Dr. Faust, a superb and incredibly necessary piece of scientific clarification. Thank you for cutting through the noise with such a clear-headed analysis of the evidence.
Your breakdown of the 2024 Swedish sibling-controlled study is a masterclass in interpreting epidemiological data. The fact that the researchers could replicate the original association without the sibling control is, as you say, a deeply compelling piece of evidence. It's the very definition of a more rigorous method identifying and correcting for confounding variables.
As a biochemist, I was also nodding along with your point on the precautionary principle. The biological reality is that a high maternal fever is a significant stressor that can have its own cascade of downstream effects on fetal development. Advising against the very tool that can mitigate that fever, based on weaker evidence, is a perfect example of how a well-intentioned but scientifically un-nuanced application of the precautionary principle can lead to a potentially worse outcome.
This is a vital defense of evidence-based policymaking. Thank you for your clarity and courage.
Not necessarily for this article, but I also think it’s important for lay people to understand that acetaminophen is one of the only over the counter pain relievers safe in pregnancy. If we know untreated depression and anxiety can have detrimental effects on a fetus, one should be able to assume that ongoing pain (chronic pre-pregnancy or pregnancy-related) could also lead to anxiety/depression and detrimental effects on the fetus. Sowing fear about taking Tylenol during pregnancy will only lead to harm for pregnant women.
I could be wrong as I have zero medicine or science background…one of the many dangerous aspects of RFK is that he also does not have a medical or science background and is leading the HHS department.
NSAIDs and Aspirin are contraindicated in at least the 3rd trimester as it can prematurely close the ductus arteriosus that the fetus needs till after delivery. Not only that, but aspirin (even a few) can make one bleed like stink with even minor trauma, much less a c-section or vaginal delivery.
Apologies for this very long post, every question is simple but the answers are complex and nuanced.
Thanks for point out the JAMA article (Ahlqvist et al, 2024). Overall, in my opinion, the article does not provide any convincing evidence that any of the exposures, including acetaminophen, have an association with autism, or the other endpoints. I wanted to share some other thoughts related to its interpretation and some of the challenges that often are lost. One really needs to look at an article in detail.
The population analysis consists of 2,489,721 children and the sibling analysis consist of 1,773,747. Both very large numbers, the siblings analysis is a subgroup of the population analysis. The large numbers will result in very narrow confidence intervals and the likelihood of increase false positive association. Interpreting a result as an association when it is not. Specific analyses have other exclusion criteria. Given the two group have large but different number of children, even when the results of the point estimate of the hazard ratio the confidence intervals will difference, assuming all other aspects of the analysis are identical.
The first topic is the size of the effect. Looking at Figure 3, the estimated HR (hazard ratio) for acetaminophen and autism for the population analysis HR 1.05 with a 95% CI (Confidence Interval) 1.03, 1.08, for the siblings analysis the results are HR 0.98, 95% CI 0.93. If we assume that there is no effect, neither of these results are 1.0. This may be due to either bias, as discussed, or noise in the sample. Just because the population analysis demonstrates a possible association does not mean that it is true; the result could be a false positive, even if there were no bias. The siblings analysis has a point estimate of 0.98, this could be due to precision (numerical noise) or due to some bias or both.
For me, an important issue is the size of the effect. Even if the true effect is a HR 1.05 as estimated in the population analysis, this indicates an increase in the HR of 5%, not a 5% increase. The analysis does not provide an estimate of the baseline hazard but lets assume the annual rate is 0.002 (0.2%) per year, a back of the envelope estimate base on some data in the article. Let’s further do some simple back of the envelope calculations – these are quick estimates. Exposure would increase this to 0.0021 (0.21%), a difference of 0.0001 per year (1 per 10,000 per year). This could contribute a maximum of an additional 3,000 cases over 10 years, from a birth cohort of 3 million, with every child exposed in utero. The total number of cases would be 63,000, of which 60,000 would occur without any exposure. Thus, the factors contributing to the majority of autism cases are not explained by this factor.
An example from Figure 3 in the article, the estimated HR (hazard ratio) for intellectual disability and antimigraine therapies for the population analysis HR 1.17 with a 95% CI (Confidence Interval) 1.07, 1.27, for the siblings analysis the results are HR 1.17, 95% CI 0.96, 1.43. We do not know truth but if we assume the true effect is a HR of 1.17 then the only difference is the less precise estimate for the sibling analysis; here, the sibling analysis is a false negative. However, if the true result is 1.0, antimigraine therapies for the population analysis would indicate a possible association a false positive finding, while the sibling analysis would indicate the opposite, no association.
The challenge is that Figure 3 has 30 analytic results and the authors interpret a 95% CI as indicating an association if that CI excludes the null value, a HR of 1.0, no difference. Effectively, the authors are doing a hypothesis test. Regardless, the number of contrasts and large sample size will likely, in my opinion, lead to many false positives. Some would use an adjustment, eg a Bonferroni, to keep the overall false positive error rate at 5%. Here, there is an additional level of complexity, the two analyses have substantially different number of children. A simple Bonferroni adjust will not address this. A Bonferroni adjustment modifies the level of statistical significance by the number of tests; so, if one wanted a 5% overall level (type I error) then one would divide this by the number, 0.05, by the number tests. Based on the methods section, my expectation was that expect many analyses would lead to the appearance of an association, that is, the 95% CI would exclude the null value a HR of 1.0. In Figure 3, 10 of the 15 population 95% CI excluded 1; only 2 of the 15 sibling analyses excluded 1.
The two siblings analysis where the 95% CI exclude 1.0 were for: (1) the relationship between intellectual disability and aspirin, the other (2) the relationship between use aspirin use and autism. What is intriguing about these two results is that both indicate a protective association between aspirin use and the outcome. For (1), the HR is 0.71, 95% CI 0.57, 0.88. For (2) the HR is 0.87, 95% CI 0.75, 0.99. The effect size of aspirin use and intellectual disability is the largest in Figure 3 but the association indicates a reduction in risk; the authors briefly mentioned in the discussion as warranting further investigation.
The use of the 5% criteria has no special properties, the 95% CI. The selection of this level is arbitrary, even though its use in health-related analyses has become expected. One may want to read Michael Oakes book Statistical Inference, where the various schools of statistics are discussed. Statistical inference is a more nuance and complex topic than discussed in Oakes’ book but this is an excellent, non-technical overview, in my opinion.
As a sidebar note, GWAS (genome-wide association study) is an area that was faced with a similar issue. When some individuals first started to undertake these studies with a similar approach, a large number of contrasts, in a large population, with genes not being independent within an individual. The problem was that nearly every contrast would exceed any of these simple criteria for association, such at 5% or even 1%. Which meant that everything warranted further investigation. To address this issue, other techniques have been developed.
Here are the court docs in the Katie Johnson rape case that specifically implicate Trump in the rape of a 13 year old (and another 12 year old). Katie dropped the case after she received death threats
I'd love to see Mr. Kennedy's science that supports these assertions. I'd love to see Mr. Kennedy's research published in a peer-reviewed journal. I have no doubt Mr. Kennedy assertions will not stand up to the light of day, much like all of the rest of his assertions about vaccines, nutrition, infectious diseases, public health, and medicine.
So, the entire article is about how (after reviewing the data and providing "a masterclass in interpreting data" yikes! your fan-base!) is that you agree with Kennedy that it needs to be looked into.
Here are the court docs in the Katie Johnson rape case that specifically implicate Trump in the rape of a 13 year old (and another 12 year old). Katie dropped the case after she received death threats
This observation is simply based on the idea that Tylenol depletes glutathione in the body. According to many element of research, autism (ASD) etiology could really be linked to low endogenous glutathione. From a biochemical point of view, RFK's theory makes sense.
The liver replenishrd the glutathione when not exceeding the recommended daily dosage of "Tylenol". In addition, those taking APAP daily should not use 4 g as a max...3200 mg from all sources.
Dr. Faust, a superb and incredibly necessary piece of scientific clarification. Thank you for cutting through the noise with such a clear-headed analysis of the evidence.
Your breakdown of the 2024 Swedish sibling-controlled study is a masterclass in interpreting epidemiological data. The fact that the researchers could replicate the original association without the sibling control is, as you say, a deeply compelling piece of evidence. It's the very definition of a more rigorous method identifying and correcting for confounding variables.
As a biochemist, I was also nodding along with your point on the precautionary principle. The biological reality is that a high maternal fever is a significant stressor that can have its own cascade of downstream effects on fetal development. Advising against the very tool that can mitigate that fever, based on weaker evidence, is a perfect example of how a well-intentioned but scientifically un-nuanced application of the precautionary principle can lead to a potentially worse outcome.
This is a vital defense of evidence-based policymaking. Thank you for your clarity and courage.
Not necessarily for this article, but I also think it’s important for lay people to understand that acetaminophen is one of the only over the counter pain relievers safe in pregnancy. If we know untreated depression and anxiety can have detrimental effects on a fetus, one should be able to assume that ongoing pain (chronic pre-pregnancy or pregnancy-related) could also lead to anxiety/depression and detrimental effects on the fetus. Sowing fear about taking Tylenol during pregnancy will only lead to harm for pregnant women.
I could be wrong as I have zero medicine or science background…one of the many dangerous aspects of RFK is that he also does not have a medical or science background and is leading the HHS department.
You don't need medical and science degrees for common sense. EVERYONE AGREES...we need to look into this.
NSAIDs and Aspirin are contraindicated in at least the 3rd trimester as it can prematurely close the ductus arteriosus that the fetus needs till after delivery. Not only that, but aspirin (even a few) can make one bleed like stink with even minor trauma, much less a c-section or vaginal delivery.
❣️Thanks Dr Faust
🤦♀️RFK JR
https://bsky.app/profile/kenaiseasky.bsky.social/post/3ly62ulmdh224
Apologies for this very long post, every question is simple but the answers are complex and nuanced.
Thanks for point out the JAMA article (Ahlqvist et al, 2024). Overall, in my opinion, the article does not provide any convincing evidence that any of the exposures, including acetaminophen, have an association with autism, or the other endpoints. I wanted to share some other thoughts related to its interpretation and some of the challenges that often are lost. One really needs to look at an article in detail.
The population analysis consists of 2,489,721 children and the sibling analysis consist of 1,773,747. Both very large numbers, the siblings analysis is a subgroup of the population analysis. The large numbers will result in very narrow confidence intervals and the likelihood of increase false positive association. Interpreting a result as an association when it is not. Specific analyses have other exclusion criteria. Given the two group have large but different number of children, even when the results of the point estimate of the hazard ratio the confidence intervals will difference, assuming all other aspects of the analysis are identical.
The first topic is the size of the effect. Looking at Figure 3, the estimated HR (hazard ratio) for acetaminophen and autism for the population analysis HR 1.05 with a 95% CI (Confidence Interval) 1.03, 1.08, for the siblings analysis the results are HR 0.98, 95% CI 0.93. If we assume that there is no effect, neither of these results are 1.0. This may be due to either bias, as discussed, or noise in the sample. Just because the population analysis demonstrates a possible association does not mean that it is true; the result could be a false positive, even if there were no bias. The siblings analysis has a point estimate of 0.98, this could be due to precision (numerical noise) or due to some bias or both.
For me, an important issue is the size of the effect. Even if the true effect is a HR 1.05 as estimated in the population analysis, this indicates an increase in the HR of 5%, not a 5% increase. The analysis does not provide an estimate of the baseline hazard but lets assume the annual rate is 0.002 (0.2%) per year, a back of the envelope estimate base on some data in the article. Let’s further do some simple back of the envelope calculations – these are quick estimates. Exposure would increase this to 0.0021 (0.21%), a difference of 0.0001 per year (1 per 10,000 per year). This could contribute a maximum of an additional 3,000 cases over 10 years, from a birth cohort of 3 million, with every child exposed in utero. The total number of cases would be 63,000, of which 60,000 would occur without any exposure. Thus, the factors contributing to the majority of autism cases are not explained by this factor.
An example from Figure 3 in the article, the estimated HR (hazard ratio) for intellectual disability and antimigraine therapies for the population analysis HR 1.17 with a 95% CI (Confidence Interval) 1.07, 1.27, for the siblings analysis the results are HR 1.17, 95% CI 0.96, 1.43. We do not know truth but if we assume the true effect is a HR of 1.17 then the only difference is the less precise estimate for the sibling analysis; here, the sibling analysis is a false negative. However, if the true result is 1.0, antimigraine therapies for the population analysis would indicate a possible association a false positive finding, while the sibling analysis would indicate the opposite, no association.
The challenge is that Figure 3 has 30 analytic results and the authors interpret a 95% CI as indicating an association if that CI excludes the null value, a HR of 1.0, no difference. Effectively, the authors are doing a hypothesis test. Regardless, the number of contrasts and large sample size will likely, in my opinion, lead to many false positives. Some would use an adjustment, eg a Bonferroni, to keep the overall false positive error rate at 5%. Here, there is an additional level of complexity, the two analyses have substantially different number of children. A simple Bonferroni adjust will not address this. A Bonferroni adjustment modifies the level of statistical significance by the number of tests; so, if one wanted a 5% overall level (type I error) then one would divide this by the number, 0.05, by the number tests. Based on the methods section, my expectation was that expect many analyses would lead to the appearance of an association, that is, the 95% CI would exclude the null value a HR of 1.0. In Figure 3, 10 of the 15 population 95% CI excluded 1; only 2 of the 15 sibling analyses excluded 1.
The two siblings analysis where the 95% CI exclude 1.0 were for: (1) the relationship between intellectual disability and aspirin, the other (2) the relationship between use aspirin use and autism. What is intriguing about these two results is that both indicate a protective association between aspirin use and the outcome. For (1), the HR is 0.71, 95% CI 0.57, 0.88. For (2) the HR is 0.87, 95% CI 0.75, 0.99. The effect size of aspirin use and intellectual disability is the largest in Figure 3 but the association indicates a reduction in risk; the authors briefly mentioned in the discussion as warranting further investigation.
The use of the 5% criteria has no special properties, the 95% CI. The selection of this level is arbitrary, even though its use in health-related analyses has become expected. One may want to read Michael Oakes book Statistical Inference, where the various schools of statistics are discussed. Statistical inference is a more nuance and complex topic than discussed in Oakes’ book but this is an excellent, non-technical overview, in my opinion.
As a sidebar note, GWAS (genome-wide association study) is an area that was faced with a similar issue. When some individuals first started to undertake these studies with a similar approach, a large number of contrasts, in a large population, with genes not being independent within an individual. The problem was that nearly every contrast would exceed any of these simple criteria for association, such at 5% or even 1%. Which meant that everything warranted further investigation. To address this issue, other techniques have been developed.
RePedoKKKunt
RapeThugLiKKKunt
SICKO!!!
PIG!!!
THIS IS WHO HE IS-SCUM!
https://veteranstoday.com/2020/10/27/blockbuster-report-trump-settlements-for-10-child-rapes-half-boys-bankruptcies-justice-department-coverup/
https://www.scribd.com/doc/316341058/Donald-Trump-Jeffrey-Epstein-Rape-Lawsuit-and-Affidavits#fullscreen?platform=hootsuite
https://nymag.com/intelligencer/2017/12/what-happened-to-trumps-16-sexual-misconduct-accusers.html
https://www.facebook.com/share/v/1QxLiXY7hR/
https://www.facebook.com/share/r/1CNYrYwBHw/ 0
Here are the court docs in the Katie Johnson rape case that specifically implicate Trump in the rape of a 13 year old (and another 12 year old). Katie dropped the case after she received death threats
https://cdn.factcheck.org/UploadedFiles/Johnson_TrumpEpstein_Calif_Lawsuit.pdf
https://www.theguardian.com/us-news/2016/feb/25/donald-trump-sexual-assault-125-million-lawsuit?CMP=Share_AndroidApp_Other
https://nymag.com/intelligencer/2017/12/what-happened-to-trumps-16-sexual-misconduct-accusers.html
https://substack.com/@newyorklovesbts/note/c-140749011?r=4rdcx6
https://www.facebook.com/share/v/1QxLiXY7hR/
https://www.facebook.com/share/r/1CNYrYwBHw/ 0
https://youtu.be/DeLoTf9QUqI?si=CGhXoIXMLLDu25KA
https://cdn.factcheck.org/UploadedFiles/Johnson_TrumpEpstein_Calif_Lawsuit.pdf
https://www.facebook.com/share/v/19d2E6HjNM/
https://www.npr.org/2016/10/13/497799354/a-list-of-donald-trumps-accusers-of-inappropriate-sexual-conduct
https://www.latimes.com/politics/la-na-pol-trump-accusers-fact-check-20161019-snap-htmlstory.html
#ReleaseTheFiles
.... has entered the chat: JFJr wants to link it to antiRFK crusade, even though the best evidence says otherwise.
It’s a pseudoargument to entrench divisions and it’s not safe or productive.
By Acadfluencers
I'd love to see Mr. Kennedy's science that supports these assertions. I'd love to see Mr. Kennedy's research published in a peer-reviewed journal. I have no doubt Mr. Kennedy assertions will not stand up to the light of day, much like all of the rest of his assertions about vaccines, nutrition, infectious diseases, public health, and medicine.
The real tragedy is not that Secretary Pedo Protector jr is in charge of disease and death promotion, the tragedy is that he hasn’t been removed
First it’s vaccines and now Tylenol. RFK just grabs at anything he can to explain autism. Next he will blame women.
The author here agrees with Kennedy that it needs to be looked into. First, you need to read the article.
So, the entire article is about how (after reviewing the data and providing "a masterclass in interpreting data" yikes! your fan-base!) is that you agree with Kennedy that it needs to be looked into.
Thank you!
RePedoKKKunt
RapeThugLiKKKunt
SICKO!!!
PIG!!!
THIS IS WHO HE IS-SCUM!
https://veteranstoday.com/2020/10/27/blockbuster-report-trump-settlements-for-10-child-rapes-half-boys-bankruptcies-justice-department-coverup/
https://www.scribd.com/doc/316341058/Donald-Trump-Jeffrey-Epstein-Rape-Lawsuit-and-Affidavits#fullscreen?platform=hootsuite
https://nymag.com/intelligencer/2017/12/what-happened-to-trumps-16-sexual-misconduct-accusers.html
https://www.facebook.com/share/v/1QxLiXY7hR/
https://www.facebook.com/share/r/1CNYrYwBHw/ 0
Here are the court docs in the Katie Johnson rape case that specifically implicate Trump in the rape of a 13 year old (and another 12 year old). Katie dropped the case after she received death threats
https://cdn.factcheck.org/UploadedFiles/Johnson_TrumpEpstein_Calif_Lawsuit.pdf
https://www.theguardian.com/us-news/2016/feb/25/donald-trump-sexual-assault-125-million-lawsuit?CMP=Share_AndroidApp_Other
https://nymag.com/intelligencer/2017/12/what-happened-to-trumps-16-sexual-misconduct-accusers.html
https://substack.com/@newyorklovesbts/note/c-140749011?r=4rdcx6
https://www.facebook.com/share/v/1QxLiXY7hR/
https://www.facebook.com/share/r/1CNYrYwBHw/ 0
https://youtu.be/DeLoTf9QUqI?si=CGhXoIXMLLDu25KA
https://cdn.factcheck.org/UploadedFiles/Johnson_TrumpEpstein_Calif_Lawsuit.pdf
https://www.facebook.com/share/v/19d2E6HjNM/
https://www.npr.org/2016/10/13/497799354/a-list-of-donald-trumps-accusers-of-inappropriate-sexual-conduct
https://www.latimes.com/politics/la-na-pol-trump-accusers-fact-check-20161019-snap-htmlstory.html
#ReleaseTheFiles
This observation is simply based on the idea that Tylenol depletes glutathione in the body. According to many element of research, autism (ASD) etiology could really be linked to low endogenous glutathione. From a biochemical point of view, RFK's theory makes sense.
The liver replenishrd the glutathione when not exceeding the recommended daily dosage of "Tylenol". In addition, those taking APAP daily should not use 4 g as a max...3200 mg from all sources.
JJF Phm 🇨🇦
He is a supremely dangerous man. Aside from spreading disinformation, he is making parents of children with ASD feel guilty FOR NO REASON.