MS1 Cost-Effectiveness Analysis of Colorectal Cancer Screening Programmes:How Omissions of Relevant Comparators CAN Lead to Inefficient Policies

Abstract

This systematic review assesses the choice of simulated screening strategies in cost-effectiveness analyses (CEA) of colorectal cancer (CRC) screening for average-risk populations. It examines the implications of strategy comparison for the identification of optimally cost-effective policies. The PubMed, Web of Science and Scopus databases were searched for model-based cost-effectiveness analysis simulations of CRC screening within average-risk populations. The interventions concerned are stool and image-based modalities of CRC screening. The analysis specifically focuses on the choice of comparisons made between screening strategies to estimate the incremental cost-effectiveness ratio (ICER). Such ratios include the cost per quality-adjusted life-year (QALY) gained or cost per life-year gained. The omission of relevant comparators and miscalculation of ICERs are the two main concerns that arose from this review. Of the reviewed studies, 87% reported results with sufficient significant figures to permit replication of ICER estimates while 76% reported the complete results for all assessed strategies. Regarding the ICERs calculation, 44% of studies reported ACERs as ICERs, 41% reported ICERs for dominated comparators, and 33% of CEAs calculated the ICERs on the basis of comparisons to dominated comparators. Approximately half (49%) of CEAs had any one of these problems. Most studies did not consider adequate screening comparators, including a limited variation in screening technologies (44%) and screening frequency (54%). Most CEAs (82%) did not simulate a sufficient large of the efficient frontier for the policy-relevant decision threshold. In addition, 5% of CEAs with sensitivity analysis evaluated some options not in the base-case analysis, and 8% excluded the no-screening option. We find the CRC screening literature to be lacking in consistent quality regarding both adequate reporting of cost-effectiveness estimates and, more fundamentally, the sufficient variation in screening alternatives simulated. This raises doubts regarding the certainty of evidence regarding optimal CRC screening policies.