Fig. 4

An example of application of Eqs. 1B and 3 to data from actual validation studies. Validation studies by Austin [30], Pladevall [29] and Kennedy [28] were selected from the systematic review by McCormick et al. [17]. A Expected positive-predictive value (PPV) of the planned database (DB) study is plotted against the disease prevalence of the DB study. B Expected risk ratio (RR) of the planned DB study is plotted against the control group risk of the DB study. The right axis is in terms of relative bias scale. In each panel, center, lower and upper lines for each study correspond to the point estimate and lower and upper bounds of 95% confidence interval of DLR⁺