'Unacceptable incompetence': CDC made dozens of basic data errors on COVID, epidemiologists find
FDA evidence for full approval of COVID antiviral Paxlovid, known for "rebound" infections in top federal officials, found wanting by scientists, including NIAID researcher.
The Facts Inside Our Reporter’s Notebook
- purporting to expose COVID vaccine misinformation
- The preprint
- Vinay Prasad
- "top 5" cause of death in children
- harmful drug interactions
- CNBC reported
- Paxlovid is not useful against household transmission
- FDA's briefing packet
- president, his COVID adviser, first lady and both FDA and CDC commissioners
- lengthy review of the FDA evidence
- Margery Smelkinson questioned
- "blueprint" for a potential COVID truth commission
- Paxlovid section of their report
- Lancet Infectious Diseases observational study
- immortal time bias
- he wrote in his newsletter
The CDC found itself hoist with its own petard by making 25 basic statistical and numerical errors related to COVID-19, particularly with regard to children, while purporting to expose COVID vaccine misinformation, according to an analysis led by University of California San Francisco epidemiologists.
The preprint, which has not yet been peer-reviewed, documented 20 errors that "exaggerated the severity of the COVID-19 situation" and three that "simultaneously exaggerated and downplayed" severity, while one each was neutral or exaggerated vaccine risks.
More than half were from 2022, but nearly as many were made in the first two months of 2023 as in all of 2021, they found. Several errors were related to the agency's COVID data tracker, which failed to align with its National Center for Health Statistics (NCHS), and the CDC corrected at least in part 13 of the 16 errors brought to its attention.
The paper emphasizes how widely CDC errors can spread even if they are later corrected, with YouTube and Spotify linking its website on videos and podcasts that discuss COVID and the wide deference to CDC guidance in schools, businesses and healthcare facilities.
"The errors are damning," coauthor Vinay Prasad, a former National Institutes of Health fellow, said on Twitter. "Basic counts of dead kids, causes of childhood death. Unacceptable incompetence."
UCSF's Alyson Haslam, a former CDC fellow who works in Prasad's lab, made the final call on CDC errors that Prasad, Tracy Beth Hoeg and independent Georgia COVID analyst Kelley Krohnert collectively agreed "were indisputable and incorrect, as a matter of fact, and not preference or opinion."
The trio conducted "real time" review of news sources, Advisory Committee on Immunization Practices (ACIP) meetings and materials, the agency's Morbidity and Mortality Weekly Report and the Twitter accounts of the CDC and its director, as well as reports sent to them by others, going back to 2021.
The errors were heavily weighted toward exaggerating COVID's risk to children. Fifteen of the 16 pertaining to children's data "enhanced the perceived risk" of the virus and more than half the total errors involved mortality statistics, with the CDC data tracker "consistently" reporting higher deaths for children and adolescents than did NCHS.
Perhaps the most consequential error was the CDC's repeated promotion of a preprint that deemed COVID a "top 5" cause of death in children, which the agency only corrected in one place months later.
That paper compared 26 months of COVID deaths, where the virus was "one of several contributing causes to deaths," to 12 months of deaths from other causes that were "identified as the single underlying cause of death ... which by design exaggerates" the COVID risk to kids, the paper says.
Not only was the claim made in ACIP and FDA Vaccines and Related Biological Products Advisory Committee meetings, but also at a White House briefing by CDC Director Rochelle Walensky and by ACIP's chair in a subsequent meeting "after the errors were identified." Only ACIP's page on "vaccination evidence for young children" includes the correction.
The agency was plain sloppy in other errors, the authors allege. It listed pediatric deaths as 4% of COVID deaths when it meant to write 0.04% and gave a lower estimated rate of pediatric infections than symptomatic illness, with some errors remaining live for seven months.
"These errors have been made repeatedly and were likely to have affected discussion of pandemic policies," particularly the CDC's guidance calling for "school closures, mask mandates, and strong recommendations for vaccinations and multiple boosters even among children who have recovered from the virus," the authors conclude.
The CDC did not respond to queries for its response to the paper.
The FDA's evidence for full approval of Pfizer's COVID antiviral Paxlovid, shared with its advisory committee on antimicrobial drugs, also came under scrutiny.
The advisers voted in mid-March to approve the two-drug treatment for high-risk adults while warning that many patients could have harmful drug interactions, CNBC reported. The agency will make the final decision in May. (Pfizer long ago acknowledged Paxlovid is not useful against household transmission.)
While the FDA's briefing packet emphasized the infrequency of Paxlovid "rebound" infections — which hit the president, his COVID adviser, first lady and both FDA and CDC commissioners, all up to date on COVID jabs — its cited evidence only applies to a sliver of potential Paxlovid patients.
The Pfizer studies gave five-day treatments to unvaccinated high-risk patients (EPIC-HR) and vaccinated high-risk or unvaccinated low-risk patients before and after the Omicron variant wave (EPIC-SR). Only EPIC-HR found "any meaningful difference" compared to a placebo, a 5.6% absolute reduction and 86% relative reduction.
The FDA emphasized the "symptom rebound" rates were similar between Paxlovid and placebo arms across both studies, around 10-16%. This shows that for a "subset" of infections, regardless of Paxlovid, "virologic and/or symptomatic rebound may occur as part of the natural progression and resolution of COVID-19 disease."
A different graph on "viral RNA rebound," however, found notably higher rates in the Paxlovid arm in EPIC-HR and EPIC-SR's Omicron period.
In a lengthy review of the FDA evidence, University of Minnesota infectious disease researcher David Boulware said it actually shows Paxlovid is "likely near zero benefit" for under-60s with a "normal immune system" and vaccine- or infection-induced immunity.
Before the Paxlovid vote, National Institutes of Allergy and Infectious Diseases scientist Margery Smelkinson questioned its emergency use authorization starting at age 12 in spite of dramatically different COVID risk by age. She noted Pfizer ended EPIC-SR early after finding no "statistically significant evidence of benefit."
A member of the Norfolk Group of scientists, physicians and policy experts that recently published a "blueprint" for a potential COVID truth commission, Smelkinson pointed to the Paxlovid section of their report, which emphasizes the lack of rigorous data on who actually benefits from the treatment.
Prasad also questioned the quality of evidence for Paxlovid, including a Lancet Infectious Diseases observational study that found "the curves already split" between treatment and placebo groups at "Day 0," which he said indicates immortal time bias.
"Non randomized evidence will forever be plagued by differences in people who get pax and those who do not (likely being rich, health literate and well connected gets you more pax!)," he wrote in his newsletter in February.
The FDA declined to comment on the criticism of its proffered evidence. "We can’t comment on pending applications," press officer Chanapa Tantibanchachai wrote in an email.