10 ways to make the vaccine look safer

When there are duplicate reports in VAERS, there seems to be a strong tendency to retain the report which makes the vaccine look more safe (and delete the one which makes it look more dangerous)

In the previous article, I discussed deleted reports in VAERS. Many of these reports were deleted because they were duplicates.

Analyzing deleted VAERS death reports

When you are deleting a report, you have two options:

• keep the report which makes the vaccine look more safe

• keep the report which makes the vaccine look less safe

(There is also a third option, which involves merging follow up information and creating a new report. I believe that is also being done in some cases)

For this particular article, I am interested in the data which is retained and which is

a) publicly queryable

b) included as part of aggregate statistics

In other words, I am interested in the current state of a given report and refer to it as the “retained” report. It is possible the retained report includes some followup information.

How to identify duplicates

So I start with the dataset from the previous article, and I only include deleted reports where all these fields are non-empty:

• calculated age (this is the age of the person on the date of vaccination)

• sex

• state

• vaccination date

From the original report (which had about ~25000 rows), this filter gave me a list of ~15100 reports.

Then I query the three datasets (CSV files) for the years 2020, 2021 and 2022 to get a list of records where these four pieces of information match.

For e.g. here is what the pandas query looks like for the 2020 dataset:

df_2020_vax_date_and_age_match: pd.DataFrame = df_2020.loc[
    (df_2020['VAX_DATE'] == vax_date) & (df_2020['CAGE_YR'] == calc_age)
    & (df_2020['STATE'] == state) & (df_2020['SEX'] == sex)]

Sometimes this provides multiple results.

I then scrape the medalerts page for the Symptom list for the deleted report (to the best of my knowledge, this is the only way to get that list).

Once I have the list of symptoms, I iterate through all the matching records from the previous step and use the one where the maximum number of symptoms match. If this number is zero (that is no symptoms match between the deleted report and the retained report), I exclude it from consideration.

10 Ways to make the vaccine look safer

There are a lot of fields in VAERS reports which indicate the seriousness of the issue.

1 SYMPTOM_TEXT

This is the narrative description of the report. If the deleted report has more information (i.e. a more detailed narrative) than the retained report, I flag this report.

Note: as you can see, this is a pretty crude way to do an analysis. A slightly better way would have been to look at the ENTITY_SCORE of the two reports. However, I am only flagging the information for further investigation, and it is certainly possible that the retained report has more useful information.

List of reports for SYMPTOM_TEXT

As you can see, there are 4655 results.

We started with a list of about 15000 reports, so this is over 30% of all reports where the report with the shorter narrative description was retained.

While this means it is quite common, we cannot simply say this automatically makes the vaccine look safer. Sometimes shorter reports are concise but still accurately report the full picture. I hope to look into this in a future article.

2 DIED

This field records if the patient died.

If the deleted report is marked Yes and the retained report has a different value, I flag this row.

List of reports for DIED

There are 41 such reports.

If anyone can provide a good reason for why this is OK, I would certainly like to know in the comments.

3 L_THREAT

This field records if the report indicates a life threatening issue.

If it is marked as Yes in the deleted report, but has a different value in the retained report, I flag this row.

List of reports for L_THREAT

There are 238 such reports

4 DISABLE

This field records if the patient became disabled.

If it is marked as Yes in the deleted report, but has a different value in the retained report, I flag this row.

List of reports for DISABLE

There are 358 such reports

5 RECOVD

This field records if the patient recovered.

In this case, if the deleted report marks this value as No, but the retained report has a different value, I flag it.

List of reports for RECOVD

There are 3311 such reports

6 ER_VISIT

This field records if the patient needed to go to the Emergency Room.

If the deleted report marks this value as No, but the retained report has a different value, I flag it.

List of reports for ER_VISIT

There is only 1 such report, so it seems like a fairly rare occurrence.

7 HOSPITAL

This field records if the patient was hospitalized.

If the deleted report marks this value as Yes, but the retained report has a different value, I flag it.

List of reports for HOSPITAL

There are 320 such reports

8 HOSPDAYS

This field records how many days the patient was hospitalized. If the deleted report has more days and the retained report has fewer days, I flag the information

List of reports for HOSPDAYS

There are 589 such reports

9 ONSET_DATE

This field records how many days it took for the symptom onset (from the date of vaccination). If the deleted report has an earlier date than the retained report, I flag the information

List of reports for ONSET_DATE

There are 1013 such reports

10 X_STAY

This field records if there was an extended hospital stay. If the deleted report says Yes, but the retained report has a different value, I flag the information

List of reports for X_STAY

There is only one such report.

How to interpret the data

Let us take a look at the reports for the cases where the DIED field was changed from Y to Unknown.

List of reports for DIED

(To be clear, this is not usually changed. Two reports were submitted. One of them has a more serious outcome. The other report leaves the outcome field as blank. The report with the less serious outcome is selected. That is the pattern I am describing here)

I am going to consider the case where DELETED_URL is

https://medalerts.org/vaersdb/findfield.php?IDNUMBER=1549750&WAYBACKHISTORY=ON

Link

and the RETAINED_URL is

https://www.openvaers.com/vaersapp/report.php?vaers_id=1332252

Link

You can read the two reports and see that they are referring to the same person.

In case you are wondering, the reason I use MedAlerts website for the deleted report is because they maintain an archive. The reason I use OpenVAERS for the retained report is because unlike the MedAlerts site, their data is kept up to date on a weekly basis. The result on MedAlerts for a single VAERS report does not seem to get updated frequently.

The report with the death information was removed from VAERS on the same week it entered into the system

The retained report. does not have any information about the eventual outcome.

You can do the same for other DELETED_URL, RETAINED_URL pairs and verify for yourself if the information I have compiled is accurate.

Please do leave a comment if you notice any errors.

This is the full list of reports:

Full list of reports

You can sort and filter through them to find additional patterns. Let me know in the comments if you find anything interesting.

Download the raw CSV data

Download the underlying CSV data