THE OLD GUARDIAN
theoldguardian.ca | Independent Investigative Journalism
The implicit contract of emergency authorization is that rigorous post-market surveillance compensates for what compressed trials cannot capture. Health Canada’s own internal documents show that surveillance was negotiated down, progressively dismantled, and in at least one documented instance unavailable in real time between the two federal agencies responsible for it. The contract was not kept.
By Chris Allen | The Old Guardian | September 2026
Primary source: ATI Release A-2025-001410, Health Canada, July 15, 2026
In June 2022, Health Canada produced an internal pharmacovigilance review of Comirnaty. By that point Canada had administered 58,112,512 doses of the Pfizer vaccine. The review was a comprehensive internal assessment of what the regulator knew about the safety of the product it had authorized.
On page 198 of that review -- obtained by The Old Guardian through federal Access to Information request A-2025-001410 -- appears a single sentence that stops the reader.
‘A detailed description of the adverse events reported to CAEFISS is not currently available to Health Canada.’
CAEFISS is the Canadian Adverse Events Following Immunization Surveillance System -- the national vaccine safety surveillance database administered by the Public Health Agency of Canada. It is the primary mechanism through which adverse events following vaccination are collected, stored, and analyzed in Canada.
Health Canada is the federal regulator responsible for authorizing vaccines and monitoring their post-market safety.
By June 2022, 58 million doses into the largest vaccination campaign in Canadian history, these two federal agencies were not sharing complete adverse event data with each other in real time. The regulator responsible for vaccine safety assessment was making decisions based on incomplete information from the surveillance system it was supposed to be relying on.
That sentence -- written by Health Canada’s own internal reviewers, in a document never intended for public release -- is the thesis of this piece. Not a conspiracy. Not deliberate concealment. Something more structural and in some ways more troubling: a documented failure of the data architecture that was supposed to tell Canadian regulators what was happening to Canadians.
It was not the only failure. It was not even the largest one. But it was the one Health Canada acknowledged in its own words.
The Contract of Emergency Authorization
When Health Canada authorized Comirnaty on December 9, 2020 under an Interim Order -- a mechanism that allowed faster authorization than the standard drug approval pathway -- it was making a specific implicit promise to Canadians.
The promise was not that the vaccine was perfectly safe. It was that the evidence available at the time supported authorization, and that rigorous ongoing monitoring would catch anything the compressed pre-authorization timeline could not.
That is the contract of emergency authorization. Speed in, compensated by vigilance after. The pre-authorization timeline was compressed from years to months. The post-authorization surveillance was supposed to be the safety net that justified that compression.
The documents in this series tell the story of what happened to that safety net.
Parts 2 and 3 documented how Health Canada’s internal signal assessments used materially different language than public communications during the same period, and how surveillance requirements were progressively negotiated down through bilateral exchanges with Pfizer. Part 4 documented the specific data architecture failures in the pregnancy monitoring system -- no observed versus expected analysis for Comirnaty, no population-level tracking of vaccination by pregnancy status, a pregnancy registry request declined by Pfizer and accepted by Health Canada.
Part 5 assembles the complete picture. Four documented surveillance failures operating simultaneously. A safety net with multiple holes, at a moment when millions of Canadians were relying on it to catch what the trials had not.
Source: ATI Release A-2025-001410, Health Canada internal pharmacovigilance review, June 2022, page 198.
Failure One: The CAEFISS Gap
The Public Health Agency of Canada administers CAEFISS. Health Canada administers Canada Vigilance -- a separate adverse event reporting database. During the COVID-19 vaccine rollout, these two systems were supposed to function as complementary pillars of Canada’s vaccine safety surveillance architecture.
The ATI documents show they were not functioning as a unified system. Health Canada’s internal reviewers were conducting weekly Canada Vigilance database searches and receiving monthly Summary Safety Reports from Pfizer. They were monitoring international regulatory signals through the Article Tracking Sheets documented in the 81 Excel files released alongside the main ATI package. They were producing internal signal assessments and issuing advisement letters to Pfizer when signals warranted action.
What they did not have -- as their own June 2022 internal review explicitly states -- was a detailed description of the adverse events reported to CAEFISS.
The practical consequence of this gap is significant. CAEFISS and Canada Vigilance captured overlapping but not identical populations and event types. Healthcare providers submitted reports to one system or the other based on their reporting pathway. Individual patients and families submitted reports through different mechanisms. The two databases together were supposed to give Health Canada and PHAC a complete picture of what was happening in Canada. Instead they were each giving their respective agencies a partial picture, and those partial pictures were not being systematically integrated in real time.
Two federal agencies. Two databases. 58 million doses administered. The regulator responsible for vaccine safety assessment acknowledged in its own internal documents that it did not have access to the national adverse event surveillance data in real time.
This is not an allegation that data was hidden or deliberately withheld. Institutional data gaps between co-existing federal agencies are a structural problem that predates COVID. The significance of this particular gap is its timing -- occurring during the largest and fastest vaccination campaign in Canadian history, at the precise moment when real-time surveillance data was most consequential.
Source: ATI Release A-2025-001410, Health Canada internal pharmacovigilance review, June 2022, page 198. BBRS Signal Tracking document, multiple entries referencing Canada Vigilance database searches.
Failure Two: The Pregnancy Tracking Gap
Part 4 of this series documented this failure in detail. Its inclusion here is brief but necessary because it belongs in the complete picture of the surveillance architecture.
Canada was not tracking vaccination status by pregnancy at the population level. The internal documents confirm this on page 229: ‘this information is not tracked in Canada.’
The 290 pregnancy-related adverse event reports recorded in Canada to May 2022 -- including 127 classified as serious and 72 spontaneous abortion reports -- existed in a denominator vacuum. The total number of pregnant Canadians who were vaccinated during this period was not precisely known. The rate of adverse pregnancy outcomes in vaccinated versus unvaccinated pregnant Canadians was not calculable from available data.
Health Canada’s request for a pregnancy registry -- which would have created the denominator -- was declined by Pfizer in August 2021. Health Canada accepted that answer. The denominator gap remained throughout the entire rollout period.
A surveillance system that cannot calculate rates cannot detect signals. A signal detection system without a denominator is a passive collection mechanism, not an active monitoring architecture. That is what Canada had for pregnancy outcomes during the COVID-19 vaccine rollout.
Source: ATI Release A-2025-001410, pages 229, 271, 273-274.
Failure Three: The Batch Variability Gap
In 2023, a peer-reviewed study published in the European Journal of Clinical Investigation examined 971,021 reports of suspected adverse effects in the EU and European Economic Area. The study, by Schmeling and colleagues, analyzed Danish national health data and found something that has significant implications for how population-level vaccine safety surveillance is interpreted.
Adverse event rates varied dramatically by vaccine batch lot number. Batches representing approximately 4% of total doses administered were associated with dramatically higher adverse event reporting rates than other batches, with R-squared values between 0.78 and 0.89 -- indicating strong statistical correlation between batch identity and adverse event reports. The remaining 96% of doses showed much lower and more uniform adverse event rates.
The mechanism for this finding is manufacturing variability. Different production runs, different manufacturing sites, and different production batches can produce vaccines with differing mRNA concentrations and lipid nanoparticle characteristics. This is a known pharmaceutical manufacturing challenge. It is why lot release testing exists as a regulatory requirement.
The Canadian connection to this finding is documented in the ATI release. The BBRS Signal Tracking spreadsheets reference manufacturing site changes for Comirnaty including the approval of a WuXi Biologics production site in Germany during the authorization period. Multiple manufacturing sites, multiple production runs, multiple potential sources of batch-level variability -- all operating simultaneously during the Canadian rollout.
The surveillance architecture implication is significant. Canada Vigilance tracked adverse events by vaccine product and dose number. The BBRS monitoring documents reviewed for this series do not reference lot-level analysis as a standard surveillance tool. If adverse event rates varied significantly by production batch -- as the Schmeling study documents in European data -- then population-level surveillance that did not systematically track lot numbers would miss a signal that lot-level analysis would detect.
If some batches produced dramatically higher adverse event rates than others, a surveillance system tracking product-level data rather than lot-level data would average those rates together -- diluting the signal from the high-adverse-event batches into the background noise of the low-adverse-event batches.
This is not a claim that Health Canada knew about specific dangerous batches and concealed them. It is a documented surveillance architecture limitation: the monitoring system was not designed to detect lot-level variation in adverse event rates, and peer-reviewed European data suggests that lot-level variation was real and significant.
Source: Schmeling M et al., Batch-dependent safety of the BNT162b2 mRNA COVID-19 vaccine. European Journal of Clinical Investigation, 2023. DOI: 10.1111/eci.13998. ATI Release A-2025-001410, BBRS Signal Tracking document, WuXi Biologics manufacturing site references.
Failure Four: The Surveillance Wind-Down
Parts 2 and 3 documented the progressive reduction of surveillance intensity through the authorization period. Its inclusion in Part 5 as a fourth surveillance failure is deliberate: it belongs in the complete picture of what the safety net looked like by the time most Canadians received their doses.
In December 2020, when Comirnaty was first authorized, the surveillance architecture was intensive. Daily Canada Vigilance database searches. Monthly Summary Safety Reports from Pfizer. Active signal monitoring across all authorized DINs. Observed versus expected analyses for fatal reports. Article Tracking Sheets monitoring international regulatory signals in real time.
By May 29, 2023, Health Canada had issued a formal memo closing the monthly safety report requirement for all Comirnaty DINs. The internal language: ‘Repeal/close the following Ts&Cs: The requirement to submit monthly safety reports for all DINs within the authorized Comirnaty product line.’ Standard COVID-19 specific reporting had been replaced with standard drug reporting requirements -- the same framework governing routine medications that had been on the market for decades.
The surveillance architecture went from maximum intensity to routine post-market monitoring in approximately three years. During that same period, the youngest age group authorization was still relatively recent. Long-term safety data remained incomplete. The batch variability question documented above had not been resolved. The pregnancy tracking gap had not been filled. The CAEFISS data sharing gap had not been publicly disclosed.
The contract of emergency authorization -- speed in, compensated by vigilance after -- had a time limit that was not disclosed to Canadians. The vigilance ended while the questions remained open.
Source: ATI Release A-2025-001410, BBRS Signal Tracking document, May 29, 2023 entry. ATI p.329, p.365.
What the Gaps Mean Together
Four surveillance failures. Each documented in Health Canada’s own internal records or in peer-reviewed literature using national health data. Each individually significant. Together, they describe a monitoring architecture that was systematically less capable of detecting rare adverse events than the public was led to believe.
The CAEFISS gap meant the regulator was working from an incomplete picture of domestic adverse events in real time. The pregnancy tracking gap meant rare adverse pregnancy outcomes could not be detected even if they were occurring. The batch variability gap meant lot-level adverse event signals were invisible to product-level surveillance. The surveillance wind-down meant the monitoring that was supposed to compensate for compressed authorization timelines was progressively dismantled before the long-term safety picture was complete.
None of these failures required bad faith to produce. They are the predictable outcomes of institutional design choices -- a federal data architecture that kept surveillance databases separate between PHAC and Health Canada, a regulatory framework that relied on manufacturer cooperation for post-market commitments, a surveillance methodology built for population-level signal detection rather than lot-level analysis, and a risk-based approach to reducing monitoring intensity that prioritized administrative efficiency over comprehensive long-term surveillance.
The cumulative effect of these design choices was a safety net that was less capable than it appeared -- and than the public was told.
Emergency authorization compressed the pre-authorization timeline. Post-market surveillance was supposed to compensate. The documents show that surveillance was itself compromised by data gaps, negotiated reductions, and progressive wind-down before the safety picture was complete. The contract was not kept.
Cameron Hitchcock: What the Gap Looked Like
Cameron Hitchcock was 29 years old, active and healthy, when he received his second Pfizer dose in 2021. His health deteriorated soon after. His body began swelling. His kidneys were losing protein at an alarming rate. He was eventually diagnosed with nephrotic syndrome. What followed was approximately five months in hospital and, ultimately, kidney failure.
Today Cameron requires dialysis six days a week. He is waiting for a kidney transplant. He arranged his dialysis schedule around travelling to Ottawa so he could testify publicly at the Allison Inquiry in September 2026.
His Vaccine Injury Support Program claim was eventually approved -- making him one of the 252 Canadians whose claims cleared the compensation threshold out of 3,557 who applied.
The government’s own compensation program acknowledged a connection between Cameron’s vaccination and his kidney failure serious enough to warrant payment. He is, by the program’s own determination, a vaccine-injured Canadian whose injury met the evidentiary standard for recognition.
And yet in September 2026, five years after his second dose, he was arranging dialysis around a trip to Ottawa to testify at a privately organized forum because no formal parliamentary inquiry, no government-commissioned review, and no institutional accountability mechanism had ever existed to publicly hear his account.
The surveillance system that was supposed to detect what happened to Cameron had documented gaps before he received his second dose. The CAEFISS data was not available to Health Canada in real time. Lot-level adverse event analysis was not a standard surveillance tool. The monitoring architecture was already being negotiated down. The data infrastructure that would have caught a signal like his -- if his experience was part of a pattern -- was less capable than the public knew.
Whether the gaps contributed to Cameron’s specific outcome cannot be established from available evidence. What can be established is that the safety net was less complete than it appeared, that its limitations were documented internally and not publicly disclosed, and that a man who arranged dialysis around a trip to Ottawa to be heard is still waiting for a kidney while the institutions that were supposed to be watching received a VISP cheque in acknowledgment of what they missed.
Editor’s Notes and Primary Sources
All findings in this piece are drawn directly from ATI Release A-2025-001410, released by Health Canada to The Old Guardian on July 15, 2026, and from peer-reviewed scientific literature. Specific documents cited:
ATI p.198
Health Canada internal pharmacovigilance review, June 2022 -- ‘A detailed description of the adverse events reported to CAEFISS is not currently available to Health Canada.’ Verbatim internal language.
ATI p.229
Canada not tracking vaccination status by pregnancy at population level -- ‘this information is not tracked in Canada.’ Verbatim internal language.
ATI p.271
August 3, 2021 pregnancy registry request to Pfizer -- declined, Health Canada accepted, no further action taken.
BBRS Signal Tracking (ATI Excel file)
Complete internal project tracking log 2020-2023. May 29, 2023 entry documents formal closure of monthly safety report requirements for all Comirnaty DINs. WuXi Biologics manufacturing site approval referenced. Weekly Canada Vigilance database search protocols documented.
ATI p.329
Health Canada formal documentation of reduced monitoring intensity -- ‘adopting a risk-based approach...reducing oversight on indications where the safety profile has been better characterized.’
ATI p.365
By 2023 -- standard COVID-19 specific reporting framework replaced with standard drug reporting requirements.
Schmeling et al. 2023
Batch-dependent safety of the BNT162b2 mRNA COVID-19 vaccine. European Journal of Clinical Investigation. DOI: 10.1111/eci.13998. Danish national health data. 971,021 EU/EEA adverse event reports analyzed. Batch lot identity correlated with adverse event rates at R2 0.78-0.89. Approximately 4% of batches associated with dramatically elevated adverse event reporting.
Cameron Hitchcock
Public testimony, Allison Inquiry, September 2026. Nephrotic syndrome following second Pfizer dose 2021. Kidney failure. Dialysis six days weekly. VISP claim approved. One of 252 approved claims from 3,557 applicants.
The Old Guardian’s six-part series What They Knew examines Health Canada’s internal COVID-19 vaccine pharmacovigilance records. Part 6 -- The Accounting -- publishes shortly. ATI file: A-2025-001410.
If you or a family member experienced a serious adverse event following COVID-19 vaccination, contact The Old Guardian at tips@theoldguardian.ca. Source confidentiality is protected. Verification is required.
The Old Guardian | theoldguardian.ca | tips@theoldguardian.ca | ATI File: A-2025-001410