Sawitree Boonpatcharanon, Jane Heffernan and Hanna Jankowski's contribution to the Discussion of ‘The Second Discussion Meeting on Statistical aspects of the Covid-19 Pandemic’

We congratulate the authors on a timely, important, and well-written paper. The effective reproductive ratio is a key characteristic that can be used during a disease outbreak to understand the spread of the disease and to gauge effectiveness of public health measures. Furthermore, real-time estimation of the effective reproductive ratio is of particular importance, so that real-time responses can be made by public health authorities. The COVID-19 pandemic has brought various measures, especially non-pharmaceutical ones, to the forefront of the public’s attention. However, public health measures have always been of great importance in monitoring and managing disease outbreaks. As zoonotic or re-emerging disease outbreaks are expected to happen from time to time, we can expect this work to have long-term impact.

The authors propose a version of a compartmental SEIR (Susceptible-Exposed-Infectious-Recovered) disease model, developed in Engebretsen et al. (2021). Their model uses both the number of cases reported based on PCR (Polymerase Chain Reaction) testing (which assumes that only a proportion of cases have been tested) as well as hospitalisation data. Certain model parameters are held static throughout, either assumed or estimated from other sources. These parameters include $π0$ and $π1$ which determine the proportion of cases that submit to PCR testing; as well as the compartmental model parameters θ which determine underlying disease dynamics. The developed method, notably, allows also to estimate the actual vs. reported number of cases; another key characteristic of interest.

One difficulty in public health management of COVID-19 has been that various inputs, such as θ and $π0,π1,$ have changed considerably over the two years of the pandemic. For example, in the province of Ontario, Canada, PCR testing was very low at the beginning of the pandemic (first half of 2020) and also starting at the height of the Omicron outbreak (early 2022 until time of writing). Indeed, starting in 2022, only select individuals from target groups (e.g., those that are severely immunocompromised) are eligible for PCR testing. The model proposed by the authors allows for considerable flexibility in modelling such scenarios, however, certain parameter choices could affect performance and accuracy of the method. Have the authors tested their methods where the proportion of the population receiving PCR testing is quite small? When PCR testing is so low, it would be particularly relevant to obtain estimates of actual vs. reported cases.

Reference

Author notes