Deploying ONC terminology standards in SNOW SHRINE i

Deploying ONC terminology standards in SNOW SHRINE i 2 b 2 data warehouse James R. Campbell MD Scott Campbell Ph. D Jay Pedersen MS Bret Gardner MS James Mc. Clay MD Nebraska Medicine University of Nebraska Medical Center

Operational Expectations of i 2 b 2 load Ø Query by instance or aggregate data sets across collaborator sites MUST occur with little or no mapping Ø Queries MUST run in collaborating datamarts without mapping or revision Ø Store facts so that query by value MUST be supported by datatype: Numeric, Code lists, structured text Ø Narrative reflects content of many EHRs but is not interoperable and relevant elements for research use cases SHOULD be structured when extracted Ø Managing and querying the datamart SHOULD not require SQL features not common to all industry operating systems

Threats to network interoperation of i 2 b 2 data Ø Structure and organization of observation facts(OF) are idiosyncratic between data sites Ø Metadata is dependent upon OF structure and must be identical across datamarts for interoperation of queries, especially aggregation Ø ONC ontologies are variably deployed across i 2 b 2 community and largely not up to date Ø Large ontologies are costly and time consuming to install as i 2 b 2 metadata Ø Change management of ONC ontologies is spotty, difficult to do and inconsistently applied at metadata install

Desiderata for interoperability of OBSERVATION_FACTs In the ONC standard information model, ontologies Ø When possible for observables, CONCEPT_CD should align have uses: reference standards as the question and with differing interoperability • Diagnoses and findingsshould are fully coordinated assertionsorof valuesets of answers be placed in NVAL_NUM fact (ICD*, SNOMED CT clinical findings) TVAL_CHAR Ø • Procedures Pre-coordinate CONCEPT_CD only if andquestion-answer regimes may be into coordinated within the valuesetinformation is small andmodel there to is no standard observable vendor populate orders, billablecode Ø events When coded ontologies are the answer. CT) , use…MODIFIER_CD or results (CPT, ICD*, SNOMED for imposing context of Observables the question are meant to be the • Clinical and laboratory Ø ‘question’ Complex data (allergy list, orders, complex and records are coordinated withmedication an ‘answer’ or result to tests) should be organized into set of facts that are organized coordinate an evaluation finding (LOINC, SNOMED CT) for context by MODIFIER_CD and linked by INSTANCE_NUM • Medications and treatments (RXNORM) may befor results should Ø Choice of TVAL_CHAR, NVAL_NUM or BLOB coordinated withinuse thepublished vendor information model toalways for reflect datatypes; standard valuesets populate orders and prescriptions. NDC are inventory codes interoperability used by pharmacists and nursing to coordinate and record dispense and med administration events.

Desiderata for interoperability of OBSERVATION_FACTs Ø When possible for observables, CONCEPT_CD should align with interoperability reference standards as the question and valuesets of answers should be placed in NVAL_NUM or TVAL_CHAR; metadataxml should support query by value Ø Choice of TVAL_CHAR, NVAL_NUM or BLOB for results should reflect datatypes; use published standard valuesets whenever possible for interoperability Ø Pre-coordinate question-answer into CONCEPT_CD only if valueset is small and there is no standard observable code Ø When coded ontologies are the finding , use…MODIFIER_CD for imposing context from the vendor information model Ø Complex data records (allergy list, medication orders, complex tests) should be organized into set of facts that are organized for context by MODIFIER_CD and linked by INSTANCE_NUM

Nebraska Medicine i 2 b 2 Data Architecture I 2 b 2 Information Class Standard Metadata Ontology ADT history Epic Facility Cancer registry ICD-O Clinical measurements Social history LOINC; SNOMED CT Demographics LOINC; SNOMED CT Diagnoses (Encounter dx; Problems; Past Med History) (ICD-9 -CM); ICD-10 -CM; SNOMED CT Encounters Epic Facility; Encounter classes Laboratory results LOINC Medications (Orders and Rx; Dispense records) RXNORM; NDC Procedures (Professional services; Hospital procedures; Procedure history ) CPT; (ICD-9 -CM); ICD-10 -PCS; HCPCS; SNOMED CT

BD 2 K Grant Summary Ø SA 1: Develop and support ONC compliant metadata tooling for procurement, installation and temporal management of i 2 b 2 datamarts Ø SA 2: Develop, test and deploy terminology extensions for structured reporting, clinical care and research in cancer including anatomic pathology and genomic/molecular data Ø SA 3: Employ structured AP and MP data in managing a research tissue biobank Ø SA 4: Install the ONC compliant metadata (SA 1) and terminology extensions (SA 2) in a research data network and demonstrate interoperability of relevant research queries across multiple cooperating datamarts

BD 2 K Grant proposals Ø SA 1: Collaborate with Harvard and enhance SCILHS metadata and tooling Ø SA 2: Identify one or more sites to collaborate on deploying structured pathology data into i 2 b 2 Ø SA 3: Identify collaborators for expanded tissue biobank indexing and management Ø SA 4: Identify SNOW SHRINE collaborators to deploy ONC-SCILHS metadata and tooling; agree on research questions for one or more demonstrations in research interoperability

UNMC Goals for SNOW SHRINE Ø Support interoperable i 2 b 2 client queries between GPC and collaborating sites Ø Support exposure of full spectrum of site data, NOT just CDMV 3 dataset Ø Support ETLs of data compliant with ONC reference terminologies for all EHRs Ø When relevant, distribute and install comprehensive ONC compliant SCILHS metadata developed in collaboration with Harvard Ø Develop distribution and maintenance architecture for collaborators supporting six month refresh cycle of metadata which can be deployed within a 24 hour install and will support site management of historicity Ø Develop tooling for interoperation of SCILHS metadata and recruit SNOW SHRINE datamarts as collaborators

SNOW SHRINE+ SCILHS 2. 02

SNOW SHRINE+ SCILHS 2. 02(2. 20) Ø Demographics: CDMV 3 only; have not deployed pre-coordinated age folders Ø Diagnosis: ICD-9 -CM and ICD-10 -CM good; SNOMED CT (conditions)not deployed Ø Lab results: 141 M/202. 4 M events; has not deployed many common tests Ø Medication: 40. 6 M/61 M events; have not evaluated modifier structure Ø Vital: CDMV 3 only

Questions? Comments?

ONC Terminology Model for Semantic Interoperability Ø Ø Ø Demographics: LOINC, HL 7/OMB code set Social and medical history: SNOMED CT Problem list: SNOMED CT Encounter and billing diagnoses: ICD-10 -CM Lab results (observables): Lab LOINC Physical findings: LOINC, SNOMED CT observables Medication orders: Rx. NORM, SNOMED CT Medication dispense & administration records: NDC Laboratory orders: LOINC Immunizations: CVX, MVX Procedures: ICD-10 -PCS, CPT, HCPCS Documents: LOINC

EPIC ETLs of Standards Data Ø Problem list – US ed SNOMED CT Ø Past medical history – US ed SNOMED CT Ø Encounter, billing diagnoses – ICD-9 -CM, ICD-10 CM Ø Procedures – CPT, ICD-9 -CM, ICD-10 -PCS Ø Surgical history – US ed SNOMED CT Ø Laboratory results – LOINC Ø Pathology and genomics – LOINC and SNOMED CT Ø Clinical findings/Vital signs/Social history – LOINC, SNOMED CT Ø Medication orders and prescriptions – RXNORM Ø Medication administration/dispense events – NDC Ø (Immunizations – CVX, MVX) Ø (Allergies – SNOMED CT and RXNORM)

I 2 b 2 Load Documentation on the User. Web https: //datahandbook. epic. com/Reports/Details/9000400 Ø UNMC i 2 b 2 ETL Procedures 20160803. docx Ø Identified (idwk) dataset procedures: – Heronloader data extracts and table builds(ETLs) – Blueheronmetadata build Ø De-identified (deid) dataset procedures: – Heronloader extracts and table builds – Blueheronmetadata – CDMV 3 extracts from deid Ø Python scripts: – Fact counter code