Karwowski et al 2018 is a new study of aluminum (Al) adjuvant from vaccines, aluminum levels in hair and blood, and infant development. Vaccine promoters are boasting about it, saying that it provides good evidence in support of Al adjuvant safety. In reality, Karwowski 2018 has several flaws that render it essentially useless as evidence for the safety of Al adjuvant.
Full Paper Karwowski et al 2018. Blood and Hair Aluminum Levels, Vaccine History, and Early Infant Development: A Cross-Sectional Study
Skeptical Raptor praises Karwowski 2018 in a May 29 blog post, stating that Karwowski provides “solid” and “powerful” evidence for the safety of aluminum adjuvant: “[Karwowski 2018] provides solid evidence that the tiny amount of aluminum in vaccines is biologically irrelevant…it provides us with powerful evidence that aluminum in vaccines has no influence on either physiological concentrations of the metal or neurodevelopmental differences in children.”
Dr Paul Offit of the Children’s Hospital of Philadelphia (CHOP) added Karwowski 2018 to a list of 6 scientific papers on Al adjuvant safety. The list includes Keith 2002 and Mitkus 2011, which are thoroughly debunked here: FDA’s Flawed Study of Al Adjuvant Toxicity.
In any debate, one must consider the strongest contrary arguments and evidence (as identified by the opponent). Failure to do so is committing the selective attention fallacy or weak man fallacy. So, I write about Karwowski 2018 because vaccine promoters say Karwowski 2018 is important and powerful. Vaccine promoters assert Karwowski 2018 as some of their best evidence for Al adjuvant safety.
What Did Karwowski do?
Karwowski looked at 85 infants, ages 9-13 months, in the USA. For each infant, the following data was collected:
1) total lifetime dose of Al adjuvant from vaccines. This would be from vaccines given at birth, and 2, 4, 6, and 12 months of age.
2) aluminum level in hair and blood. One hair and one blood sample were collected for each infant, at a single time point.
3) Bayley Scale of Infant and Toddler Development (BSID) scores. The BSID is a recognized measure of cognitive, language and motor development in infants, and includes separate scores for each category. Many different tests are combined to produce numerical values representative of infant health and development.
Karwowski looked for correlations between 1 & 2 (Al adjuvant dose and blood/hair Al levels), and between 2 & 3 (blood/hair Al levels and BSID scores). No correlations were found, unless one outlier subject is excluded from the analysis. Specifically, high Al level in hair was associated with better motor function. High hair Al level may indicates better elimination of aluminum and lower Al level inside the body (more on this below).
Vaccine promoters say the absence of correlations (between 1 & 2 and between 2 & 3) is powerful evidence for the safety of Al adjuvant. But a close look at the Karwowski paper reveals that the results are essentially meaningless for assessing Al adjuvant safety. The study has a nonsensical design and is fatally flawed. The Karwowski results are grossly inadequate for supporting the safety of Al adjuvant.
The problems include:
1) Al Adjuvant Does Not Raise Blood or Hair Levels of Aluminum.
Vaccine promoters have misconceptions about how Al adjuvant moves around the body (i.e. the kinetics, or pharmacokinetics). Dissolved Al3+ ions and solid Al adjuvant particles move around the body very differently.
Most studies of Al toxicity look only at water-soluble Al3+ (e.g. from Al lactate or Al chloride). Al exposure from food enters the body as Al3+ ions. The kidneys quickly remove this dissolved Al3+. Most of it is gone within 24 hours.
Al adjuvants are different, because they do not dissolve. They are made of insoluble solid particles of Al hydroxide, Al phosphate or Al sulfate. Because of the low solubility, Al adjuvant remains in the body and is not filtered by the kidneys. Al adjuvant particles have been detected in humans over a decade after vaccination. It does dissolve very slowly however, releasing a trickle of Al3+ ions. These Al3+ ions do appear in blood, but the amount is so small as to be undetectable (compared to baseline, normal levels).
The solid Al adjuvant particles do not appear in the blood. Al adjuvant particles from vaccines are phagocytosed (i.e. eaten) by macrophages, a type of immune system cell. Mostly, the Al adjuvant remains at the injection site, inside the macrophages.
Macrophages rarely travel through the blood, but they do travel all over the body, mostly via the lymphatic system. As macrophages travel, they carry the Al adjuvant particles inside them. This macrophage transport process occurs slowly, over months or years. Consequently, at any given time, there is little Al adjuvant in transit. Macrophages travel to organs like the spleen, liver, stomach, kidneys, lymph nodes, and brain. Al adjuvant particles cause chronic inflammation and injury wherever they go.
The transport of Al adjuvant is explained here: Vaccine Aluminum Travels Into The Brain.
1A) Aluminum From Al Adjuvant (Mostly) Does Not Appear in Blood
Flarend 1997 is a widely cited study of the kinetics of Al adjuvant. It has some shortcomings, but the study design is sound and it is generally considered to provide the best data available on the transport of Al adjuvant.
In the Flarend study, rabbits were injected intramuscularly with 850 mcg Al as Al hydroxide or Al phosphate. Blood and urine Al levels were measured many times over the subsequent 28 days. The Al level in the blood remained within normal range for the entire study, though there was a slight increase in the first couple days.
The highest blood Al level measured was 2 ng Al/g blood, which occurred about 10 hours after injection with Al phosphate. The normal range of Al concentration in human serum (serum is the clear liquid left over after blood is clotted) is about 0-6 ng/g or 0-15 ng/g (depending on who you ask). Al is indisputably toxic at blood levels exceeding about 50 ng/g. Al phosphate produced a peak blood level of 2 ng/g, and Al hydroxide produced a peak Al blood level of 1 ng/g. Al phosphate produced higher blood levels simply because it dissolves faster than Al hydroxide.
Above: Blood aluminum levels in rabbits after intramuscular injection of Al hydroxide or Al phosphate adjuvant. Chart shows blood Al level over time. Units of 1E-6 mg/g are equivalent to nanograms Al per gram of blood (ng/g), and micrograms per liter (mcg/L). Al phosphate is more soluble than Al hydroxide, which is why Al phosphate produced a higher Al level than Al hydroxide (2ng/g compared to 1ng/g). Normal Al level in human blood serum is about 0-6 or 0-15 ng/g, so the 2 ng/g spike reported in Flarend is not a cause for concern. From Flarend 1997: In vivo absorption of aluminium-containing vaccine adjuvants using 26Al
Another study (Movsas 2013) of human infants showed that vaccination at the 2-month date per the CDC schedule did not increase Al levels in blood or urine the following day. Specifically, Movsas 2013 measured urine and blood Al levels before and after vaccination of 15 infants with 1200mcg Al from vaccines. Movsas states: “No significant change in levels of urinary or serum aluminum were seen after vaccination.”
Movsas 2013: Effect of Routine Vaccination on Aluminum and Essential Element Levels in Preterm Infants
Flarend 1997 and Movsas 2013 clearly indicate that Al adjuvant has essentially no effect on blood Al levels.
So, obviously, Karwowski found no correlation between Al adjuvant dose and Al blood levels. Karwowski states:
“No correlation was found between hair-Al or blood-Al concentrations and the infant’s history of receipt of aluminum-containing immunizations, either the estimated cumulative aluminum load from previous immunizations or that from vaccines received on the date of testing. These results are similar to those reported by investigators [Movsas 2013] who studied 15 premature infants before and after they received 1200 mg aluminum in their 2-month-old immunizations and reported no changes in blood or urine aluminum levels.21”
Of course they did not find anything. Al from Al adjuvant does not appear in the blood!
Amazingly, Ref #21 is Movsas 2013. The Karwowski researchers were aware of the Movsas 2013 results, so why did they choose to measure Al blood levels? Further, why measure Al blood levels months after vaccination? This aspect of the study makes little sense because it was already known the Al does not appear in the blood.
I suppose it’s helpful to have additional research confirming that Al from vaccines does not enter the blood. However, since the Al does not appear in the blood, it is not logical to use Al blood levels to evaluate the safety of Al adjuvant.
I emailed Dr Alan Woolf (the corresponding author for the Karwowski paper) inquiring about these issues, but Dr Woolf did not respond.
1B) Al From Al Adjuvant Not Correlated With Hair Levels
A 2014 study (Bohrer 2014) compared cumulative Al adjuvant dose from vaccines to the hair Al level in 37 children aged 1-27 months. There was no correlation between Al adjuvant dose and hair Al level. As expected, Karwowski also found no correlation between Al adjuvant dose and Al hair levels.
Above: Bohrer 2014 reported that cumulative Al adjuvant dose and hair Al level are not correlated (top chart). Also, the time since the most recent Al adjuvant dose was not correlated with hair Al level. From Bohrer 2014 (full paper) Distribution of aluminum in hair of Brazilian infants and correlation to aluminum-adjuvanted vaccine exposure
The Karwowski paper did not cite Bohrer 2014, so perhaps Karwowski et al were unaware of this evidence showing that hair Al level is not influenced by Al adjuvant dose.
The range of Al doses received by children in the Bohrer study was 0.63 to 6.88 mg. By comparison, subjects in the Karwowski study received 1.43-3.55 mg Al. The wider range of Al doses in Bohrer enhances it’s ability to detect an association between Al dose and Al in hair. However, the Bohrer study had fewer subjects (37 vs 85 for Karwowski), and a wider range of subject ages (1-27 months), which will both tend to reduce the ability to detect an association.
2) No Analysis of Al Adjuvant Dose and BSID Score
The Karwowski study has a troubling omission: it did not look for, or even mention, a possible correlation between Al adjuvant dose and BSID score.
Karwowski collected 3 types of data:
1) total lifetime dose of Al adjuvant from vaccines,
2) aluminum level in hair and blood,
3) Bayley Scale of Infant and Toddler Development (BSID) scores.
There are 3 possible correlations: between 1 & 2, 2 & 3, and 1 & 3. Karwowski looked for correlations between 1 & 2 and 2 & 3, but did not look for a correlation between 1 & 3 (Al adjuvant dose and BSID score). It makes no sense to not analyze the data for a possible correlation between Al adjuvant dose and BSID score.
Karwowski 2018 was apparently motivated by concern about the safety of Al adjuvant. It is therefore inexplicable that it did not test the simplest and most obvious question: is Al adjuvant dose correlated with BSID score? Instead of answering this question, the study only looked for links to blood and hair Al levels. There is no explanation in the paper of why a correlation between 1 & 3 was not considered. The silence about this is troubling.
It has been suggested that Karwowski did not look for correlation between 1 & 3 because of the Al adjuvant dose similarity. There was little variance in the Al adjuvant dose, and this reduces the statistical power to observe an effect if the data is analyzed by Al adjuvant dose. This is a reasonable basis for looking at other measurements, like blood/hair Al. To illustrate why this is so, consider 100 people given an identical dose of a drug. The subjects will have different reactions to it. The drug causes adverse reactions in some, but looking for a correlation with dose will reveal nothing because everyone got the same dose! The drug could be linked to an adverse outcome by looking at the drug kinetics. For example, the drug may cause adverse effects only in the individuals with impaired drug clearance or with drug accumulation in a particular organ or tissue. For this reason, it is logical to look for an association between blood/hair and BSID scores.
But this reasoning does not justify not looking for an association between Al adjuvant dose and BSID score.
So, it is possible that hair Al levels may correlate with injury, even if Al adjuvant dose does not correlate with injury. For example, it is theoretically possible for hair Al level to indicate a general level of Al transport into the brain or other sensitive tissues. This is pure speculation, however.
3) Inadequate Duration Between Vaccination and BSID Testing
BSID scores were measured at 9-13 months of age. This probably does not provide enough time for the Al adjuvant to travel into the nervous system. Al adjuvant transport requires months in experimental animals. It may take even longer in humans.
Also, neurodevelopmental disorders such as autism are difficult to detect at age 9-13 months. At 12 months, perceptive parents may notice slight behavioral abnormalities. Autism is typically diagnosed at about 3-4 years.
Below are development trajectories for autistic (ASD), language delayed (LD), and normal (unaffected) children. Note that the trajectories are not very different in the age range of 9-13 months (when Karwowski measured BSID scores).
Above: Development trajectories for autistic (ASD), language delayed (LD), and normal (unimpaired) children. Karwowski measured BSID scores at 9-13 months, when developmental differences are very small and hard to detect. From Landa and Meyer 2006, paper: https://www.ncbi.nlm.nih.gov/pubmed/16712640
4) Subjects Received Similar Doses of Al adjuvant. No Unexposed Controls.
The Karwowski subjects received similar Al adjuvant doses. For example, the average Al adjuvant dose was 2.9mg, and 50% of subjects received 2.85-2.95 mg (this is calculated from the inter-quartile range (IQR) which was 0.11mg). All subjects received 1.43-3.55 mg Al adjuvant, and none received zero. The dose similarity and absence of zero-exposure (i.e. control) subjects severely impairs the ability to detect potential associations with Al adjuvant dose.
5) Other Sources of Aluminum Not Considered.
Infants are exposed to aluminum in food. Different food sources have vastly different aluminum levels. Human breast milk has the lowest levels and soy formula the highest. In the first 6 months, an infant will ingest about 7mg Al from breast milk, and 117mg from soy formula. Absorption rate of ingested Al is about 0.1-0.3%, which produces the following exposures:
Human milk: 7mg x 0.3% = 21 micrograms (0.021 mg)
Formula: 38mg x 0.3% = 114 micrograms (0.114 mg)
Soy formula: 117mg x 0.3% = 351 micrograms (0.351mg)
Karwowski did not investigate or attempt to correct for these or any other sources of aluminum.
6) Inadequate Statistical Power
The Karwowski study had 85 subjects. This is not enough to answer concerns about vaccines and autism, or other neurological disorders. For example, autism occurs in about 1-2%, which means that only 1 or 2 of the 85 subjects would be expected to become autistic. Many more subjects (several hundred) would be required to detect autism.
All the above factors together practically assure that Karwowski will not detect any affect of Al adjuvant. The dose similarity, absence of controls, too-early age of BSID testing, and inadequate number of subjects all reduce the ability of the Karwowski study to detect adverse effects of Al adjuvant.
Karwowski 2018 is grossly inadequate as evidence for the safety of Al adjuvant. Al adjuvant is given to almost all infants, including the healthy, at the beginning of life. Accordingly, evidence of safety should be especially well established. Karwowski 2018 is a far cry from what is needed to support claims of safety.
The only thing the Karwowski paper may be good for is confirming prior studies showing that Al adjuvant does not appear in the blood or hair (though the dose similarity reduces the validity of this finding). There is some value in this, because only a few studies provide data on Al levels in blood or hair after exposure to Al adjuvant. Only one prior study looked for Al in hair, so its nice to have this single result confirmed in a study with a slightly different design. So there is that.
A Comment About Journal Impact Factor
Skeptical Raptor (SR) described the journal Academic Pediatrics as a “top journal”. SR often mentions the importance of “high impact factor” journals. I don’t focus on impact factors, since they are gamed and manipulated by journals and the pharmaceutical industry. But since SR and the misnamed “skeptic” community believe impact factors are so important, lets take a look. The Academic Pediatrics journal has an impact factor of 2.720. By comparison, The Journal of Inorganic Biochemistry, which has published many papers cited here and papers critical of aluminum adjuvant safety, has an impact factor of 3.348.
Academic Pediatrics impact factor: https://www.journals.elsevier.com/academic-pediatrics?
Journal of Inorganic Biochemistry impact factor: https://www.journals.elsevier.com/journal-of-inorganic-biochemistry
Dr Chris Exley On Karwowski et al.
Dr Chris Exley sent a letter criticizing the Karwowski study to the Academic Pediatrics journal, and the Karwowski authors. Academic Pediatrics does not publish letters to the editor, and the Karwowski authors have not responded. The journal and Karwowski authors are not defending the paper.
Additionally, the Karwowski paper authors refuse to release their data, a standard requirement in scientific publishing.
Dr Exley said the following in an email to me:
“I think it is important to point out that the authors of the paper refuse all correspondence and additionally have refused to provide data used in coming to their conclusions. The latter is a prerequisite of publication but the editor also refuses to insist that the authors provide their supporting data. On this basis alone the paper should be retracted.”
Here is Dr Exley’s letter to the editor: Exley-Letter-to-the-Editor