Cervical cancer screening in low-resource settings: A cost-effectiveness framework for valuing tradeoffs between test performance and program coverage.

As cervical cancer screening programs are implemented in low-resource settings, protocols are needed to maximize health benefits under operational constraints. Our objective was to develop a framework for examining health and economic tradeoffs between screening test sensitivity, population coverage and follow-up of screen-positive women, to help decision makers identify where program investments yield the greatest value. As an illustrative example, we used an individual-based Monte Carlo simulation model of the natural history of human papillomavirus (HPV) and cervical cancer calibrated to epidemiologic data from Uganda. We assumed once in a lifetime screening at age 35 with two-visit HPV DNA testing or one-visit visual inspection with acetic acid (VIA). We assessed the health and economic tradeoffs that arise between (i) test sensitivity and screening coverage; (ii) test sensitivity and loss to follow-up (LTFU) of screen-positive women; and (iii) test sensitivity, screening coverage and LTFU simultaneously. The decline in health benefits associated with sacrificing HPV DNA test sensitivity by 20% (e.g., shifting from provider- to self-collection of specimens) could be offset by gains in coverage if coverage increased by at least 20%. When LTFU was 10%, two-visit HPV DNA testing with 80-90% sensitivity was more effective and more cost-effective than one-visit VIA with 40% sensitivity and yielded greater health benefits than VIA even as VIA sensitivity increased to 60% and HPV test sensitivity declined to 70%. As LTFU increased, two-visit HPV DNA testing became more costly and less effective than one-visit VIA. Setting-specific data on achievable test sensitivity, coverage, follow-up rates and programmatic costs are needed to guide decision making for cervical cancer screening.