Coronavirus (COVID-19): modelling the epidemic (issue no. 4)

Annex. Modelling amendments

To improve the model fit the PHS data^[1] has been substituted with NRS data.^[2] This has resulted improved fitting of modelled estimates of deaths which results in altering the estimates of incidence and subsequently R to account for all covid related deaths based on this being on the death certificate. However the logistical model, which is based on new infections, continues to be use a model fit to the PHS data as PHS is our source of positive cases. So both sources are now used.

The model has also been updated to reflect the move in Scotland to phase 1 of "Coronavirus (COVID-19): framework for decision making - Scotland's route map through and out of the crisis". Through the inclusion of additional covariates representing phase 1 - 4.

A review was undertaken of European countries to identify equivalent phase 1-4 activities (to date none of these countries, except Sweden who did not lockdown, have implemented phase 4 equivalent activities), to establish changes in data behaviour as a result of phases. Due to differential timelines between countries, it is assumed that the first implementation of a phase equivalent activity is classified as that country being in that phase.

Three additional countries of similar population size to Scotland (Slovakia, Ireland and Finland) were included to the original 13 countries to provide sampling of epidemic behaviours in similar sized populations. The case and death data for these countries are extracted from the ECDC^[3] Table 2 outlines the dates applied for each country initiated a phase equivalent activity.