Week of September 8, 2025
A patient (who likes to wear a certain red hat to our appointments) asked me to "bless" him starting Ozempic last week. By bless, he meant I should assure him that the medication "wouldn't hurt his kidneys" and that he "wouldn't go blind." His questions hinted at a complex array of thoughts - perhaps more than he realized. On the one hand, I was glad that he listened to my frequent admonitions to ask about the appropriateness and dosing of any new medications (is it safe for his degree of kidney disease?). On the other hand, he tacitly expressed his healthcare skepticism (does that other, younger physician who is recommending the "medication with a lot of hype" really know what they are doing?). Moreover, his comment is an excellent example of how complex scientific findings are filtered through various news outlets and into the minds of somewhat isolated, rural octogenarians. He said, "If you say it's okay, I trust you;" not, "I've seen news reports of vision problems with Ozempic - What is the data on the risks for a patient like me?" I have not found success in assuring people with detailed data on absolute and relative risk. So, I told him the risk of blindness was minimal (though not zero), and that he would likely benefit from the drug. He smiled, agreed to take the medication, and I wondered what he would think if he had a complication.
For reference, what I didn't say was:
"Current data indicate that glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are associated with an increased risk of certain vision-threatening adverse events, particularly in patients with diabetes and pre-existing ocular risk factors. The most robust signals concern diabetic retinopathy (DR) complications and nonarteritic anterior ischemic optic neuropathy (NAION), especially with semaglutide. However, absolute risk remains low, and causality is not established due to confounding and retrospective design from the studies in which researchers observed these trends."
https://www.openevidence.com/ask/0073dd03-abe5-4b4f-b0cc-22a65b0b8c31
---
Listen to a Google Notebook LM A.I.-generated podcast of the newsletter with two virtual "hosts."
https://drive.google.com/file/d/1HowVpcuNyjGzIBJ23UICQDvnfBEVIinC/view
------
Science and Technology Trends
Each week, I see several articles on Glucagon-like peptide-1 (GLP-1) medications (semaglutide and tirzepatide) demonstrating clinical benefits beyond glycemic control. Last week, I found the STEER trial and one other randomized controlled trial demonstrating that semaglutide is associated with a greater reduction in major adverse cardiovascular events (MACE) like heart attack and stroke compared to tirzepatide, especially in people with obesity and cardiovascular disease without diabetes ( both drugs lowered cardiovascular risk, but semaglutide appears more effective in this population). In heart failure with preserved ejection fraction (HFpEF) and cardiometabolic comorbidities, both semaglutide and tirzepatide demonstrated over a 40% reduction in hospitalizations or death, with no significant difference between the two. Additionally, early research suggests GLP-1 agonists may reduce alcohol use and show potential for treating psychiatric conditions such as depression, though findings remain preliminary. And in patients who are getting antipsychotic medications (which are often associated with increased appetite, weight gain, and diabetes), GLP-1s lowered weight and improved glycemic control. As I have said before, this is what compelling science looks like - numerous studies over time demonstrating consistent patterns of results. In this case, there is accumulating data highlighting GLP-1s offer broad benefits for cardiometabolic health, psychiatric applications, and promise in treating other conditions like alcohol use disorder.
I've included AI summaries.
Steer Study - GLP1s lower CV Risk
https://www.week' som/news/semaglutide-superior-to-tirzepatide-in-reducing-cardiovascular-event-risk/
https://x.com/therealryc/status/1962213801635569738?s=57
Semaglutide and Tirzepatide in Patients With Heart Failure With Preserved Ejection Fraction
https://jamanetwork.com/journals/jama/fullarticle/2838293
https://claude.ai/public/artifacts/424534e5-2d57-4ba7-a2ac-2048d491714f
Use in Psychiatric Patients taking Antipsychotic medication
https://x.com/therealryc/status/1963602953601757581?s=42
https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2838205
https://claude.ai/public/artifacts/9e32b03d-a9bc-4239-ad73-34f3f97019cd
Cardiovascular and kidney outcomes of GLP-1 receptor agonists in adults with obesity: A target trial emulation study
https://dom-pubs.onlinelibrary.wiley.com/doi/full/10.1111/dom.70054
https://claude.ai/public/artifacts/38199647-448e-4882-a479-eeadbe574203
Once-Weekly Semaglutide in Adults With Alcohol Use Disorder - A Randomized Clinical Trial
https://jamanetwork.com/journals/jamapsychiatry/fullarticle/2829811
https://claude.ai/public/artifacts/ca021ac8-201e-4e9c-8597-5a58c6367a4b
Australian researchers published a detailed technical review of how AI is revolutionizing protein design. Traditional versions of this work require thousands of compounds and months of testing. However, AI-driven tools, such as AlphaFold, have substantially improved efficiency by combining novel protein designs with predictions that enable researchers to identify the most likely candidate proteins to test. While there are still many limitations, these tools are rapidly scaling up, offering a promising future for protein design and other scientific fields.
Discussion on X:
https://x.com/nikomccarty/status/1962589782011453871?s=57
Article:
https://t.co/l4dRgtxNVi
AI Summary:
https://claude.ai/public/artifacts/53f52441-809d-49b7-91b6-8b196daecb5a
More comforting data about coffee consumption - "We examined the association of coffee consumption with brain MRI "parameters representing vascular brain damage, neurodegeneration, and microstructural integrity in 2316 participants in the population-based Hamburg City Health Study. After adjustment for covariates, 3–4 cups of daily coffee were associated with [denser, thicker brains as measured by MRI] compared to <1 cup." Data are imperfect - these data are from a large cross-sectional" sample of a homogeneous population (Germans in Hamburg) who self-reported their coffee consumption. In addition, it is unclear what kinds of coffee, how long one must consume coffee, or what other confounders in lifestyle or behavior might exist. Either way, it is comforting to find yet more data that coffee is not harmful and MAY be helpful.
X Discussion:
https://x.com/ntfabiano/status/1962481100095181012?s=57
Paper
https://www.mdpi.com/2072-6643/15/3/674
AI Review:
https://claude.ai/public/artifacts/02be9906-416b-4a31-9336-51ea9a401ce2
Anti-Science Articles of Note
I find it difficult to describe my sense of horror watching the video clips from last week's Senate testimony from our Secretary of Health and Human Services. If the logical fallacies were drops of water, the room would have flooded. A winning logical fallacy bingo card would have included confirmation bias, burden of proof, personal incredulity, ad hominem attacks, appeal to authority/conspiracy, cherry-picking, false dichotomy, strawman arguments, and whataboutisms. The performance would have been more entertaining, except for the fact that these ideas are shaping healthcare policy and the safety of our population.
I asked the Perplexity AI agent to watch the YouTube video and highlight any logical fallacies with timestamps. Thanks to AI, you can quickly get to the correct video locations. (I had the agent flag examples of the Senators making logically fallacious comments as well.)
https://www.perplexity.ai/search/review-this-video-and-document-LGzCNJJNSI.8T.VVQtHi1A
The various epidemiologists and public health experts I follow are far more eloquent on this topic. Former CDC director Tom Frieden's post, "Public Health is Under Assault," is a good example: https://substack.com/home/post/p-172881894
Various news outlets are reporting that HHS will soon release a paper on the association between acetaminophen use during pregnancy and autism. There is speculation that a large metaanalysis from earlier this year will be similar to what is reported by HHS. As you see the discussion in the news, please keep in mind the following:
Autism is a broad spectrum of symptoms with variable onset, variable observed symptoms, and lots of possible confounding variables; operational definitions matter - who, what, when, and how there was a diagnosis are essential.
Acetaminophen exposure during pregnancy is hard to measure without frequent biological samples from a large group of patients. Self-reported exposure is not good enough.
Association is not causation. Read the AI summary of the metaanalysis to arm yourself with intellectual weapons of understanding.
Ultimately, I am saying it is too easy to say, "acetaminophen causes autism," based on the data we have.
Article "bout the HHS report:
https://www.medpagetoday.com/neurology/autism/117336
Metaanalysis:
https://ehjournal.biomedcentral.com/articles/10.1186/s12940-025-01208-0
AI Summary of the metaanalysis
https://claude.ai/public/artifacts/6d2924e6-336c-40fd-a8d9-b00a4d153731
AI Summary of medical literature:
https://www.openevidence.com/ask/511415cd-dafd-45ce-89a3-cdb86d1ac7d3
Living with AI.
I found an interesting paper and discussion of how bias may be inherent to LLMs - and how fine-tuning with various rewards and feedback can help overcome the bias.
X Discussion:
https://x.com/rohanpaul_ai/status/1964184106910007708?s=61
Paper:
arxiv.org/abs/2504.16078v1
AI summary with some real-world examples
https://claude.ai/public/artifacts/d35e135c-ac09-4839-8388-fc05b0e992d0
Also, related to the inherent bias is an X post by Ethan Mollick, who frequently helps me understand AI in new ways:
"This paper finds LLMs' ability to understand that others have different beliefs (Theory of Mind or ToM) comes from 0.001% of their parameters. Break those specific weights & the model loses both its ability to track what others know AND language comprehension. Interesting implications." In other words, 'bad actors' could precisely degrade ToM capabilities in deployed systems targeting the 0.001% of critical parameters, making systems appear functional while impairing social reasoning, and the malicious modifications could go undetected by standard performance monitoring. Likewise, understanding and reinforcing ToM mechanisms could improve AI's ability to interpret patient concerns and emotional states during AI's clinical interactions
https://x.com/emollick/status/1963673899700211772?s=42
The paper he references:
https://www.nature.com/articles/s44387-025-00031-9
AI summary with lots of discussion of the implications
https://claude.ai/public/artifacts/6e290d6b-2c25-421c-b790-60cc8dc09e26
OpenAI is making a push into the training and job growth space. They recently published "Staying ahead in the age of AI," a five-step framework for organizational AI adoption: Align, Activate, Amplify, Accelerate, and Govern. The guide emphasizes rapid implementation while maintaining responsible practices, drawing from experiences with companies like Estée Lauder, Notion, and BBVA.
https://cdn.openai.com/pdf/ae250928-4029-4f26-9e23-afac1fcee14c/staying-ahead-in-the-age-of-ai.pdf
Related:
"OpenAI is building an AI jobs platform that could challenge Microsoft's LinkedIn"
https://www.cnbc.com/2025/09/05/openai-is-buMicrosoft'si-jobs-platform-that-could-rival-microsofts-linkedin.html
Infographics
I did not have time to find an infographic this week. Apologies.
Things I learned this week
It is about time someone made sleep apnea more fun. I found an Alien-themed face-hugger CPAP mask add-on. Christmas is right around the corner. And, which CPAP user wouldn't love to have more mask covering their face?
Given the state of the news, it feels like a good time to remind ourselves of Erich Fromm. Fromm was a philosopher who wrote extensively about love, human relationships, and the reasons why freedom and capitalism can foster feelings of isolation and alienation.
Erich Fromm on Human Nature, the Common Laziness of Optimism and Pessimism, and Why We Need Rational Faith in the Human Spirit
https://www.themarginalian.org/2016/04/04/erich-fromm-anatomy-of-human-destructiveness/
More in-depth:
It's time to revisit Erich Fromm: Why his ideas are more relevant than ever
Understanding today's social dynamics through his analysis
https://vocal.media/psyche/it-s-time-to-revisit-erich-fromm-why-his-ideas-are-more-relevant-than-ever
Erich Fromm for kids:
https://kids.kiddle.co/Erich_Fromm
There is still more shrimp news this week: Sen. John Kennedy Claims People Who Eat Foreign Shrimp Will 'Turn Into Aliens' (presumably without the CPAP facemask above, and without the super-powers like punching or nuclear fusion snaps - see last week's newsletter (https://www.whatadamisreading.com/2025/09/what-adam-is-reading-week-of-9-1-2025.html )
https://ground.news/article/sen-john-kennedy-claims-people-who-eat-foreign-shrimp-will-turn-into-aliens_6419a8
AI art of the week
A visual mashup of topics from the newsletter.
I use ChatGPT to summarize the newsletter, suggest prompts, and make the images.
Early 20th-century Art Deco advertising poster, bold WPA-style lithograph. A lively carnival midway scene: Senators as carnival barkers tossing balls labeled "Whataboutism," "Ad Hominem," and "Cherry Picking" at a giant Logical Fallacy Bingo board. Fam" lies "eating shrimp cocktails, but their shadows transform into playful aliens with antennae. A prize booth displays CPAP masks styled as cartoon alien facehuggers, wrapped with shiny holiday bows. Bright art-deco rays, bold reds, greens, and yellows with metallic gold outlines. Slogan across the top in dramatic curved lettering: "Don't Be Fooled — Eat Shrimp, Sleep Deep, Think Sharp!"
ChatGPT
https://drive.google.com/file/d/1trvNnS51k4Ujh-1xI7rqm6ekcgj8Q62C/view
Grok
https://drive.google.com/file/d/1-9SHAsVlMuhvF5T3udWxaKomimcx_dRm/view
Gemini
https://drive.google.com/file/d/1-9SHAsVlMuhvF5T3udWxaKomimcx_dRm/view
---
The Pandemic Mitigation Collaborative (PMC) utilizes wastewater viral RNA levels to forecast four-week predictions of COVID-19 rates.
Not only are national trends available, but the authors are now publishing state-level data visualizations.
Either way, COVID rates are still rising.
https://pmc19.com/data/
based upon https://biobot.io/data/
Wastewater Scan offers a multi-organism wastewater dashboard with an excellent visual display of individual treatment plant-level data.
https://data.wastewaterscan.org/
----
Clean hands and sharp minds,
Adam
Relive all the thrills and excitement - The What Adam is Reading Archive
http://www.whatadamisreading.com/
CONFIDENTIALITY NOTICE: THIS MESSAGE IS CONFIDENTIAL, INTENDED FOR THE NAMED RECIPIENT(S) AND MAY CONTAIN INFORMATION THAT IS (I) PROPRIETARY TO THE SENDER, AND/OR, (II) PRIVILEGED, CONFIDENTIAL, AND/OR OTHERWISE EXEMPT FROM DISCLOSURE UNDER APPLICABLE STATE AND FEDERAL LAW, INCLUDING, BUT NOT LIMITED TO, PRIVACY STANDARDS IMPOSED PURSUANT TO THE FEDERAL HEALTH INSURANCE PORTABILITY AND ACCOUNTABILITY ACT OF 1996 ("HIPAA"). IF YOU ARE NOT THE INTENDED RECIPIENT, OR THE EMPLOYEE OR AGENT RESPONSIBLE FOR DELIVERING THE MESSAGE TO THE INTENDED RECIPIENT, YOU ARE HEREBY NOTIFIED THAT ANY DISSEMINATION, DISTRIBUTION OR COPYING OF THIS COMMUNICATION IS STRICTLY PROHIBITED. IF YOU HAVE RECEIVED THIS TRANSMISSION IN ERROR, PLEASE (I) NOTIFY US IMMEDIATELY BY REPLY E-MAIL OR BY TELEPHONE AT (855.472.9822), (II) REMOVE IT FROM YOUR SYSTEM, AND (III) DESTROY THE ORIGINAL TRANSMISSION AND ITS ATTACHMENTS WITHOUT READING OR SAVING THEM. THANK YOU.
-DaVita Inc-
Comments
Post a Comment