Search

Journal Club: The COVID-19 Booster Study



Without a doubt, the most common question I have been getting in clinic over the past two weeks has been, “What’s your opinion on the booster shot?”


The independent advisory committee to the FDA met this week to discuss the utility of a booster vaccination against COVID-19. They recommended against the broad use of boosters for individuals over the age of 16 yo. They did recommend boosters for individuals over 65, and those considered to be at high risk for infection. High risk consisted of healthcare workers, teachers, grocery store employees, prisoners, homeless shelters, and “current residence.”


There wasn’t much transparency surrounding the decision process. One of the advisory committee members quipped, “Why not 60? I’m 60 years old. I’m going to be left out.” Also, the decision to include prisoners and homeless shelters while ignoring urban centers that lack adequate hospital beds and staff is a mystery. I digress.


You may be asking yourself…Why the confusion? Why not give the option to everyone and let them decide? That would certainly be easier, for you, and yours truly, who has been answering this question multiple times per day.


The short answer is…its complicated.


The study that has been most cited on the topic of boosters was published in NEJM two weeks ago. It is a retrospective observational study out of Israel. Israel began its booster campaign in July 2021. The study investigated individuals > 60 yo that were at least 5 months from the 2nd dose of their original vaccine. They excluded individuals for the following reasons: missing data, traveled abroad, received a booster before July 30th, vaccinated before January 16th, 2021, PCR positive COVID before July 30th.


Israel has a national vaccine, testing, and hospitalization database. They tapped into this data to identify individuals that received a booster vaccine and compared them to individuals > 60 that did not. The observation period began July 30th and ended September 2nd.


The primary endpoints were broken up into a positive test for COVID and moderate to severe disease/ hospitalization.


They excluded the first 12 days after booster as the convalescent period. Each participant did not get vaccinated on July 30th. They were vaccinated throughout the month. Since the participants had a varying number of days at risk, they used total days at risk for infection instead of number of participants as the denominator in their calculation.


In their results they found that the risk of infection was reduced 11x and the risk of moderate to severe disease/ hospitalization was reduced 19x. These numbers are very impressive, but they are on a massive scale. The study was observing hundreds of thousands of individuals that have received boosters. The comparison is on a relative scale. What that means is that they were comparing the probability of events and not absolute events.


Here is the results table.




Because they are using a relative scale and the denominator has a component of time, the denominators appear inflated. The inflation in denominators likely contributed to the significant effect size. Also, the relative comparison is misleading. This is likely what hung up the FDA. Allow me to explain.


The numbers need to be converted from relative into absolute terms.


It would have been nice to see the mean number of days at risk for each group. However, we do not have that luxury so to get to absolute terms, we need to estimate the absolute number of participants.


Given the convalescent period minus the observation period, we can assume that most individuals in the infection outcome were observed for 20 days. Given the same assumption we can assume that the observation period for severe infection/hospitalization was 15 days, since the average number of days from inoculation to symptoms is about 5.


The absolute risk reduction (AAR) for infection =


ARC = absolute risk in control group

ARE = absolute risk in experimental group


ARR = ARC – ARE = 0.017 – 0.0018 = 0.0152


Even though on relative terms, 0.017 is 11x greater than 0.0018, in absolute terms its only 1.52%. This is the difference between describing statistics in relative vs. absolute terms.


The ARR for severe infection/hospitalization =


ARR = ARC – ARE = 0.00096 – 0.000069 = 0.000891


Even though, on relative terms, 0.00096 is 15x greater than 0.000069, in absolute terms its only 0.09% difference.


To calculate the number needed to booster (NNB) we take the inverse of this absolute risk reduction.


NNB for infection = 1/0.0152 = 66

NNB for hospitalization = 1/0.000891 = 1,205


This corresponds to a number needed to booster (NNB) of 66 to prevent 1 infection and 1,205 to prevent 1 severe illness/ hospitalization.


From a population health perspective, the infection NNB isn’t that bad. In medicine we are typically dealing with numbers needed to treat between 50-100. The prevent hospitalization NNB is quite high. You would have to booster a lot of individuals over the age of 60 to prevent one hospitalization. This number is going to get even higher for younger patients without serious illness.


I would certainly be considered in a lower risk group than someone over 60. I will also be accepting the booster if I become eligible, mainly because I am exposed to COVID most days at work. The cost of my life being disrupted from a COVID infection far exceeds any risk from the vaccine.


But most people are not primary care physicians. The question of whether to get a booster is nuanced, like many of the issues surrounding COVID. Once again lines are being drawn between the pragmatists and the idealists. This was evident this week when the FDA advisory committee recommended against broad use of boosters but was overridden by the CDC director. The pragmatists continue to discuss statistics, individual risks, and allocation of resources, while the idealists point to rising cases and hospitalizations and continue to push for aggressive policies to slow COVID.


This pandemic has left our heads spinning due to its sheer size and the numbers we must comprehend. On one hand, the case numbers and hospitalizations are spun to seem so large. On the other we have slim margins in interventions due to the massive number of exposures. It’s difficult to come up with a broad sweeping policy that will be readily accepted by the public.


Individuals don’t quickly add up the NNB in their heads when trying to decide if they should get the booster. However, they do observe the world around them and come up with an educated guess. Most people have had the infection or know multiple people with the infection that were either asymptomatic or had simple colds. Some see that there are many people out and around not really concerned about COVID anymore. They hear stories about asymptomatic infections and high survival percentage and follow their instincts and behavior of those around them. Then they listen to the arguments being discussed everywhere from cable news networks to a random Facebook post reinforcing their beliefs. Others with opposite experiences will do the same. So, the push and pull between the pragmatists and idealists persists.


This scenario played out this week between the FDA, CDC, and White House with mixed messages being sent out to the American people. First, vaccines are highly effective and current infections are due to the unvaccinated, and boosters are not recommended for all. Second, most vaccinated individuals need to receive a booster. This second message is based on published data that can easily be belittled by anyone with a basic understanding of statistics. All the while, the focus is ever changing, there is no transparency regarding a reasonable path to end the pandemic, which metrics they are basing decisions on, and no consensus benchmark being used to guide their progress.


My advice to the FDA, CDC, and the White House would be the following. First, absolve this back and forth between agencies. One COVID agency (I’ll volunteer myself to be a part of it). Next, focus on the end of the pandemic and work your way back. This can be done by answering several basic policy questions. First, which metrics do we care about most and should be acted upon? Second, at what hospitalization level do regional health systems collapse? How should we intervene prior to this level? Third, how do we convey this level in an acceptable way to the American people, so they will come together when necessary? What prevalence of COVID would the majority of Americans tolerate and would work together to reach (i.e. an acceptable end)? Fourth, what reward can we offer for sustained success that will motivate enough individuals with non-immune status to become immune? Finally, can we track our progress accurately in real time?


Figuring out the "what" we are going to do, and the "why" we are going to do it, and clear benchmarks of success are much needed. I believe this strategy is far superior to the constant mixed messaging from multiple agencies and The White House.


My advice to individuals looking at these numbers is simple. If you are unvaccinated, get vaccinated. The booster data is not evidence that the vaccine is ineffective. In fact, it is the opposite. The numbers are so small after initial vaccination that the booster is just gravy.


link: https://www.nejm.org/doi/full/10.1056/NEJMoa2114255?query=featured_coronavirus


93 views0 comments