Table 2 Summary of methodological issues and suggestions to develop CEAs of screening tools

Screening/diagnostic test accuracy

Model iterations with two-way sensitivity analyses using different combinations of sensitivity and specificity to determine a threshold at which screening becomes cost-effective. Assuming 100% accuracy might overestimate cost-effectiveness estimates.

Modeling false positive and negative results

Building a pathway for false positives and false negatives that includes their costs and health outcomes. For false positives, it is important to include costs and health outcomes associated to unnecessary diagnostics and treatment. For false negatives, it is important to include the costs and health outcomes of a delayed diagnosis.

Compliance rates

Model the compliance rate of patients and healthcare delivery professionals. Compliance rates are particularly important when repeated screening is being recommended, since low compliance may mean that the costs of early testing are wasted if further testing is not done.

Prevalence/incidence

Screening programs are usually conducted repeatedly over time. Dynamic models (incidence based) can be developed to evaluate repeated screening processes while considering new at-risk patients. One-time-only screening procedures only take into account prevalent disease.

Pre-symptomatic progression rates

Population-specific progression rates are often difficult to find for pre-symptomatic disease. Extrapolation from the clinical phase, or from similar conditions, could represent a first step to tackle the uncertainty around these parameters. Sensitivity analyses should determine how progression rates are expected to affect cost-effectiveness estimates.

Sojourn time

Sojourn time determines when screening is appropriate. This is a crucial input into a screening model and there is rarely evidence to estimate it. Creating various scenarios with different sojourn times may allow the investigators to estimate its impact on cost-effectiveness estimates. Different sojourn times will affect the cost-effectiveness of different test frequencies and should be evaluated using cost-effectiveness modeling.

Treatment and health outcomes

CEAs of screening tools should always include follow-up diagnostic and treatment. Quality-adjusted life years are appropriate to account for health outcomes, but these should be specific to the population being evaluated. Every potential health outcome needs to be accounted for including side effects of screening and/or diagnostic tests.

Non-health-related spillovers

Evaluating a screening tool from a societal perspective requires the inclusion of all non-health costs and outcomes. It is important to understand the trade-offs between the different types of costs and benefits. The inclusion of non-health costs and outcomes has important distributional assumptions and will value patients differently.