Coronavirus – analysing the data makes you think we could do with more of it

If you want to understand some of the thinking behind the policy response to the spread of coronavirus, you might want to read the paper from the Imperial College COVID-19 Response Team, which is credited with accelerating the introduction of the current lockdown measures in the UK.

The paper builds a mathematical model for the spread of the infection and then applies another mathematical model for hospitalisation, treatment and death rates (both based on what experts think they might know so far about the virus’s behaviour and impact).  It models possible outcomes for three options: doing nothing, mitigation (ie, slowing down the spread), and suppression (ie, stopping the spread until a vaccine is developed), using judgements on the effectiveness of policies and actions by governments and individuals to achieve these effects.

To simplify, the assumptions put into the model suggest that more than 500,000 people in the UK could die prematurely if the virus is ignored.  Mitigating its spread could reduce this by half. The modelling suggests that suppression could reduce this into a range of 5,000 – 50,000 (compare this to the 10,000 deaths predicted for a typical flu season or the 616,000 annual deaths in the UK ).

There is of course more to it than that.  It’s worth taking on board that mitigation and suppression help reduce the peak caseload and, because not everyone can be treated at the peak, this reduces the death rate.

Also the suppression strategy is modeled on the basis of stringent measures of social distancing being in place on and off for most of the next two years.

And another key point is that the assumptions about the fatality rate for those infected vary enormously with the age of the victim: 0.03% for folk in their 20s, 9.3% for over-80s.  To put this in context, the great majority of the modeled 500,000 deaths would occur in the over-60s, and most of these in the over-70s. Or to put it another way, the infection fatality rate for over-80s is more than 60 times greater than the rate for those in their 40s.

A lay reader of the paper should be impressed.  It reads like the best expert work: clear assumptions, transparent sources, rigorous logic and modelling, identification of key uncertainties. So, it’s certainly plausible that it had the impact attributed to it.

That is also why it’s such a useful starting point for further consideration and analysis.

Because, as the paper makes clear:

“However, there are very large uncertainties around the transmission of this virus, the likely effectiveness of different policies and the extent to which the population spontaneously adopts risk reducing behaviours. This means it is difficult to be definitive about the likely initial duration of measures which will be required, except that it will be several months. Future decisions on when and for how long to relax policies will need to be informed by ongoing surveillance.”

Where might there be some good news?

First, better data on the infection and death rates will help in calibrating the response.  Particularly if it’s better than the assumptions used in the paper. But not so good, if it’s not.

Secondly, the modeling shows peak cases towering over current ICU capacity.  The capability of institutions and individuals to respond to a crisis and fill some of this gap may surprise us, for example, in measures to adapt capacity to ramp up ventilator production.

Thirdly, the possibility of better outcomes from mitigation and suppression measures.  

The policy measures and assumed impacts which the paper uses for the mitigation and suppression modeling – case isolation, household quarantine, social distancing for over-70s, social distancing for everyone, and closing educational institutions – are static (ie, changes over time are not modelled) and pretty broad brush.

Take for example, the policy of social distancing of the over-70s.  This is assumed to reduce contacts by 50% in workplaces, increase household contacts by 25% and reduce other contacts by 75% (with an overarching assumption of only 75% compliance with policy).

This is no doubt a fine initial assumption but actual behaviour is surely dynamic.  If things get as bad as modelled in the paper, one would expect contact reduction, measures to minimise contact risk, and levels of compliance to improve out of sight, which could significantly improve effectiveness.

Similarly, the model makes no allowance for any measures taken by individuals or groups to supplement (eg, better hygiene or more selective social contact) or improve the effectiveness (eg, reinforcing compliance) of policy measures, or the different impact of the virus in different parts of the country.

You might infer from this that the choice between mitigation and suppression policies is not quite as bright-edged as modelled in the paper.  Because in the face of growing infection rates, the risk of insufficient ICU capacity, frightening mortality and above all, insufficient data, the introduction of current lockdown policies looks consistent with both.

The challenge for governments, in the coming weeks and months, as they grope towards an optimal approach, will be to review and tailor their policies, deciding which ones to retain and improve, which ones to slacken or eliminate, according to what the data shows is most effective, at least cost.

The more evidence that the young and middle aged are at low risk, the more a policy of getting them back to work and a semblance of normality under appropriate hygiene and distancing protocols will suggest itself.  The stronger the evidence that the old and vulnerable are critically at risk, the more refined the isolation and support protocols will need to be, and the greater the pressure to ramp up necessary capacity (more critical care beds and ventilators).

Governments of countries like New Zealand – with few cases and easily policed borders – seem to have a better chance of implementing a suppression policy with less oppressive impacts.

But they still can’t escape the economic and social trade-offs, or the need to choose the optimal measures and implement them selectively.  And there will be more specific questions from a policy of isolationism. For example, according to Tourism NZ, the industry generates 6% of NZ GDP and 8% of employment.  How will it re-orient to no overseas visitors and New Zealanders holidaying only at home (or perhaps not at all).

The hard-edged political equation will be the impact of measures on wealth, living standards and personal autonomy.  Tough restrictions on the way one can live one’s life and lower living standards are more easily endured with a clear and shared purpose for a known duration, as shown by the experience of the second world war.  People may be less happy with more intrusive state control and a material reduction in the value of their retirement savings and wages, particularly if they feel others are not bearing the same burden.

Recall that in the aftermath of the second world war, governments in the UK, NZ and Australia, no doubt well advised by experts, sought to continue controls and rationing.  The voters thought differently, and kept the political parties responsible out of power for many years.

 

5 thoughts on “Coronavirus – analysing the data makes you think we could do with more of it

  1. This report here shows the death rate from Covid 19 matches the mortality of that age group
    https://www.statschat.org.nz/2020/03/23/another-reason-why-we-dont-know-the-covid-19-mortality-rate/
    implying that the coronovirus is jus the straw that broke the camels back.
    This set of data shows that Norway with lockdown has the same infection rate as Sweden with something close to Level 2 – in that case, what is the lockdown doing?
    https://www.statschat.org.nz/2020/03/23/another-reason-why-we-dont-know-the-covid-19-mortality-rate/
    Note a lockdown is quite different to a border closure

    Like

  2. Reblogged this on The Inquiring Mind and commented:
    An interesting post with some points to ponder. The comments on the Imperial College paper clarified a few things. It raises another issue, often missed in for example Climate Change discussions, models are only models based on assumptions and those factors which the modellers take into account. Scientists recognise this, but politicians and indeed some academics, plus activists tend to take the models as Holy Writ. Perhaps they should not.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.