welcome to impact factor, a weekly commentary on new medical research. I’m his Dr. F. Perry Wilson from Yale University School of Medicine.
I’m here today to talk about the efficacy of the Covid-19 vaccine booster in mid-2023. I want to talk about this not necessarily to delve into exactly how effective vaccines are. This is an area that has been stepped on many times. But it does give me the opportunity to talk about a neat study design called the “test-negative case-control” design. This design has some unique properties when trying to assess the effect of something outside the context of a randomized trial. .
So just a little background to remind you of where we are. These are the number of doses of the 2019-nCoV vaccine administered over time throughout the pandemic.
You can see that they are divided by age group. For example, the orange line is her adult ages 18-49. When a vaccine first comes out in early 2021, we will see a big wave of vaccinations. After that, we saw a small wave after approval of the first and her second booster, and there may be a bit of a recovery, especially among the elderly. A bivalent booster has been licensed. However, overall uptake of bivalent boosters is still very small compared to monovalent vaccines, which may suggest that the pandemic has continued to this extent of vaccine fatigue. . But it’s important to try to understand exactly how effective these new boosters are, at least for now.
what i’m talking about Initial Estimation of Efficacy of a Bivalent mRNA Booster Vaccine in Preventing Symptomatic SARS-CoV-2 Infection Caused by Omicron BA.5 and XBB/XBB.1.5-Related Substrains in Immunocompetent Adults—Community to Testing Programs. Increased Access, United States, Dec 2022–Jan 2023came out with Weekly reports of morbidity and mortality Most recently, we used this test-negative case-control design to assess the ability of a bivalent mRNA vaccine to prevent hospitalization.
The question is whether a booster dose of the bivalent COVID-19 vaccine could prevent hospitalizations, ICU stays and deaths. It may not be a question that interests everyone. I know people are interested in symptoms, work absenteeism, and infections, but this paper was looking at hospitalizations, ICU stays, and deaths.
The complication here is that the data they are using is from people who are hospitalized with various illnesses. It’s a bit counterintuitive to ask yourself, Using only data from hospitalized patients, how can we estimate the ability of a vaccine to prevent hospitalization? You might look at it superficially and say: well it is not possible. But in fact, you can with this cool Negative Test Case Management design.
Here’s how it basically works: For the population of people who have been hospitalized and confirmed to have COVID-19. Some people get vaccinated, some people don’t. Also, I don’t know much about the proportion of people who are vaccinated and who are not, because it depends on how you compare it to the proportion of the general population, for example. Let me clarify this. If 100% of the population were vaccinated, 100% of those hospitalized with the novel coronavirus would be vaccinated. That doesn’t mean vaccines are bad. In other words, if 90% of the population were vaccinated and 60% of those hospitalized with COVID-19 were vaccinated, all else being equal, the vaccine would have some effect. It will actually show that it is demonstrating So the raw percentage alone doesn’t tell you anything. Some are vaccinated and some are not. You need to understand what a baseline rate is.
The test-negative case-control design targets people who are hospitalized without being infected with COVID-19. Now, who those people are (who are the controls in this case) is something you really need to think about. For this CDC study, we used people hospitalized with illnesses similar to the novel coronavirus, including flu-like illness, respiratory illness, pneumonia, and influenza. This is a very good idea. Access to medical care. I can go to the hospital, and I’m the type of person who goes to the hospital when I’m sick. It’s better control than the general population at large, and that’s what I love about this design.
Some people who are not infected with the new coronavirus (those who are hospitalized with influenza etc.) may have been vaccinated against the new coronavirus. Some people don’t. And of course, we don’t expect the COVID-19 vaccine to necessarily prevent flu or pneumonia, but it gives us a way to standardize.
Looking at these Venn diagrams, we can see that the percentages of vaccinated and unvaccinated are exactly the same. This suggests that people who are vaccinated are just as likely to be hospitalized for COVID-19 as they are for other vaccines. Respiratory disease, which suggests that vaccines are not particularly effective.
But when you look at something like this, when you look at all the patients with the flu and other non-coronavirus diseases, far more patients were vaccinated against the novel coronavirus. I understand. What this means is that the number of hospitalized vaccinees for COVID-19 is lower than expected because it has been standardized from other respiratory infections. The number of vaccinated people who get the flu is the same, so this number of vaccinated people is expected. But their numbers are low in the COVID-19 population, which would suggest the vaccine is effective. This is the test-negative case-control design. The same can be done for ICU stays and deaths.
Here are some assumptions you may have already made. Most importantly, vaccination status is not associated with disease risk. I always think of the elderly in this context. During the pandemic, at least in the United States, older people were much more likely to get vaccinated, but they were also much more likely to be infected with and hospitalized with COVID-19. Older people are also more likely to be hospitalized for things like flu and pneumonia, so the test-negative design actually explains this in a way. So you have some control there.
However, older people are uniquely more susceptible to COVID-19 than other respiratory diseases, which could skew the results to make vaccines less effective. So the standard approach here is to coordinate these things. I think the CDC adjusted for age, gender, race, ethnicity, and a few other factors to see how effective the vaccine was.
Let’s see an example in action.
This is actual data from a CDC paper. 6,907 people were hospitalized with the new coronavirus, 26% of whom had not been vaccinated. What is the expected baseline rate of unvaccinated people? A total of 59,234 people were hospitalized for respiratory illnesses other than COVID-19, 23% of whom had not been vaccinated. So there were more unvaccinated people in the COVID-19 group than you might think. In other words, fewer people getting vaccinated suggests that the vaccine works to some extent because it keeps some people out of hospital.
Now, the 26% vs. 23% difference isn’t all that impressive. However, it is even more interesting when analyzed by vaccine type and how long individuals were vaccinated.
Let’s take a look at the “All” group in this diagram. What can be verified is the calculated efficacy of the vaccine. If we look only at the monovalent vaccine here, we can see that the efficacy of the vaccine is 20%. This means that getting people vaccinated could basically prevent 20% of hospitalizations due to COVID-19. That’s fine, but nothing special. However, if the bivalent vaccine is given within 60 days, the vaccine is much more effective.
It compared people who received the bivalent vaccine within 60 days in the COVID-19 and non-COVID-19 groups. The concern that the vaccine was most recently administered affects both groups equally and should not be biased there. The effectiveness of the vaccine is he 60 days, 60-120 days, gradual decline after 120 days. We went from 60% to 20% in 4 months. Breaking this down by age, we see a similar pattern in the 18-65 age group, with the 65+ age group likely to be more protected.
Why are vaccines less effective? Although not clear in this study, it is hypothesized that this could be an immunological effect, ie a decrease in antibodies or protective T cells over time. This may also reflect viral changes in the environment in which the virus seeks to evade certain immune responses. Overall, however, this suggests that a one-year gap between boosters may leave you exposed to infection for a significant period of time, but the lesson here is that, in general, Considering the percentage of people who are not vaccinated, bivalent vaccines are probably a good idea.
Looking at serious illness and deaths, the numbers look a little better.
It can be seen that bivalent is better than monovalent. If you received it within 60 days, it’s certainly pretty good. It tends to decline somewhat, but not by much. Considering serious illness such as ICU stays and deaths, the vaccine is still about 50% effective beyond 120 days.
Most importantly when thinking about vaccine policy, the way people are immunized against 2019-nCoV is by vaccine, by contracting 2020-nCoV, or both.
This very interesting chart from the CDC (but only updated until the third quarter of 2022) shows Americans with varying degrees of protection against the novel coronavirus, based on routine laboratory tests. showing a percentage. What we can see here is that by the third quarter of 2022, only 3.6% of people who had their blood drawn at commercial laboratories had no evidence of infection or vaccination. In other words, few were completely naive. Furthermore, 26% of people have never been infected and only have vaccine antibodies. Additionally, 22% of people had been infected but had never been vaccinated. And 50% had both. So there is a huge amount of pre-existing immunity out there.
The really interesting question about future vaccinations and future boosters is how it works behind this pattern. The CDC studies tell us nothing, and I don’t think they have the data to tell us the efficacy of the vaccine in these different groups. For example, is it more effective for people who have never had an infection? Is it more effective than people who have been vaccinated alone, than people who have been vaccinated for both, or what preventive measures? Are they more effective than those who have not been vaccinated? These are very interesting questions that need to be answered in the future as vaccine policies develop.
I hope this serves as a primer on how a test-negative case-control design can answer a question that seems a bit elusive to answer. See you next time.
F. Perry Wilson, M.D., MSCE, is an Associate Professor of Medicine and Director of the Clinical and Translational Research Accelerator at Yale University. You can find his science communication work on The Huffington Post, NPR, and here at his Medscape.he tweets @fperrywilson and his new book, how drugs work and when they don’tis currently available.
