Project Background. As part of its transformation aimed at quality improvement, the Department of Veterans Affairs (VA) began to measure and report quality-of-care performance in 1995. One of the VA performance measurements is adherence to colorectal cancer (CRC) screening, an effective and cost-effective strategy for reducing CRC deaths. Despite substantial progress made in increasing CRC screening adherence, how to appropriately measure CRC screening adherence and how to apply the measurement for evaluating innovations designed to improve CRC screening adherence remain challenging. This study represents a major new application of how VA administrative data can be used efficiently to construct veteran-level, longitudinal data measuring VA system-wide performance of CRC screening adherence. Furthermore, the investigators will use the longitudinal data to examine the effects of VA structural characteristics on CRC screening adherence, two of which will focus on access and implementation research, the priorities targeted by the Health Services Research and Development Service. Project Objectives. This project has two objectives. The first is to construct longitudinal data measuring VA- wide performance in CRC screening adherence by average-risk veterans, representing a major application of efficient use of existing VA administrative data for performance measurement. The second is to examine the effects of VA structural characteristics on CRC screening adherence. Specifically, we are interested in the following three characteristics: (1) improved access to primary care via community-based outpatient clinics (CBOCs), (2) the implementation of the CRC Oncology Watch (OncWatch) intervention within the Veterans Integrated Service Network 7, a new information technology system designed to improve CRC screening adherence, and (3) the dual-use system in which veterans receive health care from both the VA and Medicare. Project Methods. This study will use VA Medical SAS, Fee Basis, and VA-Medicare data to construct longitudinal data to measure veteran-level CRC screening adherence over a 10-year period from 2001-2010. In this study, the investigators will describe trends in adherence to CRC screening over time and by individual veteran and facility characteristics. Using the newly constructed data, they will apply longitudinal data analysis techniques (e.g., hierarchical models, natural experiment) to minimize biases arising from unobserved heterogeneity in multivariable regression analyses. The dependent variable used in the regression analyses is the status of CRC screening adherence of an average-risk veteran in each of the years from 2001-2010 for which the veteran is determined to be eligible for CRC screening. Three key independent variables, corresponding to the three VA structural characteristics, are access to CBOCs, implementation of the CRC OncWatch intervention, and dual-user of VA and Medicare services. Standard errors will be adjusted via Huber clustered standard errors correction for correlated error terms at the veteran or facility level. 1
Taking advantage of high-quality VA administrative data, this study will construct VA-wide, veteran-level, longitudinal data measuring CRC screening adherence that may be used as a potential alternative to the current chart review-based measurement. Moreover, using the newly constructed longitudinal data, this study will assess the effects of VA structural characteristics on performance of CRC screening adherence. The understanding gained from these empirical analyses may assist the VA in designing better interventions to improve CRC screening adherence by millions of veterans in the future. 1
