The parameterization of predictive value for multisite screening

In the design of a medical screening program, it is standard practice to employ the parameters of sensitivity and specificity of the test, together with the population disease prevalence, in order to obtain the predictive values of the test. Multiphasic screening, i.e. the use of two or more tests on the same occasion to screen for more than a single disease, has stimulated the need for a formulation of the joint predictive value of these tests employed in combination. In this communication a parameterization of joint predictive value is presented in the context of multisite cancer screening. The resulting parameters are related to the separate disease prevalences, as well as to the sensitivities and specificities of the separate tests, under the assumption of statistical independence among cancer screens.