Objectives: Randomized trials have shown that annual or biannual screening by guaiac-based fecal occult blood tests (gFOBTs) reduces colorectal cancer (CRC) mortality. Few clinical studies have evaluated diagnostic performance of gFOBT through validation by colonoscopy in all participants. We aimed for a comprehensive evaluation of diagnostic performance of gFOBT by age and sex under routine screening conditions.
Methods: Our analysis is based on 20,884 colonoscopies following up a positive gFOBT and 182,956 primary screening colonoscopies documented in a state-wide quality assurance program in Bavaria, Germany, in 2007-2009. Positive likelihood ratios (LR+), which represent an integrative measure of diagnostic performance, were derived, by age groups (55-59, 60-64, 65-69, 70-74 years) and sex, from a joint and comparative analysis of prevalences of colorectal neoplasms in both groups.
Results: Overall LR+ (95% confidence intervals) were 1.11 (1.06-1.15), 1.80 (1.72-1.88), and 5.04 (4.64-5.47) for non-advanced adenoma, advanced adenoma, and cancer, respectively. Assuming a specificity of gFOBT of 95.2%, as recently observed in a German study among 2,235 participants of screening colonoscopy, these LR+ would translate to sensitivities of 5.3%, 8.6%, and 24.2% for the three outcomes, respectively. Diagnostic performance was similarly poor among women and men and across age groups.
Conclusions: The performance of gFOBT under routine screening conditions is even worse than previously estimated from clinical studies. In routine screening application, gFOBTs are expected to miss more than 9 out of 10 advanced adenomas and 3 out of 4 cancers. These results underline the need and the potential for better noninvasive CRC screening tests.