web analytics

Revisiting lockdowns

Ever since the beginning of the pandemic, there has been heated discussion on the Non-Pharmaceutical Interventions (NPIs), their usefulness for stopping the spread of Covid, and their impact on society.

The revelation of the UK government’s WhatsApp messages from the early lockdown has brought this topic back.

Let’s unpack these points.

What did we know about voluntary response?

Before March 2020, there had been some research pointing out to the impact of voluntary measures.

The best evidence I am aware of is in the paper by Bayham et al, where they showed the extent of the voluntary response to the 2009 A/H1N1 epidemic.


They claim that without the response – just 2.38 minutes spent without social contact per 100,000 cases – the prevalence would have been almost 50% higher at the peak.

We also published a series of papers where we built a model that included voluntary responses to the spread of a virus (we thought of flu or norovirus).

We found that either a very strong or a very weak response is “optimal” – if the disease can be arrested, we should “panic” and stay indoors. Alternatively, if it is unlikely that we can do it, the best way is to “ignore” it and do nothing.

But, our work did not envisage a disease with 4% mortality rate for older people.

We even carried out an experiment using a computer game. We found that people changed their behaviour when faced with the risk of catching a disease but quickly restored their behaviour when the risk got lower.

Why did models ignore voluntary changes?

The models did not explicitly build in the voluntary response of susceptible individuals. I suspect the main reason was that despite the work mentioned above, we simply did not know how to model it. In our 2019 paper, we included a challenge for the modelling community:

Thirdly, epidemiological models increasingly need to be merged with an economic and behavioural framework

Kleczkowski, Hoyle, McMenemy, https://doi.org/10.1098/rstb.2018.0255

But also, there was a feeling that any voluntary measure would not be enough to stop the disease from spreading, and the government needed to be seen to take serious action.

Thirdly, I suspect the government felt that England was not like Sweden, and people would simply ignore any recommendations. I am still amazed at how much people observed the lockdown measures.

Would voluntary measures have sufficed?

We simply do not know. To answer this question, we would need to construct an “alternative” England in which no or a limited lockdown is imposed. We can only speculate.

Sweden is often quoted as an example. But, Sweden dealt with the pandemic with a mixture of relatively soft repressive NPIs and a relatively strong voluntary response. Comparing Google mobility records of Sweden and the UK shows not too dissimilar levels of reduction.

And Sweden had an appalling record of dealing with their old and vulnerable.

The future

Our 2012 paper “Controlling epidemic spread by social distancing: Do it well or not at all” has now been cited 212 times. There has also been a huge amount of modelling work done to understand every aspect of social distancing.

There is good evidence that lockdowns – as implemented not only in the UK but globally – caused a lot of harm to society. Contrary to what a lot of people say, there is also good evidence that lockdowns worked to suppress the Covid spread – as well as suppressing other infectious diseases.

I do honestly hope there could be a serious discussion when we look at both the positive and negative aspects of each NPI including the voluntary response to the epidemic.