Skip to content

Suspending The Astra-Zeneca Vaccine and The Ethics of Precaution

By Jonathan Pugh, Dominic Wilkinson, and Julian Savulescu

The authors are working on the UK Pandemic Ethics Accelerator project – @PandemicEthics_. This project was funded by the Arts and Humanities Research Council (AHRC) as part of UKRI’s Covid-19 funding.  All authors are affiliated to the University of Oxford.


Summary Points

  • Preliminary Reviews suggest that the number of thrombotic events in individuals who have received the Astra Zeneca vaccine is not greater than the number we would normally expect in this population.


  • It is crucial that we closely monitor these adverse events. The regulation of new medical interventions always requires us to manage uncertainty.


  • A precautionary approach to managing this uncertainty may be important for ensuring continued confidence in vaccination.


  • Regulators must weigh the potential risk suggested by these reports of adverse events following vaccination against the harm that suspension of the vaccine could have.
  • The harm of suspending the use of the Astra Zeneca vaccine depends on how many preventable deaths we can expect by suspending its use.


  • Amongst other things, this will depend on (i) how many people will be delayed in receiving a vaccine as a result (ii) the mortality risk of the people who would be prevented from receiving a vaccine, (iii) the prevalence of the virus at the time of the suspension, and (iv) the number of people who have received one dose of the Astra Zeneca vaccine, but not a second.


A number of European countries have suspended the use of the Astra Zeneca vaccine following reports of individuals experiencing thrombotic events after receiving the vaccine. The European Medical Association’s (EMA’s) preliminary review on the 10th March suggests that these events have included cases of pulmonary emboli (including one reported fatality), and deep vein thrombosis.

The EMA’s preliminary review goes on to claim that there is no indication that the vaccine caused these conditions, and that the “number of thromboembolic events in vaccinated people is no higher than that seen in the general population”, citing 22 cases of thromboembolic events among the 3 million people vaccinated with the AstraZeneca vaccine in the European Economic Area as of the 9th March.

More recently, the BBC reports that, as of the 16th March, there have been 37 reported thrombotic events amongst 17 million people that have received the vaccine in Europe.


Contextualising The Incidence of Thrombotic Events


The fact that thrombotic events have occurred in individuals who have received the vaccine does not mean that the vaccine caused these events. Correlation is not the same as causation.

Accordingly, it is important to place the reported thrombotic events into context, and to think about how many thrombotic events we might normally expect in a population of a comparable size to the population that has received the Astra Zeneca vaccine.

Although the incidence of venous thrombosis is very rare in young individuals (<1 per 10 000 per year), it has an incidence of around 1% per year in the elderly. This suggests that in a population of 17 million people young people (i.e. those with the lowest risk of thrombosis), you would normally expect around 1700 cases of thrombotic events.

Since aging is one of the strongest risk factors for venous thrombosis, the number of cases you would expect in a population of this size will increase sharply as the age of the population increases. According to the figures in this study, you would expect around 170’000 cases in a population of 17m elderly people.

The first Astra Zeneca vaccine was given to a patient on the 4th Jan 2021, and there have been 37 reported cases of thrombotic events, as of 15th March (70 days later).

Using these figures, if we extrapolate the 37 reported cases of a thrombotic events in a 17m population over the course of 70 days, that gives us an expected yearly rate of around 193 cases in a population of this size. This expected yearly incidence would fall a considerable way below the number of thrombotic events that you would normally expect in a population of this size, according to rates indicated by the study cited above. This is so even if you assume that the population were wholly constituted by individuals in the lowest risk age band.

In addition to the fact that there is not a clear mechanistic explanation as to why the Astra Zeneca vaccine would cause thrombotic events, the above considerations give us some reason to doubt that the vaccine is directly causing these incidents.

However, that does not answer the question of whether we should continue to use the vaccine. How we should manage these adverse events, and the uncertainty they engender, is a complex ethical question.


Managing Uncertainty


It is imperative that any vaccine approved for use meets standards of safety and efficacy. In approving it for use, regulators have already been satisfied that the Astra Zeneca vaccine meets these standards. However, since this is a new vaccine without a long-term safety record, it is very important that its performance is closely monitored. When we observe serious adverse events in the vaccinated population, it is important to establish that these events were not directly caused by the vaccine. This is why the EMA and WHO are currently reviewing these adverse events.

The regulation of a new medical intervention always requires us to manage uncertainty. We have written elsewhere about the management of the uncertainty regarding the efficacy of the Astra Zeneca vaccine earlier in the pandemic. In the current context, we now have to think about how to manage uncertainty regarding a potential safety concern with the vaccine. But in both cases, the following is true: the more evidence available, the greater the certainty that regulators can have that a vaccine is safe and that it works. But gathering evidence takes time. This means, that the higher the standard of evidence we require, the greater the delay before people can access the intervention.

One natural response to the uncertainty prompted by reports of adverse events is to adopt a precautionary approach – it is better to be safe than sorry. The precautionary approach calls for suspending the use of the vaccine until we gain a rigorous understanding of these 37 reported thrombotic events, so that we can be absolutely satisfied that they have not been caused by the vaccine.

An important reason that regulators might adopt a precautionary approach is to ensure that these reported events do not unduly undermine confidence in the vaccine. Making ‘safety’ an absolute priority in the response to these adverse events will help the public to have confidence in the vaccines that are being used in the pandemic response.

It is perhaps also tempting to think that the precautionary approach is the right one just because it ‘puts safety first’ – perhaps ‘you cannot be too careful’. In the current circumstances though, this is not obviously true. In a pandemic, time is lives; a week spent unvaccinated is a week spent living with a higher mortality risk of COVID-19. You can be too careful in minimising one kind of risk when doing so involves leaving people exposed to a much greater risk.

Accordingly, it is crucial that regulators weigh the potential risk suggested by reports of adverse events following vaccination against the harm that suspending the use of the vaccine could have.

The risks of suspending the use of the Astra Zeneca vaccine depends in part on how many preventable deaths we can expect by suspending its use. A number of factors in addition to the efficacy of the vaccine will influence this, and many of these factors will differ across countries.


Preventable Deaths


The first factor is how many people will be delayed in receiving a vaccine as a result of the suspension. Fortunately, the Astra Zeneca vaccine is not the only vaccine available in Europe, so its suspension will not wholly prevent vaccination; however, it is possible that at least some people’s vaccinations will be delayed by a suspension of the Astra Zeneca vaccine.

In the UK, by the 16th March, 1,610,280 people have received two doses of an approved COVID-19 vaccine, and 24,453,221 have received one dose. This still leaves a large number of people who have not yet received a vaccine in the UK. In other European countries, fewer people have been vaccinated.

The second factor contributing to the number of preventable deaths a suspension would cause is the mortality risk of the people who would be prevented from receiving a vaccine for the suspension period. For example, in England, those aged 56 and over are currently being invited to book appointments for vaccination. A recent study suggests that the average risk of death for (unvaccinated) 55-59 year-olds infected with coronavirus is 0.323%. So, suspending the use of the vaccine would be delaying protection for a relatively low risk group. But even preventing these individuals from receiving a vaccine would still translate to some preventable deaths. In the week ending 5th March 2021 alone, ONS data suggests that there were 53 deaths that included COVID-19 on the death certificate in this age-band alone in England and Wales.

Notably, in countries that have not yet vaccinated older age groups, the risks of a suspension in this regard will be higher. The study cited above suggests that (unvaccinated) 70-74 year olds infected with the coronavirus have an average risk of death of 1.674% – for those over 80, the risk is 8.292%.

The risks in the above passage concern an infected individuals’ mortality risk. A third factor that is relevant to calculating the number of preventable deaths is the prevalence of the virus at the time of the suspension. In times of higher prevalence, we can expect a larger number of infections and therefore deaths. According to the latest data release, the ONS estimates that there is currently a low prevalence of the virus in England, around 1 in 270 (0.37%). However, in countries with a higher prevalence, suspension of the vaccine would lead to more deaths. Crucially, there is a currently a wide prevalence range in countries across Europe according to data from the European Centre for Disease Prevention and Control, with prevalence rates ranging from 4 to 1’572 per 100’000 people (0.004% – 1.572%).

A final factor to consider is the number of people who have received one dose of the Astra Zeneca vaccine, but not a second. The reason for this is that, depending on its length, a suspension of the vaccine may extend the dosing period for these individuals, potentially diminishing the effect of the vaccine for some of them. The UK government figures do not provide a fine-grained breakdown of the data in this regard, but the figures suggest that 22’842’941 people have had one dose of an approved COVID-19 vaccine, but not a second.


Share on

8 Comment on this post

  1. Alberto Giubilini

    Thanks for this. Very helpful to make some clarity in the current discussion and to put fears in the right perspective.

    It would also be relevant to compare any risk of serious adverse effects of the AstraZeneca vaccine with the risks of serious side effects we consider acceptable in the case of other vaccines (including the other approved COVID-19 vaccines). This is never 0. No approved vaccine has absolute 0 risk. We always accept at least some very small risk, so even if there was a causal correlation here, the question would be if that level of risk is acceptable, given what we normally accept with other vaccines or drugs (e.g. birth control pill has some small risk of thrombosis). There does not seem any good reason to keep a COVID-19 vaccine, or the AstraZeneca vaccine among the alternative ones, to a higher standard (if anything, given the global situation, we might want to accept a slightly higher riks).

    The other relevant thing is that the data mentioned here refer to thrombotic events in general, but Germany has identified a specific cerebral venous thrombosis that has a higher prevalence than normal after the AstraZeneca vaccine: 7 cases in 1.6 million doses, 6 of them in women, though not all of them fatal. But even in this case, assuming there is a causal link, we would be talking of a 0.0004% risk (if I am doing the maths right). So the question again is whether this is enough to warrant the suspension, given the level of risk we would inevitably have to accept with any vaccine or drug.

    The question of whether these cases warrant investigation is not the same as the question of whether they warrant suspension. Side effects are constantly monitored and can be assessed without having to suspend the roll out. Investigating them is certainly warranted. But given the extremely low rate of such side effects, the suspension might do more harm than good, including unnecessarily increasing people’s hesitancy.

    1. Alberto Giubilini

      this for example is one consequence that probably should have been given more consideration before suspending it UK patients cancel AstraZeneca Covid jabs as Europe’s mixed messages sow confusion:

    2. There’s a low risk of death immediately following the vaccine, but it may be worth suspending and doing extra testing to see if there are as yet undiagnosed harms being caused- ie if these are just the tip of the iceberg. You don’t want thrombotic events to suddenly increase in 6 months time for example. (Not a doctor so this is speculative).

      Also, a very small risk is still relevant. If you are 20 and already have COVID you have a very low risk yourself from COVID and are vaccinating as a duty of easy rescue to others. But a risk of death however small starts to make the rescue way less easy. Whereas at least the risk of thrombosis from birth control is lower than the risk of thrombosis from pregnancy- not to mention other risks in pregnancy. So even though both are low risk, their benefit profile to certain individuals is VERY different.

      1. Alberto Giubilini

        I agree on both points, but I am not sure they warrant the suspension. As for the first point, if it takes 2 days to investigate the matter but at least 2 weeks to catch up with delays caused, the low risk suggests that it might be better to have the investigation without the suspension. So it would not be 6 months.
        As for the second point, that is true, but European countries still have to vaccinate the very elderly. Any risk of that size is probably worth taking for many of them. So maybe a suspension only for the young would have been better.

      2. There is a real need to stop repeating the assumption that young people and children are simply “low risk”.

        In reality a lot of young and even athletic people have developed heart or lung changes that could be seen on a CT or Ultrasound after an asymptomatic case of Covid-19.

        Then you have the issue of “long Covid-19” where a notable minority of people who suffer from mild to moderate Covid-19 and who were never hospitalized simply fail to recover and continue to suffer for months with non-life-threatening but disabling symptoms. If this is the same condition as the long mysterious CFS/ME as some people suspect, then some of them might remain alive but very disabled or largely bedridden for decades. This “Long hauler” syndrome seems to affect a non-rare minority of not just younger adults but teenagers and children.

        There have also been reports of bizarre side effects in people who were not severely ill nor high risk ranging from kidney damage to Alzheimer’s like memories problems, or stuttering (whey they’d never had that problem before or as kids), or psychotic episodes.

        Even some not-rare side effects of mild “covid-19” such as “loss of taste” or smell can be not only a major problem for somebody’s quality of life, some neuroscientists worry it can be a portent for “problems down the road” such as early onset dementia or development of mental illness in the genetically predisposed. Again, not just adults of all ages, but loads of children have lost their sense of taste from the virus-often in people with no other symptoms.

        Finally, they’ve seen that scary “Multisystem Inflammatory Syndrome” in some children with both symptomatic and asymptomatic Covid-19.

        In short, this upfront-assumption and relentless repetition of the idea that kids and young people can just get Covid-19 and it won’t be that big of a deal really needs to stop.

      3. The assumption that anyone in their 20’s would only get the vaccine as a “duty of easy rescue to others” is a mistaken one. First of all, as I’ve outlined while it’s true a slim minority of younger adults and kids have been the ones dying in the hospital or getting hooked to ventilators, it still is probably a higher risk than with the thrombotic events. Plus younger adults and children who have developed Covid-19 while unlikely to be hooked to a ventilator or die of acute Covid-19 may still face serious risks of becoming “long haulers”, since many people who have become long-haulers, lost their sense of taste or smell, or have shown up with lung, heart or kidney damage despite mild or no symptoms are young adult or even kids.

        Also since thrombotic events increase with age, it is not unlikely that younger people may be less likely to experience that side effect from the vaccine as well.

        Then if you want to keep the comparison to oral contraception in this, there’s a parallel argument. The risks of thrombotic events have been accepted even when options such as condoms, sponges, or abstinence exist. Similarly even if we were to “roll with” the flawed assumption that young adults and kids face no real dangers from Covid-19 itself, they can experience very real lifestyle, social, education, and economic benefits from vaccination.

        This can include ability to go back to work, ability to attend school from early grades, Kindergarten or university up through college/university, ability to have a normal social life, ride on public transit, visit grandparents, parents, or other older relatives and so on.

        1. You have misrepresented my comment: I specified a 20 year old who has already had COVID, not just any 20 year old or child/ young person. The risks you outline have already been taken on by that person, there is no way to undo it. But the evidence is good that natural immunity works well, especially in the young. In fact, it has been more stringently tested than some vaccines, since the Public Health England study included for asymptomatic infection, (there were a few, but many fewer than the control group) many of the vaccine trials only tested upon symptoms. The pre-infected young group is the group who are then asked to have a vaccine as a part of doubling down on non-transmission but can expect virtually no proven personal benefit, though there may be a theoretical benefit (if we think vaccine immunity will last longer, or be more complete, though there is not much evidence that this in fact is the case yet).

          Moreover, you are comparing the risks of a disease which has been studied extensively for a year now, with a vaccine which is just being released into the field. Noone has scanned the hearts of people who have had the vaccine and had no ill effects- why would they? Likewise, just as there are a range of bizarre cases following COVID, a range of bizarre symptoms have been reported after vaccine that may or may not (likely not of course) be linked to the vaccine. My (poorly explained) concern about the current state of knowledge around thrombotic events is that the reports of immediate and catastrophic clots may simply be the outliers- those with the most severe responses. Maybe there is a more common but less severe thickening effect that will not show up symptomatically until later on. As far as I know, there is no systematic checking other than self- or doctor- reporting of suspected linked events.

          It’s true that there are alternatives to both pregnancy and the pill, and indeed many people DO choose not to take it because of its side effects, including thrombosis. In this case, it is fair to offer the vaccine with the information about the thrombotic risk, I agree. But at this stage, we don’t have much information about what that risk really is. We can continue to be quite confident for some groups that it is safer than the alternative. But can we be sure about young people with pre-existing natural immunity?

          It is clear that eliminating COVID is best for individuals, and best for society. I don’t intend to downplay the benefit of that. But equally, I think pretending we know more about the risks of the vaccine than we do to encourage those who have less reason to fear COVID than others to take it is bad for trust in the longterm.

  2. The authors overlook two important elements. The first is that the reported adverse event rates come from adverse event reporting systems, similar to the yellow card system used in the UK. These are known to significantly under-report adverse event rates, yet the authors seem happy to accept them as conclusive epidemiological rates and compare against epidemiology from well conducted studies. This is a dangerous assumption. The second is to limit the perceived risk to the specific case type reported. If this was a vaccine related event affecting the vasculature, there could also be an increase in cases of myocardial infarction, stroke and many other conditions which are less distinct in their aetiology and generally more common, therefore more likely to be missed or assumed to be ‘common or garden’ cases.

    A safety surveillance exercise for an adverse event signal seeks, among other things, to identify the true rate of the event, which cannot be done relying on adverse event reports, but also to look for other adverse events that may reasonably occur in other body systems.

    It is a fine line between the benefit-risk assessment presented and the question ‘Why did you continue to allow use of this medicine/vaccine before underatdning the magnitude or extent of the broader risk?’, and much easier to judge looking back than forward in time.

Comments are closed.