Did the UK lockdown too late ?

I have been running  the Imperial College Covid-19 model for different scenarios. The published  parameter files (on GITHUB) use an R0 value of 3.0 with an IFR of ~1%, higher than that published in February (2.4). This predicts a total number  of 620,000 deaths in the UK if the disease were allowed to run its course through to “herd immunity” without any mitigation. These are the figures that SAGE must have been  considering in early March, when the UK policy then was to only test arrivals from China and flatten infections  towards an eventual  herd immunity.  On the 13th March however the advice was suddenly changed:

The new advice issued by the Chief Medical Officer is as follows:
Stay at home for 7 days if you have either:

  • a high temperature
  • a new continuous cough

Do not go to a GP surgery, pharmacy or hospital.

Something dramatic changed very quickly and by March 23rd a full lockdown was imposed – Places (schools, shops, restaurants, pubs, businesses ) were closed, travel restricted  and social distancing measures applied. The new slogan was “stay at home, protect the NHS, save lives”.

I have used Neil Ferguson’s (ICL) model to investigate exactly why this sudden policy change occurred. It seems that SAGE had by then concluded that the current R value  was ~3.0 and infections exploding mainly in London. ICL’s  model was now predicting 620,000 deaths without government intervention. This was way too much to “take it on the chin” by “slowing the curve”. But what would have happened if the decision for lockdown had instead been taken  a week earlier or a week later ? This is what the ICL model says.

 

Different ICL lockdown timing simulations

Figure 1. How the timing of lockdown measures affected final UK death rates. Note also the  strange effect that the models and data all coincide on day 100 with 10,000 deaths (10th April). This appears to be because Ferguson forces the model to agree with the actual data  on this day.

It would have been far worse to delay the decision a week than to advance it a week! Hindsight is a wonderful luxury for all armchair critics. Fergusson’s model  predicts that if the lockdown had been imposed a week earlier, then it might have saved up to 15000 lives  in the short term. However if it had instead been applied a week afterwards it may instead have cost an extra 40,000 deaths !

The IC model is driven by various parameter files which are very obtuse and difficult to fathom out without any proper documentation. So instead I simply delayed all  intervention start  dates by ~ 1 week to get these results. However this results in the day 100 effect as described above. ATTP has an ad hoc  fix for this but I am not sure if this doesn’t perhaps also bias the result.

 

This entry was posted in Public Health and tagged . Bookmark the permalink.

10 Responses to Did the UK lockdown too late ?

  1. dpy6629 says:

    At least in the US, the rationale for mitigation was always to flatten the curve so hospitals wouldn’t be overwhelmed. All the epidemiologists understand that total ultimate infections will not be reduced much. In New York state, recent evidence suggests that the number of “detected cases” is 15 times less than those infected. That would indicate that 5.556 million (28% of the population) have already had the disease. In New York City, It seems possible that 50% have had it already and herd immunity might not be far away.

    So flattening the curve won’t prevent getting to herd immunity and perhaps that’s already happened in London too. I don’t understand why the UK doesn’t do random serological testing to find out.

    All these serological studies indicate that Ferguson’s IFR is perhaps 2-5 times too high. 620,000 sounds a huge number, but its probably way off and the true number is close to 200,000. I personally think the lockdown didn’t make much difference. People were already voluntarily socially distancing and working from home if they could. I can’t find it right now but someone did an analysis of R in every US state before and after lockdown and found little difference.

    • Clive Best says:

      Of course you’re right. This pandemic will only end once the global population achieves herd immunity. That can either happen through infection or through a vaccine. Countries that appear to have won the battle with COVID-19 by eliminating the virus e.g. New Zealand will have to self isolated from the rest of the world until an effective vaccine becomes available. Either way their economies will suffer.

      There have been some initial serological results. These imply the infection rate in London is ~ 25% and ~10% elsewhere.

      • MarkR says:

        Where’d you get those infection rates from? Hancock said 17% in London, 5% outside. https://www.bbc.com/news/live/world-52749186

        Excess mortality in London was ~0.1% of everybody through 2020-05-08 so naively that’s 0.6% IFR (Hancock) or 0.4% (you) for London.

        ONS gives ~90% of excess mortality in 65+ group, and UK outside London is older (~60% more 65+) so maybe non-London UK IFR would be 0.9-1.0% (Hancock) or 0.6-0.7% (you).

        But if you take *total* excess mortality and and Hancock’s London/non-London infection rates (17%/5%), you get IFR=1.4 %. Yours (25%/10%) give 0.8%.

        That difference is still a terrifying number of lives.

  2. Clive,
    I’m not sure why you think my fix would introduce a bias. The problem with leaving it as is (i.e., 10000 deaths on day 100) is that this is after the interventions were applied and is only really suitable for the intervention that actually happened (23 March). By leaving it as is, it means that you’re assuming that whatever the date of intervention, there will be around 10000 deaths on day 100, which would clearly not be the case if the intervention were very early (for example, in my runs the an intervention starting 1 week earlier wouldn’t even get to 10000 deaths). Similarly, if the intervention were very late, you’d expect there to have been more than 10000 on this date, but the model would still try to go through this point.

    • Clive Best says:

      ATTP,

      I agree that 10,000 is wrong for the early and late scenarios. However if you look at the unmitigated result (without any interventions) then you find that that too passes through 10,000 at day 100.
      So I think I am going to try a run without any constraints. I also now understand why my Sweden run gave such high death rates – it was also using 10k deaths at day 100. The correct figure is 400 !

      I might try running it without any constraints – i.e. simply delete it and see what happens.

      • Clive Best says:

        My run without any constraints failed and produced null results. That means that a normalisation point is compulsory to run the model.

        Also if you chose a date too early on when there are say just 5 or 6 deaths the results are crazy. So it seems that the model has to be put on the “correct” path first.

  3. John Bradley says:

    Many thanks Clive.

    Have you come across any other attempts to run the ICL model? I haven’t. I would have thought that there should be people swarming all over it, running different cases, looking at sensitivity of the key parameters, figuring out what drives what, in view of its importance in policy decisions.

    By making the lockdown start on March 16, rather than March 23, 17k fewer deaths would occur per Fig. 1. I presume that this is because 1.7 million fewer infections occurred. Similarly delaying the lockdown to March 30 would result in 35k more deaths because 3.5 million more people become infected as a result of the delay in lockdown. The effect of a change in the date of the lockdown is a function of the number of people who would have been infected in that time.

    This assumes of course that lockdown is not lifted, that deaths are prevented, not deferred.

    A question about the model that keeps bugging me is: how many deaths in the unmitigated scenario are caused by critical care capacity being exceeded?

    • John,
      I have done some too. They show the same picture as Clive’s results, but suggest more avoided deaths (more like 30000) because of how I’ve normalised the simulations (see comment above). However, the assumptions are not entirely realistic since they assume that we did nothing until the lockdown occured, which is not quite what happened (i.e., there were some suggsted interventions in place before 23 March). If I try to account for this, then it reduces the difference by about 5000 – 10000 (maybe it’s more like 20000 avoided deaths, which is similar to what Clive is getting).

      As to your question at the end, my understanding of the model is that it simply assumes that a certain fraction of those who get infected will die (with a strong age dependence). Hence, it’s not actually modelling the impact of exceeding the critical care capacity, it is simply illustrating what we would probably need to do to avoid it being exceeded.

  4. MarkR says:

    Does this model allow for arbitrary changes in R and IFR through time, or as a function of e.g. total caseload?

    For IFR it seems possible that it might go down with time as we get better at treatment, but it should also probably go up if the caseload is really high.

    I don’t know anything really about treatment, but physicians insist to me that early proning and oxygen seems to help, and the Lancet hydroxychloroquine (HCQ) results suggest that HCQ greatly increases deaths, while physician polls (e.g. https://www.sermo.com/covid-19-press-releases/#post-9053) hint that a lot of hospitals were using it for already sick patients.

    Assuming the Lancet results bear out, it looks like a decent reduction in in-hospital fatality rate would happen if they just stop giving HCQ.

  5. Hugo says:

    The thing what matters most is the viral load..
    Putting lots of people together in a confined space at home was not a good solution. Has killed many elderly couples.
    Spacing outside.
    UV has 2 benefits. its disinfects / kills viruses and it supports the immune system with vitamin D.
    Which seems to lack with groups most affected.

Leave a Reply to Clive Best Cancel reply