A critical debate about Covid-19 vaccines is when does protection wane, by how much, why, and what does this mean for controlling the pandemic and the impacts of infections.
Depending on the studies or headlines you read it can be confusing. Some report declining vaccine effectiveness, and others don’t. Some report a lot, and others a little. Like most things associated with the pandemic, few of them provide certainty.
In this post I’ll summarise what’s emerging from recent studies, and challenges in establishing if effectiveness is decreasing, and what causes it.
The post is long because it’s a complicated issue.
Long story short: Vaccine effectiveness, at least against infections, can decline after 5 or 6 months. This may be more likely due to immunity decreasing, rather than just the Delta variant being better at avoiding immune responses. However, the observational nature of these studies mean that a variety of factors may influence the results. There is evidence that increasing the interval between vaccine doses may improve effectiveness, but further research is needed. There is evidence that an additional “booster” Pfizer dose increases effectiveness over the short term (a few months), but more research is needed on this too.
From clinical studies to the real world
There is a minor sense of disappointment that after some great results from vaccine clinical trials less than a year ago, discussions are already shifting to the need (or not) for an extra dose.
Results from the clinical trials for some vaccines exceeded expectations. Effectiveness against symptomatic infections were higher than 90% for the two mRNA vaccines (Pfizer/BioNTech’s BNT162b2 and Moderna’s mRNA-1273). And effectiveness against severe Covid-19 was nearly 100%, at least for those under 65.
For two adenovirus-based vaccines effectiveness against infection in clinical trials was nearly 80% for the AstraZeneca/Oxford vaccine (aka AZD1222 or ChAdOx1; and now called Vaxzevria or Covishield in some countries), and 66% for Johnson & Johnson/Janssen’s single dose vaccine (Ad26.COV2.S).
These trials all assessed effectiveness within a few months of vaccinations. Some of those levels of effectiveness don’t appear to hold up over longer time periods.
Real world effectiveness
As vaccination programmes role out “real world” observational studies have taken over from clinical trials in assessing vaccine effectiveness.
Early reports from Israel (for the Pfizer/BioNTech vaccine) and the UK (for both the Pfizer/BioNTech and AstraZeneca/Oxford vaccines) indicated that real world effectiveness can decline five to six months after getting the second dose. These studies reported decreasing effectiveness for infections, hospitalisations and deaths. Declines in effectiveness were largest for prevention of infections, and for those over 65.
Other studies subsequently showed large declines in effectiveness against infections, but not necessarily against protection from severe Covid-19 in the five to six months after full vaccination.
For example, in the US where effectiveness against hospital admissions remained around 93% for six months (a similar result was reported for New York State, at around 95% in Qatar (both of these looked at the Pfizer/BioNTech vaccine.)
In the UK, analyses by Public Health England found 95% effectiveness against hospitalisation for at least 5 months for the Pfizer vaccine, and just under 80% for AstraZeneca’s.
In British Columbia, mRNA vaccines were more than 90% effective in preventing hospitalisations over at least four months. While a six month follow-up of an international clinical trial of the Pfizer/BioNTech vaccine found around 91% effectiveness against symptomatic infection.
Similar results, published in The New England Journal of Medicine, have been found for the Moderna vaccine too. Five months after the second dose vaccine efficacy in preventing Covid-19 symptoms and severe disease remained at over 90%, while effectiveness against asymptomatic infection declined to 63%.
However, a Scottish study found that effectiveness against hospitalisation declined quickly over a few weeks after the second dose. However, the level of protection then stabilised to around 80%. This applied to both mRNA vaccines and AstraZeneca’s.
Reducing biases and confounding effects
Are these declines are real, or are they due to biases or limitations in the studies?
There are a range of factors that muddy the immunity waters, making it challenging to identify waning immunity, distinguish it from other causes, and determine the scale and significance of decreases in vaccine effectiveness.
A news item in Science discussed the possibilities of waning vaccine effectiveness against infections. A similar article was published in Nature. They note that factors other than waning vaccine effectiveness or immune escape could contribute to a rise in post-vaccination infections. Biases associated with measuring effectiveness are also outlined by Public Health England, and an editorial in The British Medical Journal.
The real world studies have the advantage of including very large numbers of people, and seeing how infection and disease rates change in the months after vaccination.
However, they can’t control for some biases and confounding effects as well as clinical trials. For example, there is no randomisation for treatment, so differences between vaccinated and unvaccinated people, and the infected and uninfected may affect results. For reliable conclusions the groups need to be as similar as possible – age, sex, health, etc.
Real world studies also have to contend with changes in behaviour after people have been vaccinated. This can change exposure risks and so give misleading results about the level and duration of protection.
There are a variety of methods that studies can use to address these types of confounding effects. However, there isn’t a standard protocol, and studies may focus on controlling a few different potential biases, so differences in results aren’t unexpected.
Other factors can be involved as well, some of which I also look at below.
Lab experiments show changes in immune responses with time too
In research, it always helps to have independent lines of evidence.
While disappointing, decline in vaccine effectiveness over time isn’t a surprise to public health scientists. Waning vaccine effectiveness is common for many, if not most, vaccines. Some remain effective for decades or many years, others lose effectiveness in months. The immunological reasons for the waning are still not well understood.
Real world observations of declining effectiveness are supported by lab studies of immune cell populations and activities. For the Pfizer/BioNTech vaccine antibody levels and neutralisation activity declined from three months after full vaccination, while T cell levels rise more slowly but then remain relatively stable.
The same trend is seen for the Moderna vaccine too. For the single dose Johnson & Johnson/Janssen vaccine antibody levels were lower compared with the two mRNA vaccines, but did not decline so levels for all three after 8 months were similar. (Although in this study only 8 people who had the Johnson & Johnson vaccine were included).
The immune system is naturally dynamic
Such declines and stabilities are normal for many types of vaccines, and non-vaccine responses to infectious diseases. The immune system is dynamic; there is an early “call to arms” to respond to an infection, followed by development of an immune memory that can rapidly produce more antibodies and T cells if an infection reoccurs.
Generally, antibodies’ roles are to stop infections, while T cells and other components of the immune system target infected cells, attempt to reduce damage done by infections, and help establish a long lasting immune memory.
There is, though, an expectation that as neutralising antibody activity declines the virus has a greater chance of infecting a person. So, declining effectiveness against infections may be due to declining levels of neutralising antibodies.
Research is underway to determine if particular levels of antibodies, at specific points in time, provide a reliable guide to the degree of protection a vaccine provides. While there are some indications for such “correlates of protection”, further research is necessary.
However, other factors may play a role as well.
Is decreasing effectiveness due to the Delta variant?
One of those is the increasing dominance of the Delta variant since vaccinations began.
It is known to be more infectious than the Alpha and other variants. Compared with the Alpha variant, neutralising antibodies are also less effective against it. Either, or both, of these factors may enable it to be better able to avoid immune surveillance and control.
Several studies, though, have not found evidence to support this.
A US study (published in The Lancet) found similar declines in effectiveness against infection for both Delta and non-Delta variants over five months.
Studies in the US (The New England Journal of Medicine) and Israel (not yet peer reviewed) both found lower infection rates for those who had been vaccinated four months earlier, compared with those who were vaccinated six months earlier. Variant type appeared to have no effect on the trend.
However, a large study from New York State (not yet peer reviewed) concluded that the spread of the Delta variant rather than waning immunity was the cause of declining vaccine effectiveness. There was evidence for waning vaccine immunity, but a stronger correlation was seen with the increasing prevalence of Delta than time since vaccination.
Another US study also reported increases in infection rates correlated with the spread of the Delta variant.
The weight of evidence currently appears to be supporting waning immunity rather than immune avoidance. Still, none of the studies are perfect. And just because a study is very large doesn’t mean it is better at controlling for bias.
Does the timing between first and second doses play a role?
Another aspect that can make comparing different studies difficult is that, in some places, the times between first and second vaccine doses differ.
For the Pfizer, Moderna, and AstraZeneca vaccines clinical trials used three to four weeks between doses. This allowed information on effectiveness to be obtained relatively quickly.
In Israel, most people had had that interval for the Pfizer/BioNTech vaccine. In the UK, and some other places, timing between doses (for Pfizer/BioNTech and AstraZeneca/Oxford vaccines) was extended after the vaccination campaigns started to ensure more people got a first dose quickly as new variants spread.
Lab studies subsequently reported that antibody responses can be stronger if there is a longer time between doses. This may in part be due to better development of immune memory. This could lead to better effectiveness against infection and/or severe Covid-19.
For example, peak antibody levels were 3.5-fold higher in those over 80 after a 12 week interval compared with a 3 week interval. However, T cell responses were about 3.5-fold lower.
A separate study of healthcare workers found that a 6-14 week interval doubled the levels of neutralising antibodies, compared with a 3-4 week interval. T cell levels were lower overall following the longer interval, but had proportionally more helper T cells, which help develop immune memory.
Observations of infections tend to support these laboratory studies.
Increased time (6 weeks or longer) between doses were found to improve vaccine effectiveness for the Pfizer/BioNTech vaccine in one UK study (not yet peer reviewed).
Canadian data (also not peer reviewed) also found that increasing the time between the first and second dose for the Pfizer/BioNTech vaccine increased effectiveness against infection from 82% (for a 3-4 week gap) to 90% or higher (7 weeks or longer).
But on this issue, too, other studies don’t come to the same conclusion.
A different UK study (not yet peer reviewed) did not find that vaccine effectiveness was affected when it compared infection rates between those whose interval between first and second doses was less than 9 weeks, and those who had a longer interval.
An obvious difficulty in comparing these studies is the differences in vaccine intervals.
Compared to studies of the effect of the Delta variant there are fewer real world observations of the effect of increasing dosing intervals. Expect more studies investigating this the influence of variants. They may clarify (or further obscure) what effect(s), if any, a longer interval between doses has.
So, a key message for vaccine waning studies is don’t be convinced by the results of just one study.
Recent data from Israel (not yet peer reviewed) shows at least short term benefits from third doses of the Pfizer/BioNTech vaccine for all age groups. The third dose was given at least five months after the second.
Several more months of data collection in Israel will be required to see if effectiveness wanes again. The British Medical Journal editorial suggests that evidence for everyone getting an extra vaccine dose is currently weak.
Discussions about who benefits from additional doses, and when, will continue.
Ideal vs real vaccines
A vaccine that prevents infection creates what is called “sterilizing immunity.” This is the perfect vaccine, but may be more ideal than real.
Respiratory viruses like coronaviruses, usually do not produce long or lifelong immunity after an infection, so we shouldn’t expect current vaccines to either according to a paper in PLOS Pathogens.
This immune avoidance is possibly due to a variety of factors. Such as how quickly the virus mutates, and the fact that they infect mucosa (such as the linings of the respiratory and digestive tracts) where immune systems seem to be weaker at controlling viral infections. Research is underway to develop oral mucosal vaccines which may be more effective.
In the meantime, we have to make the most of the vaccines we have. The studies I’ve covered in this post all reinforce the message that a vaccine is only on aspect of controlling the pandemic. While SARS-CoV-2 is still circulating we’ll still need to take other public health precautions too.
Every week I produce a Research Tracker for the Science Media Centre summarising some recent Covid research highlights.