Reducing contamination risk in cluster-randomized infectious disease-intervention trials

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Cluster-randomized trials (CRTs) are common for infectious disease interventions.
• Geographic sampling units in CRTs can lead to contamination risks.
• The 'fried-egg' design, a common method, has limitations including reduced precision and ignoring spatial heterogeneity.

Purpose of the Study:

• To present a novel approach for cluster-randomized trial design.
• To address limitations of existing methods like the 'fried-egg' design.
• To optimize CRT design for settings with contamination risks and spatial heterogeneity.

Main Methods:

• A new algorithm for CRT design is proposed.
• The approach fully includes or excludes clusters within a study region.
• It identifies the maximum number of clusters while maintaining randomness in selection and allocation.

Main Results:

• The novel approach was successfully applied to a CRT for malaria vector-control in Malawi.
• It provides a framework for reducing contamination risk in CRTs.
• It maintains randomization validity for hypothesis testing.

Conclusions:

• The presented approach offers a valuable framework for designing CRTs.
• It is particularly useful in settings prioritizing contamination risk reduction and where spatial heterogeneity is likely.
• This method ensures a randomization-valid test of the primary trial hypothesis.