented.
• Using large language models (LLMs) to screen and identify cases for second opinions can help reduce biases and improve the effectiveness of diagnostic cross-checks.
• Often happens naturally at times of handoffs, such as weekend coverage or team switch.

"If you have an open workroom, it works well. The person needs to identify that they need a second opinion, but it's biased by how you tell the story to someone."

Patient Engagement

Conveying diagnostic information, including diagnostic uncertainty, is a key part of the diagnostic process. Tools to help elicit questions about diagnoses from patients as well as frameworks to improve communication about diagnostic processes may be helpful.

• Targeting the most vulnerable patients, such as those with limited English proficiency or dementia, can ensure that patient engagement efforts are both impactful and efficient.
• Timing patient engagement discussions 1-2 days before discharge to review the course and discharge plan can help identify missed diagnoses and improve patient satisfaction.
• Involving nursing staff in reinforcing diagnostic communication can help alleviate time pressures on providers and enhance patient understanding and retention of medical information.

"There's incredible value in finding those cases where the patient or family feels things are not going well or 'something is just not right.'"

Conclusion

These three programs have significant promise as ways to reduce diagnostic errors and improve patient safety, and are worthy of pilot testing. Results from this workshop helped lead the ADEPT team to develop refinements to these approaches which will be part of ongoing implementation work.

Our next HOMERuN meeting will be on September 12, 2025.

Image Attributions: Icon images from https://www.flaticon.com