It didn’t take long. Less than 24 hours after the government announced it was to use an algorithm to help refine the Covid vaccine priority list, a healthy 32-year-old revealed he had been offered a jab because his medical record said he was just 6 inches tall.
“I was feeling weird about why I’d been selected ahead of others so rang GP to check,” tweeted Liam Thorp, political editor of the Liverpool Echo. “Turns out they had my height as 6.2cm rather than 6 ft 2, giving me a BMI of 28,000.”
The tweet quickly went viral, attracting over 100,000 likes and hundreds of replies, the best among them a suggestion that Mr Thorp is re-titled “political editor of the Lilliput Echo”.
To be clear, the error had nothing to do with the government’s new algorithm, which will only come into play over the next few days and weeks. But it does underline the fact that algorithms of any sort are only as good as the data they feed on.
So, how good is the new system being used by the NHS? And how worried should we be that it might all unravel, like the GCSE and A-level algorithms did last summer?
We can take heart from the fact that the new vaccine algorithm was overseen by experts, rather than a troubled quango and a former fireplace salesman. The QCovid Risk Assessment was developed by researchers at the University of Oxford, at the behest of the Chief Medical Officer Professor Chris Whitty.
The aim was to create a tool which could automatically trawl our medical records and other datasets to accurately predict who had the greatest risk of being hospitalised or dying of Covid-19. To do this, researchers used data from six million patients to assess what combinations of conditions – from obesity to heart disease and cancer, along with risks such as age, ethnicity and deprivation – were most likely to result in hospital treatment or death as a result of the disease.
A detailed peer-reviewed paper published in the British Medical Journal in October found the algorithm worked. In the top 5 per cent of patients with the highest predicted risks of death, the tool was found to be 75.7 per cent accurate. People in the top 20 per cent of the same metric accounted for 94 per cent of all deaths from Covid-19.
The system has been further honed and validated by, among others, the Office for National Statistics. You can access a version of it here to calculate your own score. I put in my height as seven centimetres and it very wisely refused the entry, advising that it should be in the range “140 to 210” centimetres.
In a letter sent to GPs on Tuesday, the Department of Health and Social Care said the algorithm had identified an additional 1.7 million vulnerable people who would now be added to the shielding list. Of those, 820,000 are under the age of 70 and have not previously been prioritised for vaccination.
This group will now be moved to the top of vaccine priority group six (those aged between 18 to 69) and notified and they will receive a priority invitation for vaccination. “High-risk flags” for these patients will also appear on GP systems. They will be reviewed locally to weed out any obvious errors.
“I think the algorithm will work well on population health basis to identify patients who may otherwise not have been picked up but there will always be outliers or errors,” one GP told The Telegraph yesterday. “Hopefully the review process will mean that the bulk of those are picked up.”
So why is the new system being introduced now? Age alone correlates well for Covid risk, but its power fades as you come down the age range. At the same time, a host of other factors, medical and social, take on greater relative importance.
For GPs, this presents a logistical problem. Trawling through thousands of patients records one-by-one is a time consuming and difficult task. The algorithm can do it automatically and so will save many millions of hours of work.
Prof Whitty is also reported by the BBC to be concerned GPs have not been using the flexibility granted to them by the Joint Committee on Vaccination and Immunisation (JCVI) to add obviously vulnerable patients to the vaccine priority list where necessary. GPs counter that such flexibility has never been explicitly granted, and that all the communications they have received from NHS England essentially tell them to “stick to the rules, or else”.
Either way, is it vital all those most at risk of Covid-19 are vaccinated quickly. The ultimate aim of the campaign is to reduce hospitalisations and deaths, so the threat of the NHS being overwhelmed fades and society can gradually start to be opened up again.
Even small percentages of people missed add up to very large national totals. If the new algorithm works to bring that number down – and most experts expect it will – it can only be a good thing.
Protect yourself and your family by learning more about Global Health Security