: Researchers in colorectal cancer care have focused on hospital and physician volume to explain variation in quality of care. Understanding how other hospital and physician factors influence the process of care and patient outcomes is paramount to improving the quality of the health care delivery system. The primary objectives of this study are (1) to examine a comprehensive set of hospital and physician factors as potential determinants of variation in outcomes for patients with colorectal cancer in the United States, and (2) to examine the extent to which processes of care for these patients mediate the relationship between hospital and physician factors and outcomes, adjusting for market factors as well as patient socio-demographic and clinical characteristics. A secondary objective is to quantify hospital- and physician-level variability using appropriate statistical methods for clustered data. A population-based cohort study design will be employed using incident colorectal cancer cases diagnosed between 1995 and 1999, identified from the Surveillance Epidemiology and End Results (SEER) registry, and prospectively followed through 2001 using linked Medicare claims. Hospital, physician, and market factors will be included through linkage with the National Cancer Institute's Hospital File, American Medical Association data, and the Area Resource File, respectively. Patients who received primary resection will be followed for receipt of adjuvant therapy and other processes related to coordination of care as well as for morbidity and mortality outcomes. Considering colon and rectal cancers separately, bivariate analyses will be used to explore the association between hospital and physician factors and outcomes and to build multivariable models adjusting for market factors as well as patient sociodemographic and clinical characteristics. The mediating effect of processes will also be examined using multivariable analyses. Hospital- and physician-level variability in outcomes will be quantified.
