The joint winner of the HealthWatch Award 2021 was Christina Pagel, Professor of Operational Research, Clinical Operational Research Unit, University College London. Presenting the award, Nick Ross said, “How politicians reach their views on healthcare is of huge importance, especially at the moment. Christina Pagel has numeracy, literacy, an understanding of risk, a managing of uncertainty, and a disarming way of taking things that to many of us are extraordinarily complex and finding ways to unravel them and make them seem simple as daylight.” Professor Pagel shares the 2021 award with David Spiegelhalter.

The following is a lightly edited version of her talk, given at the HealthWatch AGM held at the Royal Society of Medicine, London, on 6 October 2021. The full recording of the HealthWatch Annual General Meeting 2021 and awards, including this presentation, can be experienced on the HealthWatch YouTube channel

DSC09828

Independent SAGE

I was asked to join Independent SAGE when they started back in May 2020, based on my expertise in mathematical modeling and decision making. But it quickly became clear that was not what I was going to be doing. We have some of the world’s most brilliant infectious disease modelers already in this country. What I felt I could do, was to talk to the public, to try and communicate the data, and to show them how different contexts, demographics and policy, affect the data.

Since last summer we have been giving weekly briefings to the media. We still get thousands of people tuning every week and I hope it is making a difference. COVID data has inundated our daily lives, and we are all used to looking at cases, hospital admissions, deaths, and vaccinations. But the numbers don’t speak for themselves. So, what do you need to know to interpret that data?

COVID data – what lies beneath

To know what exactly we are measuring, we need to understand how this disease progresses.

With COVID, you are infected on Day 0. You might become infectious to people 3-5 days after that. But it is not until days 5-7 symptoms that may appear, and that is when you first appear in the data on numbers of cases – as long as a week after you get infected, and then only if you get a test. If you don’t get symptoms, or don’t recognize them as COVID, you won’t get tested. It also relies on you knowing how to get a test, wanting to get a test, and having time to get tested, and having agency to act on a positive result. These are all biases that affected data on case numbers.

This has changed since we started doing asymptomatic testing, particularly in schools, although the accuracy does depend on whether the test is being done properly.

The NHS website offers a free PCR test which is the gold standard to check for coronavirus. It is offered to those with symptoms of a high temperature, a new, continuous cough, or loss of sense of taste of smell. But these are no longer the most common symptoms of COVID. The ZOE symptom tracker app lists the top five symptoms in children and vaccinated adults as being runny nose, headache, sneezing, sore throat, and loss of smell. This has not been communicated at all, and is affecting people’s likelihood of seeking testing.

Ten to fourteen days after infection, if it gets worse and you are admitted to hospital, that is when you show up in hospital data. The advantage of this data is, it doesn’t rely on your testing behaviour. But it does skew to an older, sicker and now unvaccinated population. So, while hospital data tells you something about the burden of COVID in your community, it misses a lot of cases. Last year, hospital admission data missed the rise in cases among students, and only showed later as their infections spread to the older people.

Finally, maybe three weeks later, someone might die if they get sick enough. It takes time for deaths to be registered, which is when they show up on the dashboard. All countries have good death registries but the data takes a very long time to come through, which makes daily deaths a very poor indicator of the state of a pandemic and, again, it skews to the oldest and sickest and unvaccinated.

Finally there is long COVID, with symptoms usually 12 or more weeks after the initial infection, and can occur whether you’ve been asymptomatic, mild disease, or hospitalized. It’s common, it is really hard to measure, so we don’t measure it routinely. Which means that its burden is not feeding into policy decisions.

Age, ethnicity and geography

Contexts and demographics are important. Take age. Pre-vaccination, hospital admissions and deaths were very much concentrated in older people. It is that relationship that informed the vaccine priority rollout, which had a really great impact protecting the elderly.

This summer, as the school term ended, the Delta variant was running rampant, with very high case rates among school children, and through the summer among young adults as festivals and nightclubs opened. Currently we have high case rates among 10 to 14-year-olds, and among the 40 to 50-year-olds – their parents.

DSC 0133We have all been affected by COVID, but we haven’t all been affected equally. The most deprived communities have suffered disproportionately, with a combination of workplace exposure, and more having to work outside the home or in public-facing jobs. They are less likely to be able to isolate, living in overcrowded housing and without access to green space. They are more likely to get it, more likely to be hospitalized, more likely to die or get long COVID.

They are also less protected – there is a 20 per cent difference in the vaccination rates between the most and least deprived populations. Looking at ethnicity, over 90% of white English over-50’s are fully vaccinated, compared with about 65% of black English people.

When we compare ourselves to other countries, here in the UK we have persistently had a much higher level of weekly confirmed cases than the rest of Europe. Our vaccination levels are the same or not much lower than other European countries. Yet some of our most similar neighbours have much lower case rates, and many have not seen a back-to-school spike. That is partly because they have vaccinated their teens over the summer, but it is also because of “vaccination plus” – they still have mask mandates for indoor spaces, many require COVID passes for entry to crowded venues. We should look around other countries to see what is working.

COVID data – interacting contexts

Look at what is happening around England. The North, the Midlands, and particularly the North East have consistently had higher case rates and hospitalizations than anywhere else. The impact of geography and deprivation on the disease is toxic politically, especially considering the regional restrictions last year.

Care homes is another really tragic interacting context. Our most vulnerable populations, who bore such disproportionate numbers of deaths are being cared for by people from our most deprived populations – many could not afford to isolate if infected because they got no sick pay. People who might work agency shifts, working from care home to care home, spreading COVID very effectively. SAGE did not understand this initially when they were modelling it, so could not foresee how to protect care homes during wave one.

The last context I want to talk about is hospital admissions. Although we’ve had quite a high burden this summer, it is nowhere near where we were last year, and that is entirely down to the power of vaccination.

But hospital admissions for children, are now as high as they have ever been. Cases stayed low until the Delta variant came in during the summer, and mask mandates were removed. We have had over 4,000 under-18s admitted to hospital with COVID since 1 May 2021. So, for children, the risk right now is the worst time of their pandemic. Even though, their risk is low compared to adults, I don’t feel that we should forget that.

Final thoughts

It has been a strange year and a half, life-changing in some ways. I never expected to be working other than behind the scenes. Mostly it has been more public.

But the principles I have used in my “day job” have carried over into my role in Independent SAGE. I do a lot of looking at data in hospitals, communicating it to clinicians and also to patients and the public. And I’ve directly taken the learning from that into this role, but I’ve also learned so much about, what are the pitfalls in COVID data? What is important to understand? Where is the story hiding in the data? How much detail do I need to tell people?

A running theme has been combining knowledge, evidence and expertise from as many different areas to try to make sense of the whole, and then communicate it as honestly as I can, the good news as well as the bad.

Data is not neutral, understanding is hard, and you have to make decisions in an uncertain situation. For example, on child vaccination, the government’s Joint Committee on Vaccination and Immunisation may have wanted to wait for six months for more safety data. But, by that time, most children will have been infected.

It has meant a huge amount of work on top of my day job. It has changed how I interact with people online, I have a lot more Twitter followers so if I make a mistake it will be very public.

But it has been a privilege and a responsibility and one I take very seriously.

DSC 0122