BACKGROUND/RATIONALE:
Many Veterans receiving medical care in VA have multiple coexisting chronic conditions; yet, performance measures and clinical decision support (CDS) typically provide recommendations about only one disease at a time. Veteran populations are aging nation-wide, increasing the prevalence of complex comorbidity. New technologies are needed that account for clinical complexity.

OBJECTIVE(S):
The overall objective of this project is to develop new informatics methods to automate quality improvement through CDS for care of patients with complex clinical scenarios. The Specific Aims are (1) To create automated knowledge bases (KBs) with clinical detail about evidence-based recommendations; (2) To conduct stakeholder interviews with key VA program offices and quality managers regarding prioritization and coordination of recommendations for patients who have multiple comorbidities; (3) To elaborate KBs with clinical knowledge about recommendations in the context of comorbidities; (4) To develop systems to provide CDS across multiple chronic conditions to health professionals in the Patient-Aligned Care Teams (PACTs) to improve performance.

METHODS:
The project used informatics development methods to encode the clinical KBs, building on prior CDS work and considering comorbid conditions in analyzing recommended next steps in clinical management. Stakeholder interviews with key staff in quality management and in several domains of clinical care were synthesized to guide prioritization and coordination of recommendations. The VISN21 Pharmacy Benefits Management (PBM) clinical dashboard served as the model for the test environment for providing CDS to PACT health professionals providing patient-centric care for Veterans with complex conditions. Agile development procedures were employed based on frequent interaction with developers, using input from potential end-users, in iterative design and development cycles. Informatics tools developed in this project have been and are continuing to be evaluated by constraint verification, testing of accuracy of recommendations, assessment of suitability to clinical workflow, and stakeholder input.

FINDINGS/RESULTS:
The project team developed or updated the KBs for 5 clinical domains (diabetes, hyperlipidemia, hypertension, chronic kidney disease, and heart failure). They developed functionality for the CDS to process guidelines for all five KBs sequentially for the same patient.

We created wireframes of the graphical user interface (GUI) incorporating end-user recommendations. We enabled the display of a table of sorted drug recommendations, where each drug recommendations is also shown with its associated indications and contraindications.

We conducted stakeholder meetings to determine how best to design the system for the needs of the PACT team, which PACT team members will be using the system most, what features in the CDS would be most useful for PACT team members, and priority areas for VHA Office of Analytics and Business Intelligence (OABI). We found that it would not be sensible to formulate different recommendations by clinical role (for example, medical assistant vs nurse) because the actual work done by different individuals varies considerably across clinics, and because who uses the dashboard, by role, also varies.

The existing VISN 21 Dashboard already automated many of the VA performance measures in three of our projects clinical domains: hypertension (HTN), hyperlipidemia (HL), and diabetes (DM). We designed the system to trigger CDS for cases in which the patient data indicates that the performance measure has not been met. Since the PACT dashboard currently does not have performance measures for heart failure (HF) and chronic kidney disease (CKD), we designed the system to evaluate HF and CKD if one or more of HTN, DM, or HL were triggered.

We implemented the system in a test environment that mimics the data structures of the VISN 21 PBM clinical dashboard. Project plans called for development in such a test environment, to prepare for future potential use in the VISN 21 clinical dashboard. Near the end of the project, as a test of capability to run in the clinical dashboard, we provided the CDS system to staff to test if it would run in the VISN 21 dashboard's own test environment. The system ran and generated output, that is, from a system architecture viewpoint, the system is ready to take to a next project that will run within the actual VISN dashboard.

IMPACT:
This informatics research project developed CDS tools designed for presentation in the context of a performance measurement dashboard to aid in the management of five chronic conditions: hypertension, hyperlipidemia, chronic kidney disease, heart disease, and diabetes, taking account of comorbidity. The project developed informatics insights regarding the differences in design for CDS as compared with performance measurements, and has developed some extensions of the framework for steps in moving from guideline knowledge to implementation.

---

External Links for this Project

NIH Reporter

Grant Number: I01HX000620-01
Link: https://reporter.nih.gov/project-details/8196140

---

PUBLICATIONS:

Journal Articles

1. Tso GJ, Tu SW, Oshiro C, Martins S, Ashcraft M, Yuen KW, Wang D, Robinson A, Heidenreich PA, Goldstein MK. Automating Guidelines for Clinical Decision Support: Knowledge Engineering and Implementation. AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2017 Feb 10; 2016:1189-1198. [view]
2. Tso GJ, Tu SW, Musen MA, Goldstein MK. High-Risk Drug-Drug Interactions Between Clinical Practice Guidelines for Management of Chronic Conditions. AMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science. 2017 Jul 26; 2017:531-539. [view]

Conference Presentations

1. Goldstein MK, Corley AM, Martins SB, Tu SW, Furman AE, Oshiro CM. Stakeholder Input to Clinical Decision Support (CDS) for Complex Chronic Disease. Poster session presented at: Society for Medical Decision Making Annual Meeting; 2012 Oct 19; Phoenix, AZ. [view]

DRA: Health Systems, Diabetes and Other Endocrine Disorders, Other Conditions
DRE: Diagnosis, Treatment - Observational
Keywords: none
MeSH Terms: none