Using Choice Data in Research
HERC has launched a new series of cyberseminars, “HERC Q&A,” which have an emphasis on audience discussion and practical application. The first HERC Q&A, led by HSR&D Investigators Amy Rosen, Megan Vanneman, and Todd Wagner, focused on using Choice data in research, which is a growing area of interest for those evaluating non-VA care provided through the 2014 Veterans’ Choice Program. Investigations into the Choice data are still in the early stages; the purpose of this cyberseminar was to provide a starting point for understanding how to work with these data. The Q&A opened with a brief discussion of available Choice data sources, followed by a question and answer session with participants.
Drs. Wagner and Vanneman presented an overview of three sources of Choice utilization data: Third Party Administer (TPA) Monthly Choice tables, Fee Basis Claims System (FBCS) Data, and the Program Integrity Tool (PIT). Researchers interested in evaluating the administrative process, such as clinic wait times, can find detailed records in the TPA Monthly Choice tables, which can be found within the VACAA tables. Dr. Wagner cautioned that the VACAA tables are “live,” which means that the data are constantly updated. In addition, the TPA tables do not contain clinical information, such as diagnosis and procedure codes. Researchers looking for clinical data can use the “obligation number” variable to identify Choice care in the Fee Basis or FBCS Data. Drs. Wagner and Vanneman also discussed a new source of Choice data: The Program Integrity Tool (PIT). PIT, which was created to check for fraud, waste, and abuse, provides comprehensive data on all Choice utilization. This data is not yet available in CDW, but interested operations users can request access through ePAS.
In the second half of the cyberseminar, Drs. Rosen, Vanneman, and Wagner answered audience questions about Choice data. Included below is a summary of questions discussed in the Q&A portion of the seminar.
Q: Is there a way to see if patients have been referred but have not yet seen the provider?
A: Researchers can look at referrals in the VACAA tables; however, it is difficult to link referrals to visits, one to one. Researchers should give themselves a reasonable amount of time for the patient to have received care. There is no variable that specifically indicates a patient received care.
Q: Are you aware of any nationally available data to track success metrics?
A: The Office of Community Care has rolled out several tools to aid with care coordination and evaluate quality. These evaluations are in the early stages, so they are just developing the metrics now, and data is not yet available. Visit the Office of Community Care intranet site to learn more about their programs.
Q: Are the Choice data available on CDW raw or production servers?
A: The VACAA tables and the FBCS data are available on the CDW raw server (A06). PIT data is not currently available on CDW.
Q: Can a researcher determine whether a Veteran qualified for Choice due to mileage vs wait time?
A: Determining how a Veteran qualified for Choice can be complicated. In Dr. Vanneman’s project, she used patient level information in the VACAA tables, and linked the VACAA patient level data to the visit data. Investigators should remember that the VACAA tables are on the raw server, so the information is constantly updated. This means that researchers will only see the reason a Veteran is currently eligible for Choice and will not know if the Veteran was originally eligible for a different reason.
See the 2017 paper by Vanneman et al, published in Medical Care, for more information on Dr. Vanneman’s methods: Vanneman ME, Harris AHS, Asch SM, et al. Iraq and Afghanistan Veterans' Use of Veterans Health Administration and Purchased Care Before and After Veterans Choice Program Implementation. Med Care, 2017. 55 (Suppl 7 Suppl 1): S37-S44.
Q: Is Fee Basis data distinct from FBCS data?
Fee Basis (formally known as Purchased Care or SAS Fee) and FBCS are two distinct schemas within CDW, although they have many similarities. Both Fee and FBCS were developed to track VA’s purchased care and contain much of the clinical and payment data typically in administrative claims datasets. The schemas also have many overlapping variables. However, the way information gets into the schemas and how they are updated is different. Fee data are local and entered through VistA. FBCS is a national system, which means that there are over 30 SQL servers aggregated into FBCS on CDW. Fee Basis data are static, meaning that the datasets are updated annually, and data is available for each fiscal year. Fee Basis data live on the CDW production servers. FBCS has live data, which means that the data are updated frequently, and datasets can change. In addition, live datasets, which are available on the CDW raw server, can be difficult to analyze since they have not been cleaned by the CDW team.
Q: Are data dictionaries available for Choice data?
A: VIReC and HERC are in the process of developing data dictionaries. Due to the considerable interest in Choice data, the resource centers are working to develop resources that can be released faster than a traditional guidebook. No data dictionaries exist right now, which is why crowd sourcing and input from other investigators working with Choice data is so important.
If you are working with Choice data, please email email@example.com. This will help the team build a community of colleagues who can share information and grow the Choice data knowledge base.