Introduction

This is some work to implement ASSA COVID-19 Model assumptions in R using Imperlial College MRC unit's squire package.

This works repeats the model as is approximately but it has age structure so results do not correspond broadly. Theis model also uses slightly different incidence parameters compared to ASSA.

Why do it?

Several reasons:

• Check reasonability of ASSA model.
• Produces output by age band which the ASSA model doesn't do.
• Corrects for double counting of an assymptiomatic assumption in ASSA.
• Corrects for an aggressive assymptomatic assumption in ASSA.
• Allows modelling of hospital and ICU bed capacity constraints.

Assumptions

The code adjusts default assumptions provided by squire. To explore the default assumptions for South Africa they can be extracted by parameters_explicit_SEEIR and one can then examine the object. Below we do so and plot the per age-band hospitalisaion probabilities.

parameters_default <-
  parameters_explicit_SEEIR(country = "South Africa")

parameters_default$prop_hosp

Scenarios

Multiple scenarios are run corresponding to the original 4 ASSA scenarios:

• sX_incorred should be closest to the relevant scenario from ASSA.
• sX_fix1 fixes the double counting of assymptomatic assumptions in the above.
• sX_fix2 reverts to the implicit assymptomatic assumptions in Verity as used in squire.
• sX models excess deaths due to hospital and ICU bed capacity constraints as well.

Capacity Constraints

We use the squire model's aassumptions for death if requireing ICU and not receiving it is 95%. Similarly if hospital bed would be required but not available we assume 60%. You can read more about these here. They state it's based on expert clinical opinion.

Installation

To run this you need the odin and squire packages:

Install odin:

install.packages("drat")
drat:::add("mrc-ide")
install.packages("odin")

Install squire:

install.packages("devtools")
devtools::install_github("mrc-ide/squire")