R, R, WHERE R'T THOU R? (Updated 29 May)

R has gone AWOL.  The king of the COVID metrics is missing, presumed dead.  Yet so important. 

That's what I wrote early on 28 May.  I then found something that made me fall off my chair.  I'm now returning to the topic late morning on 29 May:

R is so important as an indicator of how well this coronavirus is being kept under control, and a key part of assessing whether lockdown restrictions can be eased or indeed need to be tightened .  But R is only useful if reflecting the transmission rate now.  The problem is that it isn't.  I am shocked!


RECENT ANNOUNCEMENTS OF R

On 15 May, Sir Patrick Vallance, joint chair of SAGE, tweeted this:

We'll go to that web page in a moment.  But before that....

Sir Patrick has tweeted nothing about R since 15 May (as at 10 pm on 29 May).  Nor has Professor Chris Whitty, the other joint chair, presumably as his brief is medical matters.    .

So I looked at all the slides from the daily press conferences since 15 May.  Didn't take long, and interesting to see how the messages and focus in general changed between 15 May and 28 May (last night).  I only found these few mentions of R.  Each buried in slides crowded with other information:

Presser 15 May

Presser 22 & 25 May

Then alongside PM Johnson announcing on 28 May further relaxations in lockdown restrictions, and schools beginning to go back, we had this.  No mention of the 1.0 limit, I've checked:

Presser 28 May

So the estimate of R has gone down from an average of 0.85 to 0.8.  Perilously close to that 1.0 limit.   I would be far happier if R was under 0.5, which frankly should be the upper limit anyway to drive new COVID-19 infections down near zero.  As this graph shows:

You would be excused for interpreting "Current R" as being something reflecting the rate of transmission shortly before the presentation.  But when I looked at the web page that Sir Patrick tweeted. as above. "To find out more about R, go to" , I fell off my chair.  Shocked.  Dismayed.  Head in my hands!

Firstly it said that R is only calculated weekly (so there was an offical update to 0.7-0.9 late on 29 May).  Then I saw this (being a screen shot of the web page - spot the repeated words):

You what?  Not "Current"?   You're saying the "Current R" announced on 15 May was two or three weeks out of date?  Because you back-calculate it, rather than attempt to observe it directly?

We've been hoodwinked about it being "Current".  Who looks at the small print?  Not many!

I hope you don't mind me saying this.  "FFS".

This makes it particularly interesting that Dr Jenny Harries answered a question at a recent press conference with words to the effect that 'R will be shortly calculated directly rather than estimated, when it will be possible to know by region'.  Rather sheepishly I thought.  As R hasn't been publicised by region, I am assuming the figure in the 28 May presser was estimated using the old method.  As set out on that same page:

Two or three weeks out of date?  So the "Current R" on 28 May is back when PM Johnson announced the change from "Stay at home" to "Stay Alert" and eased some of the lockdown restrictions.  It doesn't reflect changes since.  Again, FFS !


AFTER A LIE DOWN

In my professional work, I get involved in designing Performance Dashboards for Boards of FTSE, AIM and other larger businesses.   I interview members of the Board.  Very often it is the CEO who has initiated the project.  I can just imagine the conversation:

Me: "So what are your most important decisions that would be helped by data?"
CEO: "Whether I need to ease or tighten the lockdown restrictions due to this coronavirus"
Me: "What do you want to see?"
CEO: "The latest R number.  Oh, and the current total of people with COVID-19 infection.  And the number of new infections would be useful too."
Me:  "How often do you want to see those?"
CEO "Every morning  at 8am.  Every day 24x7"
Me: "How about graphs?"
CEO: "Yes, two graphs by day since we started measuring them.  Marked with a note of major events such as the easings on 11 May and 28 May"

I hope you've got a good imagination, because we are not seeing anything like that.  Maybe SAGE and the PM are seeing something better, and are just hiding it from the public.  But I suspect that these figures and graphs simply don't exist.  A weekly estimate of R is just not frequent enough.

But the figures and graphs could exist:

• The current level of infection can be estimated from the daily 'New Cases'  data, already presented in the press conferences.  Though ' New Cases' lag 'New Infections' by about a week.
• There are better ways of calculating R.  Let's investigate...

I would leave my meeting with the CEO and think about it.  We have a simple sequence of events, with several days' lag between these events for each person contracting COVID-19:

• Contracting disease
• Become infectious
• Displaying symptoms
• If serious, admitted to hospital
• If treatment fails, death

These last two steps are the backbone of the graphs in the slides presented in the press conferences.  Important, but too late to support decisions about adjusting the lockdown:

• The number of 'Deaths' is some weeks after infection
• The number of 'Hospital Admissions' is a couple of weeks after infection.  PM Johnson's COVID-19 illness is a good example.
• The number of 'New Cases' is at least a week after infection.  Symptoms don't usually appear for at least 5 days after infection.  Swab testing for viral RNA won't show positive until at least a couple of days after infection  (I'm helping a research project into how long before)

We need to know the rate of infection when people are getting infected..

So we need to know R when people are actually contracting the disease.  There's no point back-calculating R from the subsequent steps.  Indeed what purpose does that serve?  We can see from the fall in new cases and hospital admissions that R must have been below 1.0 a week or two earlier.  But that's too late to make decisions now.

I don't know what methods the boffins have developed to more directly estimate R on a more timely basis.  Maybe they have a better way that they can announce shortly.  Let's hope so.


SOMETHING ELSE USEFUL

Maybe it is based on this.  I have found something else which we could usefully add to the dashboard.  This would identify 'hotspots' around the country, as well as being averaged to produce an overall UK picture.  Current viral RNA testing can be kept for when individuals need to be tested.  To read on, I'm hoping you're not eating your meal!

There appears to be a strong correlation between testing for the SARS2 coronavirus RNA in sewage and subsequent hospital admissions.  This is being examined in more detail.  Indeed the FT reported on 21 May that the UK government is considering doing that type of testing.  Maybe what Dr Harries had in mind.

Yet the idea has been around for some time.  There is an academic paper from Wuhun that pre-dates this article in Nature on 8 April.   Here is a graph showing that detection of viral RNA in sewage corresponds closer to when people contract the disease in a community, three to four weeks before hospitalisations:

If I was the CEO (PM) of the UK I would want to see this RNA count at a national level  But also at the local level, as would the Local Council .  

The test has the advantage of being able to identify local hotspots at the earliest possible time, which can then be charted in a similar way to this:

So let's be shown a new dashboard at the daily press conferences with three graphs:

1. Current level of cases, and/or new infections, and
2. Estimates of the real current R, and
3. Viral tests from waste water   

Plus a hotspots graphic like the one above.

Wouldn't that nake a difference to our understanding of what's happening with COVID-19, both nationally and locally?