"Patient has not tested positive for COVID-19 since having the vaccine"

We see this in the narrative text for a lot of UK VAERS reports. What can we infer from it?

In case you did not know, the CDC has removed the narrative text (description of vaccine injury) from a large fraction of VAERS foreign reports.

The lost myocarditis, death, neuropathy and pulmonary embolism safety signals as part of the free text purge from the VAERS Foreign data set

So this analysis is based on the last dataset which had the narrative text (I got a copy from a fellow VAERS analyst).

Note: I do not directly provide the SYMPTOM_TEXT in my analysis, but for every report, there is a link to the MedAlerts website, which has a history of the report, including what was omitted.

One of the very interesting things about the foreign dataset (specifically the UK one) is the number of reports where you can see some variant of the phrase

“Patient has not tested positive for COVID-19 since having the vaccine”

Given that a lot of adverse reactions are considered to be caused by COVID-19 itself, let us think about the implication of this phrase.

Even if the patient did have COVID-19 at some point, given the fact that the vaccination was the most recent event before the adverse reaction, we cannot rule out the vaccination as the cause of the adverse reaction.

But there are actually two more signals you can use which make this assessment more concrete.

Many reports from the UK which include that phrase also mention two more things:

a) did the patient ever experience COVID-19 symptoms? If the answer is No, then it makes it even more likely that the vaccine caused the adverse reaction.

b) who reported this adverse reaction? If it is a contactable consumer or medical professional and not some prankster, then this also makes it more likely that the vaccine caused the adverse reaction

So I took all the foreign VAERS reports related to COVID19 where the SYMPTOM_TEXT field contained the following phrase

“not tested positive“

and did all my analysis using that dataset as the starting point.

While the phrase does not match the full title of this article, I wanted to shorten it to the smallest subphrase which is likely to not yield false positives.

What are false positives in this context?

A false positive is a report which contains a subphrase you are looking for, but the subphrase does not mean what you might think it does.

As an example, just because you see the phrase “not tested positive” it does not mean it is referring to COVID19, and also it doesn’t mean this was after taking the vaccine.

But as you will see in the dataset, for the most part, searching for this phrase yields matches for the full sentence.

While there are probably going to be some false positives, a quick glance will tell you that it is a very small fraction of the total number of matches.

Additional signals

So in addition to the phrase “has not tested positive”, I am also looking for who reported the adverse reaction, and whether the patient has had symptoms related to COVID19 at any point of time until the report.

For the first signal, I just search for the phrase “contactable”. Not only does it have a remarkably high hit rate (in other words, a very large fraction of foreign reports in the dataset seems to be made by a contactable consumer or contactable health care professional), it also provides at-a-glance information on who made the report. You will understand what I mean by this when you go through the dataset.

For the second signal, I search for the phrase ‘not had symptoms associated with covid-19’ within the SYMPTOM_TEXT. Given that this is a much longer phrase and also much more specific, when this phrase matches, you can be nearly 100% certain the patient did not have any COVID19 symptoms till date.

What about asymptomatic COVID19?

Even if the patient did not have symptoms, it is possible they might have had asymptomatic COVID19 at some point. However, if the vaccination came later than the asymptomatic COVID19, we should attribute the adverse reaction to the vaccine and not to the disease. (Or that’s how it is usually supposed to work).

Additional fields

I have added the following fields to the CSV file representing the VAERS data (it is usually called NonDomesticVAERSDATA.csv).

POSITIVE: The sentence which matches the phrase “not tested positive” in the SYMPTOM_TEXT field

COVID_SYMPTOMS: The sentence which matches the phrase “not had symptoms associated with covid-19“ in the SYMPTOM_TEXT field

REPORTED_BY: The previous 5 and following 5 words after the word “contactable” in the SYMPTOM_TEXT field

In addition to the above, I have added the following fields to the dataset:

DERIVED_AGE: This is based on parsing the SYMPTOM_TEXT field to see if I can extract the age. Unfortunately, quite a lot of foreign reports report the age inside the narrative text but do not translate the value into the AGE_YRS field.

Why is the age field missing in so many foreign VAERS reports?

NUMDAYS_BOUND: There are a few reports where you can see that both the vaccination date (VAX_DATE) and the symptom onset date (ONSET_DATE) are mentioned, but the days-to-onset is not translated. This is because sometimes only the month of the onset date is known. When this happens, we can still establish an upper bound on the days-to-onset. I explained in a previous article how I did this, and use that same approach to add the field in this dataset.

A simple way to improve VAERS analysis

SYMPTOMS: I also look at the corresponding symptoms from the SYMPTOMS dataset and add it as a comma separated list into the SYMPTOMS column

URL: Finally, I provide a link for the history of the particular VAERS report on the MedAlerts website so you can click through and verify everything that I have provided in the dataset.

Results

Out of the 550K COVID19 related foreign reports, 78836 had the phrase “not tested positive” in the SYMPTOM_TEXT

This is the base dataset. You can glance at the POSITIVE field and see that nearly all of these have the sentence “Patient has not tested positive for COVID-19 since having the vaccine”

Of these, 49818 also have the had no COVID-19 symptoms.

Of these, 44368 were made by contactable consumers or contactable healthcare professionals.

In my view, these 44K reports are making a pretty strong case that the vaccine either caused the adverse reaction or cannot be ruled out.

You can do more filtering too - according to CDC’s own guidelines all adverse reactions within a month are considered temporally associated with the vaccination. Of these, when I filter NUMDAYS_BOUND < 31, we get 38133 reports which are also very close to the vaccination date.

I will also be doing a follow up to this article where I look at reports where this temporal association is mentioned by the reporter themselves as a reason not to exclude the vaccine as the cause of the adverse reaction.

Click here to see the dataset