1. Epidemiological Modeling to Guide Efficacy Study Design Evaluating Vaccines to Prevent Emerging Diseases
An Vandebosch, PhD
Joint Statistical meetings, Vancouver, July

2. Background
Efficacy of prophylactic interventions is evaluated through the reduction in disease attack or infection rates between subjects receiving the intervention versus a control group
Long duration: event-driven and potentially lengthy follow-up until required number of infections occurred
Costly: sample size depends on the number of events and infection rate
Design and sample size of the trial requires accurate assumption of the infection rate in the population under study
Sample size
Follow-up time
Study population and study area: exposed to the infection
A thorough understanding of the transmission of the infection could support study design and sample size justification

3. Study population and endpoint may vary in space and time
Seasonal diseases
Variation in annual infections, strain, severity
Influenza, RSV
Emerging diseases
hotspots, outbreaks, epidemics
Dengue, Zika, Ebola, …
Finland
Figure: # positive respiratory pathoghen results in Finland between , Gunell. EJCMID 2016
Ebola outbreak West-Africa

4. Infectious disease transmission models
Infectious disease transmission models allow to mathematically capture disease processes
Evaluate short-term incidence
Real-time modeling allows to characterize potential future long-term trends of an emerging epidemic
Can be extended to allow evaluation of intervention strategies and/or preventive measures
Metcalf and Lessler, Science 357 (2017)
O'Neill, Statistics in Medicine (2010)
This presentation illustrates two case studies and how information retrieved from the models was employed to guide study design and planning for efficacy trials in the context of Ebola and Dengue

5. Case 1 - Simulation‐guided phase 3 trial design to evaluate vaccine effectiveness to prevent Ebola virus disease infection
Evaluate effectiveness of a prime-boost regimen in preventing laboratory-confirmed EVD in an outbreak setting in Sierra Leone
Large-scale population-based approach: up to 400,000 prime and boost doses available  ± 160 clusters of 5000 participants
Cluster-randomized controlled trial. 1:1 randomization to
Immediate vaccination
Control (e.g. delayed vaccination)
Clusters=geographical, administrative divisions
Vandebosch et al. Clinical Trials 2016

6. Challenges in the Tail Phase of the Epidemic
Declining EVD incidence
Herd immunity effect in case of mass vaccination
WHO SitRep Sierra Leone up to Jan 21, 2015
Static predictions unlikely to capture epidemiological picture
Spatial heterogeneity
Small local outbreaks
Competitive landscape

7. Study power and Adaptive Rules
Adaptive rule to identify primary endpoint
Study power and type I error maintained

8. Real-Time Fitting and Forecasting
Situation on Feb 15, 2015

9. Real-Time Fitting and Forecasting
Effect of the start date of the trial
Cumulative number of cases for control (blue) and vaccine arms
Persistence probability
Situation on Feb 15, 2015

10. Case 2: Modeling for Dengue Prophylaxis
Antiviral against dengue virus: pre-exposure prophylaxis against dengue infection in travellers and endemic populations
Aim: Estimate the transmission rates and dynamics in the target study population
to inform planning and design of clinical efficacy trials
Collaboration with London School of Hygiene and Tropical Medicine to develop an infectious disease model to characterize dengue outbreaks geographically and over time
Model designed to inform
on the number and geographical spread of secondary infections following the detection of a first "index" infected case
number of secondary cases that can be prevented through a prophylactic intervention by protecting individuals against infection and limiting disease transmission
by age group (adults, children) and timing

11. Case 2: Dengue Cluster-randomized ring-prophylactic design
Employs the mechanism of an infection spreading into the population
Simulations to
identify optimal cluster size (balance cost new cluster vs additional subject)
determine sample size
with geographical spread and clustering (e.g. intra-class coefficient) from
disease transmission model
Index case
Contacts of
index case
Clusters randomized
to intervention
to control
Radius of ~40m
is optimal
Nina D’Hollander, Joris Menten

12. Future for Epi Models in Support of Pharmaceutical Development
Mechanistic models well established for use in evaluation of vaccination programs
Implemented for trial design in 2 case studies
Their use in trial design, planning, and analysis, is a relatively new and growing area of research:
THANK YOU!
“Rapid modeling exercises can be critical in making timely decisions and guiding interventions and field studies in a rapidly changing environment. For instance, models played a critical role in the design of vaccine trials during the Ebola outbreak (Camacho et al. Vaccine 2017). However, such real-time efforts remain sporadic and ad hoc.”