Transdisciplinary Learning Lab To Eliminate Patient Harm and Reduce Waste

Principal Investigator: Adam Sapirstein, M.D., Johns Hopkins University, Baltimore, MD
AHRQ Grant No.: HS23553
Project Period: 09/30/14–03/29/19

Description: The goal of the Johns Hopkins Armstrong Institute Learning Lab was to use systems engineering methods to partner with patients, patients’ families, and others to eliminate preventable harm, optimize patient outcomes and experience, and reduce waste in healthcare.

The specific aims were to:

1. Develop high-level design requirements for an ideal intensive care unit (ICU), using design thinking and systems engineering methods.
2. Leverage open-application programming interfaces to engineer interoperability between electronic health records (EHRs) and infusion pumps.
3. Develop and implement an indicator of unit-level stress in an engineered care system to predict and mitigate risk.

The team created a graphic roadmap for using the quality function deployment process to share with organizations that want to become high-reliability hospitals.^1,2 The roadmap encompasses four phases of work:

1. Eliminating a single harm.
2. Developing an interactive situational awareness platform to eliminate multiple harms.
3. Focusing on methods, visualization, and control systems; and
4. Building what was envisioned.

The team automated and validated the nurse-managed insulin infusion protocol at Johns Hopkins to illustrate the potential of optimizing safety, efficiency, and workload via system interoperability.^2-4 The Smart Agent system dynamically controls insulin infusions to reduce the time and workload burden on clinical staff, improves the accuracy and efficiency of the protocol, and prevents calculation errors that can lead to patient harm.

Nurses found the Smart Agent system more efficient, safe, and usable than the current standard-of-care insulin infusion process. They found the new system less demanding in terms of workload. They also found the Smart Agent as trustworthy as the standard process. Researchers believe this approach can be replicated for a number of medications.

The research team conducted the first application of statistical and machine learning (ML) techniques to develop a prediction model for the susceptibility of ICU patients to preventable harms. Of five prediction models (four ML and one logistic regression [baseline]), three of the ML algorithms outperformed the baseline model.² The two best-performing models had prediction accuracies for the composite harm outcome of 88 percent and 86 percent.⁵ These highly encouraging results suggest that ML may be useful for predicting common ICU harms.

This PSLL's work has resulted in at least six peer-reviewed journal publications, as well as posters and presentations at conferences across the United States.

References

1. Matthews S, et al. Prioritizing healthcare solutions using the quality function deployment process. Crit Care Med 2019. 47(1):661. https://cdn.journals.lww.com/ccmjournal/Citation/2019/01001/1369__PRIORITIZING_HEALTHCARE_SOLUTIONS_USING_THE.1324.aspx. Accessed June 12, 2020.
2. Sapirstein A. Final Report: Transdisciplinary Learning Lab To Eliminate Patient Harm and Reduce Waste. Baltimore, MD: Johns Hopkins Medicine; 2019.
3. Griffiths SM, et al. Automated, web-based solution for bidirectional EHR-infusion pump communication. Biomed Instrum Technol 2019;53(1):30-7.
4. Barasch N, et al. Automation and interoperability of a nurse-managed insulin infusion protocol as a model to improve safety and efficiency in the delivery of high-alert medications. J Patient Saf Risk Manage 2019 Dec 10. https://journals.sagepub.com/doi/abs/10.1177/2516043519893228. Accessed June 12, 2020.
5. AFYA. Internal Communication. PSLL CONNECT Newsletter: Issue 6. AHRQ, Rockville, MD, 2019.

Grant Publications

All the links below were accessed June 12, 2020.

2019

• Barasch, N., et al., Automation and interoperability of a nurse-managed insulin infusion protocol as a model to improve safety and efficiency in the delivery of high-alert medications. J Patient Saf Risk Manage 2019 Dec 10.
• Matthews S, et al. Prioritizing healthcare solutions using the quality function deployment process. Crit Care Med 2019;47(1):661.
• Griffiths SM, et al, Automated, web-based solution for bidirectional EHR infusion pump communication. Biomed Instrum Technol 2019;53(1):30-7.

2018

• Day RM, et al. Operating management system for high reliability: Leadership, accountability, learning and innovation in healthcare. J Patient Saf Risk Manage 2018;23(4):155-66. https://journals.sagepub.com/doi/abs/10.1177/2516043518790720. Accessed June 12, 2020.
• Romig M, et al. Developing a comprehensive model of intensive care unit processes: concept of operations. J Patient Saf 2018;14(4):187-92.
• Rosen MA, et al. Sensor-based measurement of critical care nursing workload: unobtrusive measures of nursing activity complement traditional task and patient level indicators of workload to predict perceived exertion. PLoS One 2018;13(10):e0204819.