Should VHA "Make or Buy" Elective Coronary Revascularization Procedures?
The VA Patient-Centered Community Care (PC3) program provides Veterans with care that is not available from a nearby VA provider. We are completing an HS&RD supported study of the quality, access, and costs of elective coronary revascularization procedures to assess how VA care compares to care purchased by VA.
The study identified 23,003 elective procedures provided to 20,755 Veterans under age 65 from 2009 to 2011. Coronary artery bypass grafting (CABG) accounted for 28 percent of the procedures and percutaneous coronary intervention (PCI) for 72 percent of the procedures.
Claims from the VA Community Care program contained complete cost, diagnosis, and procedure codes and allowed us to identify elective coronary revascularization procedures obtained from non-VA community providers and their cost to VA. These claims represented 22 percent of the cardiac procedures in our study, with the remaining procedures provided at VA facilities. Diagnoses recorded in the claims data enabled us to identify the risk factors of cohort members. Although the final results of this investigation are not yet ready for release, this article presents several lessons learned from our analysis that may be useful to other VA investigators.
In order to measure ease of access, we needed to know the location of the hospital where the Veterans received care. The zip code in the claims data is the address where remittances were sent, but it is not necessarily the providing hospital's location. Therefore, we obtained a database of hospital addresses from Medicare. We also obtained information on hospital performance from Hospital Compare and the annual volume of cardiac procedures from various surveys.
Each of these sources identifies hospitals using the Medicare hospital identification number. The hospital identifier is available in most VA Community Care program inpatient claims, as it is used to determine reimbursement at standard Medicare rates. When the reimbursement is based on a negotiated contract, or when the care is provided in an outpatient setting, the identifier is not used to determine the payment and thus not found in the claim. We needed a method to find the Medicare hospital identifier for these claims in order to identify the location and characteristics of the hospital. This was especially important for PCI claims; more than half of the Community Care PCIs are provided on an outpatient basis, and without the Medicare hospital identifier, we had no information on hospital location, volume, or performance.
We were able to assign a Medicare identifier to almost every procedure purchased from Community Care providers in our study. We took advantage of the fact that a VA vendor identification number is assigned to every claim. This VA identifier may represent a hospital, a health system, or managed care organization. We studied inpatient claims paid by each VA medical center to find the Medicare hospital identifier associated with the VA vendor number at that VA facility during the year of service. There were some cases where there was more than one hospital Medicare identifier associated with a VA vendor number. When this occurred, we found the correct hospital through a record-byrecord lookup of the name of the hospital in the detailed claims data by Community Care program staff.
Properly accounting for the full costs of the procedures presented additional challenges. Like all healthcare payers, the VHA Community Care program makes separate payment to hospitals and individual physicians. Payments to physicians represent about 20 percent of hospital payments.1 We needed to account for these physician payments, as the cost of physician services is part of the cost determined by the VA Managerial Cost Accounting System. Finding the physician payment for each inpatient stay is as challenging in the Community Care data as it is for Medicare and other claims systems. We found the guide to Community Care from the HSR&D's Health Economics Resource Center (HERC) to be a helpful reference.2 Analytic issues described in this article are more fully discussed in the HERC Bulletin.3
It will be increasingly important for VA to improve access without compromising quality or incurring excessive costs. For example, referring Veterans to the provider closest to their home is not ideal if that provider provides lower quality care than another provider located farther away. One approach to determining the best balance of quality, access, and cost is a cost-effectiveness analysis. Outcomes, expressed in quality-adjusted life years, can be compared to costs that include not only the actual procedural costs but also the travel expenses and time of both Veterans and their caregivers.
We hope that our early experiences with analyzing VA and Community Care data will benefit future research projects, and look forward to disseminating the peerreviewed findings in the near future.
1. Peterson C, et al. "Professional Fee Ratios for US Hospital Discharge Data," Medical Care 2015; 53:10, s 840-9.
2. Gidwani R, Hong J, Murrell S. Fee Basis Data: A Guide for Researchers. Menlo Park, CA. VA Palo Alto, Health Economics Resource Center, November 2015. http://vaww.herc.research.va.gov/include/page.asp?id=guidebook-fee-basis
3. VA Health Economics Bulletin. Health Economics Resource Center, Vol 16, No. 1 and No. 2 http://vaww.herc.research.va.gov/include/page.asp?id=bulletins