This study investigates real-world data of all vaccinated individuals in Scotland up until 14 April 2021 and examines the association between receiving the first dose of the AstraZeneca/Oxford and Pfizer/BioNTech vaccines and potential rare adverse effects related to clotting and bleeding disorders.  The paper suggests a possible link between the AstraZeneca/Oxford vaccine and the development of very rare cases of idiopathic thrombocytopenic purpura (ITP), an autoimmune disorder characterised by low platelet counts.

“The occurrence of ITP after the first dose of the AstraZeneca/Oxford vaccine appears to be extremely rare – around one in 100,000 – and, as with other potential side effects of the COVID-19 vaccines, this risk remains far lower than the risks of serious health effects associated with COVID-19.

“It is important to note the limitations of the study, including the stage of the vaccination programme at the point of analysis (some individuals had received the second dose of the vaccine and few young people had been vaccinated), limited access to hospital medication records and the difficulty of ITP diagnosis.  We therefore need to gather and analyse more robust data before we can know whether the AstraZeneca/Oxford vaccine induces this specific disorder.

“All the approved COVID-19 vaccines in the UK have been through rigorous clinical trials to ensure their safety and continue to be closely monitored during the rollout by the UK regulator, the Medicines and Healthcare products Regulatory Agency (MHRA).  Vaccination is the safest and most effective way to protect yourself from falling ill with COVID-19 and we continue to encourage people to accept the offer of both doses of the vaccine.

This study that suggests that the Oxford/Astra Zeneca vaccine causes the bleeding disorder called ITP has been carried out well.  Its interpretation may not be as good, as the accompanying editorial suggests.  The study suggests that an excess of 1 case per 100,000 first doses of the vaccine results from this vaccine.

This bleeding disorder occurs very frequently in Covid-19 itself, perhaps as much as 340 per 100,000 people getting the disease.  It is also known, as pointed out in the article, that it has been reported with other vaccines, again generally very rarely.

In vaccine epidemiology it has been acknowledged that comparisons made within individuals over time is the most reliable method for studying rare effects and the authors have also used such method, the “Self-controlled case series” (SCCS), in their study.  For every adverse event they have studied the results in the SCCS analysis gives a smaller effect size compared with the main method they have used.  This certainly suggests that the SCCS results may be more reliable and it is strange that they continue to suggest possible adverse events (arterial thromboembolism and bleeding) as being associated with the vaccine when the SCCS results indicate no association.

The emphasised rate ratio of 5.77 (with an uncertainty of 2.4 to 13) is reduced to 1.98 (with an uncertainty of 1.3 to 3) using the SCCS method for the Ox/AZ vaccine.  The equivalent result for the Pfizer vaccine has a value of 0.54 (uncertainty 0.1 to 3) and for the SCCS method the result is 0.90 (uncertainty 0.47  to 1.7).  While there is no evidence from this study that the Pfizer vaccine has associated ITP, there are about one-third of the numbers in the Pfizer group, and there remains uncertainty round the Pfizer results in this study.  We do know that the Ox/Az vaccine has much less reactogenicity following a second dose while possibly the opposite is true for the Pfizer vaccine.  Hence, it is important not to over-emphasise a very rare effect seen with the first dose of the Ox/AZ vaccine when it’s possible it could occur with other vaccines, possibly after second doses.  It is also possible that the rate of the effect is over-estimated in this study, and clearly uncertainty exists in either direction.

The overall message is that, even if the Ox/AZ vaccine does have an increased risk of ITP its benefit outweighs its risk.  For the majority of people ITP does not cause serious problems but it is not the case for everyone.  The ITP support association has a very good website and they have surveyed patients with ITP who have had their platelets measured before and after vaccination: https://www.itpsupport.org.uk/index.php/en/16-home/238-itp-patient-covid-19-vaccine-survey-spring-2021-initial-findings  While this was not a published study the results do not show a notable difference in the vaccines and do not raise concerns about either of them.

Simpson and colleagues report an apparent association between the AZ vaccine and rare cases of ITP in Scotland using two different methods.  Thrombocytopaenia represents low numbers of circulating platelets in the blood and ITP is often diagnosed when no other cause for this abnormality can be found.  It may cause bleeding and treatments are available to prevent this but in many cases the condition is mild and self-limiting.

“The paper also reports weaker evidence of similarly rare bleeding events and clotting events but the study did not detect occurrence of the thrombosis thrombocytopaenia syndrome following the AZ vaccine that has been reported elsewhere.

“There were no associations observed between administration of the Pfizer-BioNTech vaccine and any of these problems.

“Overall, this study adds somewhat to the understanding of the haematological problems that occur in very small numbers of recipients of this vaccine and, given its importance to the global effort to control the pandemic, ongoing efforts to clarify the characteristics, causes and mechanisms of these events are extremely important.

Much of this report is concerned with medical and haematological aspects that are beyond my expertise as a statistician, so I’ll stick to the statistics.  The statistical methods used in the study are, in my view appropriate.  In those terms, the study is statistically strong.  And it’s based on data from several million people in Scotland.  However, there are inevitable limitations, arising from the kind of data that were available, which means that it can’t tell us everything we might be by interested in.

“One issue is that the study is observational.  People weren’t allocated to get the vaccination, or not, at random by the researchers.  They were vaccinated, or not, in line with the vaccine roll-out in Scotland, and to some extent because of their own choice to accept vaccination.  So there will be differences between vaccinated and unvaccinated people, other than whether they were vaccinated, and any difference in how often symptoms or diseases arise after vaccination might be caused by those other differences rather than by the vaccine.  The researchers did make statistical adjustments to allow for some of these differences, where they had data, but as they point out in the report, it’s possible that there may be important differences on which they had no data, so could not make an adjustment.

“A statistical strength of the study is that the researchers made two different kinds of statistical analysis.  In the main analysis, they compared how many people experienced the possible adverse effects in people who were vaccinated and people who were not.  But they also compared the occurrence of the adverse effects, in vaccinated people, before and after their vaccination, using what’s called an SCCS (self-controlled case series) analysis.  That can remove some effects of differences that have nothing to do with the actual effect of the vaccine, but again this can’t be perfect, because it doesn’t remove the effects of differences between how a person is before and after vaccination that have nothing to do with the vaccination.  The method also generally involves slightly stronger statistical assumptions than the main analysis.  So it’s good to have done this SCCS analysis, and if the results had looked very different from the main analysis, that might have cast some doubt, but it doesn’t get over the problem that we can’t be sure about cause and effect.

“Finally in terms of identifying cause and effect, the linked News item in the journal points out that the condition that was found to be associated with the Oxford AZ vaccine first dose, ITP (idiopathic thrombocytopenic purpura), is not always diagnosed, because there is no specific test for it and because it isn’t necessarily serious.  That might make it harder to distinguish between cases that began before vaccination and those that started after, again muddying the waters of cause and effect.

“Because of all this, more work needs to be done to investigate whether the Oxford AZ vaccine really does cause it to be more likely for people to develop ITP, and it’s reported that that work is being done.  The evidence linking the vaccine to other blood clotting disorders from this study is much weaker, generally, and if there are increases in risk, they are pretty small.  And I should point out that ITP, though it involves platelets in the blood, is not the same as vaccine-induced thrombotic thrombocytopenia (VITT), the condition that also involved platelets and can lead to serious blood clotting disorders.  It’s concern about that condition, not ITP, that has led to the recommendations in several countries (including the UK) that, where possible, the Pfizer or Moderna vaccines should be used for younger people rather than Oxford AZ.  These VITT incidents are just too rare even to show up clearly in this Scottish study of over 2.5 million people.  The key limitation for being able to detect associations between vaccines and possible adverse effects is not the total number of people involved, but how many of them (vaccinated or not) experience the possible adverse effects, and for VITT that’s just far too small a number in this study.  The number of people who had ITP was larger, but still not large in comparison to 2.5 million, and that’s why the estimated rate of ITP after the first dose of the Oxford AZ vaccine is actually very small, only about 11 in every million first vaccine doses.  That’s a very small risk, even if it turns out that it is caused by the vaccine, and it’s tiny compared to the risks arising if someone is not vaccinated and developed Covid-19.

The associated editorial is very balanced.

Unlike the rare syndrome of Vaccine Induced Immune Thrombocytopenia and Thrombosis (VITT) described after vaccination with the AZ and J&J COVID vaccines, Immune Thrombocytopenia (ITP) is generally manageable.

ITP is a condition where the immune system destroys a persons own platelets.  Platelets help prevent bleeding.  Bleeding symptoms are usually mild and death from ITP is very rare.

ITP has been described after other types of vaccination (e.g. MMR) and there is currently no evidence that any specific type of vaccine for COVID is any more risky than any other or any more risky than other historical vaccines.

In addition to bleeding, patients with ITP have a slightly higher background risk of blood clots (thrombosis) for reasons that are not completely understood.  However the risk of thrombosis in ITP is minimal in comparison to the rare syndrome of VITT which is usually associated with a specific antibody in the blood against PF4.

Population studies like this run the risk of mixing up patients with ITP and VITT as they do not have the patient specific level of detail required to separate these conditions.

What is ITP?

“The background rate in the normal population of immune thrombocytopenia (ITP) in adults is around 6 per 100,000.  Although the hallmark of ITP is a low platelet count and an increased risk of bleeding it is also associated with a small increased risk of thrombosis – in general an increase of around 20% over the general population where the risk is 1 per thousand per year, so that increase looks large but in real terms still small.  Thrombosis is increased by certain things like age, surgery, obesity, diabetes and these increase the risk in both ITP and in the normal population; in the elderly with ITP (over 60) and post-surgery or any procedure this may increase 3 fold.

“Thrombocytopenia following vaccinations is well recognised and has been calculated at around 10 per million (these in general are the standard childhood vaccinations) and are usually mild and transient.  It is no surprise that a similar response would be seen following vaccination against Covid-19 and in view of the intense rate of vaccination over the last 5 months any reactions would be highlighted.  By early June over 68 million doses had been administered and thrombocytopenia has been described with both commonly used vaccines.

“The rate with the Pfizer vaccine is around 10 per million vaccinations and following the Astra Zeneca vaccine around 20 per million vaccinations.  If you compare that with the likelihood of thrombocytopenia (ITP) occurring in the normal population (at 60 per million) over the first 5 months of this year we would have expected around 535 new diagnoses.  So it is difficult to be certain whether these are vaccine related or chance associations, although the close association of some to the vaccine suggests that it is likely to be related.  Neither the Moderna or Janssen had been given to sufficient numbers to produce meaningful figures.  To make matters rather more complicated around one third of adults with ITP are picked up incidentally and have no problems.  In ITP the platelet count may fall post vaccine (in around 5%) so we do not know in patients who present post vaccine whether this was a new problem or highlighting and underlying disorder.

“The good news is, however, that treatment is not always required and if it is the count usually responds quickly and is not associated with long-term disease.  It is uncertain how to proceed with the 2nd vaccination when there is a reaction to the first and this must be considered on an individual basis.  We have had patients who have had both of the same without recurrence but most doctors would suggest swapping to the alternate.

“The ITP Support Association, the patient support group, has been surveying platelet counts in their members with ITP post Covid-19 vaccination.  The survey covers the 220 people who completed the survey from the end of February 2021 to 12th May 2021.  Of that total 41 ITP Patients received a pre and post vaccine Platelet Count, split, almost 50/50 between Pfizer and Astra Zeneca.  Five patients in all had a reduction in their platelet count; 2 and 3 respectively.  In none was the fall sufficiently severe to require hospitalisation.  Interestingly 5 patients showed an increased count.  Probably a non-specific reaction to the vaccine along with the other, non-specific, reactions reported in about 1 in 10 with both vaccines used mainly in the UK.   None of this group fell into the category of Vaccine induced thrombocytopenia and thrombosis (VITT) that is associated with the cerebral venous sinus thrombosis syndrome and is clearly a separate entity.

“In a similar study from the USA of 52 consecutive patients 15% had no worsening of ITP symptoms but no count measured; 73% had no new symptoms and no change in platelet count.  However, 12% had a drop in the platelet count but all either bounced back or responded to treatment and the majority were above 30 x109 /L within 3 days.  The changes in platelet count occurred independent of remission status, concurrent ITP treatment or vaccine type.