Personal COVID Infection and Death Risk Models – and one you can play around with

The risk of death resulting from exposure to SARS-CoV-2 virus is a producty of personal vulnerability, exposure level and duration.  In the current presentation and exercise the focus is indoor exposure.  The vulnerability of an individual depends on immunity status, age, gender, underlying health condition plus other more or less understood factors that include, e.g., blood group.  Exposure depends on the incidence of COVID-19 in the surrounding society, the duration, number, proximity to and physical contacts with other people in the same indoor space, its air exchange rate and the use, and efficiency of personal protection.  Most of these factors act in a multiplicative way, i.e. doubling the infection incidence in the society and doubling the number of infected people in the space, and cutting the air exchange by 50 % would increase the resulting infection risk by a factor of 8.  Adding everyone face masks that would reduce person to person virus transmission by 50 %, would drop the overall infection risk increase to 4.  Separately the impact of doubling or halving each exposure factor on the risk is rather intuitive, but the combined impact of changing multiple counter- and coacting exposure factors is not – and for better or worse the resulting impact may be far bigger than perceived.  The risk may increase but also decrease much more than our intuition tells us.  Modelling tells us what our intuition does not.

A two part ISIAQ webinar: Modelling Infection Risk from Indoor Aerosol Exposure to SARS-CoV-2 was organised by professorsrs Lidia Morawska from University of Brisbane and Brad Prezant from Valid Air Sciences Ltd. in Victoria, Australia.  The First part took place on 30.11. for Asian and European participants and the second on 2.12. for American and Pacific participants.  Thirteen different models for COVID-19 infection risk in indoor environments were presented, most of them available in the internet [and I will provide the links to as many of them as possible in the coming days].  You will find more information about the products of this groudbreaking webinar in the ISIAQ (International Society of Indoor Air Quality and Climate) website.

The models focus on many different questions from personal to more general risks, from relative to absolute risks, from single space to multiple rooms and transfers between them, from residential to school and occupational exposure scenarios.  Therefore some of the 13 presented models may answer your specific need much better than some others.  Where multiple models would fit, I advice you to try each, because such testing would indicate some of the model limitations and uncertainties, and deepen your understanding of the sources of these uncertainties.

In the Asian-European session I presented the Personal Relative COVID Risk Modellling Tool that estimates the COVID death risk in a particular indoor exposure scenario described by certain exposure factors relative to another, reference scenario. This model is a simple-to-use 3 sheet Excel Workbook and the following short example is actually an introduction to how you can use it and what for. I therefore suggest that you download it, so that you can view and play around with it as you read on.

Personal relative COVID risk modelling tool and data.GENERAL_PERSONAL_CASE

The model assumes that the effects of the four input factors – personal (physiological and behavioural), microenvironmental (inhalation and dermal), social and epidemiological – on the risk are multiplicative, i.e. doubling or zeroing any factor would double or zero the overall risk.  Consequently the model cannot compare scenarios where any of the exposure factors is zero.  This tool does not estimate an absolute risk, its output is the ratio of a case scenario risk to the reference scenario risk. 

  • On top left of the Excel sheet, rows 2 through 7, are brief user instructions.
  • Below is the simple model itself with the input values on column E in red and model calculated intermediate values on column F and exposure factor values on column G in black. Enter data only in the red cells.
  • To the right, I13…AB54 ,you find background data and link suggestions that you may use to choose input values for the model.

This is the first sheet, the General Reference sheet.  Its input data in the example represent European population level conditions in the summer of 2020.  In the model these data produce a reference risk of 5.05E-10.  This is merely a dimensionless numerical reference value.

The next, Personal Reference sheet represents a set of one individual’s physical, behavioural, microenvironmental and social factors plus the average epidemic factors during one day in July 2020 in Finland.  His age and gender (E15) increase his risk, blood group (E22) reduces it, spacious house (E31), low occupancy (E44, himself and spouse) and low local COVID incidence (E54, July in Kuopio/Finland), pull the risk downwards.  His resulting [local] Personal Reference risk estimate turns out to be only 6 % of the [European] General Reference risk at the same time.

In the third, Personal Case sheet his personal and microenvironmental factors remain unchanged, but his social factors now represent a family reunion – adding 9 visitors who spread the virus into the same exchange air volume and reducing the social distances.  The 50 times higher personal risk that the model predicts from the 9 visitors could be an overestimation, though, because the model treats decreased proximity and decreased air exchange per capita as multiplicaruve risk factors.  My guestimate is that in real indoor settings their sum effect would be less than multiplicative but more than additive.

The local epidemic factor reflects the 44 times higher local COVID incidence in November compared to July in Finland. 

His resulting personal total COVID death risk is now 50 x 44 = 2 200 times higher than in July!

Because the model objective is relative, not absolute risk, its data requirements are smaller and validity does not depend on many rarely measured COVID variables, such as the concentration, virulence or exposure/dose/response of the Sars-COV-2 virus in indoor air and surfaces.

A different limitation of the model is that it only compares the risks from exposures of equal duration in one stationary indoor setting. Scenarios involving different exposure durations and mobility between indoor and outdoor, or different indoor environments fall out of the scope of this model, but can be built up with it.