Risk factors for advanced disease in colorectal cancer.

Abstract

The goal of this study was to identify predictors of presenting with late-stage colorectal cancer with a focus on potentially modifiable factors. This was a multicenter, case-based study of patients with colorectal cancer. Detailed information about the cancer was abstracted from the tumor registries, pathology reports, and medical records. The remaining information was obtained by telephone interview. Inclusion criteria were age 40-85 yr with a first diagnosis of histologically proven colorectal cancer between July 1, 1997 and January 1, 2001. Simple contingency table methods were used to examine the relationship between potential risk factors for early versus advanced-stage disease. Logistic regression was performed to simultaneously control for potential confounding factors. There was complete information for 549 respondents. Approximately, 43% of the sample presented with late-stage colorectal cancer. In univariate analysis, lacking a usual source of health care (doctor's office or clinic), no participation in any colorectal cancer screening test in the prior 10 yr, symptoms of blood in stool, and unexplained weight loss were associated with late-stage colorectal cancer. In the logistic regression model, only lacking a usual source of healthcare and unexplained weight loss were associated with late-stage colorectal cancer with odds ratios (95% confidence intervals) of 0.4 (0.2-0.6) and 1.9 (1.2-3.0), respectively. These results suggest that system changes in the VA health-care system that increase access to and improve utilization of primary care may reduce presentation with late-stage colorectal cancer and thus, reduce mortality from colorectal cancer in veterans.