Cervical cancer screening involves a multi-step process that includes initial screening, follow-up testing, diagnosis, and treatment of precancer. “Leakages” in this process can diminish the effectiveness and cost-effectiveness of screening. Model-based evaluations of the long-term health impact and cost-effectiveness of alternative cervical cancer prevention strategies in the United States often make optimistic assumptions of perfect coverage and follow-up compliance. However, there is evidence that cervical cancer screening practice is imperfect and inefficient. Integration of this evidence into policy models are imperative to understand the real-world impact and cost-effectiveness of screening strategies.
We describe the challenges of estimating the practice of cervical cancer screening in the United States. We estimate the frequency with which women in the U.S. receive cervical cancer screening, as well as compliance to referrals for diagnostic procedures and precancer treatment using different sources of empiric data, including a national survey in which respondents self-report on screening practices, large health organizations, and a state-wide lab-based screening registry, the New Mexico HPV Pap Registry (NMHPVPR). Using a microsimulation model, we provide examples of how different inputs for screening practice yield different estimates of the long-term health impact and cost-effectiveness.
We found that several sources of national surveys reported on the frequency of cervical cancer screening cross-sectionally, but longitudinal data and data on follow-up visits were limited. Screening intensity reported in the national surveys were much higher than suggested by lab-based data in the NMHPVPR over the same time period. National surveys typically did not collect information beyond the initial screening visit. When we integrated data from the NMHPVPR on the full spectrum of cervical cancer screening into the microsimulation model, including both underutilization and overutilization of diagnostic procedures and precancer treatment, we found that estimates of life-years gained and costs deviate widely from scenarios under ideal assumptions of perfect coverage and compliance.
The challenges in modeling cervical cancer screening practice limit our ability to estimate the true health and economic impacts of screening. Understanding patterns of screening in different health care settings is a priority for the CISNET-Cervical working group to overcome these challenges and identify ways in which to make improvements to maximize the impact of cervical cancer screening in the United States.