30 July 2020

Lies, Damned Lies and Coronavirus

Test and Trace

by David Chilvers

This week*, we look at the issue of statistics relating to test and trace.  This system is meant:

• to follow up individuals who have tested positive for COVID-19
• to identify those with whom they have recently been in contact
• to contact those individuals
• to ask them to self-isolate for 14 days

Before looking at the issues with statistics about test and trace, it is worth recollecting the history of this system. It was meant to combine:

1. Personal contact with those that had tested positive, to generate a list of those with whom the individual had been in close contact. By definition, these individuals would be people the contact would know and hopefully have contact details
2. A Smartphone based app, which would additionally identify individuals that had been in close contact with the person that had tested positive (as well as those covered in 1 above) and would include people the individual did not know (e.g. was close to in a shop, on public transport etc)

The Smartphone app was developed by NHS-IX as a new product, in spite of the fact that other countries had already developed such apps, some with the assistance of Apple and/or Google.  The idea that coders in the NHS would have the skills to quickly develop an app that would be superior to what was already available was dubious from the outset, and was confirmed when the app was discontinued following trials on the Isle of Wight.

The reason given for veering away from the app was that people actually preferred contact with healthcare officials, rather than receiving a text or email from an anonymous system.  However, only around 3,000 of the 25,000 recruited as contact tracers were healthcare professionals and whether contact from a call centre run by SERCO would be seen as preferable to receiving a text or email must at least be worth some debate.  And, by definition, this system would miss those not known to the person testing positive.

The “world class test and trace system” announced by Boris Johnson in late May has thus become a manual system, with many of the 25,000 contact tracers having very little to do, yet racking up a £10 billion bill as at the 9^th July.  It is hard to see where this money has gone as 25,000 tracers even at £1,000 per week would only burn through £250 million in 10 weeks and the bill is 40 times that – must be all those management consultants setting up the apparently dozen or so separate systems to record all the data.

Given this complexity, statistics from the system come out some time after the event; for example data for the latest period ending 15^th July was published on 23^rd July.  Some key highlights over the first seven weeks of the system include (averaging over the seven weeks):

• 77% of those with a positive test were contacted by NHS Test and Trace, with the percentage slowly increasing
• 80% of those contacted provided details of recent contacts, again increasing
• Just 54% of those with a positive test have been both contacted and provided details of recent contacts, increasing over time
• 84% of the contacts provided have been reached, but this is decreasing

Around 200k contacts have been generated by the system, three quarters in different households from those of the person with the positive test

The good news from this data is that around 150k individuals that would not have otherwise been reached have been asked to self-isolate and that the percentage of those providing contact details is increasing over time.  The bad news is that no contacts have been generated from nearly half of those having positive tests and that the proportion of contacts being reached is declining.  Add to this the fact that no contacts not known to the person are being generated due to the absence of the Smartphone app and it is clear that the “world class” system is a long way short of this label.  Whether this system is worth the £10bn that has been spent on it remains an open question.

Next time, we look at the factors affecting death rates from COVID-19 which, in spite of repeated assurances from politicians, are still not as clear as they should be some five months into the pandemic.

*This article is one of a series, find last week’s article “Testing Statistics” here.

Follow the Shaw Sheet on