My colleagues and I just published a paper in eClinicalMedicine evaluating effects of vaccination on long COVID risks in children and adolescents during the Delta and early Omicron periods.

These data were from the RECOVER network including 21 pediatric hospital networks from all over the USA, including 112,590 adolescents during the Delta period, and 84,735 adolescents and 188,894 children during the early Omicron period.

Long COVID-19 (post-acute sequelae of SARS-CoV-2, PASC, or multi-system inflammatory syndrom, MIS) was defined using a symptom-based computable phenotype definition based on five body systems.

Our analyses utilized propensity score weighting to adjust for confounding from age, demographics, medical co-morbidities as well as healthcare utilization including past COVID-19 testing practices, and we used proximal analyses with negative control exposures and outcomes to investigate and adjust for potential residual bias from unmeasured confounders.

In adolescents 12-20yrs, we found vaccination resulted in 95.4% reduced risk of long COVID-19 during the Delta period, and 75.1% during the Omicron period.

In children 5-11yrs, we found vaccination resulted in 60.2% reduced risk of long COVID-19 during he Omicron period.

To evaluate how much of this vaccine protection was from reduced risk of infection and how much was reduced risk of long COVID-19 independent of any effect in reducing infection, we performed a causal mediation analysis to split the total vaccine effect into indirect effects, mediated through reducing risk of infection, and direct effects, independent of any reduced risk of infection.

Again, propensity score weighting was used to carefully adjust for potential confounders.

We found that the protective effect of vaccines on long COVID-19 was almost wholly mediated through its reduced risk of infection.

Various sensitivity analyses were done and included in the online supplement along with a detailed description and explanation of all methods and modeling decisions.

@chenyong1203

sciencedirect.com/science/articl…

These data were from the RECOVER network including 21 pediatric hospital networks from all over the USA, including 112,590 adolescents during the Delta period, and 84,735 adolescents and 188,894 children during the early Omicron period

Earlier this month, a research group from University of Southern California published a paper studying long term (>1000 days) risk of major adverse cardiac events (MACE, including myocardial infarction, stroke, and all cause mortality) after documented COVID-19 infections from population level data from the UK Biobank.

Confirming many other studies, they found increased risk of MACE after COVID-19, 2.09x higher for all levels of severity and 3.65x higher in hospitalized COVID-19.

Notably, they found that hospitalization for COVID-19 was a coronary artery disease risk equivalent, such that a person with no history of cardiovascular disease who got hospitalized COVID-19 was higher risk than a person with history of cardiovascular disease without hospitalized infection.

This, of course, validates what has been found in many other studies -- that COVID-19 infected have increased risk of severe cardiovascular events even after recovery, especially if they had severe/hospitalized COVID-19, and explains the increased cardiac risk in the past few years.

BTW, this study covers infections in 2020, before vaccines were available.

ahajournals.org/doi/10.1161/AT… The study was quite well designed, using n=10,005 COVID-19 cases between February 1, 2020 and December 31, 2020 in the UK Biobank, compared with population controls (n=217,730) and matched controls (n=-38,660) based on propensity scores including age, sex, ethnicity, education, diabetes, asthma, smoking status, and use of lipid-lowering or antihypertension medication.

The comparison with the propensity score-matched controls is more valid for estimating COVID-19 effects given that it adjusts for confounders.

They found that with >1000 days of followup, the risk of MACE was 3.65x higher in hospitalized COVID-19 group than matched controls.

A group of French researchers published a paper in JAMA today assessing long term cardiac outcome with 18 months after vaccine-caused myocarditis after mRNA vaccines and compared with post-covid-19 mycarditis and conventional myocarditis using a large whole-country cohort covering the entire 12-49 year old population of France.

They found 558 cases of post-vaccine myocarditis out of ~64 million doses of vaccine in this age group (376 after 2nd dose), for an incidence of 1 in 115k doses (1 in 64k after 2nd dose), and 298 with post-COVID myocarditis and 3779 with conventional myocarditis.

A total of 18/15/1 of the 558 post-vaccine myocarditis patients were rehospitalized for myocarditis, had other cardiovascular events, or all cause death, versus 12/22/4 of 298 post-covid myocarditis patients and 277/49/17 of 3779 conventional myocarditis patients.

After rigorously adjusting for confounders including age, sex, region, SES, lifestyle habits, comorbidities, vaccination history, and medications using propensity score weighting to calibrate all populatons to the conventional myocarditis group (the comparison group), they statistically compared the post-vaccine and post-covid groups to the conventional myocarditis group with respect to
(1) Rehospitalization with myocarditis
(2) Cardiac events, including heart failure, heart rhythm and conduction disorders, and cardiomyopathy
(3) All cause death
(4) Rehospitalization for any reason
(5) Composite outcomes (1)-(3)
(6) Composite outcomes (1)-(4)

From this comparison, they found the risk of all of these 6 events were all lower after post-vaccine myocarditis than conventional myocarditis by 25%/46%/47%/31%/45%/36%, respectively, while risk after post-covid myocarditis was similar to conventional myocarditis, at the higher levels.

Myocarditis is the key minority harm risk for mRNA vaccines, especially for young men and especially after 2nd dose.

Before this study, there were very little looking at moderate to long term cardiac outcomes of myocarditis, so this paper is an important addition to the literature.

More study of this important question is needed.

In this thread, I discuss a few of the details and strengths and limitations of this study. SARS-CoV-2 vaccines have been shown by many studies to greatly reduce the morbidity and mortality from COVID-19 during the pandemic, especially during 2021-2022. mRNA vaccines have been the predominate type used in many places around the world, and has consistently shown the highest effectiveness in most studies.

However, within a few months of rollout, a key minority harm risk of myo/pericarditis was discovered to occur in a subset of individuals soon after vaccination, with rates highest in teen boys and young men, especially after 2nd dose given shortly after 1st dose, and with Moderna having the highest rate. It is important to weight this risk against the benefits, especially in young men who have lower risks of severe/fatal COVID-19.

While most cases appeared to be mild and quickly resolved, myocarditis is serious and there is always a risk of cardiovascular sequelae after myocarditis, so long term follow up studies are critical to assess potential long term cardiac damage in those experiencing post-vaccine myocarditis.

How can excess deaths be higher in 2021 in 2020 if vaccines had any benefit?

This is a good question I see many ask.

Some conclude from this question that vaccines must have been completely ineffective, or perhaps even have made things worse.

In this thread, I will show that this is not the contradiction that it appears to be.

If you look at the available data and studies and think through the various relevant factors, it is clear that the following are simultaneously true:
1. Excess deaths in the world were higher in 2021 than 2020
2. Vaccines were highly effective in reducing risk of covid death, the primary driver of excess deaths.
3. The excess deaths in 2021 would have been MUCH WORSE had there not been covid vaccines.

The key factors I will highlight in this thread include:
1. Far more people were exposed to COVID-19 in 2021 than 2020
2. In most places, the strict containment measures of 2020 were lifted in 2021
3. The variants emerging in 2021 were demonstrably more transmissible than those in 2020
4. The vast majority of people in the world were unvaccinated for most of 2021
5. Vaccines were highly effective in 2021, but not perfect. 1. Far more people were exposed to COVID-19 in 2021 than 2020

While the pandemic first emerged and was most disruptive in 2020, relatively few people were actually exposed to SARS-CoV-2 in 2020.

Below are plots of confirmed COVID-19 cases over time in the world from OWID

We see that only 1% (10k per 1m) of the world had confirmed COVID-19 cases in 2020, while another 2.5% (25k per 1m) had confirmed COVID-19 cases in 2021.

So we see 2.5x more confirmed COVID-19 cases in 2021 than 2020, with 2021 a full pandemic year and 2020 only a partial pandemic year.

Of course, most SARS-CoV-2 infections are not formally documented as confirmed cases, so more than 1% were exposed to SARS-CoV-2 in 2020.

Serology studies from around the world provide more information about what % were exposed to SARS-CoV-2.

Mostert et al. published a paper in BMJ Public Health “Excess Mortality across countries in the Western World since the COVID-19 pandemic: ‘Our World in Data’ estimates of January 2020 to December 2022”
()

The paper presents excess death estimates for 47 Western countries from the publicly available “World Mortality Dataset” (Karlinsky and Kobak, 2021) and reports excess deaths from 2020-2022, with 13 countries peaking in 2020, 21 in 2021, and 12 in 2022.

Their conclusion expresses surprise that excess deaths continued from 2020 into 2021 and 2022 “despite the implementation of containment measures and COVID-19 vaccines” claiming “this raises serious concerns.”

They discuss several potential causes of excess deaths, but do not delve into any detail about the evidence or lack thereof for any of them based on existing literature and data.

Their conclusion is that “government leaders and policymakers need to thoroughly investigate the underlying causes of persistent excess mortality”, implying there is not already substantial evidence for what the primary driving factor is, and that studies investigating potential causes of excess deaths are not being done.

Overall, I welcome any call for more studies characterizing how the pandemic affected mortality rates and investigating potential primary and secondary contributing factors. I’m always in favor of more data and studies (and that they are appropriately analyzed using valid statistical approaches).

However, this study does not accurately represent the existing understanding about sources of excess deaths, downplaying the COVID-19 deaths that are clearly the driving factor throughout 2020-2022, as I will show, and implicitly magnifying the potential role of vaccines beyond what is supported by the data.

Also, many popular media articles about this study, including the Telegraph article () entitled “Covid vaccines may have helped fuel rise in excess deaths”, blatantly misrepresent the content of the paper.

They make it sound as if the paper was primarily about vaccines, which it is not, or provides evidence for vaccines being a potential driving factor, which it does not.

This misrepresentation is so egregious that BMJ felt the need to post a statement about it() and, who knows, may be considering retraction of the article.

In this thread, I will discuss some of these points.bmjpublichealth.bmj.com/content/2/1/e0…
telegraph.co.uk/news/2024/06/0…

https://x.com/bmj_company/status/1798751750046019666

First, I am not sure why this article is classified as “original research” and not a “narrative review” or “commentary”.

There is no primary data collection or original data analysis in this paper.

The study design and text and equations for the methods are almost verbatim from the public repository for the World Mortality Data set (WMD, Karlinsky and Kobak 2021 ) which is also where they got their data. The authors cite the paper, but it is highly unusual in an “original research” article to have such large sections match so closely with an existing publication and with nothing new added.

Their results are simply the publicly available excess deaths from Karlinsky and Kobak’s WMD for 47 countries, aggregating by month and year to get total excess counts in 2020, 2021, and 2022, with the results simply showing that there were excess deaths in 2020, 2021, and 2022, which has been well known for a long time.

I have no idea how BMJ Public Health considered this manuscript sufficiently novel for publication as an “original article”.

The remainder of the paper consists of a narrative-driven review of various aspects of the pandemic and a commentary suggesting further investigation of causes of excess deaths, following the format of a “narrative review” or “commentary”, not an original research article.elifesciences.org/articles/69336

In this thread I will make some comments on the recently published paper in Cureus entitled "Increased age-adjusted cancer mortality after the third mRNA lipid nanoparticle vaccine dose during the covid-19 pandemic"

The title is extremely misleading, given their data do not show an increase in age adjusted cancer mortality, and the paper provides no evidence relating the cancer mortality trends to the 3rd doses of mRNA vaccines given in the country outside of simple post hoc arguments based on the fact they were given in 2022.

Their paper shows that the age-adjusted cancer mortality remains stable and in fact slightly decreases, but it decreases at a slower rate than predicted by linearly extrapolating pre-pandemic decreases in cancer mortality from 2010-2019.

They don't provide any evidence that the vaccinated or those receiving 3rd dose boosters are at any higher risk of cancer death than those of similar age/co-morbidity status who didn't, or otherwise provide any evidence that vaccination is responsible for the arrest of the pre-pandemic decreasing trend and not any one of the many other factors greatly impacting life in Japan during the pandemic.

While the paper summarizes a large and important data set and does some nice things (like compute age-adjusted rates), the general conclusions they make, and certainly the conclusions that people are representing about the paper on social media, are not supported by the data.

cureus.com/articles/19627… First, some positives.

This paper analyzes a large data set of population-level cancer death rates in a large country, Japan, and this type of large data set is valuable for assessing trends during the pandemic.

The authors duly recognize that changing age distributions over time impact the cancer incidence and death rates, so compute age-adjusted cancer mortality estimates which is the right approach to assess changes over time.

They also looked at subgroup analyses by cancer type and age, which provide useful breakdowns in smaller groups that can shed light on the trends.

New paper in Heart compares vaccinated to unvaccinated with respect to cardiovascular events including myocardial infarction, heart failure, arrhythmia/cardiac arrest, myocarditis and clotting events including strokes, thrombosis deep vein thrombosis, pulmonary embolism, and venous thromboembolism through 1, 2, 6, and 12 months in electronic medical record cohorts of 20 million from UK, Spain and Estonian.

They adjust for measured cofounders using propensity scores weighting, adjust for unmeasured confounders using negative controls, and federated learning to combine data across countries.

All events had significantly lower risk through 1m including myocarditis, and most through 12 months, providing strong evidence vaccination rollouts substantially reduced risk of cardiovascular events.

heart.bmj.com/content/early/… I should have clarified these cardiovascular events being compared are post SARS-CoV-2 infection

Also should have specified it covers the time period starting early 2021 when rollouts started and the various data sources ending December 2021, February 2022, June 2022, and December 2022, so is predominately in pre-Omicron era, with a lot of early Omicron era and a small part later in 2022 Omicron era.

Recall that from many dozens of studies vaccines strongly reduced risk of infection throughout 2021, and this effect was much weaker in 2022 with Omicron variants, but looking across the dozens of studies during the Omicron era in 2022 reduced risk of infection was still evident.

Here I will briefly discuss this paper you linked talking about negative effectiveness observed in some subgroups for a small subset of studies

I will first summarize their points, and then go through every study they mention.

It is not a reasonable viewpoint that vaccines have negative effectiveness and make covid risks worse based on the many published studies and data Here is their overall conclusions: that these rare cases where negative VE is observed is likely causedThey discuss some of these ,biases by confounding bias in settings where true VE is very small such as Omicron vs infection long after vaccination

An Epoch Times article discusses a recent paper by FDA researchers assessing safety of Pfizer and Moderna bivalent boosters given to 8,638,661 and 5,240,178 individuals, respectively, comparing incidence rate of 18 different serious adverse events of special interest (including various cardiovascular events) in data bases from Carelon Research, CVS Health, and Optum.

Out of all 18 events, they only identified safety signals for:
1. anaphylaxis for both vaccines 18-64yr in 1 out of the 3 data bases (but not the other 2), and
2. myocarditis/pericarditis for 18-35yr for Pfizer vaccine in 1 out of the 3 data bases (but not the other 2)
They concluded results were consistent with previous studies and supporting the safety profile for these vaccines

However, the Epoch Times article highlights the following numbers, suggesting they are from this paper:
1. anaphylaxis rate was 74.5 cases per 100k person-years following Pfizer vaccination
2. anaphylaxis rate was 109.4 cases per 100k person-years following Moderna vaccination
3. myo/pericarditis rate was 131.4 cases per 100k person-years
These figures to a novice reader might make one think that the rate of anaphylaxis per person is
1 per 1342 (100,000/74.5) after Pfizer,
1 per 914 (100,000/109.4) after Moderna,
and the rate of myo/pericarditis is
1 per 761 (100,000/131.4) after Pfizer,
which would seem alarmingly high, seemingly contradicting the conclusions of the paper. However, this is completely false.

As I will show in this thread, these figures:
1. were not even reported in the paper, but computed for the Epoch Times article from Table 3.
2. were actually mis-computed from Table 3 in the paper
3. based on rate per person-year is not the most useful summary for incidence after vaccination given the time frame for these events varies from 2 days (anaphylaxis) to 28 or 42 days after vaccination, and easily misconstrued/misinterpreted.

Looking at the paper, what they actually found for incidence was:
1. anaphylaxis from d0-d1 after bivalent vaccine booster occurred at rate between 1/4.3m and 1/430k for Pfizer, and between 1/2.6m and 1/260k for Moderna
2. myo/peri-carditis from d0-d21 after bivalent vaccine booster occurred at rate of 1/27k for Pfizer and 1/29k for Moderna
3. For 5-17yr olds, the myo/peri-carditis rate from d0-d21 after bivalent vaccine booster occured at rate between 1/535k and 1/54k for Pfizer, and <1/50k for Moderna
4. For 18-64yr olds, the myo/peri-carditis rates from d0-d21 after bivalent boosters was 1/53k for Pfizer and 1/55k for Moderna

These are in line with previous literature, and not alarming -- the calculation and inclusion of the rates per person-year in the Epoch article appear to be trying to exaggerate the risk

I'll transparently explain where I get these numbers from in this thread...

theepochtimes.com/health/fda-fin… First, let's consider the anaphylaxis data. Here is the part of Table 3 containing the number of cases and person-years of follow up for two vaccines for the different age groups.

Privacy rules forbid them from publishing any raw counts <11 in the table, so "<11" means some unknown number between 1 and 10.

Based on that I compute the event rate per 100,000 person years.

For Pfizer, we see that combining age groups, the event rate is between 4.3 and 42.7 anaphlaxis events per 100k person-years

This is significantly lower than the 74.5 per 100k person-years that Epoch put in their article -- not sure where they got that from.

However, this is not a very meaningful measure since this event is only defined at d0-d1 after vaccination, meaning a person is only at risk for 2 days for this event.

It would be more meaningful for us to consider the proportion of individuals who experience anaphylaxis in d0-d1 after the vaccine, i.e. at the individual level, not person-years.

1/n
Annals of Internal Medicine just published our epidemiological study of vaccine effectiveness in children and teens during the delta and omicron waves based on large USA pediatric cohorts accounting for nearly 4% of the USA pediatric population.

The study found strong vaccine effectiveness vs. infection, severe disease, and ICU during the delta wave for adolescents and the omicron wave for children and adolescents, with no evidence of increased risk of cardiac outcomes.

Senior authors are Yong Chen, Christopher Forrest, and Jeffrey S Morris, and Lead authors are Qiong Wu and Jiayi (Jesse) Tong.

#PEDSNet #COVID_19 #COVIDVaccine @ChildrensPhila @UPennDBEI @PennMedicine

acpjournals.org/doi/10.7326/M2…
2/n
Study Data
This study was done using electronic medical records data from PEDSnet, a network of 8 USA pediatric health systems across 23 states and the District of Columbia, capturing nearly 4% of all children and adolescents in the USA, including large primary care systems, including the Children’s Hospital of Philadelphia, Nationwide Children’s Hospital, and Nemours Children Health.

This study included a total of 77,392 adolescents (45,007 vaccinated) during the Delta wave, a total of 111,539 children (50,398 vaccinated) and 56,080 adolescents (21,180 vaccinated) during the Omicron wave, with extensive follow up of these cohorts providing the longest follow up among studies of pediatric vaccine effectiveness.

@TracyBethHoeg, Duresetti, and @VPrasadMDMPH published a commentary in @NEJM yesterday pointing out unmeasured confounding due to the healthy vaccinee effect (HVE) in a December 2021 @NEJM paper published by a group led by Clalit researcher @ArbelRonen.

They pointed out a 94.8%… https://t.co/KAUrLUtaL4twitter.com/i/web/status/1…
The healthy vaccinee effect, or healthy user effect, is a well-known phenomenon that people who follow public health recommendations tend to be healthier in general, and this can manifest in vaccine effectiveness studies as unmeasured confounders beyond those included in the… https://t.co/EJ9J23oqRitwitter.com/i/web/status/1…

I have seen many people forwarding this graphic around, claiming that the UK ONS data show that the death rate for vaccinated individuals was much higher than unvaccinated, suggesting vaccines were dangerous and killing people.

This plot is inaccurate and misleading.

In this 🧵, I will demonstrate that the UK ONS data show consistently lower death rates for the ever vaccinated than the unvaccinated throughout the pandemic, for COVID-19 deaths, non-COVID-19 deaths, as well as all cause deaths.

The only way people can spin these data to suggest vaccinated have higher death rates is if they magnify the very small vaccine subgroups and minimize the majority of the vaccinated, as I will show. @Tony48781320 @DuttyMonkey_ Here is the data for 40-49

https://twitter.com/jsm2334/status/1679994146277453824?s=20

@1onetenthdegree @czssschhrsxcf @TheChiefNerd @Hammersmith84 @_aussie17 The paper is a basic ecological analysis, simply plotting deaths over time, estimating excess deaths based on some baseline pre-pandemic death rate assumptions, and then trying to read into the variability of these data in various age groups over time, in particular the fact that… twitter.com/i/web/status/1… @1onetenthdegree @czssschhrsxcf @TheChiefNerd @Hammersmith84 @_aussie17 They don't test any hypotheses or perform any real inference in this paper, but just plot descriptive summaries and try to interpret what they see and speculate about potential causes.

Today UK ONS released a vaccine effectiveness analysis based on their new national data set linked to 2021 census estimating VE vs Covid hospitalization and death and effect on risk of all cause death while adjusting for many key confounders.

https://twitter.com/sarahcaul_ons/status/1633408733945507840

They started 3/31/21 since it was not possible to consider earlier dates given the data set was based on the census, and measured using a 1 year period through 3/31/22

Myocarditis is well known to be a key risk of mRNA vaccines, especially for young men, as well as COVID-19 infection.

This large Nordic study compares clinical outcomes after vaccine-associated, COVID-19 associated, and conventional myocarditis:

bmjmedicine.bmj.com/content/2/1/e0… Assessing clinical sequelae of myocarditis of different causes has been a critical under-addressed scientific question.
This paper does so using registry data from 4 Nordic countries, Norway, Sweden, Denmark, and Finland, taking all cases of clinical myocarditis from 2018-2022.

@FLSurgeonGen 1. Since death events clearly prevent future exposure periods, this violates the assumptions underlying SCCS. Have you made adjustments to your analysis to take this into account and mitigate the bias it causes?

From the article you linked: @FLSurgeonGen 2. Also, by not including all exposure times but only the most recent one before death, do you think this can be another source of bias?

@EthicalSkeptic still doesn't want to show exactly where his results come from, but instead wants to continue personal attacks & misrepresenting posts of those who are skeptical about his claims and asking him how he reaches his conclusions when raw data show no increase in 2022 I'm glad to see my erudite friend's dictionary word of the day was "perdocent" and he got to use it trying more off base ad hominem attacks against me instead of responding to my questions
I am not "appealing to credential" or throwing "tantrums" just asking him to show his work

@EthicalSkeptic You missed the point of my post.

Twitter makes it hard to follow back and forth responses so I will number my responses
(1/n) @EthicalSkeptic If you read the report, it is clear my goal was not to do a detailed de novo analysis taking all factors into account, but to assess whether the data support your conclusions, & to identify which of your assumptions (stated/unstated) drive them.
(2/n)

I've seen recent strong claims that there have been substantial excess cancer deaths since Spring 2021 based on the CDC cause-specific death data from
(1) prepandemic (2014-2019) to establish baseline rates
(2) provisional deaths during pandemic (2020-2022)
Let's take a look. Here I consider the 8/17/22 CDC release.
1. First I combined 2014-2019 & 2020-2022 and plotted the raw reported weekly cancer deaths, stopping at 3/17/22 since the provisional nature of the 2020-2022 data set means not all cancer deaths in recent months have been uploaded yet.

This video plots % excess deaths for USA states vs. the state's "propensity to vaccinate", i.e. % fully vaccinated as of 7/31/22, for 1 month windows of time starting just before the pandemic and continuing through 7/31/22.

Let's assess these associations over time. (1/n) First, I need to clarify that we are not plotting vaccination rate over time, but rather the "propensity to vaccinate" for each state given by the total % fully vaccinated (2 doses mRNA or 1 dose viral vector) as of 7/31/22. (2/n)

I was curious to see how excess deaths related to vaccination rate by state in the USA.

Here is % excess deaths (3/1/20-7/31/22) vs. % fully vaccinated (7/31/22) from CDC all cause death data.

Nonlinear smooth fit shows strong negative correlation (Spearman r=-0.58, p<0.00001). As I have repeatedly emphasized, one cannot infer individual vaccine effects from scatterplots of aggregated event vs. vaccination rates b/c of ecological fallacy, since systematic state-to-state differences are confounded w/ vaccine effects, especially when for short time period