COVID has high heterogeneity in its infectiosity, which has consequences on:
- reliability of predictive modeling;
- contact tracing strategy.
This paper, by Susan Holmes and Claire Donnat, focuses on the uncertainty in modeling:
These two papers insist on the relevance of backward tracing:
This idea is rising to prominence in Germany as well.
This is great, but it feels like the intersection of the two ideas is missing. The backward contact tracing process has an even bigger heterogeneity, due to the inherent difficulty in conducting it. (You start interviewing index case X, which has multiple contexts c of interaction, each with different risk level, different probability of recall, and different difficulty in contacting co-participants - how do you balance priorities???).
Anticipating a bit the direction things will go, it would be very good to understand how these uncertainties in the contact tracing process would affect models, including staffing needs for contact tracing.