Background: Information on use of colorectal cancer tests, particularly for the purpose of population surveillance, is often obtained through self-report. The Behavioral Risk Factor Surveillance System (BRFSS) is a major source for population-based estimates and is used by health professionals, public health organizations, and researchers to identify and quantify self-reported utilization of screening procedures.
Methods: We provide estimates of the reliability of responses among persons age > or = 50 to questions on the 1999 BRFSS questionnaire addressing two colorectal cancer testing procedures, fecal occult blood test (FOBT), and sigmoidoscopy or colonoscopy (endoscopy), based on responses of 868 persons who responded to a callback survey.
Results: We found moderate reliability for questions addressing ever having an FOBT exam, (Kappa [K] = 0.55, 95% confidence interval [95% CI]: 0.49-0.61) and good reliability for questions addressing ever having an endoscopy exam (K = 0.69, 95% CI: 0.65-0.74). Questions addressing the timing of the most recent exam were only slightly less reliable (K = 0.49, 95% CI: 0.43-0.55 and K = 0.62, 95% CI: 0.57-0.67, respectively). We observed comparable reliability across levels of most demographic and risk factor characteristics for both ever having and recency of exam.
Conclusion: Our results suggest that colorectal cancer testing questions on the BRFSS display a reasonable level of test-retest reliability.