Is litigation a source of over-reporting for COVID19 vaccines?
The case for vaccine date science - Part 2
This is Part 2 of my Case for Vaccine Data Science series.
In a recent article, David Gorski points out that litigation itself could be a potential source of VAERS over-reporting.
Moreover, contrary to Mercola’s claim that we can be “quite certain there’s no over-reporting going on”, we can actually be quite certain that VAERS has been gamed multiple times in its 30 year history. For instance, one of the earliest times I wrote about VAERS was in 2006, when I discussed a study that examined how vaccine litigation could influence VAERS reports. Using VAERS reports from 1990 to 2003, the study found “most case reports to VAERS that were related to overdose, neuropathy, and thimerosal were related to litigation”, as were “many cases” that were related to “autism and mental retardation”. Since we now know with a great degree of certainty that vaccination is unrelated to autism, neuropathy, and mental retardation, we know with a great deal of confidence that these reports represented, if not overreporting, misreporting of AEs not related to vaccination as though they were. The study concluded, “This review shows a previously undisclosed rise in the number of reports to the VAERS related to pending litigation for vaccine injury.” In other words, this is not a new problem with VAERS.
Then I followed the link provided to the original Orac Knows article, which discusses the Methods used by the authors in more detail
In the study, the authors, Michael J. Goodman and James Nordin, did something incredibly simple that no one had done before. They took data from the VAERS database from 1990 through 2003 and imported it into SAS data files for analysis. Then they searched the database using key words to look for reports associated with litigation, particularly with regards to autism. They searched for records containing "thimerosal," "mercury," or "autism" in their fields, especially when coupled with terms like "lawyer," "legal," "attorney," or "litigate," while excluding records containing "legal" coupled with the term "guardian" that did not relate to litigation. They also excluded cases related to well characterized allergic reactions to thimerosal. Finally, they compared records from nonlitigation cases to those from litigation cases regarding symptomatology reported.
Not surprisingly, beginning in 2001, they noted a dramatic increase in the number of non-Lyme disease VAERS reports related to litigation, from only 7 in 2000 to 213 in 2002 and 108 in 2003. (They attributed the decline in 2003 reports to processing delays in creating public use files.) Next, they examined symptom sets related to symptom sets. For autism, they observed a dramatic increase in the percentage of litigation-related reports from 0% of the reports related to litigation in 1999 to over one-third (35%) in 2002. For records mentioning thimerosal that weren't related to allergic reactions, the rise was even more dramatic, from 0% of these reports related to litigation in 2000 to 87% in 2002.
In this article, I will compare the dataset studied by the paper authors, another dataset I constructed which is after the period of study, and finally two more datasets for the COVID19 vaccine era. And then I will list some takeaways at the end of the article which show how we can do better VAERS analysis by using Machine Learning techniques.
I followed the exact process outlined by David Gorski in the above quoted text (the actual journal article is now paywalled).
I wrote some code to combine all the VAERS reports for the years 1990-2003 into one dataset (DS1), the years 2004-2019 into a second dataset (DS2), the non-COVID19 vaccine reports for the years 2020-2023 into a third dataset (DS3) and finally the Covid19 vaccine reports for the years 2020-2023 into a fourth dataset (DS4).
Within these 4 datasets, I filtered for reports where the writeup had any of the following words: “lawyer”, “legal”, ”attorney” or “litigate”. Then I did a second check to omit the phrases “legal guardian” (as the paper suggests) and also to omit two more phrases I found during my analysis (“legally blind” and “illegal drugs”).
So let us first look at the raw numbers. They already explain what is going on.
For the period between 1990-2003 (DS1), there are 704 VAERS reports which are related to litigation, out of a total of 155209 VAERS reports. This amounts to 704/155209 = 0.45% of the total number of reports.
For the period between 2004-2019 (DS2), there are 6224 reports which are related to litigation, out of a total of 513332 VAERS reports. This amounts to 1.21% of the total number of reports.
For the period between 2020-today (DS3), there are 2479 reports related to litigation out of a total of 122614 non-COVID19 reports. This amounts to 2.02% of the total number of reports.
But there are only 420 litigation related reports for the COVID19 vaccine out of a total of 940872 COVID19 VAERS reports (DS4). This amounts to only 0.045% of the total number of reports.
In other words, even as the percentage of litigation related reports is steadily increasing in VAERS over time for the non-COVID19 vaccines, litigation could not have been a cause for over-reporting in the case of the COVID19 vaccine adverse event reports.
My takeaways from these datasets
1 The percentage of reports related to litigation is definitely increasing, and this means you cannot rule out the possibility that there might be some over-reporting in VAERS.
2 However, there are not that many litigation-related VAERS reports if you consider all reports for all vaccines together. (Although it is certainly possible that a lot of VAERS reports for a specific vaccine could be related to litigation).
3 I also added in both the year of the report and the corresponding year of vaccination to see if they are different. As you can see from all the datasets, the gap between the vaccination date and the VAERS report is quite often measured in years for litigation related reports. This suggests that we should be able to use the difference in days between the RECVDATE and the VAX_DATE as a predictor of whether the VAERS report is related to litigation.
4 As the original paper says, it is possible that there could be other reports in VAERS which are also related to litigation but they were not identified because the right keyword was not used during search. I believe it should be possible to use the sentences identified by my Python script as “seed phrases” to create a Machine Learning model to predict if a given VAERS report is related to litigation. One big advantage of understanding how Machine Learning works with Natural Language Processing is that you can go well beyond basic keyword search and start using word vectors etc. for more advanced analysis.
5 By combining ML models and human review, it should also be possible to decide whether or not litigation is actually a source of VAERS over-reporting.
So what do you think?
Both of these studies do not constitute evidence of any type of reporting bias. The only way to measure reporting bias is to follow individual vaccinees, and note which experienced adverse effects and which reported to VAERS. Lazarus' Pilgrim Health study (https://digital.ahrq.gov/ahrq-funded-projects/electronic-support-public-health-vaccine-adverse-event-reporting-system) is the most recent and rigorous example of this kind of evaluation (incidentally, he found dramatic underreporting). Overreporting exists only when there is a report to VAERS that is not matched by independent evidence of adverse effects -- that is, an "overreport" is a fake report. Changes in the proportion and number of VAERS reports that might be related to litigation and exemptions may simply reflect changes in litigation behavior and exemption-seeking independent of VAERS (in fact, legal changes, as in the California law, spur such reactions). My sense is that both of the latter have increased dramatically over time.
Also, even some of the reports flagged as litigation- or exemption-related clearly aren't (such as '"legally" blind' and "injury lawyers know how"). I think there's no replacement for actual human beings to read and code the content, especially when the number of relevant cases is so small.
I define the hallucinatory "over-reporting" as 100% of vaccine harm having been reported in VAERS PLUS additional reports to announce legal action.
This is of course impossible because, along with the large number of reports that CDC deletes, they would not allow a duplicate, no?
I'm seeing 1,461,998 covid reports through 2022 so the mere 420 for covid is less than 29 thousandths of 1 percent.
And that is a meager sliver of an almost non-existent percentage far far far-and-away LOWER than previous years with other vaccines as evidenced by the data presented, is it not?
Perhaps a better question might be: Is it now legal to destroy lives with vaccines?
The answer is yes.
And the reason for that, to cut to the chase: Life is now cheap. We have 385,000 new babies per day. (Google kindly highlights the result when asked). Everything hinges on that.
So we find ourselves in a situation here, where people in power may want to try to reduce the growth rate, may have been convinced by actuary tables projecting our future planetary conditions that it would be beneficial to reduce that growth, and may have a difficult time letting a good crisis go to waste, believing earth's survival is at stake. You argue with them folks, I would find it rather tough.
VAERS remains synonymous with under-reporting.
Unless Gorski's attacking CDC as irresponsibly allowing "over-reporting" in VAERS, he made it up as an act of desperation to hide truth.
I wonder how the pay compares for manufacturing false impressions versus truth telling (a route our keepers have never tried).