Adaptation of Practice Guidelines for Clinical Decision Support: A Case Study of Diabetic Foot Care Mor Peleg 1, Dongwen Wang 2, Adriana Fodor 3, Sagi Keren 4 and Eddy Karnieli 3 1 Department of Management Information systems, University of Haifa, Israel; 2 Department of Biomedical Informatics, Columbia University, NY 3 Inst. of Endocrinology, Diabetes & Metabolism, Rambam Medical Center, and RB. Faculty of Medicine, Technion 4 Department of Computer Science, University of Haifa, Israel
What are clinical guidelines? A recommended strategy for management of a medical problem in order to –Improve outcomes –Reduce practice variation –Reduce inappropriate use of resources Computer-interpretable Guidelines can deliver patient-specific advice during encounters GLIF3 is a CIG formalism dev. by InterMed
Guideline Sharing: the GLIF approach Database of CIGs Encoded in GLIF Central Server to Support Browsing and Downloading of CIGs Tools for Encoding CIGs, Validating, & Testing them Internet Local Adaptation of CIG Integration with Local Application (e.g., EPR, order-entry system, Other decision-support system)
Reasons for Local Adaptation/changes Variations among settings due to –Institution type (hospital vs. physician office) –Location (e.g., urban vs. rural) Availability of resources Dissimilarity of patient population (prevalence) Local policies Practice patterns Consideration of EMR schema and data availability
Research purpose Characterize a tool-supported process of encoding guidelines as DSSs that supports local adaptation and EMR integration Identify and classify the types of changes in guideline encoding during a local adaptation process
Methods Guideline: Diabetes foot care –By the American College of Foot and Ankle Surgeons Guideline encoding language: GLIF3 Authoring tool: Protégé-2000 Guideline execution/simulation tool: GLEE EMR: Web-based interface to an Oracle DB Analysis of changes that have been made during the encoding and adaptation process
Guideline encoding and adaptation Narrative Guideline encoding Abstract flowchart in GLIF3 informaticians
GLIF3’ guideline process model (Diabetes) Created using Protégé-2000
Hierarchical model
Guideline encoding and adaptation Narrative Guideline encoding Abstract flowchart in GLIF3 Analysis of Local Practice informaticiansInformatician+ Experts Needed changes+ Concept defs Encoding Revision & Formalization Local CIG Mapped to EMR
Hierarchical model
Computable specification Note the different naming conventions
Guideline encoding and adaptation Narrative Guideline encoding Abstract flowchart in GLIF3 Analysis of Local Practice informaticiansInformatician+ Experts Needed changes+ Concept defs Encoding Revision & Formalization Local CIG Mapped to EMR Manual Validation Validation by Execution of test-cases Iterative changes
GLIF Execution Engine
Validation using GLEE Executed: –14 real patient cases from the EMR –6 simulated cases, which covered all paths through the algorithm The validation considered 22 branching points and recommendations At the end of the validation, all 22 criteria matched with the expected results
Types of changes made Defining concepts –2 of 10 concepts not defined in original GL –6 definitions restated according to available data Adjusting to local setting –GPs don’t give parenteral antibiotics (4 changes) Defining workflow –Two courses of antibiotics may be given (4) Matching with local practice –e.g. EMG should be ordered (4)
The EMR schema & data availability affected encoding of decision criteria Multiple guideline concepts mapped to 1 EMR data item (e.g., abscess & fluctuance) A single guideline concept mapped to multiple EMR data (e.g., “ulcer present”) Guideline concepts were not always available in the EMR schema (restate decision criteria) Unavailable data (e.g., “ulcer present”) Mismatches in data types and normal ranges (e.g., a>3 vs. “a_gt_3.4”)
Summary We suggest a tool-supported process for encoding a narrative guideline as a DSS in a local institution We analyzed changes made throughout this process
Discussion Encoding by informatician was done before consulting clinicians re: localization – Presenting an abstract flowchart to them eases communication –But involving clinicians early saves time Ongoing work: –Integration of the decision support functions within the web-based interface to the EMR –a mapping ontology that would allow encoding the guideline in GLIF through clinical abstractions and mapping to the actual EMR tables
Thanks!
Changes made during encoding Versions Knowledge Item OriginalV1 V2 V3 Decision steps Action steps Decision criteria Data items