Why we need Pillar 2 testing data to be shared to help stop local outbreaks

June 29, 2020 • Reading time 4 minutes

Follow us on Twitter for more frequent updates (@georgebatchelor @edge_health_)

Published 30 June 2020

Hospital-based Covid-19 mortality figures from early March tell a chilling story of one of the first (if not the first) outbreak of Covid-19 in England. No one saw it coming, and I am not sure it is fully recognised now.

On the 17th of March, a total of 46 deaths were attributed to Covid-19. More than 10% of these were at The Royal Wolverhampton – an 850-bed hospital in the city of Wolverhampton (population 250,000). Compared to the regional average, there were about five additional deaths per day at the hospital in the early weeks of the pandemic.

The likely cause of that outbreak, which may well have seeded other outbreaks that started at events like Cheltenham Races (a few weeks later and an hours drive away)?

On the 20th and 27th of February, there were two Europa League clashes between Wolves and Espanyol. Game 1 took place in Wolverhampton and Game 2 in Catalonia – a region in Spain that was severely hit by Covid-19. Deaths were relatively higher at the Royal Wolverhampton about three weeks after Game 1 – see chart below.*

This was back in February when the idea of catching Covid-19 was not taken seriously, and testing numbers were low.

Testing has now increased and gets recorded under four Pillars. Pillars 1 and 2 report swab testing, which captures current infections – so these data are vital to spotting current people with the disease like the one in Leicester. Data for Pillar 1, which covers testing for those with a clinical need (often in hospital), are reported nationally (this is good). Pillar 2, which reports testing in the community, are not published – some local health officials get these data if they have signed Data Protection Agreements (this is bad).

Pillar 1 test data show no clear evidence of an outbreak in Leicester (see our chart below based on publicly available information). Pillar 2 data, on the other hand, appears to hold the key to spotting new outbreaks (See FT chart below, unclear how they got hold of these data that are not yet public).

It is also unclear how many of these reported infections will end up in hospital cases or, worse, deaths.** So I imagine staff at The Royal Wolverhampton are feeling a bit concerned and uncertain about what to expect over the next few weeks. Hopefully, they have a good stock of PPE and have put in place other reassuring controls to prevent hospital-acquired infections.

Possibly these data are not shared for fear of alarming the public, or concern over the quality. But never before has the case for being transparent and open with data been stronger:

  1. Not sharing data causes confusion (e.g. big argument in Leicester overextending the lockdown)
  2. Not sharing data stops organisations like us providing independent analysis to challenging the development of groupthink
  3. Data that has a public interest will eventually be leaked to the press (e.g. FT chart above), so it is pointless and obstructive not to share
  4. Sharing the data helps control its use and interpretation
  5. The cost of sharing the data is meagre

… the list could go on. But to put it another way: if data had been collected and shared earlier on care home deaths, would national policy and support for care homes have changed sooner – potentially saving lives? Presumably, some analyst somewhere saw what was going on in care homes, but lacked the voice to communicate the worrying conclusion more widely.

As we enter Phase 2 of the Covid-19 saga, data in the national interest must be shared more openly. Not in a tick-box way like the current test and trace data. No. PHE needs to go further than just sharing summary data for Pillar 2 (which it should be doing already) and start sharing detailed data on testing. This should include:

  • number of tests by date
  • confirmed infections
  • location (postcode region)
  • demographic information for tested individuals (age band, gender, occupation, ethnicity, etc.)
  • information on where the infection is suspected of having happened

Everyone could plan better if these data were shared. Surely this is a good thing if we want to reinvigorate the economy and stay safe at the same time?

At least three countries are already ahead of us:

So it is possible to share more data openly. And it will help save lives and livelihoods – not to mention the faff of trying to work out what is going on from snippets of information leaked here and there. It doesn’t need to be in yet another fancy interactive dashboard, but a simple downloadable CSV file. It is maddening these data are not shared already, and sadly, I fear, it is the same mindset and set of reasons that have hampered the response in England from the start of 2020.


* Previously we have shared analysis of the impact of Cheltenham and The Champions League game between Liverpool / Atletico Madrid (https://www.edgehealth.co.uk/post/how-antisocial-is-covid-19), which used the same methodology.

** Possibly we can examine the ratio of tests to hospital admissions in early March (before tests were focused on hospitals only) and assume a similar rate of convergence. Still, we would need to account for changes in the rate of testing… and that wouldn’t take into account differences in demographics, etc.

Edge Health

Edge Health are a specialist UK healthcare analytics consultancy that use data and insights to improve the delivery of health and care services, so that better outcomes can be delivered more efficiently.