Wikipedia:Reference desk/Archives/Science/2021 April 14

= April 14 =

COVID-19 Vaccines and Blood Clots
It's been reported that there might be a causal relationship between receiving the Johnson & Johnson and AstraZeneca COVID-19 vaccines and getting blood clots, but the actual incident range of those dying appears to be extremely low - about one in a million (or so). My question is this: What is the expected number of people getting killed from blood clots among every million people? A Quest For Knowledge (talk) 19:36, 14 April 2021 (UTC)


 * This article in The Lancet gives stats from Denmark. PaleCloudedWhite (talk) 20:03, 14 April 2021 (UTC)
 * From the UK Medicines and Healthcare products Regulatory Agency on 7 April: "By 31 March 20.2 million doses of the COVID-19 Vaccine AstraZeneca had been given in the UK meaning the overall risk of these blood clots is approximately 4 people in a million who receive the vaccine".  Alansplodge (talk) 21:58, 14 April 2021 (UTC)
 * My understanding of the OP's question is that they are asking for the incidence of blood clot deaths generally, unconnected to Covid vaccinations. PaleCloudedWhite (talk) 22:12, 14 April 2021 (UTC)

I was about to ask a similar question. Maybe someone can answer both.

I have a friend in his late 60s who I just found out is refusing to be vaccinated. He is a physicist, and like many scientists, he likes to analyze and figure out things rather than believing experts, but feeding him raw statistics works very well.

I want to write up an analysis that compares three things: The current infection rate in Los Angeles County, the current fatality rate among 60 to 69 year olds, and the fatality rate for the vaccines available in LA county. I am hoping to be able to show that, statistically, he is more likely to die from not being vaccinated than from being vaccinated.

I am just starting to research this. Has anyone gathered those numbers with citations? --Guy Macon (talk) 23:51, 14 April 2021 (UTC)
 * While this is something that has interested me since the early days of Astra Zenica's in late February/early March since it's true most news reports never seemed to really provide figures or how they were derived at best saying it's higher than expected or within expected, note that the expected frequency of blood clots in the population who've received the vaccine probably isn't as meaningful as it seems. As I understand it, one particular reason why blood clots associated with these vaccines have received attention is because most of them have been specific types of blood clots that are fairly rare compared to other forms of blood clots such deep vein thrombosis. See e.g. what the EMA said [//www.ema.europa.eu/en/news/astrazenecas-covid-19-vaccine-ema-finds-possible-link-very-rare-cases-unusual-blood-clots-low-blood] or these the news source discussions [//www.cidrap.umn.edu/news-perspective/2021/04/studies-suggest-link-between-blood-clots-astrazeneca-covid-vaccine]/[//www.bmj.com/content/373/bmj.n954] [//abc7news.com/covid-19-vaccine-side-effects-cdc/10512795/]. (There's obviously a lot better information on AstraZenica than Johnson and Johnson given as I said earlier there have been significant concern for the former for over a month but it's only recently we heard similar concerns for the latter. Note as some of those sources mention, at least for AstraZenica the most similar condition seems to be Heparin-induced thrombocytopenia and there is a tiny mention of the vaccine in that article.) It's possible you will still see a rise in over numbers of blood clots unless the vaccine is associated with a lower risk of other types, but it's also possible the rise is lost in noise. Maybe most likely is you'll have difficulty calculating a figure useful for comparison as you'd need to take into account differences between the subset of people who've been vaccinated and the general population. Age differences is one obvious difference, but also probably others like pre-existing conditions and existing medication usage, socio-economic status and access to healthcare, possibly even sex differences. (I don't know if there is a sex bias in vaccinations but for Astra Zenica thrombocytopenia and blood clots seem to happen in women a lot more than men as the sources mention. It's still early days for Johnson and Johnson but at per sources [//www.bbc.com/news/world-us-canada-56733715] all the reported cases were women. But per our article for DVT as an example, frequency is higher in men than women although it depends on age.) Nil Einne (talk) 04:56, 15 April 2021 (UTC)


 * I understand your more relevant question about general blood clot occurrence. What I have seen, are articles comparing blood clot occurence among vaccines, e.g., here. Apparently, blood clot occurence for BiontechPfizer and Moderna are not higher-than-expected, so that gives a rough (!) idea about general occurrence. (And it saves you the methodological problems mentioned by Nil Einne: comparing a general population with the sample that gets vaccinated.) I see why you would also want to have general figures, though.
 * As for the basic question: Do I risk more getting vaccinated than not getting vaccinated? From general media outlets (and an experienced pulmonologist who works with Covid patients at a specialized lung clinic, whom I personally know) I understand that figures exist that, yes, it's still safer to get vaccinated than not, esp. for the elderly. This includes--at least for the elderly--AstraZeneca, given that there are few incidences of older people having problems with the vaccine (BTW: no men above 60, to my knowledge). I understand that this is what your physicist will not believe (relying on experts' expertise), yet I thought I'd mention that at least others have done those stats already. :-) Like, you, I'd wish though that general media outlets would simply publish the figures... --Ibn Battuta (talk) 07:37, 15 April 2021 (UTC)
 * It doesn't have to be general media outlets. I feel that the manufacturers ought to make the data publicly available that they provide to health authorities, and that health authorities have a responsibility to publish the arguments, including hypothesis testing with regard to safety and efficacy, that lead to their recommendations, and make them available to public scrutiny. A problem in convincing someone who is science-savvy of accepting the risk because of the greater benefit is that they have no influence over the risk of a thrombotic event following vaccination, whereas they have a great deal of influence, through their behaviour, on the risk of contracting COVID-19. So comparing the risk of vaccination with the risk of contracting COVID-19 and falling critically ill, if the latter is based on the rates among the general, possibly not sufficiently cautious population, may not be entirely convincing. There is another issue, that of making rational decisions given existing non-null risks – it is not rational to avoid something beneficial because of a small risk if something less beneficial with a higher risk is gladly accepted. I have heard it said that for someone over 40 the incidence rate of life-threatening thrombosis following a long-distance flight is higher than the reported rate following COVID-19 vaccination among younger women. However, I did not find citable incidence rates in the literature. --Lambiam 10:06, 15 April 2021 (UTC)


 * As I understand, you would like to show whether P(dying from Covid19|not vaccinated) > P(dying from Covid19|vaccinated) + P(dying from vaccine side-effects|vaccinated) for a given age group and geographical location. (Note that this excludes non-fatal effects from the computation, which might be too simplistic.)
 * The historical attack rate for LA (i.e. what fraction of the population has been infected) could probably be found on the official websites, but obviously this will not tell you the prospective attack rate (what fraction of the population will be infected in the future), and it is the latter that matters for the probability computation. Of course, there is a whole host of prisoner's dilemna considerations; if your friend is afraid of vaccination, they probably still want everyone else to get vaccinated so that the attack rate plummets and they are safe without getting the vaccine themselves (see: mumps resurgence in recent years, Herd immunity, etc. etc.).
 * The IFR (infection fatality rate, i.e. P(dying|infected)) for a male between 65 and 70 is about 2% (?) from this paper (if you want the exact number, you probably have to look at the supplementary material, I eyeballed the figure but it's a log scale so all I can say for sure is that it is above 1% and below 10%).
 * The probability of dying of Covid19 when vaccinated is unknown, and cannot be known with the clinical trial data because the trial sizes were not chosen for that (read the "double zero" paragraph). It seems reasonable (in a null-hypothesis sense) to assume that deaths are cut by about as much as symptomatic infections are (which is around 60 to 90% depending on which vaccine, risk profile etc. etc.), but it could be either much better (if vaccination decreases the severity of symptoms about the same for everyone, those who risk to die have the most marginal benefit) or much worse (if vaccination either prevents infection entirely or fails altogether, it tends to fail exactly on people who are at a higher risk to die).
 * For blood clots I am pretty sure the numbers are too low to be able to give a precise number. This general news article tends to hover at 0.001% or less of "serious harm" (not necessarily death, so not directly comparable to the covid IFR) but it lacks real sources. Tigraan Click here to contact me 09:33, 15 April 2021 (UTC)
 * I've only done a quick "back of the napkin" calculation, but based on 6 cases of clotting for the J&J vaccine out of about 7 million vaccinated in the United States, that comes to 8.5*10^-5% chance of having the blood clot complication. That's assuming that we have actual good data on the number of people having clotting complications, and that 100% of them were actually caused by the vaccine, and not what would already be generally expected for that many people (since I don't have that data). By contrast, with 564402 deaths out of 31421360 cases according toe John's Hopkins, that's a fatality rate of about 1.8%. So, if you are infected with SARS-CoV-2, that's 1.8% chance of dying (ignoring factors like age, for the purposes of this napkin calculation). If you get the J&J vaccine, that's 8.5*10^-5% chance of getting a blood clot. That's worst case, assuming all of these were caused by the vaccine. That means, worst case, you are ~21000 times more likely to die of COVID19 if you are infected with the virus than you are to get a blood clot complication if you take the J&J vaccine. Not die from the vaccine, just clotting complication (which can be life threatening, of course). Given those odds, where I'm 21000 times more at risk of dying from the virus than I am of even having a blood clot from the vaccine that protects me from the virus, it's easy to see what I would choose to do. --OuroborosCobra (talk) 19:20, 15 April 2021 (UTC)
 * For a comparison of fatality risks based on this calculation, you have to multiply the number ~21000 by the probability of becoming infected. For someone who is extremely careful, that will bring down the ratio considerably, although most likely not into the ~1 ballpark. --Lambiam 10:03, 18 April 2021 (UTC)
 * This is likely to undercount the true incidence rate. First, roughly half of J&J doses were given in the last two weeks, and only by raising this issue will we get a fuller accounting of the issue. Second, the vaccine is similar to the Oxford one, whose clotting problem was also at first dismissed. Now we know at least 222 cases in Europe, or they say 1 in 100,000 vaccinations. Imagine Reason (talk) 19:26, 15 April 2021 (UTC)
 * I just heard on Radio 4's Inside Science programme that the blood clot complications associated with the AstraZeneca vaccine have, so far, showed up between 5 and 28 days after vaccination. PaleCloudedWhite (talk) 20:30, 15 April 2021 (UTC)
 * In this study https://www.bbc.com/news/health-56760163, they compare cerebral sinus vein thrombosis (CSVT) blood clot who develop this type of blood clot within 2 weeks of vaccination(Pfizer or Moderna) or 2 weeks of being diagnosed with COVID. In this article they say for COVID cases there were 39 per million (so 3.9 per 100,000) and the vaccinated group is 8 to 10 times less (0.4 per 100000). AstraZenica was not in this study so can't compare directly. But the European Medical Agency say there were 5 in a million (0.5 per 100000) of a certain type of CSVT. Dja1979 (talk) 21:46, 15 April 2021 (UTC)