Specialty Lab Informatics and its role in a large academic medical center Zoltan N. Oltvai, M.D. Associate Professor Department of Pathology University of Pittsburgh
Disclosures I have no financial interest, arrangement or affiliation with any organizations that could be perceived as a real or apparent conflict of interest in the context of the subject of this presentation I do not anticipate discussing the unapproved or investigative use of a commercial product/device during this activity or presentation
Specialty Lab Informatics (SLI) Team Information Systems Division UPMC Informatics support for clinical labs that do not fit under the AP and CP umbrella Tim Fennell Manager Zoltan N. Oltvai, M.D. SLI Medical Director Mary Zahorchak Tony Kondisko Julie Hill Geoff Shullo Marco Lukacs Denine Maglicco Dean Brown Analyst Analyst Analyst Analyst Analyst DBA Programmer
SLI-supported clinical laboratories Molecular & Genomic Pathology Lab Hill/ Kondisko Flow Cytometry Lab Brown/ Kondisko HLA/ Tissue Typing Lab Zahorchak/ Lukacs/ Kondisko Hematopoietic Stem Cell Lab Shullo Immuno-Monitoring Lab Shullo/ Brown/ Maglicco Cellular Products Lab Shullo/ Brown/ Maglicco Tissue Procurement Facility Lab Shullo/ Brown/ Maglicco Microbiology Lab (artial support only) Hill
SLI Team Services Installing vendor supplied LIS Custom code for special applications Install/ develop instrument interfaces to LIS Install ADT, Billing, Orders and Result interfaces
SLI Team Services (cont ) Handle daily maintenance issues Long term maintenance (Bug fixes, Work-arounds and Enhancements) Integrated testing with Enterprise level EMR Integrate with external downstream systems
LIS system flow in UPMC MediPac Patient Registrations LIS Cerner Powerchart Orders Results Results DB Motion Results Results Children s Powerchart Orders Epic
Daily Maintenance tasks Correct lab mistakes made during data entry Correct patient registration issues Perform patient merges Perform integrity checks Addition of new physicians to receive results Verify that results are being sent Addition of new Send-out tests Lot number/qc updates
Additional Maintenance Test Events Regression testing Validation testing Miscellaneous troubleshooting
Development 1. a. Installing vendor-supplied software b. Custom code for special applications 2. Integrated software testing with Enterprise level EMR
Vendor-based vs. in-house development Locally Custom Built Solution Pros: No annual support expenses More control over application s future capabilities Develop and implement bug fixes and enhancements on your own timeframe Enhancements can be targeted to exactly what the lab requests Users get closer attention to support issues Easier to generate custom reports No contract administration Cons: Required to keep staff on site to retain domain knowledge Danger of losing domain knowledge and continuity of code during staff turnover of key informatics personnel Need to own the development infrastructure, and development and testing tools Liability issues!
Vendor Based Solution Pros: Expertise/ knowledge of latest trends reside with the vendor Expert level knowledge is not required in-house Vendor handles bug fixes/enhancements Do not have to pay for in-house expert level support No development startup costs, development and testing tools Cons: Expense of annual support Do not have direct control over the application s future capabilities Can request, but not guaranteed of enhancements Longer wait for enhancements and level of compliance with request (large vendors can take 12 to 18 months to turn out an enhancement) Longer wait for bug fixes (average time for Cerner 4 to 12 months) Some bug fixes get pushed out to next major release Longer attention to support issues Have to create and monitor contracts
Installing vendor-supplied software list of recently introduced products Vendor StemSoft Cerner SystemLink Product StemLab Helix HistoTrak User Multiple labs Micro-, MGP labs HLA lab
Custom coding for special applications Software build FDRSS IMbase Various custom codes User Flow cytometry lab Immuno monitoring Molecular lab
Steps Developing and Installing New LIS Understand current workflow in the lab and represent it as workflow charts Decide if existing (legacy) data to be converted to new system (depends in part on the capabilities of the new system) Design and implement data conversion (can be large endeavor depends on cleanness of data)
Design New Implementation Design Interfaces - Registration, Orders, Results, Billing Design in-lab worksheets/ calculations Design how results are to be presented Design both in-lab and published reports
Implementation Steps Build out (i.e., implement) individual entities for each test in new system (e.g., for Helix) Create: Worksheets for Lab use Custom Reports used in-lab and published to customers Instrument interfaces Registration, Orders, Billing, and Result HL7 interfaces Perform legacy data conversion (can be very resource intensive!)
Additional steps to be made Create tests dictionary Create tray dictionary Create instrument interfaces Define Security Roles
HLA/ Tissue Typing Laboratory Management Software System Reasons why converting to HistoTrac: Bugs in the Cerner PathNet HLA system forcing several work-arounds by the lab PathNet HLA module was outdated and not keeping up with technology advancements in the HLA testing area Not receiving any updates from Cerner, bug fixes or enhancements for the PathNet HLA LIS No future plans to enhance the Cerner PathNet HLA Positive experience at external pretesting of product
First step of implementation of HistoTrak: creating global settings for whole application
Create Tests Dictionary
Create Tray Dictionary (items in dictionary) Tray codes Reagents Dilution codes Well characteristics
Example: Defining Tray Layout
Example: Defining QCs and Dilutions Blue = QC wells. Red = dilutions
Example: Crossmatch Scoring Tray (in an actual run)
Define Security Roles Define roles (administrator, supervisor, technologist) and access for role
In-house built custom software Software build SeqReporter Dr. Somak Roy User NextGen Sequencing in Molecular & Genomic Pathology Lab
Clinical nextgen sequencing (NGS) > explosive growth during last 3 years Multigene panels Whole exome Whole genome Ion Torrent PGM Chip sizes: 1.2, 6.2 & 11.1 million wells Illumina MiSeq Illumina HiSeq 2500 Ion Torrent Proton Chip sizes: 165 & 660 million wells
Bioinformatics workflow of NGS data Sequencing reaction Signal processing Pipeline #1 (Torrent suite) Pipeline #2 (NextGENe) Alignment FASTQ / BAM Alignment QC QC Variant caller Variant caller VCF VCF
Physician request NGS test order in LIS by case pathologist Transmission to EMR Receipt of test order and sample/testing material QM/QC Signout in LIS Target marking and % tumor load assessment Microdissection and DNA extraction Variant knowledgebase management Report generation Variant review and interpretation by pathologist Library preparation & QC Sequencing reaction NGS test clinical workflow cycle Signal processing and base calling Variant calling Adapter trimming, Alignment and mapping QC
Test request & sample acquisition Data workflow Nucleic acid extraction Library prep Sequencing reaction WGS 1 o and 2 o bioinformatics processing WES Targeted Seq Tighter bottleneck Variant interpretation Reporting results in LIS Transmission to EMR Bottleneck in clinical informatics and data repository domains
Aim of SeqReporter custom software 1. Create middleware between NGS analysis systems and existing LIS with minimal human input 2. Develop a comprehensive and highly customized knowledge base and streamline variant classification 3. Minimize human errors arising from redundant data entry and manual report synthesis during lab workflow 4. Automate clinical report synthesis by incorporating multiple annotations and appropriate clinical comments 5. Develop a database management system for managing clinical test information and results to establish ongoing QM/QC practices and improve overall laboratory workflow. Roy et al. J Mol. Diagn., 2014.
Roy et al. J. Mol. Diagn., 2014. Input VCF file and raw coverage information Sample and Run information Historic report lookup, QM/QC report SeqReporter Algorithm Sample and run information SIR module #1 VC classifier module #2 In-house knowledgebase Report levels 1-5 VC Mgmt module #3 External variant database Preliminary report for Pathologist review Report synthesis module #4 Signoff Knowledge base training module #5 LIS compatible report
App Implementation Lab production server institution s central production server SeqReporter App and DB server
Enhanced variant display << Level 1 << Level 2 << Level 3 << Level 4 << Level 5
Report synthesis module
All variant data is instantly available for QA/QC review Variant trend characteristic of a given variant
Common implementation concerns Understand the laboratory s detailed needs upfront Be flexible but protect against Scope Creep Build first, Enhance later (Don t get caught up in attempting to create the perfect product all at once) Make sure install team communicate well with outside teams, EHR, Interface teams, Results teams and Laboratory staff
Recommendations In-house complete software development is an excellent solution for clinical labs when concepts develop and instrument platforms change rapidly For more mature laboratories implementation of carefully selected and extensively pretested commercial products is the optimal solution Beside software development tasks, there is a need for a Specialty Lab Informatics-like team at all major medical centers to provide expertise, routine maintenance, small scale programming etc., for disparate lab needs
Acknowledgments - Tim Fennell - Somak Roy
Report design NGS report is very complex (numerous data fields) Balance content and readability Input from pathologists and clinicians Preserve report structure in EMR Appropriate font format Elimination of special characters Pseudo-tables (fixed width font and spaces) Appropriate background texts and disclaimers Variant nomenclature: HGVS recommendations