Screening participation, is more necessarily better? Accounting for competing risk factors in three screening programs.

Abstract: Participation in screening programs is typically below 100 percent, hence, it is believed that not all potential benefits are fully exploited, and increasing uptake from the ex ante (insurer) perspective could be socially beneficial. However, simply arguing to increase participation may not be straight forward.

Traditionally, the use of decision analytic models to evaluate the costs and benefits of increasing participation in cancer screening programs often assume equal risk factors among participants and nonparticipants while  only a few have captured differences in cancer risk among the two groups. We argue that several factors should be considered when weighing the costs and benefits of increasing uptake, i.e., the incremental size of the uptake, the cost of increasing uptake and the differential characteristics of the nonparticipants with regard to cancer risk and the correlation with comorbidities.

By not taking all these factors into account, policy decisions may be erroneously based on a very simplified assumption: that those individuals who decide not to participate are inherently no different than those who decide to participate which could result in overuse of health care services without conferring benefit.

Within a simplified framework, screening for cervical, breast and colorectal cancer are modeled under various risk assumptions to determine whether additional resources should be invested to increase screening participation, and illustrate how different risks and correlations have implications for costs and expected life years gained.

The marginal effects of changes in uptake was explored theoretically considering four different scenarios: 1) equal risk of cancer and comorbidities, 2) greater (lower) risk of cancer among non-participants, 3) greater risk of cancer considering comorbidity affecting all cause mortality, and 4) greater risk of cancer considering comorbidity affecting both all cause mortality and cancer specific mortality. The findings were tested empirically  through a series of stochastic simulations using decision trees. Model input parameters were based on literature reviews.

Preliminary findings indicate that the benefit of increasing uptake varies by scenario and screening program. There is a tendency that as the correlation between comorbidities, all-cause mortality and cancer mortality increases, the value of increasing uptake decreases. It seems less likely that an increase in uptake would be cost-effective when comorbidities increase the risk of cancer deaths and all-cause mortality more than 10% and 15%, respectively, in nonparticipants compared to participants.

When increasing uptake, decision makers should be cautious regarding the background characteristics among the nonparticipants, and rather, targeted uptake measures among nonparticipants with fewer comorbidities may be warranted.

Authors: Eline Aas and Emily A. Burger