Healthcare providers face the challenge of growing complexity in medical science and technology, and are inundated with information about how to produce good outcomes in patient care. Clinical practice guidelines (CPGs) define evidence-based practices for heart failure management that reduce morbidity and mortality. Health technology in the VA is well-poised to support patient-specific evidence-based recommendations displayed at the point of clinical decision making. Such systems require a computable format of clinical knowledge.
The objective of this project was to build on previous work on the ATHENA-Clinical Decision Support (CDS) system to develop a prototype computable knowledge base (KB) of recommendations for managing heart failure (HF). The aims were to develop a guideline specification for computer representation of key steps in HF management; to encode the specification in a KB, ATHENA-HF; and to integrate other guideline KBs for commonly co-occurring conditions.
VA QUERI HF experts selected the ACC/AHA guidelines for HF as the CPG to encode. In collaboration with the VA Chronic Heart Failure (CHF) QUERI, knowledge engineers encoded the recommendations using the open-source knowledge acquisition system Protege with the EON model, a model designed to encode CPGs. Clinical concepts were extracted from each recommendation and represented as classes in Protege, and the source for patient data for them were identified. Subject matter experts clarified sections of the guideline that were ambiguous and added information necessary to make the recommendations actionable. Accuracy of encoding was evaluated using the testing environment in Protege. We explored methods to run other previously encoded ATHENA-KBs for patients with co-morbidity.
We encoded 53 recommendations for Stages A, B, and C of HF. Stage D (refractory HF) was defined as out of scope. These recommendations included 123 clinical concepts. The data required for the majority of concepts is available as structured data elements in the VistA electronic medical record (e.g., medications) while a few would require either natural language processing of free-text data (e.g., ejection fraction) or user input at the time of the patient visit (e.g., current symptoms). Preliminary testing with 28 test cases confirmed all recommendations were generated for appropriate cases. We successfully loaded and ran additional guidelines with ATHENA-HF providing recommendations across multiple conditions for patients with these comorbidities: diabetes (DM) glycemic control, foot care, and eye care; hypertension; hyperlipidemia; and chronic kidney disease. The system first evaluates patient's eligibility for each guideline and then generates recommendations for managing each condition.
The development of computable clinical best practices can potentially improve quality of care and outcomes for veterans, in this case Veterans with HF, by providing evidence-based patient specific recommendations at the time of clinical decision-making. This computable knowledge can be used in different settings such as primary care or heart failure teams as well as in quality improvement initiatives. Recommendations can be made for multiple diseases concurrently.
- Tsai V, Goldstein MK, Hsia HH, Wang Y, Curtis J, Heidenreich PA, National Cardiovascular Data Registry. Age Differences in Primary Prevention Implantable Cardioverter-Defibrillator Use in U.S. Individuals. Journal of the American Geriatrics Society. 2011 Sep 1; 59(9):1589-1595.
- Tsai V, Goldstein MK, Hsia HH, Wang Y, Curtis J, Heidenreich PA, National Cardiovascular Data's ICD Registry. Influence of Age on Perioperative Complications Among Patients Undergoing Implantable Cardioverter-Defibrillators for Primary Prevention in the United States. Circulation. Cardiovascular quality and outcomes. 2011 Sep 1; 4(5):549-556.
- Goldstein MK. ATHENA Clinical Decision Support System: Aspects of Design and Implementation. [Cyberseminar]. 2010 Feb 18.
- Goldstein MK, Martins SB, Tu SW, Heidenreich PA, Massie BM, Furmaga E. Developing Heart Failure (HF) Clinical Decision Support for Patient with Comorbidities. Poster session presented at: AcademyHealth Annual Research Meeting; 2011 Jun 12; Seattle, WA.
- Goldstein MK, Martins SB, Tu SW, Heidenreich PA, Massie BM, Furmaga E. Developing Heart Failure (HF) Clinical Decision Support for Patient with Comorbidities. Poster session presented at: American Geriatrics Society Annual Meeting; 2011 May 13; Washington, DC.
- Goldstein MK. From Evidence to Patient Care: Decision Support with Actionable Guidelines. Paper presented at: American Medical Informatics Association Annual Symposium; 2009 Nov 15; San Francisco, CA.
- Goldstein MK. Health Informatics Tutorial: Knowledge-Based Decision Support Systems for Implementing Clinical Practice Guidelines. Paper presented at: American Medical Informatics Association Annual Symposium; 2009 Nov 14; San Francisco, CA.
Clinical practice guidelines, Comorbidity, Decision support