We don’t know how much Test and Trace affects the spread of Covid-19

By | February 18, 2021

Over the past few weeks, the chair of NHS Test and Trace, Baroness Dido Harding, told two MPs’ committees that the programme significantly reduced the spread of Covid-19 last October.

This was based on research by the Department for Health and Social Care (DHSC), which has since been published.

The research is uncertain about the effectiveness of Test and Trace specifically, which means that Baroness Harding’s claim is something we cannot be sure about.

She has partly clarified this point in a subsequent letter.

What did Baroness Harding say?

On 18 January, Baroness Harding told the Public Accounts Committee: “Independently verified analysis suggested that as of October, Test and Trace was impacting R by between 0.3 and 0.6.”

The R is the reproduction number, which shows how fast the virus is spreading. It represents the average number of people that each infected person goes on to infect.

In another session on 3 February, she told the Select Committee on Science and Technology: “In October, Test and Trace was reducing R by between 0.3 and 0.6.”

Later, when she was asked why the analysis behind estimates like this could not be published quickly, she said it was a matter of technical complexity. Explaining this further, she said: “In order to do that modelling you need to establish the counterfactual. What would happen if you weren’t doing Test and Trace?”

What did the analysis say?

When the analysis was subsequently published, on 11 February, it did show an estimated effect of 18-33% reduction in the R number in October, which meant a reduction of 0.3-0.6 in the overall R number at the time.

However, this was not an estimate of the specific effect of Test and Trace, but of its combined effect along with people with symptoms self-isolating.

This is important, because people with Covid symptoms, and their households, have been required to self-isolate since March 2020, before Test and Trace was launched (on 28 May 2020).

As the name suggests, Test and Trace is a combined system of testing, both in hospitals and in the community, and tracing, whereby people who have been close to an infectious person are found and told to self-isolate. Some of the tasks involved are performed by private companies, others by local authorities or the NHS.

The 11 February analysis estimated that the specific effect on R of the contact tracing element of Test and Trace was small, at about 1.7-4.6%.

The analysis did not estimate the effect of testing, because it “depends on key behavioural parameters that we are not yet confident we can quantify”.

Altogether, this means that almost all the reduction of R that Baroness Harding mentioned comes from the combined effect of testing (which Test and Trace provides) and self-isolation with symptoms (which it may influence, but is not solely responsible for).

In short, it is hard to say how much of the effect was caused by the Test and Trace programme, and how much would have happened without it.

Baroness Harding did not include this point in her evidence at either session. However, she did make it in a subsequent letter on 11 February, in which she said: “Reducing the R number should therefore be attributed to a combination of the work we are doing within NHS Test and Trace as well as the efforts of the public when they self-isolate upon experiencing Covid-19 symptoms… I recognise that this may not have been made explicit at the hearing.”

The Test and Trace Business Plan, published on 10 December, also explains this, saying that “in October, testing, tracing and self-isolation (on symptom onset or following contact by Test and Trace) reduced the R number by around 0.3–0.6, compared to a scenario with only social distancing, restrictions and no self-isolation.”

The “explainer” document accompanying the full analysis also emphasises this specific point.

Full Fact raised this with the statistician Professor Sir David Spiegelhalter, who told us: “It’s vitally important to make the comparator clear in any evaluation, and the explainer to this report literally underlines that the R reduction is estimated to be 18-33%, ‘compared to a scenario with only social distancing restrictions and no self-isolation’. It’s unfortunate but understandable that they felt unable to estimate the influence of TTI [test, trace and isolate] over and above other strategies to encourage self-isolation.”

We asked the DHSC whether Baroness Harding agreed that the evidence made her claim uncertain, but it did not respond with an answer to our question.

What is really going on?

Untangling all the different effects in order to work out the true impact of Test and Trace would be extremely complicated. On the face of it, the fact that people self-isolate when they have Covid symptoms seems to be a very important factor on its own.

The Test and Trace Business Plan itself says in a footnote: “In our model, the largest part of the transmission reduction (90%) is due to isolation of positive cases from symptom onset, assumed to increase after a positive test. We assume that some of this isolation occurs because testing contributes to better self-isolation adherence both before and after the test. The structure of the model means that, if we were to assume people would isolate on the basis of symptoms without a test, then a large proportion of the modelled transmission reduction could be achieved from the public guidance to isolate after symptom onset alone.”

However, as this suggests, it is possible that receiving news of a positive test makes people more likely to self-isolate correctly, which would be an effect of testing. It is possible that hearing news of a positive test in someone close to you could help as well.

These estimates also describe the effect in what the analysis calls an “October-like environment … with symptomatic testing only”. This means it does not include any effect of routine or regular testing of people without symptoms, which was already part of Test and Trace before October, although it has increased since.

Baroness Harding also mentioned Test and Trace’s other benefits, beyond directly reducing transmission, such as enabling people who test negative to work and live more normally, and to provide data to inform government decisions. This data itself can be used to measure R, so it arguably plays an indirect role in helping to reduce it too.

Even so, there is a lot more uncertainty in the estimate than Baroness Harding indicated in the select committee hearings. As a result, the true effect of Test and Trace in reducing the spread of Covid might be much less than she claimed.

Source