Frequently Asked Questions

What data are you using for my country?

To enable the software to be as up to date as possible, we input daily updates of the number of deaths from COVID-19 in each country as compiled by the European Centre for Disease Control. The model is calibrated to this number. For countries not included in the European Centres for Disease Control, we source deaths from worldometers.

Local demography is used for each country based on the UN World Population Projections. Contact patterns are based on representative data for the setting but are not country-specific due to limited data.

Other epidemiological parameters in the model are based on our best current understanding of the virus and will be updated regularly – please see the Research page.

The timing of major interventions is currently taken from the Oxford Government Response Tracker. We are working to incorporate Google movement data as a more nuanced way to estimate the impact of interventions up to the current time.

Is transmission between countries captured?

No, the model assumes homogeneous transmission within each country. This is clearly an approximation, especially for large countries, and it is likely that transmission will occur across international borders.

Why can I not change the model parameters?

The underlying model parameters are detailed in the “Research” tab. These are our current best estimates based on extensive and ongoing epidemiological analyses of SARS-Cov-2 virus transmission. Varying individual parameters can lead to combinations that no longer correctly represent transmission. Researchers who are familiar with modelling should consider the equivalent R package “squire” which is available here (https://github.com/mrc-ide/squire).

Why can’t I model specific interventions such as school closure or face masks?

Types of intervention vary greatly by country (e.g., different levels of mixing permitted outside, inside, within households) and in terms of level of enforcement, and countries are opting for various combinations of these. For the most flexible approach, we have enabled the user to change the reproductive number, R_t, which reflects the reduction in transmission implied by the combined effect of all interventions currently in force. We have provided guidance on the typical reductions in R_t that accompany some of the most common interventions, such as school closures and use of face masks, on our Research tab, so the user can use these to arrive at an informed estimate of the likely reduction in R_t that accompanies their chosen suite of interventions. The model allows for multiple phases, that is, a different R_t value for each phase in the epidemic, so the user can alter their estimate of R_t over time according to what interventions are in force during each phase. This allows the user to adapt the model to the current situation in their country of interest.

Why are R_t and R_eff plotted?

Each infectious disease can be described by how quickly it spreads. This is determined by two epidemiological parameters:

1. The generation time: the time between the infection of individual A and the infection of individual B by individual A.
2. The reproductive number, R: the average number of secondary infections caused by an individual infection.

At the beginning of the epidemic (Time = 0 days), the reproductive number is equal to R₀. As the epidemic spreads, the value of R will change over time. R_t represents the average number of secondary infections caused by an infected individual at time t. If interventions are put in place or human behaviour changes leading to a reduction in transmission, R_t will fall. R_t can also increase if interventions are relaxed or human behaviour relaxes.

As the epidemic spreads, infected individuals who recover from infection will become immune and will no longer be infected if exposed to infection. The depletion of susceptible individuals in the population will decrease the effective number of secondary infections. This is referred to as R_eff. R_eff is the average number of secondary infections after accounting for immunity at time t. R_eff is what is being estimated from repeated cross-sectional infection prevalence surveys. At the beginning of the epidemic R_eff is equal to R_t. However, if the epidemic is allowed to spread throughout the population, R_eff will fall below R_t. When R_eff falls below 1, the epidemic is slowing down.

Currently, in covidsim.org, we assume that after recovering from COVID-19, individuals develop immunity which protects them from reinfection and remains for the duration of the scenarios simulated. However, if immunity wanes, transmission and R_eff will increase. We will be exploring the impact of waning immunity in a later version of covidsim.org