85% are Serious in VAERS Reports, NOT 20%, Study Reveals
Sound crazy? Because such work is mighty hard.
The respected British Medical Journal (BMJ) published a report by Jennifer Block (good going) in the Investigation category (Nov 2023) about VAERS like “isn’t operating as intended … signals are being missed” … “not meeting its own standards” [True]. It addresses serious reports that sound an alarm like “people, including physicians and a state medical examiner … have filed [serious reports] … never contacted”. (CDC says they follow up on serious reports by the way. But those are all kept secret.)
The trouble is that a sentence jumped off the page at me:
“Nearly one in five meet the criteria of serious.” [False]
This deepdots dive is to ascertain a more true number of serious reports in VAERS. And to offer some methods/actions for analysts, the goal is for their future reports to be more accurate. Analysts are being outsmarted by the obfuscation being handed to them.
Because it is really hard to get it right. Plus grey areas.
Gathering conclusions from VAERS is like refining ore, not easy. Some steps are described below and they can be improved upon but this does provide a ballpark closer real value for serious adverse events, 85%, the two in red here together.
The figure of 85% (serious total) arrived at is a safe baseline. In reality it is most likely much higher because many serious reports have been set aside here, affording a large amount of leeway for critics. Some tough calls were made as sometimes reports are a mix of serious and not serious. The most common symptom, headache, is not regarded as serious although migraine is. Symptoms in the top 100 such as the following are being taken as NOT SERIOUS: Headache, Fatigue/Asthenia/Malaise/Lethargy/Weakness, Fever/Pyrexia, Nausea/Vomiting, Dizziness/Vertigo, Rash/Pruritus (itching)/Erythema, Cough, Diarrhoea, Swelling/Lymphadenopathy, Insomnia, Chills/Cold or Feeling Hot, Gait disturbance, Mobility decreased, Fall, Anxiety, Herpes, Rhinorrhoea/Runny nose, Insomnia, Fainting/Syncope, Sweating/Hyperhidrosis
What is Serious?
A serious adverse event is defined in the Code of Federal Regulations (CFR).
Fatal, life-threatening, disabling, hospitalization, birth defect and some others including required ongoing at home treatment or developed drug dependency or abuse.
In the serious category, I’m including cancers plus emergency room action as serious (among those officially marked as such in VAERS) and critics are free to tell people their ambulance trip emergencies from the injections are not serious.
Notice the Code offers examples of others that are serious … allergic bronchospasm requiring intensive treatment … at home OR convulsions that do not result in inpatient hospitalization, or the development of drug dependency or drug abuse. I’m omitting those in this study and thus the real figure is surely higher than 85%.
The FDA also writes about a serious "adverse event" (SAE), it helps clarify and includes other serious medically important event[s] … substantial disruption of a person's ability to conduct normal life functions … disruption in […] quality of life. Medical staff have their view. The patient would also have an opinion.
28% of VAERS reports are check-marked in one of the category-columns that indicate a serious adverse event (excluding purely admin reports, see below). That all by itself without considering all of the other serious reports not check-marked are more than one in four, not the mere one in five or 20%.
There are 855,010 VAERS reports that are NOT check-marked in any of the categories considered serious while actually serious, even Myocarditis, Thrombosis (clot), Miscarriage and so on flying under the radar.
Critics have to accept that having covid post-injection supposedly protecting them from it must also be considered serious because they shut down the world for fear of it.
On filing a VAERS report online (not the PDF version), some of these fields represent columns in the data …
Here, all except Office Visit are taken as serious regarding the check boxes in the VAERS data. Straight from the downloadable spreadsheet below …
Another spreadsheet for symptom entry counts provides background. 127 of the 15K+ official symptom entries by CDC have the word pain. In the distillation process, I’m often using the word pain as serious but not mere arm pain at the injection site (those are neutralized in this process to let them fall away).
For VAERS Analysts
Tip 1: Learn Regular Expressions. Powerful codes for pattern matching in searches, and to massage data. It’s like a new language. Once you start to get the hang of it, the tool I prefer for testing them is https://regex101.com/. Although I was at Microsoft for 8 years, since LibreOffice has built-in regex, I find L.O. more productive than Excel in this area. They are very similar. A lot of people are exceptionally skilled with Excel, great, but LibreOffice can also be there alongside for certain tasks at least for regex.
For example, the \b is code for word-boundary in this regular expression I’m using at the moment for extracting other serious reports (not indicated by the reporter or CDC). Do keep in mind by the way that a string like … not serious … is first screened out. The pipe symbol means ‘OR’. This can use some honing …
regex_serious_1 = r'(?:very serious|severe|critical|extreme|intensive|emergency|excruciat|extensive|cardi|thromb|\bclot\b|pulmon|embol|cerebr|arrhythm|chest|tomogram|vascular|extrasystole|heart rate increased|heart rate decreased|heart rate irregular|heart rate abnormal|guillain\-barre|le cells|abortion spontaneous|uterin|sterili|placent|resuscitat|neuro|monkeypox|myasthenia gravis|disability|gallbladder|fibrin|onia\b|etria\b|ulia\b|thria\b|veolar|demyelin|\bboil\b|urticaria|cleft|collapse|choking|chimerism|asis\b|bleeding|pneumonia|alopecia|asthma|sperm|hallucinat|esis\b|opia\b|dynia|xia\b|ground glass|\bggo\b|exia|hoea|uria|abscess|carotid|sudden\b|noea\b|tinnitus|spasm|oma\b|thickening|sepsis|vaso|lesion|unresponsive|transfusion|convuls|valve|polyp|intensive care|\bicu\b|hyper|hypo|autopsy|failure|shock|openia|lgia|itis|osis|sia\b|plasty|ymia|pathy|agia|iasis|lysis|usis\b|phism|axia|aemia|blister|pemphigoid|dysfunct|angina|respiratory|gastr|abdom[ei]n|intestin|kidney|egia|cyto|encepha|bloody|glob|axia|defib|trophy|haemo|dosis|leuk|ventr|paraly|immobili|esophag|tomy|myoma|coma\b|tremor|crisis|brain|nerv|guillain|blind|migraine|infertil|neurop|pneumothorax)'
regex_serious_2 = r'\b(?:renal|nephrotic syndrome|amyloidosis|aneurysm|aort[io]|artery|atrial|atrio|beriberi|bundle branch|chagas|chordae tendinae|coronar|coxsackie|dressler|dyssynchrony|ecg|ekg|mri\b|lqts|torsades|magnetic resonance imaging|entero|eosinophilic|fistula|glycogen storage|haemorrhage|histoplasma|holt\-oram|iron overload|ischaemic|kearns\-sayre|oesophageal|orthostatic|papillary|peripartum|pleuroperi|pulseless|rheumat|rupture|sarcoidosis|septal|septum|shoshin|shunt|sigmoid|\bsirs|steatosis|syphilitic|tamponade|uraemic|valve disorder|vein|venous|fibril|purpur|lung|pte\b|stroke|miscar|stillb|palsy|copd|paralys|palpitat|death|pain|hives*\b|lupus|bile\b|ataxia|polyserosit|proctitis|fo*et[auo]|necros|infarc|arte``ri|seizure|autoimmun|scrot|epilep|vascul|bifida|suicid|disease|deaf|arrest|immuno|cancer|neoplasm|tumou*r|malignant|metast|stage \d|chemoth|cytoma|carcino|melanom|cyst\b|mass\b|mast|myeloma|lymph|neuroma|sarcom|nodule|dysplas|hyperplas|lipoma|neoplasm|adenoma|oncol|anaphyla|mutat|myelodysplastic|acute|blood cell|visual loss|loss of|hypertens|vitiligo|ovar|heavy menstrual bleeding|testicular swelling|bedridden|impaired quality of life|impaired work ability|screaming|specialist consult)'
I’m working in Python by the way with pandas dataframes for the data but regular expressions are just about everywhere now, although not simple, if at all, in Excel, let me know if there’s actually an easy plugin that works well now.
Tip 2: Try the flat file for input if/when covid-only is ok, each line contains everything about that report. An important point to understand that affects this entire page is that it contains a column called ‘symptoms’ which is a combination of SYMPTOM_TEXT and the symptom entries brought in from the symptom files. To improve on this, the LAB_DATA and CUR_ILL columns could also be utilized. Currently just reports officially tagged as covid by the reporter or CDC. In other code I rescue and gain extra covid reports (based on the presence of lot/batch codes in the writeups in part plus some other techniques like designation of Pfizer, Moderna etc) and could do so also in the flat file except for lack of interest so far. It also rescues over 30K reports that were deleted, makes note of the changes CDC has made after records are published and keeps content otherwise disappeared in the 11-11 purge in 2022 of 31 European countries (nearly a Half Gig of Data unzipped, not the small figure-impression an analyst out there creates wrongly, bless her heart).
Also: Flatfile output is divided into multiple sheets in the XLSX Excel file. Spreadsheets can only handle 1,048,576 rows per sheet, we’re at 1.6 million. I might wind up being forced to buy a laptop with 32 GB RAM but so far with 16 I’m able to cope. Every analyst should become familiar with excellent tools out there to make the data just what you need and want to be working with if not already. Concat or split, grep et. al., I use tools like those via cygwin etc. and I also like CompareIt and —personal preference—Notepad++.
Tip 3: Pre-screening of terms. Nullify strings first that would otherwise produce false positives in the search, with regex. Mine are extensive, countless days of time and experimentation.
Remove sentences containing the word history. Sentences also with dates using [^\.]*?\b[12][90][12]\d\b[^\.]*?(?=\.\s)
to blank out any like … myocarditis in 2019.
I’m removing symptoms ending in ‘normal’. Etc.
Tip 4: Screen out administrative reports (list) like ‘storage error’ where no human was actually harmed. I strongly suggest to the world that they have no business being there in the first place by US Code (VAERS mission). That’s my opinion. Congress did not say, also harness the VAERS database for procedural errors and flood it with those so people can’t really decipher a signal from the noise to be able to tell how harmful the vaccines are. CDC can and should sideline those to a separate file if they want to be respected. All reports involving humans experiencing harm can remain in place.
At the moment, for screening, I’m using a regex with ~147 symptom entries that look like admin indicators. A bit complex though, because some actual-human reports and some that are even checked as serious have those admin entries too. Separate actual harm first in the definitely-serious category (or undetermined) set. Example admin report: 2654493
Current Results
1,637,278 covid total (only those correctly tagged containing ‘COVID’).
152,702 admin only. Administrative reports that don’t belong there.
1,484,576 excluding admin-only (1,637,278 - 152,702). Using that figure for percentages.
410,784 (27.7%) marked as serious (incomplete). Officially marked as serious with ‘Y’ in one of those columns.
183,037 (12.3%) additionally got covid anyway thus serious.
671,973 (45.3%) additionally ALSO serious. Not marked, lacking the “serious” flags.
85.3% serious overall excluding admin-only (marked + got_covid + also_serious).
A few counts (ONLY IN THOSE NOT MARKED AS SERIOUS BY CDC:
41,472 cardi
22,353 palpitat
17,912 thromb|clot
12,623 tachycard
10,145 pulmon|lung|pte|embol
8,454 tremor
6,915 paralys
6,767 myocard
5,251 palsy
2,152 stroke
1,510 infarc
1,374 death
1,068 miscar|stillb
991 guillain
The Process
Like chipping away in making a sculpture or purifying gold.
Open the data file.
Move those flagged as ‘Y’ in the serious columns to marked_serious (410,784).
Move those that got-covid thus serious (use the symptom entry counts file/link above).
Move purely admin noise to its own separate admin_only file.
Blank out sentences with history, injection, arm, sore, and terms indicating non-serious symptoms.
Move serious reports based on a search using terms determined as serious.
What remains should be non-admin, did not contract covid and non-serious (or at least undetermined).
Print summary.
Save all of the separate sets for manual review, make improvements, re-run it.
Summary
The figure could even go UP due to others that should be added as serious. Look through the results I have labeled as supposedly not serious and you’ll find a lot that are pretty serious such as “she had a headache that lasted for 7 weeks after her second Moderna“. It’s a question of whether they meet the bar in FDA’s disruption in […] quality of life definition.
Do the loved ones of those affected consider their case “serious”? How serious?
We have to sit around wondering how long-term the harm is. Very bad? Not too awfully terrible? Unknowns foster a toxic world. Could a FOIA for the CDC followup reports clear the air?
With this level of high complexity in the output, mistakes were made surely. How many?
Either way, inarguably, the point is that this study finds …
Serious VAERS reports are at least 4 out of 5 and potentially over 90%, not 1 in 5.
If any analysts adopt methods like this, then the next major journal citing a seriousness figure ought to be quoting them instead of the measly 20% from an unstated source at BMJ.
Output Data and Code
Based on 2023-10-27 dropped to the public on 2023-11-03.
https://univaers.com/download/special/hawk_vaers_other_serious.xlsx
Python code available on request.
Your Mission
Find fault, prove me wrong, find reports that aren’t really serious and vice versa to move the needle 1% or more. I moved it 65 from 20 to 85.
All you have to do is find 14,845 regarded here as serious that are not, to move it 1%.
(And then pray someone doesn’t counter with 14,845 from not_serious which on further review actually are).
Other
Cited serious percentages in VAERS:
85% Gary Hawkins, deepdots on substack & https://univaers.com/
19.6% Canada
17.9% US National Institutes of Health
9.2% AVAC, adultvaccinesnow.org
9% American Association of Family Physicians
A related page at vaccinedatascience with focus on foreign reports: What the BMJ VAERS investigation missed
David Gorski upset about the BMJ articles uses the single 2006 deliberately fake report about “Hulk” against the Flu Vaccine 17 years ago (a test submission) to lamely attempt to devalidate all of VAERS today. The report was deleted by CDC, you won’t find it. Use your time more productively David, attack this 85%.
There’s also another term: Serious adverse events of special interest” (SAESIs)
Recently I discovered a paper written by the CDC itself (which is about as official as it gets):
"During this same time period VAERS averaged around 6,000 foreign source reports annually. Vaccine manufacturers, which accounted for >99% of foreign source reporting, are required by law to submit foreign source adverse event reports that are both serious and unexpected [21], but not other types of foreign source reports. Given the vaccine manufacturer reporting requirements and the minimal amount of direct public reporting, it is not surprising that a relatively high percentage (48%) of foreign source reports are classified as serious. This likely represents selective reporting based on regulatory requirements rather than any substantial differences in safety profiles of foreign vaccines."
https://pubmed.ncbi.nlm.nih.gov/26209838/
While every foreign country only reported "serious and unexpected" cases as per the CDC guidelines, I think Austria was an exception. I wrote an article about it:
https://vaccinedatascience.substack.com/p/why-does-austria-have-an-anomalously
But we can just ignore reports from Austria for this particular discussion, and what you have written here will be true for all the rest anyway.