Disclaimer: This information was collected as part of a research project and is not updated regularly. This information was accurate as of the last updated date at bottom of page.
Resources on All-Payer Claims Data for VA data users:
- All-Payer Claims Data on the VHA Data Portal
- Cyberseminar Requesting and Including State Data, Including All-Payer Claims Databases (APCDs), in VA Studies and Evaluations (1/8/2024)
To understand the full spectrum of care a VA patient receives, studies may look to include state All-Payer Claims Data (APCD), which contain claims from public and private insurance. State health care claims data may contain vital information, as many younger Veterans use private healthcare in addition to their VA-provided care. When combined with data from VA-provided care, VA-purchased care, and CMS care, researchers can have a nearly complete picture of Veteran health care use.
The table below aggregates much of the information about requesting state health care claims data. Questions? Email HERC@va.gov.
| State | Links to Patient-level VA Data |
Notes |
|---|---|---|
| California | Our facility (VA Palo Alto) only qualifies for a limited dataset with no patient identifiers. | |
| Colorado | X | Data is provisioned by the Center for Improving Value in Health Care (CIVHC). CIVHC is already a VA vendor and is willing to work with research teams. Must send a finder file of VA patients to CIVHC to match. The data is expensive and will require a contract. |
| Connecticut | Connecticut does not release name and SSN. | |
| Florida | X | Data is provisioned by the Florida Agency for Health Care Administration. There is no cost for federal agencies. AHCA requires the team to execute a new DUA each year, as the DUA is only good for the length of the IRB approval. Data is locked from use between the IRB expiration date and the new DUA execution date. The DUA process each year takes 2-3 months. |
| Georgia | Georgia does not collect patient-level data, only aggregate data. | |
| Illinois | Illinois does not allow linkages at the person level and the data cannot be linked to other data sources (e.g., you cannot link Illinois data to data from VA, Medicare, and other states). | |
| Louisiana | X | Data is provisioned by the Louisiana Department of Health, Bureau of Health Informatics. The process for approval includes state IRB approval and can be lengthy. |
| Massachusetts | Massachusetts does not collect patient identifiers (name, date of birth, SSN). They offer probabilistic matching but it can take 1-2 years. | |
| Minnesota | Minnesota does not collect patient-level data, only aggregate data. | |
| Missouri | X | Data is provisioned by the Missouri Department of Department of Health and Senior Services. The application process can be lengthy, as it requires state IRB approval. Each year the project must submit an updated application the month prior to your IRB expiration date. |
| New York | X | Data is provisioned by the New York State Department of Health, Statewide Planning and Research Cooperative System (SPARCS). There is no cost for VA studies. |
| Oregon | X | Data is provisioned by the Oregon Health Authority, Office of Health Analytics. There is no cost for VA studies. |
| Pennsylvania | X | Data is provisioned by the Pennsylvania Health Care Cost Containment Council (PHC4). The cost for data depends on the cohort size and dataset characteristics (which files/variables, years, etc.). PHC4 was unwilling to become a registered federal vendor and payment may be challenging. |
| South Carolina | X | Data is provisioned by the South Carolina Revenue and Fiscal Affairs Office. The data fees are relatively inexpensive and the application process takes only a few months. |
| Tennessee | State was unresponsive to inquiry requests. | |
| Utah | X | Data is provisioned by the Utah Department of Heatlh, Office of Health Care Statistics. The office was willing to match using a finder file but the per year cost for each file was very expensive. |
| Virginia | Virginia does not collect patient-level data, only aggregate data. | |
| Washington | Very expensive for each year of data. Because of the data costs posted online, we did not contact the state to find the possibility of matching using a finder file. |
Challenges
While state healthcare claims data can be informative, there are challenges to obtaining and using the data, namely:
- feasibility of linking to VA data, especially at the patient-level
- cost
- time required to complete requests
Feasibility of Linking to VA Data
Many researchers want to link state healthcare claims data to patient-level VA data but this is not always possible. While this is feasible for some states, it is not possible for others (e.g., MA, IL, VA, GA), either because the state does not include patient-level identifiers (such as name, SSN, date of birth) or their data restrictions do not allow linkage to other datasets.
Cost
The cost of state healthcare claims data varies and can be a limiting step for projects. Some states provide data free of charge (e.g., OR, FL, NY) or at low cost (e.g., LA, SC, MO), whereas other states have very high data fees (e.g., CO, PA). The cost of data is dependent on how much data you need (years and types) and whether you want a standard or custom dataset. To purchase state data, the provisioning agency must be a federal vendor (completing the vendorization process at SAM.gov), and payment is through a non-competitive contract directly between VA and the provisioning agency. Some agencies may be unwilling to complete this process.
Time Required to Complete Requests
In addition to the time required to set-up payment, each state has its own request process, which take several months of communication between the project team and state data administration to obtain approvals and set-up the data transfer process. A time-consuming aspect of the request process can be the data use agreement (DUA). Most states require the data team to sign a DUA, and teams should work with their local research privacy officer and information system security officers (ISSOs) to carefully review the DUA before signing. In our experience, there are a few common issues, such as the process by which VA sends a file of VA enrollees to a state for matching and language that binds VA to state law or creates an unfunded liability (often found in an indemnity clause).
Additionally, if the study team plans to send a cohort finder file to the state, it needs to be explicitly stated in your IRB protocol, and a DUA to share VA data with the state will need to be executed (for a draft DUA between a federal and non-federal agency for research, see the VHA Data Portal’s DUA page (VA intranet only) and VHA Handbook 1200.12: Use of Data and Data Repositories in VHA Research).
HERC has requested state data for several projects. Recently, we looked into 18 states; we were unable to get data from 8 due to data linkage or request process barriers (CA, CT, GA, IL, MA, MN, TN, and VA) and 2 were too expensive for the project (UT and WA). We successfully requested data from 8 states (CO, FL, LA, MO, NY, OR, PA, and SC).
If you plan to include state all payer claims data in your study, here are our top 4 tips to improve efficiency:
- Feasibility: Reach out to the state data agency contact to determine whether linking to VA data is feasible.
- Budget: Determine how much data you'll need. Visit the state data website and reach out to the provisioning agency for a cost estimate.
- Compliance: Work with your local research privacy officer and information system security officers (ISSOs) to ensure all DUAs meet VA requirements before you sign.
- Timeline: Plan for a minimum of 9-12 months to complete the request process for each state.
APCD Resources
- Visit the All-Payer Claims Data page on the VHA Data Portal for an overview of APCD and a list of helpful resources.
- Cyberseminar: Requesting and Including State Data, Including All-Payer Claims Databases (APCDs), in VA Studies and Evaluations (January 8, 2024)
