BTRIS The NIH Biomedical Translational Research Information System

BTRIS: The NIH Biomedical Translational Research Information System James J. Cimino Chief, Laboratory for Informatics Development NIH Clinical Center

National Institutes of Health Clinical Center In-patient beds - 234 Day hospital and out-patient facilities Active protocols - 1800 Terminated protocols - 7100 Clinical researchers - 4700 All patients are on a protocol

Clinical Data at NIH Institute System EHR Lab System Personal “System”

Clinical Data at NIH Institute System EHR Lab System Personal System

Clinical Data at NIH Institute System EHR Lab System BTRIS Personal System

Biomedical Translational Research Information System (BTRIS) Database Data Standards (RED) Preferences Data Access Security

Architecture Overview • • • Generic data model in SQL-Server (Microsoft) Data acquisition: HL 7, ODBC, batch Standard extraction-translation-loading Encoding with Research Entities Dictionary (RED) Terminology Development Editor (Apelon) National Cancer Institute extensions to TDE Cognos (IBM) business intelligence tool Javascript extensions to Cognos Home-grown system for user data entry

NIAAA CRIS, MIS 33 NIAID

BTRIS – Two Applications

BTRIS – Two Applications

BTRIS – Two Applications BTRIS Data Access

What is in BTRIS? • Clinical Center MIS (1976 -2004) and CRIS (2004 -) • Demographics • Vital signs • Laboratory results • Medications (orders and administration) • Problems and diagnoses • Reports (admission, progress, discharge, radiology, cardiology, PFTs) • National Institute of Allergy and Infectious Disease • Medication lists • Laboratory results • Problems • National Institute of Alcohol Abuse and Alcoholism • Clinical assessments

BTRIS Data Growth M i l l i o n s o f R o w s

BTRIS Data Access • • • Reports IRB Inclusion CBC Panel Chem 20 Microbiology Demographics Individual Lab Panels Medications Vital Signs Diagnoses/Problems • • • Lists Individual Lab Test Lab Panels Medications Subjects Vital Signs

33 years of Data

BTRIS Reports per Week

BTRIS Users and Subjects 115 BTRIS Users thru March 2010 619 Unique Protocols + 130 Non. BTRIS PIs = 245 BTRIS Beneficiaries 80, 073 Attributed Subjects (of 395, 005 attributions, or 20. 27%)

Subject-Protocol Attributions • 395, 005 total attributions • 126, 533 verified by Medical Records • 44, 142 verified by IC systems • 1, 966 verified by users • 363 unverified subjects “not on protocol” • 236 verified subjects “not on protocol”

Re-using Data in De-Identified Form • Look for unexpected correlations • Pose hypothetical research questions • Determine potential subject sample sizes • Find potential collaborators

Access to De-identified Data • De-identified data available to NIH intramural research community • NIH researchers wanted access policy to ensure protection of intellectual property and first rights to publication • Resolved through three means: – Association of data with an NIH PI – Status of protocol – Age of data

Access to De-identified (Coded) Data b) Terminated Protocol – PI Gone a) Data Outside Any Protocol Period c) Terminated Protocol – PI at NIH d) Active Protocol

Data Available for De-Identified Reports Total Subjects: 430, 196 Attributed to Protocol: 181, 068 Terminated > 5 yrs: 36, 467 Not attributed to any protocol: 249, 128

Data Available for De-Identified Reports Available Subjects – 285, 595 (66. 4%)

OHSR Exemption Process • Required for all de-identified data queries • Automated process replaces OHSR “Form 1” paper process for exemption

Serum Albumin Trends

Using BTRIS For Clinical Research Identify Potential Subjects Identify Potential Controls Obtain Clinical Data Potential Subject Cases Potential Control Cases Include Cases with Pathology Specimens Subject Cases Control Cases Assign Case Numbers Specimens Obtained from Pathology Department Send Case Numbers and MRNs to Pathology Deidentify Cases Deidentified Subject Cases with Phenomic and Genomic Data SNPs Sequenced

Re-using BTRIS For Clinical Research Investigator Office of Human Subjects Research Trusted Broker Develop Deidentified Query Perform Query in Identified Form Obtain Clinical Data Deidentified Subject Data Merging Records De-identified Text Reports and Other Data Identified Text Reports Manual Scrubbing De-identified Text Reports

Capabilities of Data Model • Storage of like data in meaningful model • Preservation of original data details • Can “promote” commonalities to main table • Preservation of original meanings • Queries based on users’ aggregations

Informatics Challenges • • • Understanding data sources Finding the right balance for unified data model Modeling in the Research Entities Dictionary Organizing the Research Entities Dictionary Understanding researchers’ information needs User interface (including Cognos customization) Keeping up with report requests Integration into multiple research workflows Access to deidentified data New policies on contribution and use

So What? • • • Easier access to protocol data from EHR Easier access to archived data Protocol data integrated from multiple sources User empowerment Concept-based queries Data feeds to institute systems Data model flexible but not too flexible Rapid development timeline (under budget) User adoption can be considered good High user satisfaction Success with NIH policy Success with data sharing

Future Directions • Finish historical data • Add more institutes and centers

NIAAA Other CC Sources Radiology Images CRIS, MIS N C I NINDS NID DK NHL BI N HG RI N R 33 NIAID

Future Directions • • Finish historical data Add more institutes and centers Images “-omic” data Specimen identification and location New reports and analytic tools Clinical Trials. gov reporting Beyond NIH

btris. nih. gov