Designing Interoperable Clinical Practice Guidelines Ensuring Quality Care

Designing Interoperable Clinical Practice Guidelines Ensuring Quality Care to Veterans by Integrating BPM+, Solor, and ANF into the Integrated Health Practice Strategy August 26, 2020

AGENDA 1 Context 2 Our Approach 3 Work Deliverables 4 Takeaways and Recommendations 5 Q&A 2

Context

TRANSFORM COVID-19 CLINICAL WORKFLOWS WHY HOW WHAT VA is at the forefront of innovation bringing in different specialists, provider consultations, the Director of Informatics Architecture, and direct reports to the KBS Terminology Domain Director to develop COVID-19 workflows. Transforming the 10 COVID-19 Patient Treatment Workflows to a computable format With provider consultation, utilize BPM+, Solor, Analysis Normal Form To standardize procedural knowledge by transforming clinical decision into computable inputs and outputs 4

MOTIVATION The development of COVID-19 clinical workflows presented a prime opportunity to explore frameworks for business process standardization enabling systems and applications to assist in improving Veteran care. Leadership Commitment Robust & Continuous Process Improvement HIGH RELIABILITY ORGANIZATION O ET DG CE LE AN OW RM KN RFO PE D KN ATA OW TO LE DG E Safety Culture Health Problem of Interest PERFORMANCE TO KNOWLEDGE LEARNING HEALTH SYSTEM 5

PROBLEM & OBJECTIVES Clinical Practice Guidelines (CPGs) are not generally scalable, posing a risk of misinterpretation. Stable, standard business architecture and operations provides clinicians with rapid access to reliable information in a standardized and consistent format. THE ISSUE THE NEED Clinical Practice Guidelines are communicated in narrative text, which is generally not scalable and poses the risk of misinterpretation To deliver the best care possible for Veterans, clinicians need rapid access to clear, concise, and reliable information THE SOLUTION Reduce the clinical burden Standardize application of workflows and clinical decision support Create consistency in representing clinical data Improve both quality of health IT systems and care given to veterans 5 6

Our Approach

APPROACH STRATEGY The KBS Informatics Knowledge Architecture provides guidance for the creation of informatics artifacts used to simplify clinical workflows, reduce complexity within clinical data, and improve accuracy of clinical decision support. Informatics Knowledge Architecture Informatics Solutions Business Process Model and Notation (BPMN) – provides a modeling language to standardize how to collaborate and create shareable clinical pathways Decision Model and Notation (DMN) – standard approach for describing and modeling shareable decisions Health Level Seven International (HL 7) Fast Healthcare Interoperability Resource (FHIR) Questionnaire/Response Resource – structured questions and responses to collect clinical data in a FHIR-specified structure Analysis Normal Form (ANF) – defines a normalized data model that promotes simpler clinical data analysis processes Integrated Terminology – harmonizes terminological content in an extensible and controlled ecosystem Collaboration System – enables and supports distributed development through version control, extensions, and modularity capabilities 8

APPROACH IN ACTION Our approach aims to develop VHA informatics artifacts and guidance that could be operationalized to address complexity and implementation roadblocks in a robust manner to effectively roll out clinical workflow products. Approach Methodology: Architecture-Centric, Use Case Driven, Iterative and Incremental Other Standards and Structures Value Sets FHIR Observation Assertional Knowledge BPMN Integrated Terminology FHIR Questionnaire & Response Field Clinicians and Physicians Identify COVID-19 Guidance/Standard of Practice for Enterprise clinical processes Translate COVID-19 clinical workflow Visio diagrams into BPMN artifact(s) Evaluate available clinical questionnaires to capture data for DMN tasks Analysis Normal form Assess terminology required for clinical questionnaires and responses Transform FHIR Questionnaire Responses to ANF FHIR Observation clinical statements DMN Incorporate normalized clinical decision support rules into shareable process models 9

Work Deliverables

BPMN TO STANDARDIZE COVID-19 WORKFLOWS Work products can be found on Git. Hub and Confluence. We used BPMN to standardize the documentation of COVID-19 clinical workflows around screening, testing, and treatment. 11

COVID-19 MEDICAL TRIAGE QUESTIONNAIRE Work products can be found on Git. Hub and Confluence. We standardized the representation of clinical questionnaires for COVID-19 triage workflows by normalizing the questions and answers in existing questionnaires so that they are understandable, efficient and predictable. Questionnaire • • • Exposure Information: Analyzed questions and looked for areas of improvement Did the patient travel outside of the state of their normal residence? Example: What is the patient’s temperature? Did the patient travel on a cruise ship or vessel? Terminology • Snippet of COVID-19 Medical Triage Questionnaire Restricted answers to eliminate ambiguity in codes and concepts Example: Fever measurement versus feels hot/feverish versus feeling feverish Did the patient travel internationally? Was the patient exposed to an animal with susepcted COVID-19? What type of animal were they exposed to? Was the patient exposed in any other way? Symptoms Did the patient experience a measured fever >100. 4 F (38 C)? Did the patient develop pneumonia? Did the patient have acute respiratory distress syndrome? Did the patient have an abnormal chest x-ray? Geographic Location Value Set 373066001 373067005 Yes (qualifier value) No (qualifier value) 129902007 223369002 ……. Region of United States of America (geographic location) Country (geographic location) Yes/ No / Unknown Value Set 373066001 373067005 261665006 ……. . Yes (qualifier value) No (qualifier value) Unknown (qualifier value) Animal Value Set 387961004 ……. . Kingdom Animalia (Organism) Fever Value Set 426000000 103001002 ……. . Fever greater than 100. 4 Fahrenheit (finding) Feeling Feverish Symptoms Value Set 373066001 Yes (qualifier value) 373067005 No (qualifier value) 261665006 Unknown (qualifier value) 233604007 Pneumonia (disorder) 67782005 Acute respiratory distress syndrome (disorder) 168734001 Standard chest X-ray abnormal (finding) ……. . 12

TRANSFORMING HL 7 FHIR QUESTIONNAIRE USING ANF Work products can be found on Git. Hub and Confluence. Our team represented clinical data elements for COVID-19 workflows in a consistent format (utilizing ANF) underpinned by a fully defined terminology model and standard data model (HL 7 FHIR Questionnaire Response). Data is collected via Questionnaires Data is represented as HL 7 FHIR Resources Data is transformed to ANF and Solor enabled FHIR Resources EHR 1 Temp 101. 0° F Temp 38. 3° C Current FHIR Representation code: LOINC [8333 -7] Tympanic membrane temperature. . . body. Site: tympanic membrane: laterality = left 8/12/2020 2: 11 pm Tympanic Thermometer Ear Left Side EHR 2 Temp 101. 0° F Temp 38. 3° C 8/12/2020 2: 11 pm Tympanic Thermometer Left Ear Statements are not comparable code: LOINC [8333 -7] Tympanic membrane temperature. . . body. Site: left tympanic membrane Normalized FHIR Representation category: [Observation procedure] Method: [Examination – action] Has focus: [Tympanic temperature] Procedure site – Direct: [Structure of left tympanic membrane] Using Device: [Tympanic thermometer] z category: [Observation procedure] Method: [Examination – action] Has focus: [Tympanic temperature] Procedure site – Direct: [Structure of left tympanic membrane] Using Device: [Tympanic thermometer] There are many ways to say the same thing within FHIR 13

DMN TO STANDARDIZE DECISION MAKING Work products can be found on Git. Hub and Confluence. DMN’s Rules Table enabled us to represent decisions based on answers to clinical questions identified within BPMN. A B C … Z Speaking in short phrases Evident respiratory distress Using accessory muscles of respiration … Medical Triage Findings F 1 … X Present Absent Unknown Severe Moderate Mild Asymptomatic Error Present - Severe Absent … Asymptomatic Decision Findings from clinical questions within BPMN Categorization of disease severity 14

Takeaways and Recommendations

LESSONS LEARNED AND IMPLICATIONS Lessons learned and key implications for the Integrated Health Practice Alignment address terminology limitations, questionnaire issues, and FHIR challenges. Category Terminology Limitations Questionnaire Issues FHIR Challenges Lessons Learned Implications • Incorrect use of concepts – Feeling Feverish vs Feels hot/feverish vs Fever measures • Inverse concepts – Patient is symptomatic (yes) vs Patient is not symptomatic (no) • Overlapping value sets – race ethnicity codes exist to some extent in SNOMED CT, CDC, and other sources • Inaccuracies in patient data • Compound questions – questions have too much information in one bucket (e. g. , did the patient travel outside their own state) • Recording results without raw measurement – answers are recorded as a heavily precoordinated result (e. g. , Fever) versus the temperature value (e. g. , 103 ° vs 99° Fahrenheit) • Value sets too large to be usable – dropdown answers would require too much scrolling (e. g. , Kingdom Animalia for animal exposure) • Data is not recorded with the right granularity to facilitate precise analysis • Many ways to say the same thing – Lack of consistent guidance regarding which data structure contains specific clinical data values • Inability to aggregate data effectively Recommendations • Guidance is needed for when to use one concept vs another • Changes to terminology standards (harmonization via Solor) • Include clinicians, terminologists, informaticists early in questionnaire design • Restrict value sets to only pertinent concepts • Guidance is needed for when to use one data element vs another • Changes to harmonize FHIR via ANF 16

RECOMMENDATION In order to reduce clinical burden, harmonize clinical data, and standardize clinical workflows, we suggest forming a KBS Tiger Team to assist Integrated Clinical Communities (ICCs) apply a Knowledge Architecture driven approach to the Integrated Health Practice Process. KBS Tiger Team The KBS Tiger Team will assist in the preparation of informatics artifacts and guidance to accompany Health Practice Pattern Parameters: 1. Translate existing artifacts using KBS Informatics Tooling to increase understandability, reproducibility, and usefulness for implementation in Health IT Systems (e. g. , Cerner, Vist. A, other private care provider system) 2. Evaluate existing enterprise clinical procedure in terms of overall compliance with the KBS Informatics Knowledge Architecture interoperability principles 3. Provide guidance and design parameters to increase interoperability and adherence to the knowledge architecture ICCs set enterprise clinical procedure KBS Informatics Knowledge Architecture Enterprise Standard team assembled Readiness Assessment Implementation Planning Compile Parameters for Solution Design 17

Q&A

APPENDIX

AHRQ EVIDENCE-BASED CARE TRANSFORMATION SUPPORT (ACTS) This is an opportunity to collaborate with the ACTS Team as VHA transforms into an HRO and execute an efficient and effective clinical practice guidelines supply chain by combining standards that are fit for purpose at multiple levels. ACTS COVID-19 Guidance to Action Collaborative Agency Collaboration “…to support those working along the COVID-19 'knowledge supply chain' (data-to evidence-to knowledge-to guidance-to action) to improve clinical guidance development and workflow integration for COVID-19 (and beyond) to make the knowledge supply chain more efficient and effective” 1 Fostering collaboration among groups (e. g. the CDC) developing COVID clinical decision support (CDS) knowledge artifacts Guideline Automation Interested in interoperable CDS that could enable, among other things, automated update of guidelines By investing resources, we can create implementation guides in collaboration with ACTS 1. https: //digital. ahrq. gov/acts 20

ANALYSIS NORMAL FORM (ANF) OVERVIEW ANF is an HL 7 clinical data model used to transform clinical data so it can be shared and interpreted consistently across systems thereby facilitating analysis. Helps structure clinical information so that it is consistently represented when exchanged among systems Medication Condition Normalizes a variety of clinical data formats to a common normal form Vital Signs ANF Statement Transforms clinical data so that it can be shared and interpreted consistently across the health care continuum Adheres to the principles of simplicity, consistency, reusability, and reducing variability, so that clinical statement representations become understandable, reproducible, and useful The ANF HL 7 Informative Ballot was successfully published in January 2020 21

SOLOR OVERVIEW An open-source ecosystem that brings together data-encoding standards with an extensible, single model to help health care organizations avoid errors in the interpretation of clinical data. Key Benefits of Solor Saves time and money from maintaining standards by streamlining data integration Enhances product development by focusing on collaboration in an open-source ecosystem Improves patient safety by reducing reliance on errorprone mapping practices Enables interoperability of sharing clinical data across the entire health care ecosystem 22

BPM+ HEALTH OVERVIEW BPM+ Health is a community initiative to improve the quality and consistency of healthcare delivery. Is a cross-discipline group of professional organizations, clinical societies, and healthcare providers working together to develop and pilot standards-driven healthcare process automation techniques to fully realize the benefits of Healthcare IT Achieves improved quality by applying business process modeling standards to clinical best practices, care pathways and workflows directly at the point of care Combines BPMN™ with Case Management Model and Notation (CMMN™) and Decision Model and Notations (DMN™) to disseminate and leverage evidence-based best-practices 23

CONFLUENCE OVERVIEW https: //logica. atlassian. net/wiki/spaces/SOLOR/pages/924680274/Business+Process+Automation+BPA+Implementation+Guide+IG Designing Interoperable Clinical Practice Guidelines • • • Background Mission and Value Statement Objective Operational Readiness – BPM+ – Solor Terminology – Analysis Normal Form – Alignment to Existing Standards and Initiatives (e. g. , HRO, LHS, FHIR) Work Products – Notional Workflow – BPMN – Questionnaire – DMN 24