Using transmission models to inform trial design for dengue


We are interested in conducting a randomized controlled trial to assess the impact of a intervention on dengue transmission. Given how few infections result in disease, we are going to use first infection* as our primary endpoint. To design the trial, we must have an estimate of the probability of seroconversion in the study population, so we can calculate how many fewer infections we expect in the treatment group. This will be used in the Power calculations to identify the minimal sample size.

If we only have dengue disease data, how can we use models to estimate the probability of seroconversion in the study population. Potential complication: We are restricted by the ages of individuals we can enroll in the trial. Individuals who live in a place with a lot of transmission may have already had an infection by the time they are enrolled. Conversely, places without a lot of transmission may not have enough events to power the trial.

Things to consider

  • When deciding on your interest in this project, please consider the following:
    • The faculty member leading this project is based in Seattle (nine hours earlier than Cape Town).

    • This project requires no background on sample size calculations for RCTs (but any background is welcome)

    • The models used in this project may be “simple” or may be “complex” depending on the coding skills / interest of the participants

      • ‘Dynamic’ model approaches require someone with medium to advanced coding

      • ‘Catalytic’ model approaches should be approachable for all participants

    • This project will not include mosquitoes or mosquito population dynamics :D

      • Folks can certainly take than and run with it, but that creates considerable complexity
  • This group is recommended for:

    • Participants interested in using models in trial design
    • Participants interested in nuances between disease and infection
    • Participants interested in emerging relationships between age and disease when there may not be any difference in probability of disease given infection by age
  • This group will have the opportunity to engage in the following:
    • Literature review about host jumps and disease establishment
    • Coding and analyzing stochastic simulations


  • We will use some studies that contain both disease and infection data as anchors in our model.