It’s always nice to have your predictions come true. Or is it? Here’s what I wrote in the New Statesman on 2 December last year:
“We can be absolutely certain that some people who’ve taken the vaccine will later develop health problems – even if these have nothing to do with the vaccine. If you vaccinate this many people, many of whom would go on to get ill no matter what, this is guaranteed. And yet, inevitably, the anti-vaxxers (and some aspects of the media) will seize on such cases… likely prolonging the pandemic.”
I was correct that people would make this mistake. But I completely failed to predict who would be making it: not just swivel-eyed anti-vaxxers, not just irresponsible journalists looking for clicks, but most of the major European governments.
As I write this, Germany, France, Italy, Spain and 11 other European countries have temporarily suspended the use of the Oxford/AstraZeneca Covid vaccine (others, including Greece, Belgium and Czechia have not so far). In Germany, the specific reason given was that several cases of cerebral venous sinus thrombosis – a blood clot in the brain that produces similar symptoms to a stroke – have occurred in people who recently had their shot. Since this is a rare condition, Germany is pulling the plug.
Clearly the German (and other European) regulators think the logic I describe in the quote above doesn’t apply; they think they have more blood clot cases than they would see in a world without the vaccine. But it all depends on the denominator.
Jens Spahn, the German health minister, said seven cases occurred among the 1.6 million people in Germany who’d had the AstraZeneca jab, suggesting that figure is unexpectedly high. This is true in a sense, but is also quite misleading: 17 million people in the EU and UK (mainly the latter) have now had the vaccine, and there have been less than 40 such blood clots reported in total. It might even be that the total number of blood clots is lower than we’d expect ordinarily.
But even if the numbers were higher than expected, there’s another denominator to consider. Because the appearance of a wide range of different adverse reactions, not just blood clots, would have tripped the alarm, there are many more chances for a false-positive and to see correlations that do not exist. Whereas it might be rare for a cluster of blood clot cases to appear, it’s less rare for at least one of the following to do so: a cluster of blood clot cases; a cluster of allergic reactions; a cluster of fevers; a cluster of heart attacks… and so on, through a long list of possible conditions.
As the medical scientist Paul Hunter points out, this problem complicates the usual reasoning about probabilities, since it multiplies the likelihood that at least one rare adverse event will occur in a vaccinated group.
We should also zoom out and look at the wider context. In the UK we’ve seen more notifications of blood clots after the Pfizer vaccine than the AstraZeneca jab (after roughly the same overall numbers of doses given) – but no more than the normal level in the population. And even in cases where we do think a medicine increases the risk of blood clots – as, for example, the contraceptive pill appears to by around two to four times – we weigh those problems up against the benefits.
You don’t have to be a hardcore utilitarian to understand the balance of harms. There are double-figures in the whole of Europe for reports of blood clots, against a background of rising Covid cases in some European countries – and the potential hundreds or thousands of deaths they will lead to. Each day we do not vaccinate the population adds more potential deaths in future. Suspending vaccination on such shaky statistical grounds – and stoking fear in the minds of many about the safety of the jab – does terrible damage to our efforts to end the pandemic.
AstraZeneca deserves some of the blame. Unlike Pfizer and Moderna, the company failed to include very many older people in its trials, opening up an avenue of criticism that several European leaders have happily followed. Not only that, but its vaccine research has been dogged by mistakes and confusions. It’s little wonder that the company’s product is regarded with some suspicion. Happily, it seems the entirely new trial AstraZeneca set up in the US (in large part to try to iron out those confusions) is nearly ready to report its data.
Which raises a final point: although the observational data from across Europe is important, we should weight the evidence from randomised controlled trials much more heavily. In observational data, we might, for example, see higher rates of apparent side effects simply because older, less healthy people tend to get the vaccine first, and the “adverse” effects are due to their being sicker in general, not due to the vaccine. In a trial, randomisation means such biases won’t occur: whether people are given the treatment or the control depends on a random number generator, rather than any personal characteristics.
The new AstraZeneca trial will be instructive, but we can look back at the old ones too: in these controlled circumstances, the tiny number of blood clots showed few differences between the vaccine and the placebo group (again, if anything, the placebo group had more of them).
In other circumstances, the calculation might be different: we might want a lower bar to suspend a treatment due to adverse reactions. But the situation is dire. Not only are cases creeping up, the EU has also been struggling to get shots into arms. After this new, unforced error, that struggle is going to become all the more difficult.