I like your model.
I am not an expert of any kind, so hearing the public discussion and seeing the numbers as well as models leaves me with many questions that I don’t yet know the answer to. I do not know if my questions will lead to any insight or whether it is relevant or doable to include them in your model.
So, first off, I have come to understand that SARS-CoV-2 seems to be similar to SARS-CoV in that the individual propensity to transmit the disease varies greatly between infected individuals (overdispersion, they called it, you genii surely know it, I did not), with some individuals hardly transmitting it at all and others causing super-spreading-events. I think it was something along the lines of 10% of infected individuals causin 80-90% of the transmission. To me it seems like this could be an opening for a targeted public health response. Could one accurately model for this phenomenon? It seems easy to do in a deterministic model at baseline, but what if public health policy specifically reduced the opportunities for super-spreading events? Wouldn’t that be easier to include in an agent based model? If one could design public health policy to specifically reduce SSEs (both their number and the size of each one), as bans on larger gatherings does, wouldn’t this in practice be a very efficient policy (high reduction/restriction ratio)? I believe similar effects could be achieved by compartmentalizing public spaces, workforces as well as groups of students. Would it be possible to, in addition to the effect of restrictions, to include efficiency of restrictions in a model?
Second, WRT contact-tracing: Isn’t the breadth of quarantine policy very relevant to accuracy, like a shotgun effect? For example, tracing single contacts and isolating them surely has great effect, but is very difficult to accomplish, as the workload grows huge and the likelihood of being able to reach each contact shrinks, in practice, as the size of the epidemic grows. Couldn’t this be compensated by increasing the breadth from just isolating the infected individual to quarantining the entire household? Couldn’t it be taken even further? I know Norway at the peak had 10 000 hospital workers quarantined. What if we extended quarantines to other settings than the just the household? If this was done for workplaces, it would mean a great deal of the tracing could be offloaded to the managers of workplaces. This surely would be bothersome, but surely the gains should be huge as compared to longer lockdowns? And what about schools (maybe)? And possibly other settings? Could this be modeled?
Third, wouldn’t it be possible to reach conclusions about optimal restrictions and optimal breadth of quarantine in combination with a contact tracing policy, by considering transmission potential in different settings? If, hypothetically, we reduced all contacts outside the workplace, the home and school to 0, wouldn’t contact tracing and quarantining become much easier, as one would only need to reach one member of a household (the positive case himself), one manager of a workplace and one headmaster of a school to achieve most of the effect? I know this black/white hypothetical isn’t realistic, but I believe such matters should be taken into consideration. If they haven’t already, that is. Could this be modeled?
Finally, again, I like your model. Very much. It is encouraging to see people working on more flexible models than the standard ones. I appreciate it. Thank you!
Olov Waldemar Hansson
Citizen of Finland, currently residing in Sweden, first year medical student