New CDC report on myocarditis after infection or vaccination from health systems covering 20% of USA

Myocarditis has been validated as a minority harm risk of mRNA vaccination. This risk was discovered early in 2021, and was added to the Pfizer vaccine insert in June 2021 for public disclosure. Numerous studies have been done in the past year to validate and characterize the risk of myocarditis/pericarditis after mRNA vaccination, which is especially notable in teen boys and young men. It is also a risk after COVID-19 infection, as has been established by numerous studies in the past 18 months. Whether caused by infection or vaccination, myocarditis/pericarditis results from an overactive immune response producing inflammation of the heart. Further characterizing these risks is important to assess the benefit and risk of vaccination in children and teens, especially boys.

On April 1st, the CDC released a report tracking and comparing rates of cardiac issues (myo/pericarditis) after infection or mRNA vaccination, combining information across electronic medical records (EMR) from 40 health systems in the USA covering a total of 66 million Americans, 1/5 of the national population.

From these data, they constructed cohorts of ~800k with documented and confirmed COVID-19 infections, ~2.5m with documented 1st dose vaccination with known rate, ~2.5m with documented 2nd dose vaccination, 17.m with documented vaccination of unspecified dose, and 6.7m with any vaccine dose. Here are the demographics of the various groups.

The study estimated rates of myocarditis/pericarditis per 100k between 1/1/21 and 1/31/22 within 7 days or within 21 days of infection, 1st vaccine dose, 2nd vaccine dose, unspecified vaccine dose, or "any" vaccine dose. The denominators were the cohort sizes, and numerators were subset of the cohort with diagnosis of myocarditis/pericarditis within the specified time range. They computed the relative risk ratio (RRR) of infection to each of the vaccination cohorts by taking the ratio of the corresponding rates per 100k.

With multi-inflammatory syndrome (MIS-C) being a key post-COVID-19 complication in children, they also compared the rates of myocarditis/periocarditis/MIS-C in infected vs. rates of myocarditis/pericarditis in the various vaccinated groups using 7d, 21d, and 42d time frames.

Here, I will first focus on summarizing the results of incidence of myocarditis/pericarditis per 100k within 21 days of baseline (infection or jab date) for boys, the group at highest risk.

Age 5-11yr:

The rates were 17.6 per 100k within 21d of infection, and 4.0/0.0/6.5 per 100k within 21d of 1st/2nd/unspecified dose of mRNA vaccination. This corresponds to a relative risk ratio (RRR) of COVID-19 infection to 1st dose of 4.4 (95% confidence interval 0.5-35.7), meaning the rate within 21d of infection was 4.4x higher than within 21d of1st dose.

The rate of myocarditis+pericarditis+MIS-C within 21d of infection was 103.0, with corresponding RRR vs. myocarditis+pericarditis within 21d of 1st dose of 25.7 (3.5-187.0), meaning the rate of myocarditis+pericarditis+MIS-C within 21d of infection in this age group was >25x higher than the rate of myocarditis+pericarditis within 21d of 1st dose vaccination.

The number of events for this age group is very small, which is why the confidence intervals are so wide.

Age 12-17yr:

The rates were 64.9 per 100k within 21d of infection, and 3.3, 35.9, and 29.7 per 100k within 21d of 1st/2nd/unspecified dose of mRNA vaccination. This corresponds to an RRR of 19.8 (5.9-66.2)/1.8(1.0-3.1)/2.2(1.1-4.2) of risk within 21d of infection to within 21d of 1st/2nd/unspecified dose.

The rate of myocarditis+pericarditis+MIS-C within 21d of infection was 159.3, with corresponding RRR vs. myocarditis+pericarditis within 21d of 1st/2nd/unspecified vaccine dose of 48.7 (15.2-155.7), 4.4 (2.9-6.9), and 5.4 (3.1-9.4).

Age 18-29yr:

The rates were 100.6 per 100k within 21d of infection, and 8.1, 15.0, and 27.8 per 100k within 21d of 1st/2nd/unspecified dose of mRNA vaccination. This corresponds to an RRR of 12.5 (6.2-25.2), 6.7 (3.9-11.7), and 3.6 (2.3-5.8) of risk within 21d of infection to within 21d of 1st/2nd/unspecified dose.

The rate of myocarditis+pericarditis+MIS-C within 21d of infection was 112.3, with corresponding RRR vs. myocarditis+pericarditis within 21d of 1st/2nd/unspecified vaccine dose of 13.9 (7.0-28.0), 7.5 (4.4-13.0), and 4.0 (2.5-6.4).

Age 30yr+:

The rates were 114.0 per 100k within 21d of infection, and 7.3, 7.3, and 20.1 per 100k within 21d of 1st/2nd/unspecified dose of mRNA vaccination. This corresponds to an RRR of 15.6 (11.8-207), 15.6 (11.7-20.7), and 5.7 (4.5-7.1) of risk within 21d of infection to within 21d of 1st/2nd/unspecified dose.

The rate of myocarditis+pericarditis+MIS-C within 21d of infection was 123.0, with corresponding RRR vs. myocarditis+pericarditis within 21d of 1st/2nd/unspecified vaccine dose of 16.8 (12.7-22.3), 16.8 (12.7-22.2), and 6.1 (4.9-7.7).

Thus, we see that in all age groups, the rate of myocarditis/pericarditis in boys/men was greater within 21 days of infection than within 21 days of any dose of mRNA vaccination, with the relative risk ratios ranging from 4x up to nearly 50x.

The rates of myocarditis/pericarditis were much lower in females, but again 6-10x higher within 21 days of infection than from any dose of vaccination.

This shows that in the USA as represented by these 40 health systems covering 1/5 of the population, while the mRNA vaccines clearly present some risk of myocarditis/pericarditis in young men within 3 weeks of the jab, a confirmed COVID-19 infection carries a much higher risk than a shot of mRNA vaccine. This study provides additional valuable data to consider when evaluating benefit/risk of mRNA vaccination for children, teens, and young adults.

Some study limitations:

There are a number of limitations of this study. First, it doesn't split out by vaccine type. Given other studies have shown Moderna has higher rates of myocarditis, that would have been useful to see.

Second, it did not have a control group that was unvaccinated and not previously infected to use to establish the excess rate of myocarditis in recently vaccinated or infected.

Third, the infected cohort included both unvaccinated and vaccinated individuals. Had the data been available, it would have been instructive to see whether the breakthrough infections have higher or lower rate of infection-based myocarditis. Arguments could be made either way -- some could suggest the rate of infection-based myocarditis is higher in breakthrough infections as the risk of myocarditis from vaccination and infection could accumulate, while others could suggest that it is lower given the fact that vaccines are efficacious in reducing risk of severe or critical virus, and severe or critical disease may be more likely to lead to inflammatory complications like myocarditis/pericarditis. It would have been good to have data to assess this subgroup analysis.

Fourth, they did not compute cumulative risk of multiple doses, or multiple infections. They computed the risk after each dose, and their "any dose" category just summed the numerators and denominators for each of the dose categories. Summing the incidence of myocarditis after 1st or 2nd dose as the numerator divided by the total number given 2 doses in the denominator would have provided a cumulative risk. Also, many teens have been infected by COVID-19 multiple times, and similarly the cumulative risk of multiple infections could have been computed by summing the incidence of myocarditis after any documented infection in the numerator by the total number of individuals in the denominator.

Fifth, there may be selection bias in this study given that only infections and vaccinations documented within the health system EMR are considered in the denominator of the myocarditis rates. This has led some on social media to dismiss the study, for example:

The argument they make is that the post-infection myocarditis rate is biased upwards, implying that this bias is the reason why the post-infection myocarditis rates are higher than the post-vaccination myocarditis rates.

Because this is a subtle but important issue, I will discuss this question of selection bias, denominators, and representativeness of data from health systems.

Selection bias, denominators, and representative sampling:

As I will now explain, the claim that "wrong denominators" were used is not accurate, although those making this claim are noting a type of selection bias that is relevant and requires further thought.

Strictly speaking, the study used the correct denominators. For myocarditis rate after infection, the denominator used was the size of the infection cohort, which was ~800k, and this is the right denominator to use with the numerator that was the subset of this cohort with myocarditis within 21d of the respective vaccine dose. Inflating this denominator would be wholly inappropriate. So the denominators are perfectly legitimate and the correct ones to use in this study.

The question is whether the subset of confirmed infections documented in the EMR of these health systems are representative of all infections, and most importantly, whether the rate of myocarditis computed from this subset is a biased estimate of the rate of myocarditis within 21d of any confirmed infection. This is a valid concern.

The infection cohort of this study only included ~800k confirmed infections documented within the EMR of one of these 40 health care systems covering 66m USA residents, 1/5 of the population. From https://ourworldindata.org we can see that between 1/1/21 and 1/31/22, there were ~55m confirmed infections in the USA. Given these 40 health systems cover 1/5 of the USA population, we would expect ~11m confirmed infections among USA residents in these health systems. Thus, the infection cohort comprises a select subset of only ~800k/11m = 7.3% of total confirmed infections. Is this a representative subset? Almost certainly not -- since the documented infections should include those detected at the hospital among those who were presenting with a medical conditions, we would expect this subset to be overrepresented by more severe infections, which could also mean a higher rate of post-infection myocarditis. This would suggest the post-infection myocarditis rates reported in this study may be biased upwards as an estimate of the overall myocarditis rates within 21d of confirmed infection, validating their concern. However, those stating that concern don't account for the fact that the vaccination cohort is subject to the same potential bias.

The 1st dose vaccination cohort in this study includes ~2.5m whose 1st dose vaccination is documented in the EMR in one of these 40 health systems. From https://ourworldindata.org we can see that between 1/1/21 and 1/31/22, there were ~250m USA residents who were given their 1st mRNA vaccine shots. Given these 40 health systems cover 1/5 of the USA population, we would expect ~50m of the USA residents in these health systems to have received their 1st dose during this time frame. Thus, the "1st dose vaccination cohort" comprises a select subset of only ~2.5/50m = 5.0% of those receiving their 1st vaccine dose. Thus, it appears the proportion of 1st dose vaccinations documented in the EMR here is even smaller than the proportion of confirmed infections documented in the EMR. Is this a representative subset of all individuals receiving their 1st dose of vaccine? Almost certainly not -- A vaccinated person with health problems (including myocarditis) is more likely to come to hospital, and then more likely to have vaccination status uploaded to their EMR than someone vaccinated at CVS who never has to go to the hospital. This would suggest the post-vaccination myocarditis rates reported in this study may be biased upwards as an estimate of the overall myocarditis rates within 21d of 1st dose vaccination in the population. That is, there is the same potential selection bias in the vaccinated cohort.

There is always risk of this type of selection bias in health system based observational studies, and this potential is exacerbated in health systems with more incomplete records of confirmed infections and vaccinations. Countries with centralized healthcare systems like the UK or Israel have more complete documentation of infections and vaccinations, as their system is more integrated and it is more likely that vaccinations and positive PCR tests will get integrated into the patients' EMR than decentralized countries like the USA with many different medical systems, and for which many testing and vaccination sites did not automatically send results to a centralized EMR. This is one reason why it is the UK and Israel that have provided some of the more complete and rigorous observational studies in the pandemic, and why the USA is underrepresented in spite of its wealth and intensive biomedical research system. While there is a concern that there might be some selection bias in the small subset of infections (~7.3%) documented in these health care systems, there is also concern of the same type of selection bias in the small subset of vaccinations (~5.0%) documented in the same systems. Thus, this concern is not enough to dismiss the study, as many critics have been trying to do.

Conclusions:

The well-documented myocarditis risk, especially in young men and teen boys, has been suggested by some as a reason to not pursue vaccination in these age groups. But the data in this study show that the benefit of vaccination for young adults and minors should not be dismissed based on the documented risk of myocarditis/pericarditis in young males, but that the even higher risk if infected should be taken into account, as well. The CDC study does not attempt to perform a complete cost/benefit analysis for any of these age groups nor does it make any prescription for mandates, but rather concludes "these findings provide important context for balancing risks and benefits for mRNA COVID-19 vaccination among eligible persons >=5yr." Full cost-benefit analyses require more data and detailed analyses.

But the demonstration that the major validated risk of vaccination, myocarditis, is even higher after infection, along with the benefit of vaccination in reducing risk of severe/critical/fatal COVID-19 that has been documented in many studies provides support for vaccinating these age groups. These benefits are seen even against Omicron, as seen in the recent NEJM study just published focusing on children and adolescents.

This paper used a test negative design to estimate vaccine effectiveness vs. hospitalized and vs critical COVID-19 in children and teens from 31 children's hospitals across 23 states from July 2021 to February 2022.

For 12-18yr found VE=92%/96% vs. hospitalized/critical COVID-19 during Delta wave and 40%/79% during Omicron wave. For 5-11yr, VE=68% vs. hospitalized COVID-19 during Omicron wave.

While of course the risk of severe/critical/fatal COVID-19 is much lower in young adults, teens, and children than older adults, we still see severe/critical/fatal cases occur, and also we see MIS occurring in children. This provides support for the availability of vaccines in these age groups.

More studies are needed to quantify long term risks of myocarditis after infection or vaccination, and the net benefit of vaccination in preventing infection-based myocarditis and other sequelae of COVID-19 infection relative to the vaccine-based myocarditis to provide a more complete picture that will help inform vaccination decisions for children and adolescents. Appendix: "If myocarditis rates after infection are so high, then why did we not hear about myocarditis in 2020 before vaccination?"

Some may ask: "If myocarditis rates are higher after infection than vaccination, why did we not hear about myocarditis in 2020, before vaccination?" Good question. Let's check the data.

The study summarized above estimated for 12-17yr males the rate of myocarditis within 21d of:

1. COVID infection was 64.9 per 100k

2. 1st dose of mRNA vaccine was 3.3 per 100k

3. 2nd dose of mRNA vaccine was 35.9 per 100k.

Let's consider how many cases of myocarditis we'd expect in 2020, and how many we'd expect since 1/1/2021, if the estimates reported in this paper were correct. From the report from the American Academy of Pediatrics (AAP), we can see that through 12/31/20, the number of children with confirmed COVID-19 infections was 2,128,587, which was 12.4% of the total number of cases in the USA 17,137,295. We can see from aggregated case statistics by age to date that among those confirmed COVID-=19 cases <18yr, 43.2% are 12-17yr, which would give us an estimated: 43.2% x 2,128,587 = 919,867 cases 12-17yr in 2020.

Assuming ~1/2 are boys, that would would be 459,934. The myocarditis rates of 64.9 per 100k within 21 days of infection would imply we'd expect to have: 64.9 x 4.59934 = 298 myocarditis cases from infection in 12-17yr boys in 2020. How does this compare to how many we'd expect since 1/1/21?

Well, from the AAP report we can see that from 1/1/21-3/24/22, there have been 10,685,061 confirmed COVID cases in <18yr group out of a total of 50,245,525 confirmed COVID cases in the USA, which is 21.3% of all cases. So we can see that <18yr have made up a larger proportion of total cases since 2021 (21.3%) than in 2020 (12.4%), plus we have had a massive wave in January 2021, Delta waves in summer/fall 2021, and the Omicron wave in winter 2021 that affected teens much more than 2020 waves

Given 43.2% of <18yr cases are 12-17yr, this means that we expect to have seen: 43.2% x 10,685,061 = 4,615,946 cases 12-17yr after 2020, say ~2,307,973 boys. The 64.9 per 100k myocarditis rate would imply: 64.9 x 23.07973 = 1498 myocarditis cases within 21d of infection since 2020. Consider the vaccination rates of 12-17yr. We have ~54% of 12-17yr in USA have been fully vaccinated (2 doses) Given there are ~12.5m 12-17yr boys in the USA, which means we have: 54% x 12.5m = 6.75m 12-17yr boys vaccinated with 1st & 2nd dose in USA

Based on the estimated myocarditis incidence of 3.3 per 100k after 1st dose and 35.9 per 100k after 2nd dose, we would expect: 3.3 x 67.5 = 223 myocarditis cases within 21d of 1st dose 35.9 x 67.5 = 2423 myocarditis cases within 21d of 2nd dose, and up to

39.2 x 67.5 = 2646 myocarditis cases within 21d of either 1st or 2nd dose.

Thus, we'd expect 2646 myocarditis cases within 21d of 1st or 2nd dose of vaccine in the 12-17yr male age group, and 1498 within 21d of confirmed COVID-19 infection, for a total of 4144 myocarditis cases in 12-17yr males since 1/1/21. This 4144 is almost 14x higher than the 298 expected in 2020, which would be far more noticeable. It seems plausible that ~300 extra myocarditis cases after infection in 2020 might go unnoticed given background rate of myocarditis in the population, but 4144 extra cases coming from the much higher numbers of 12-17yr infected with COVID since 2020 along with those coming from vaccinating half the cohort could not. So this argument that "we would have noticed myocarditis in 2020 if the post-infection myocarditis rates were correct" is not necessarily true, so we cannot dismiss these results on the basis of this argument.

Again, as emphasized above, myocarditis is not the only risk of COVID-19 (or vaccination), so these numbers are only one piece of the puzzle in assessing cost/benefit of vaccination in teens.