The preprint Effectiveness of Covid-19 Vaccination Against Risk of Symptomatic Infection, Hospitalization, and Death Up to 9 Months: A Swedish Total-Population Cohort Study by Nordström, Ballin and Nordström suggests that not only does the effectiveness of the COVID-19 vaccinations decline over time (which is not too surprising), but in the case of double-dosed AstraZeneca vaccine recipients (ChAdOx1 nCoV-19), the effectiveness of reducing infect is negative 19%, which might be interpreted as a higher risk of infection than without any vaccination.

Source: Spektrum magazine (German)

The value is explained as "Adjusted for age, baseline date, sex, home maker service, place of birth, education, and comorbidities [...]" and as the Spektrum article explains, there are multiple explanations for the value such as more risky behaviour or different testing patterns.

While that sounds sensible, I still wonder: Does the AstraZeneca COVID-19 vaccine increase infection rates? Has any vaccine ever been shown to do that?

  • 10
    Although a mod answered this, it's not a good fit here as it asks for theoretical (mechanism) speculation. Should probably be migrated to biology SE. Also, the question "quote" is actually the OP's own thinking, not something from the article. So, it's disprove-MY-theory kind of question. Commented Nov 30, 2021 at 2:09
  • 2
    @Fizz I've seen other people getting similarly confused by exactly this figure, and arriving at the same questions ("does this particular vaccine increase risk of infection" and "are negative values possible here at all" which is equivalent to "can vaccines increase infection risk"). I think both of these would be on-topic here, and the question written here mostly just sounds more speculative than it actually is.
    – Mad Scientist
    Commented Nov 30, 2021 at 7:34
  • 3
    This question is not a good fit for skeptics.SE, because OP is asking about additional information related to the article not questioning the claims made in the article. The source makes it very clear that the effect could be caused by the behavior of vaccinated people instead of the vaccine itself. It does not say in any way that "AstraZeneca COVID-19 vaccine (or any known vaccine) increases infection rates."
    – Chris
    Commented Nov 30, 2021 at 10:20
  • 4
    The graph is dodgy to start with: the X-axis is drawn with evenly-spaced markers, that aren't equal in time. Probably an error, but a misleading one. That would make me very suspicious about assuming other features of the plot and the analysis aren't also wrong. Note that the plot, and the numbers it directly shows are from the magazine article, not the scientific one the journalists cite
    – Chris H
    Commented Nov 30, 2021 at 14:34
  • 4
    There are a lot of other factors in play here, exactly what makes this kind of studies so difficult. As double vaccinated and living in Sweden, I might add things like: behaviour (once vaccinated I and several others basically ignored restrictions), group selection (as Astra Zeneca vaccine was the first available vaccine it was given mostly to elderly and other risk groups). And possibly more factors as the usage of AstraZeneca vaccine basically has been phased out from Swedish usage.
    – ghellquist
    Commented Nov 30, 2021 at 14:49

1 Answer 1


I'll answer the question about AstraZeneca first, because it's important to be precise here about what this means for the SARS-CoV2 vaccines we're using right now. The answer there is that there is no evidence at all that any of them can increase the infection risk or the severity of the disease*.

To your general question about all vaccines: the answer is yes. There is a phenomenon called Antibody-dependent enhancement where a previous infection or vaccination can lead to a worse outcome in later infections:

In general, virus-specific antibodies are considered antiviral and play an important role in the control of virus infections in a number of ways. However, in some instances, the presence of specific antibodies can be beneficial to the virus. This activity is known as antibody-dependent enhancement (ADE) of virus infection. The ADE of virus infection is a phenomenon in which virus-specific antibodies enhance the entry of virus, and in some cases the replication of virus, into monocytes/macrophages and granulocytic cells through interaction with Fc and/or complement receptors.

Antibody-Dependent Enhancement of Virus Infection and Disease

This is a known phenomenon, and unfortunately there are some diseases where this happens and leads to worse outcomes in a second infection. One example is dengue, where a second infection with a different strain of the virus has a much higher chance for very serious disease progression. The reason in this particular case is that dengue infects macrophages, and the immune response brings the virus into close contact with them. If the antibodies don't neutralize the virus, they're essentially helping it to bind to its targets (https://www.nature.com/scitable/content/model-of-antibody-dependent-enhancement-of-dengue-22403433/).

But the question you probably should have been asking about the AstraZeneca result is "what's the confidence interval on that negative value?" This illustration is bad, I've seen it linked a few times already and it's just really annoying how misleading it is.

If you look at table 2 in the paper you already linked, you'll find that the confidence interval for the AZ vaccine at >120 days ranges from -97 to 28. The interval is huge and still includes positive values. You cannot conclude from this that the vaccine increases the risk of infection, that data point is simply not robust enough to read anything like that from it. And even if it were statistically significant, as you mentioned there are other potential confounding factors.

* Here's a response from the German institute responsible for vaccine safety:

aus klinischen Prüfungen mit mRNA-Impfstoffen gibt es keinerlei Hinweise auf eine verstärkte COVID-19-Erkrankung VAED (Vaccination Associated Enhanced Disease) bei geimpften Personen.

Translated this means:

clinical studies on mRNA vaccines give no hints at all for amplified COVID-19 disease (Vaccination Associated Enhanced Disease) in vaccinated persons.

  • 9
    Beat me to the answer! Also, in addition to the antibody dependent enhancement (since there isn't much evidence that COVID works that way), you might also want to mention something about vaccinated status leading to changes in behavior. Waning immunity coupled with higher exposure risk due to lax preventative measures can lead to more infections.
    – rjzii
    Commented Nov 29, 2021 at 18:14
  • 7
    @rjzii there's so many potential confounders with this kind of study, that would probably bloat the answer to much, and the one you mentioned is also already mentioned in the question itself. I'm a bit more skeptical personally about the background incidence of COVID that isn't excluded because the people were never officially tested. That alone is also likely to lead to an apparent decrease in vaccine efficiacy as more people had the infection at later points in time.
    – Mad Scientist
    Commented Nov 29, 2021 at 18:27
  • 2
    Indeed, given that this is a preprint I suspect there are going to be a fair number of revisions when it gets back from the peer reviewers.
    – rjzii
    Commented Nov 29, 2021 at 19:19
  • 12
    It may improve the readability of your answer for many if you briefly explain the meaning of confidence interval. I suspect that many people are not familiar with the concept. Commented Nov 30, 2021 at 8:50
  • 2
    @TobiasKienzler There are other potential mechanisms for ADE, the dengue one is just the most dangerous one as far as I understand. You can see in the linked response from the PEI that they mention the modified Spike in the pre-fusion state as a mechanism against potential ADE. The idea here is probably that antibodies against the active state could potentially stabilize the virus in a more infectious conformation. But that is again hypothetical as the response said.
    – Mad Scientist
    Commented Nov 30, 2021 at 21:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .