25.05.2020

Estimation in emerging epidemics: Biases and remedies

Modeling InfectiologyTransversal
Britton T et al
J Royal Soc Interface

Main result

This work has for aim to quantify the effect of different biases and to see how to adjust the estimates. It is based on a numerical example and simulations for the Ebola case.

Takeaways

The estimation of key parameters (R0...) at the beginning of the epidemic presents different biases and difficulties due to sampling, multiple contacts, temporality... Their impact and the first elements of correction are given in this work.

Strength of evidence Moderate

This work is very well written and gives an interesting first overview of estimation difficulties, with theoretical elements and clear numerical examples. The model emphasizes the randomness of the epidemic onset and of the times that appear and neglects the question of the population heterogeneity, the effect of which is surely still poorly understood.

Objectives

This work focuses on the estimation of key epidemiological parameters: growth rate, R0, time distribution related to the infection.

Method

These parameters are sensitive and have predictive or interpretive consequences. Difficulties discussed are related to biases induced by "uniformly random selection in the population", the limited time interval for observations, the possible discrepancy between the actual time of infection and the time of symptom onset, or the multiplicity of potential infecting contacts. The model is in medium field (mixed population, homogeneous) with random times for the different transitions with memory (gamma laws in particular).

 

Read paper

bibliovid.org and its content are bibliovid property.

Legal Notice