Is it Safe? On vaccine safety

Lots being done, and you can help.

If you find this post interesting and think others might do so, spread the word.


Safety in vaccines is tricky. Unlike most drugs, we give vaccines to healthy people, including a lot of childhood vaccinations, so any harms of a vaccine are extremely important. They influence the threshold of whether a regulator would licence a vaccine at a population level or choose to take a vaccine at an individual level.

The background rate of disease you might otherwise suffer is also important. Yellow Fever vaccine is mandated in many countries, and currently the only disease for which proof of vaccination may be required for travel, but it is associated with serious side effects with a prevalence of 1 in 100,000 - more common in those over 60 years of age. So, even if you are in good health, if the disease in question is serious enough, and prevalent enough, at an individual level some risk can be acceptable.

In the case of the COVID-19 vaccines, the rare reports of anaphylaxis and more common vaccine side effects like feeling groggy, headaches and sore arms seem a worthwhile risk compared to the risk of COVID-19 and its various outcomes. That’s even before you factor in the more widespread societal benefits of herd immunity and opening up society. I mean, we all want to see that Bond film eventually…

Clinical trials of vaccines can find the frequency of more common adverse effects, and can follow-up on unexpected events that might occur in the cohort months down the line. That said, there have been times when clinical trials have found issues with vaccines (a good example is rotavirus vaccine) that required increased vigilance when a vaccine was released more widely. The fact that the COVID-19 trials had no such safety signals to start with is good news.

It is only after wider use of vaccines in the real world that we can find rare and novel events which might exist. So far in the UK, the data from the MHRA’s Yellow Card scheme surveillance of the COVID-19 vaccines shows that the vaccines are safe. The Yellow Card scheme depends on reporters (either patients, carers, or healthcare professionals) reporting suspected adverse drug reactions via the Yellow Card Website. It’s a quick system, so this is great news that nothing has arisen.

However, this is just one part of the safety monitoring. The MHRA had been planning since May 2020 for the roll-out of these vaccines:

  1. To rapidly detect, confirm, characterise and quantify any risks not detected in clinical trials, and weigh these against expected benefits.

  2. To maintain public confidence by rapidly establishing if any serious events occurring after vaccination are truly casual in nature, and not just co-incidence.

The first point I have already covered, and the second point is extremely important. In any population there is a background rate of events happening all the time. So far there have been over 17 million people vaccinated, mostly in priority groups. Just looking at care home residents alone, it is fair to say there there will have been no end of events which have happened in this cohort of vaccinated people after they received the vaccine. Deaths, new diagnoses, all sorts of rare events will have happened. The key is knowing which of these events is linked to the vaccines, and which are merely co-incidents. Clearly, co-incidental events can be exploited by anti-vaccine groups and state disinformation campaigns (such as those run by China and Russia).

The MHRA are using 4 main approaches to safety monitoring.

1. Enhanced Passive Surveillance

This is essentially the Yellow Card scheme. I don’t like to call it passive myself since, spontaneous reporting depends on active efforts of reporters, and happens in real-time (unlike pharmacoepidemiology studies which retrospectively mine databases which have lain passive until they are looked at), but that’s just me. I wish we didn't undersell spontaneous reporting with the word passive. I think we use the word passive entirely the wrong way round in this respect.

They have enhanced the Yellow Card Scheme by tying in background rates of events from the Clinical Practice Research Database (a database created from GP electronic healthcare data) and having good information on how many people and which type of people have been vaccinated.

Let’s imagine a number of cases of people developing a totally imaginary ‘blue nose” adverse effect which is reported on Yellow Card reports. The MHRA can go away and see how many cases of ‘blue noses” turn up naturally in the CPRD. If the Yellow Card “blue nose” cases are larger than might be expected given the naturally occurring frequency of ‘blue noses’ in the CPRD, then it would be more likely it would be a real effect of the vaccine. Otherwise, it might be just a co-incidence. Those people were going to get blue noses anyway. So this observed versus expected is really useful on sifting out real signals of safety from the chaff of false signals. The MHRA also have to take into account the under-reporting in the Yellow Card scheme, which is common.

2. Electronic healthcare record surveillance

So, this is slower than Yellow Card data, but doesn’t require people to report a suspected adverse effect to a vaccine (so avoids the risk of something not being reported). As your vaccination status is updated in your GP record, the MHRA will start to run Rapid Cycle Analysis, which will be looking for theoretically possible adverse events each week. They can choose new types of possible adverse events to look as they go forward (so if they had reports of ‘blue noses” from Yellow Cards, they could start to look for those). This system should also be even better at filtering out false signals. The MHRA will also be able to look at events in large population cohorts based on the prioritisation groups, they can then compare events that happen pre-vaccination and post-vaccination (this is called an ecological analysis).

3. Targeted active monitoring

The MHRA are also going to active go looking for certain groups of vaccinees, especially those groups of patients who were not recruited into the clinical trials. A random number of people will be selected (about equivalent to a clinical trial sized group) and they will be contacted to see if they have experienced an adverse drug reaction at set intervals of time. This won’t find rare events (for the same reason clinical trials don’t, so no “blue noses” here), but it should lead to a comparator group of patients who didn't fit in the clinical trials (like those who are immunocompromised - who would have been excluded from clinical trials to test vaccine efficacy for obvious reasons). This will add additional safety information for those groups.

4. Formal epidemiological studies

These are the standard safety studies run with drugs and vaccines after marketing. They tend to be used when you think you have a safety signal. “Hey, we have some sort of signal about blue noses, let’s set up a pharmacoepidemiogical study to see if this is real or not.”

The MHRA have a lot of resources to run these, but they can also be set-up with academic units and other organisations who are specialist at running these sorts of studies.

If you want to know more about the MHRA’s plans, all of the above is in greater detail at the MHRA website here. The MHRA have put a lot of thought, planning, and effort in trying to ensure they can find any adverse effects to the COVID-19 vaccines. Firstly, we need to be able to find any such effects, if they exist, in order to minimise their effects (by finding which groups are more at risk). Equally importantly, this helps ensure that there is public confidence in the vaccines and the monitoring, and that if there are new adverse effects we are actively looking for them, and able to counter any incorrect information about adverse effects that may arise (for example from some patient anecdote in a newspaper or social media).

If you have had any of the COVID-19 vaccines, and are in the UK, you can report any adverse effects you experience here on a Yellow Card.

You can find the weekly COVID-19 vaccine adverse reactions page here.

How else can you help find adverse effects to the COVID-19 vaccines?

The ZOE app

Those using the ZOE app, as part of the COVID symptom study can log their vaccines in the App. They have reported fairly mild short-lived reactions found in the clinical trials, and more symptoms after the second dose (as you would expect).

After getting the Pfizer jab, around four in ten people experienced some local after effects, including pain and swelling on the arm where they received the injection. 

Around one in seven also recorded systemic after effects including headache and fatigue, which are in line with the after effects reported in the clinical trial. The reported after effects were short-lived, lasting just a day or two, and we haven’t seen any evidence of longer term health effects.

Interestingly, people recorded more symptoms after the second dose of the vaccine compared with the first. And people who had previously had COVID-19 were also more likely to have after effects following a single jab, compared with those who had never had the disease.

The DSRU study

The Drug Safety Research Unit (DSRU) are running the Active Monitoring of the Safety of COVID-19 Vaccines Study.

The study will recruit at least 10,000 people in the Phase I vaccination group, which includes all those aged over 50, health and social care workers and people who are clinically extremely vulnerable.

The study will use short questionnaires which include questions about the vaccination and your health and any symptoms/conditions following vaccination.

Register your interest here. They are specifically looking for those who have had the Oxford/AZ vaccine at the moment.

So there’s a lot of work being done on safety with COVID-19 vaccines, which should be massively re-assuring. So far, they appear to be hugely effective, even with one dose, and reduce transmission. There is a light at the end of the tunnel.

Quick bits…

Bad-mouthing of the Oxford vaccine by EU politicians comes back to haunt them.

Single dose Pfizer-BioNTech vaccine has 92.6% efficacy, which backs up the JCVI decision. NEJM.

A 3 month gap between AZ/Oxford vaccine doses is better then 3 weeks, which backs up the JCVI decision. The Lancet.

Also big contributions from the EU, the US, and Germany to COVAX. Most of the UK surplus will go to the poorest countries.

The best way to reduce the chance of variants arising is to vaccinate everywhere as quickly as possible and drive down transmission. It’s entirely reasonable that national governments consider their citizens first—that’s realpolitik.

Once the UK has vaccinated our most vulnerable communities and healthcare workers, however, we should make vaccines available to other countries. It’s enlightened self-interest, as well as the right ethical thing to do.

Jeremy Farrar in the BMJ.

The British public agrees.

That’s all for this week and stay safe.


Don’t forget to report your suspected adverse effects from medicines and vaccines. In the UK, this means using the Yellow Card Scheme.