Statistical and economic models for analysis and optimal design of laboratory screening policies for cervical cancer

This paper develops several statistical and economic models in order to analyze the accuracy and total cost of laboratory procedures commonly used in examining Pap smears for early indications of cervical cancer. In particular, the expected values, variances, and probability distributions are derived for the overall system sensitivity, overall system specificity, workload, and total system cost of several possible laboratory screening policies. These models are useful for analytically comparing possible alternative screening processes and for identifying the optimal minimum societal cost policy. Additionally, these models are used to show that the overall screening process never can have higher sensitivity nor lower specificity than the pathologist, that a rescreening rate of either 0% or 100% always will be economically optimal, and that total costs are highly variable period to period. While a primary objective of the present paper is to develop analytic and research tools, preliminary applied results suggest that significant improvements in sensitivity and reductions in total societal costs are possible by changing the policy currently required by the congressional Clinical Laboratory Improvement Amendments Act of 1988, typically with one to three repeated cytotechnologist screenings of presumed negative smears. Sensitivity analysis indicates these results are very dependent on cost and other inputs, and several directions for continued research therefore are described. Use of these models at the more micro laboratory and insurer levels, incorporation of automated rescreening devices, and extensions to mammography and other laboratory tests also are briefly discussed.