When a body like the UKHSA releases data or reports, the assumption is that they are meant to be useful: to doctors, public health practitioners, and - perhaps - to the public. They are not in any way meant to be a comprehensive summary of what the UKHSA knows or suspects.
But they are meant to be useful. So it’s interesting to trace through what they might reflect, and to try to infer what is understood.
Three views into a complicated situation
Roughly, there are three sources of contemporary data that the UKHSA release publicly that tell us about notifiable diseases like measles.
In roughly ascending order of reliability, and descending order of timeliness:
NOIDS weekly reports - essentially every report of a statutory notifiable disease that UKHSA has received. These come principally from GPs, and can be from symptoms alone - no lab test or extensive case investigation needed. Shown by reporting date.
NOIDS causative agents reports - the lab notifications when a test shows the presence of a notifiable disease. As you might expect, these tend to run lower than the first type of reports (historically, about 1/6th of the ongoing level for measles - kids get a lot of rashes.). Not de-duplicated or investigated on a case level (e.g., separate tests from the same individual might be reported multiple times). Shown by reporting date.
Incident report - when there is an infectious disease outbreak, a team produces a (roughly) weekly report where they investigate case history, de-duplicate the tests on an individual level, and (often) retest local lab results in national reference laboratories. One was triggered at the start of the year for the Birmingham measles outbreak. Shown by sample date.
So, in “how to read” terms:
Take the first with a massive pinch of salt - only a small fraction turn out to be real; if it grows - especially in one area, this is a flashing light to look further.
The second is more useful to reflect trends, but beware of reporting backlogs and multiple tests-per-case.
The third is UKHSA’s considered view of what’s going on - or at least the summary that they think it useful the public should know.1
Was there an explosion in measles cases in late February in England?
Previously, there was a well-reported outbreak of measles cases in Birmingham (West Midlands), which spiked around the New Year 2023, and then receded a little. Then things quietened down in that area, while further clusters of reports (according to Source 1) started to trickle in from the rest of the country.
Then, last week, source 2 - the causative agent report - showed a sudden jump in cases. 140 new ones reported in Week 9 of 2024.
This looked worrying.
This looked like the explosion of cases you might expect from a high R_0 disease like measles if it had found its way into a set of under-vaccinated people (of whom there are worryingly many - not only in East Midlands, where this outbreak had started, but also in cities in the North West and in London).
Next, puzzlingly, the NOIDS weekly report (source 1) failed to show a similar signal (if it were real, you’d expect a bunch of GPs to have turned up cases), and they did not yet show up in the incident report (source 3) either - though it was quite early for this.
And then UKHSA released a new causative agents report (source 2), which was updated to delete almost all of the cases from week 9 and had Week 10 as more or less continuing the previous plateau. This happened without explanation, and both reports are still up (here and here).
However, a few days later, the incident report came out to cover these weeks … and confirmed a reassuring-looking picture of a fall-off:
It included some some eyebrow-raising notes:
The first of these points looks a odd, but it’s what the incident report is supposed to do: to follow up and confirm reports with reference labs. Throwing out fully 34 previously confirmed seems like a lot, … but … OK.
But then there’s an additional 21 individuals who do not fall into this category but were “erroneously reported as measles cases”? Goodness knows what this means.
Of course none of these numbers match the 83 positive measles tests that just vanished from the causative agents report, but maybe - if you allow for multiple tests for each individual - they represent some kind of explanation for most of them?
So there’s some kind of screw-up in lab testing or reporting, which may have somehow given rise to a large number of erroneous reports of lab confirmed cases.
So, we’re fine? The Midlands outbreak is over, and there’s no rise in the rest of the country? We’re out of the woods?
Read the small print
Let’s check carefully in the current report, and the chart from the week before:
While the incident report came out a week after the previous one, it only advanced the reports by two days, from the 4th to the 6th of March. So all the new reports - apparently - came in those 48 hours.
This is unusual. The reports usually advance reports by one week, each week (the one before these two took reports to 28 February for example). But this time it was update covering only updates received in two days. In those of two days, that bunch of reports were removed from the reporting especially in the North West, and the East Midlands numbers continued to come down … but the London ones started gearing up.
To be precise, 19 new reports from London were added. 14 of them in patients under 5 years of age, apparently most of them with sample dates in the last two weeks in February.
If we naively take a “reports coming in per day” then this suggests a big jump up in London. But honestly, I don’t think we can do that. There’s obviously a big set of cleaning, investigating and - by - sound of it - sorting out reporting errors going on. The result is that these two days looks very unusual - we should be hesitant about inferring anything from it. It’s a hint, no more.
And as for why they’d only update two days of reports … this seems to be yet more of an indication that there’s something seriously amiss with the testing, reporting or tracking of cases. Which is not great.
What’s going on - on balance.
The trouble is - from an external point of view - with all this reporting and re-reporting, is remains very difficult to understand what is going on with measles. It appears clear that:
The explosion in tests reported in week 9 of 2024 was not real - it was a mistake, or perhaps a series of mistakes, albeit mistakes that need to remain in the reports for statutory reasons.
There is some kind of issue with testing, or reporting, or de-duplicating of cases, which is impacting numbers, and the ability to update usually regular reporting.
The outbreak in the East Midlands is (probably) not growing further; with new growth in London is being hinted at, but the hints are far from conclusive.
This is of course just a guess from the outside, using only the data that UKHSA are choosing to release. And outbreaks of an infectious disease will be fast-moving - reporting oddities and corrections are to be expected.
The hope is that within UKHSA - and from there, in the public health community - it is much more clear what is happening than the mess that is shown in these reports. I much prefer to believe this - and that they could do a slightly better job communicating it - to the alternative.
Two of these reports cover England and Wales. The third is for England alone. I imagine this is a) for reasons of devolved politics in healthcare and b) to keep people on their toes.
Thanks for doing the deep dive in to the numbers. As you rightly point out the 'devil is often found in the footnotes'.
You may want to consider contacting UKHSA with your findings. In my experience, they have generally responded positively and have even updated subsequent reports based on my feedback. Your input could assist them in enhancing their communication.
In some ways I am surprised given covid and prior downgrading of public health we still have a functioning Public Health community and a sufficiently reactive lab testing system. (NHS testing has been under severe pressure we are told.) Can I imagine kit being pulled out of the cupboard and admin protocols fished from files?
And measles is really dangerous to the youngest - serious stuff.