Randomization for the susceptibility effect of an infectious disease intervention

Area of Science:

• Epidemiology
• Biostatistics
• Infectious Disease Modeling

Background:

• Randomized trials for infectious diseases often involve interconnected individuals.
• Interference, where treatment of one affects another's outcome, is common with transmissible pathogens.
• Key parameters include the 'susceptibility effect' (controlled exposure) and 'direct effect' (unconditional risk).

Purpose of the Study:

• To demonstrate that the 'direct effect' may inaccurately reflect the 'susceptibility effect' in contagious disease trials.
• To highlight potential misinterpretations of intervention efficacy due to interference.

Main Methods:

• Utilized structural features of infectious disease transmission to define the susceptibility effect.
• Employed a novel probabilistic coupling argument to establish stochastic dominance between potential infection outcomes.
• Analyzed widely recommended randomization designs in the context of contagious outcomes.

Main Results:

• The 'direct effect' may fail to recover the true sign of the 'susceptibility effect' under standard randomization.
• Estimating the 'direct effect' can lead to incorrect conclusions about an intervention's true impact.
• Misleading conclusions may suggest beneficial interventions are harmful, or vice versa.

Conclusions:

• Standard estimation of the 'direct effect' in randomized trials with contagious outcomes can be misleading.
• Researchers must carefully consider interference when interpreting results of infectious disease interventions.
• Alternative analytical approaches may be needed to accurately assess intervention effects in such settings.