Figure 2. Output from our analytical framework showing the relationship between VE_I and VE_C.

For this example, we assumed the following serotype-specific efficacies against infection: VE_I1 = 0.6 (i.e., VE_I for DENV1 = 0.6), VE_I3 = 0.8, and VE_I4 = 0.9. We explored a range of VE_I against dengue-2 (VE_I2) (y-axis). The x-axis is the ratio of the probabilities of developing clinical disease in people with and without prior immunity, and hence reflects the extent to which prior heterologous immunity (natural or vaccine acquired) modifies the probability of clinical disease. The background color represents expected vaccine efficacy against clinically apparent infection for all serotypes (left panel) and for dengue-2 (right panel). Solid black contours indicate the clinical vaccine efficacy VE_C observed in the Ratchaburi trial [1] according to the intention-to-treat analysis. Dashed white contours indicate levels of the ratio VE_C2:VE_I2. Thus, the 0.5 contour divides the regions where VE_C2 underestimates VE_I2 by a factor greater (above) or lower (below) than 0.5. In this particular simulation, VE_C always underestimates VE_I.