Colorectal neoplasia screening with virtual colonoscopy: when, at what cost, and with what national impact?

Abstract

When optimized, virtual colonoscopy may be highly sensitive for colorectal neoplasia. We evaluated the effectiveness and cost-effectiveness of virtual colonoscopy screening (VC) vs. colonoscopy screening (COLO) and the potential impact at the national level. Using a Markov model, we estimated the clinical and economic consequences of VC and COLO from ages 50 to 80 years. Using census data, we made projections to the national level. In the best case considered (95%, 94%, and 87% sensitivity for colorectal cancer [CRC], polyps > or =10 mm, and polyps <10 mm), VC was nearly as effective as COLO. However, if test costs were equal, total cost per person was 15% greater for VC than COLO, making COLO dominant. When test cost for VC was < or =60% of test cost for COLO, the small benefit of COLO vs. VC cost >200,000 US dollars/incremental life-year. The greater the likelihood of being referred for colonoscopy after VC, the greater the advantage of COLO. With 75% screening adherence in the United States, VC and COLO could decrease CRC incidence by 46%-54%, with COLO requiring 6.9 million colonoscopies/yr, and VC, 3.2 million colonoscopies/yr, plus 5.4 million virtual colonoscopies/yr with VC. Even if screening test sensitivities were similar, COLO is likely to be preferred over VC unless virtual colonoscopy costs significantly less than colonoscopy. VC may be most appropriate in persons unlikely to need colonoscopy, such as those at low CRC risk. If VC were substituted for COLO, the demand on resources would shift from endoscopic to radiologic services, but would not diminish.