Contact email

lzhou@bwh.harvard.edu

Overview

  • This project at an academic health system utilized re-assessment of current rule authoring environments to improve enterprise-level electronic medical record function. 
  • Partners HealthCare is re-assessing the current rule authoring environments and requirements in order to convert medical knowledge into a scalable, comprehensive tool to manage enterprise-level rules.

Department

Department of Medicine

Collaborators

Neelima Karipineni
Janet Lewis
Saverio M Maviglia
Amanda Fairbankds
Tonya Hongsermeier
Blackford Middleton
Roberto A Rocha

Status/Stage of Development

Planning

Funding Sources

Funding was provided internally by Partners HealthCare.

Practice Setting

Academic Medical Center

National/Policy Context

  • One core function of the Electronic Health Record (EHR) is to perform clinical decision support (CDS), which is crucial in serving as a billing and documentation tool in addition to patient care and management. 
  • However, converting medical knowledge into programmable, executable CDS rules has proven complex and difficult.

Local/Organizational Context

  • Because Partners HealthCare is a Boston-wide integrated healthcare system, knowledge sharing and technology play a crucial role. 
  • Currently there are no shared patient data models, knowledge representation formalisms, or computer-interpretable guideline models established to design and implement rule authoring environments (RAE) for the support of CDS function in the EHR.

Patient Population Served and Payor Information

  • The Partners Healthcare health system treats more than ⅓ of hospital patients in the Boston metropolitan area. 

Leadership

  • Li Zhou and Roberto Rocha conceived the study, participated in designing the study, and drafted the manuscript. 

Project Research + Planning

  • The team identified currently active and supported clinical decision support (CDS) rule authoring environments (RAE) used at Partners HealthCare and divided them into two categories:
    • Reminder Rule Editors: these are RAEs for managing clinical reminder rules when managing chronic disease and offering guidance with general health maintenance
      • Examples of these reminders include overdue immunizations, recommended screening tests and lab orders.
    • Medication Rule Editors: These RAEs set guidelines for medication management CDS 
      • Examples include warnings for acetaminophen dosing in liver disease and safety warnings for antidepressants in pregnant patients.

Tech Involved
  • Electronic medical record
Workflow Steps

  • For each RAE the engineering team identified the following information:
    • Software platform on which the RAE was developed
    • Clinical setting for the rules implemented (inpatient vs. ambulatory) 
    • Clinical system in which tool is implemented 
  • The team also conducted informal meetings with Knowledge Engineers familiar with the RAEs to identify strengths and weaknesses of different systems.
  • Evaluation checklists were created for each determined critical success factor. A successful scalable RAE needed to have the following critical success factors: 
    • Formal knowledge representation and standards
    • Metadata support
    • Terminology integration
    • Collaboration support 
    • User interface
    • Integration with EHR systems 
    • Dedicated testing environment 
    • Report-generating capabilities 

Outcomes

  • The following measures for each RAE was evaluated: 
    • Scope of use: 
      • RAEs were found to be diverse and have unique features in their design and workflow process. 
    • Limitations: 
      • Some RAEs were isolated within their inpatient/ambulatory EHR systems and did not interface with a centralized knowledge repository. 
      • RAEs were often not sharable across different EHR systems. 
      • Furthermore even if sharing RAEs was feasible, standardization would be needed in order to express rules efficiently. 
  • After evaluating all RAEs used at Partners with the critical success factor checklists, there is no single RAE that meets the requirements for every critical success factor listed above. 

Future Outcomes

  • Future goals are to develop a scalable integrated rule authoring environment that can support all necessary key requirements and functions such as medical knowledge integration, collaboration, and EHR system embedding. 
  • Other measures to be evaluated include: metadata, terminology, authoring collaboration, user interface, integration with electronic health record systems, testing, and reporting.

Benefits

  • The team was able to identify key success factors and functions needed in order to establish a RAE in a scalable way:
  1. Create Knowledge Specifications
  2. Integrate with Terminology
  3. Author Rules 
  4. Test Rules
  5. Publish Rules
  6. Reporting
  • The following limitations for RAEs in general were identified:
    • Isolation: There is no centralized knowledge repository, leading to redundancies. 
    • Nonsharability: The rules developed for each RAE are specific to sites.
    • Non-standardization: Most rules are expressed using local dictionaries and non-standardized knowledge bases, making duplication and sharing of rules inefficient. 
    • Disjointed: The RAEs are not integrated with a collaboration platform where once a rule is validated and approved, the KE must manually transfer the rule details to the applicable RAE. 
    • Cost: Specialized KEs and SEs are fundamental to developing and maintaining RAEs. However, availabilities of these experts are increasingly scarce. 

Intervention-Specific Challenges

  • One limitation was that the RAEs analyzed are limited to the Partners system. Therefore these findings cannot be generalized to other institutions. 
  • RAEs were also limited to the areas of reminders and medications CDS. As a result, other areas such as order sets, templates and info-buttons were not evaluated. 
  • RAEs are costly to develop and maintain, requiring the abilities of highly skilled KEs and SEs. Hiring specialized KEs and SEs is a major challenge at most institutions.

Glossary

  • Clinical Decision Support (CDS): provides clinicians, staff, and patients with knowledge and person-specific information, filtered or presented at appropriate times, to enhance health and health care.
  • Rule authoring environment (RAE): “suite of tools that manage the end-to-end process of creating specifications for rules, integrating with terminology, and authoring, testing, publishing, and reporting on those rules.” 
  • Knowledge Representation Formalisms: field of artificial intelligence (AI) dedicated to representing information about the world in a form that a computer system can utilize to solve complex tasks such as diagnosing a medical condition or having a dialog in a natural language.

Sources

  • Zhou L, Karipineni N, Lewis J, Maviglia SM, Fairbanks A, Hongsermeier T, Middleton B, Rocha RA. A study of diverse clinical decision support rule authoring environments and requirements for integration. BMC Med Inform Decis Mak. 2012 Nov 12;12:128. doi: 10.1186/1472-6947-12-128. PubMed PMID: 23145874; PubMed Central PMCID: PMC3554596
  • (2018, April 10) Clinical Decision Support. Retrieved from https://www.healthit.gov/topic/safety/clinical-decision-support

Innovators
  • Li Zhou Ph.D., M.D.
Talk to the Innovators