The 2019 Annual Meeting of the AVAS, held April 27-30, 2019, at the Motif Hotel Seattle, WA, was a resounding success.
The program committee, led by...
2018 Panel Discussion: Nuts and Bolts of VHA Data Warehouse Outcomes Research: What Do I Need to be Successful?
June 12, 2018
The 2018 Annual Meeting of the AVAS was blessed with several enlightening panel discussions. The second panel session on “Nuts and bolts of vha data warehouse outcomes research: what do i need to be successful? “was moderated by Panos Kougias, MD and George Sarosi, MD.
Laura Graham, PhD (HSR&D fellow Center for implementation to Innovation, Palo Alto VA HCS) began the educational discussion by describing the “Medication/pharmacy database”. Dr Graham explained that of the four sources of pharmacy data (Corporate Data Warehouse (CDW), Pharmacy Benefits Management (PBM), Pharmacy Managerial Cost Accounting Ntional Data Extract (MCA NDE) and Non-VA meds) the CDW has the most granular information on prescriptions. She noted that the database structure can be challenging to work with. The CDW has information on inpatient and outpatient medications. The PBM has data on prescriptions dispensed within the VA since 1999. MCA-NDE has pharmacy costs and workload for inpatient and outpatient VHA. Non-VA meds data is entirely reliant on the medication reconciliation process so dates are not exact. Dr. Graham concluded by providing the audience with several data resource links.
Daniel E Hall, MD (Surgeon and Core investigator, Center for health equity research and promotion VA Pittsburgh HCS) next explained his experience and lessons in “Working with local and national VASQIP data for quality improvement”. Dr Hall explained that the VASQIP database has high quality data including outcomes beyond 30 days. He explained that the data contains useful information in several surgical specialties. Dr. Hall outlined the data access process. Dr. Hall explained that if National data is sought for research purposes then the National Data System (NDS) and Data Access Request Tracker (DART) are useful tools.He reminded us that if the purpose is soley for quality improvement we can access our local data through the Surgical quality nurse. If the goal is for the performance of research then an IRB would be required. He defined research as a systematic investigation designed to develop or contribute to generalizable knowlege. To further expand on the various components determining whether a project was purely for QI as opposed to research, Dr. Hall explained the details found in handbook 1058.05. He explained that any learning activity can be published in a peer reviewed journal. If you have identified a problem, clinical champions, cooperation with administrative leadership and collected data for QI purposes then you can always submit an IRB for conducting research later on.
Nader Massarweh, MD, MPH (VA HSR&D Center for Innovations Effectiveness and Safety, Michael E. DeBakey VAMC) finished off the session explaining that “There is no perfect dataset—optimizing your research by linking VA data sources”. Dr. Massarweh explained to the audience that the CDW is the national central repository of VA data (clinical EHR and administrative data). The data warehouse obtains data from multiple VHA sources that then allows one to query for desired subsets (i.e. diabetes, OEF/OIF etc). Dr. Massarweh, proceeded to provide the audience with a detailed description of observational research designs and sources. He explained the important considerations regarding External validity and Internal validity. The audience was provided a detailed synopsis of the potential for bias (the achilles heel of observational research) and two methods to address bias- statistical methodology and study design. Dr. Massarweh explained linking data sources using common variables across different data sources- administrative/operational data, surveys, EHR, Financial, quality improvement, CMS and cancer registry. He then went on to explain several examples of how to put a study together looking at the primary cohort, VASQIP and CDW. Finally, he explained that VA data can support a robust, informative and impactful observational and comparative effectiveness research enterprise. In addition, linkage of VA data sources is a useful and available research technique to provide more robust observational research data.