The recent FDA Panel on SSRIs and Pregnancy attracted a scathing Media Response. See Unsafe Safety Systems for 25 responses including the American Psychiatric Association and Britain’s College of Psychiatrists.
For the most part the responses were remarkably similar, displayed complete amnesia for previously accepted facts, and downplayed any risks with phrases like earlier smaller studies hinted at a slight increase in risks but bigger more recent and better conducted studies have cast doubt on this so that nothing is firmly established whereas the fact that suicide is the leading cause of maternal death is established. There is no hint this might be caused by the SSRI the woman is taking.
As Adam Urato one of the FDA panelists later put it:
This @wbur segment What is the risk of taking antidepressants during pregnancy? running for 6 minutes doesn’t clearly note ANY risks and even walk back Paxil concerns.
This is an example of what I mean when I say that the public is not being accurately informed about the risks of SSRIs in pregnancy.
Those risks are miscarriage, birth defects, preterm birth, low birthweight, preeclampsia, postpartum hemorrhage, & poor neonatal adaptation. The SSRIs also alter fetal brain development, with evidence showing long-term effects on the children including speech/language difficulties, depression, & other neurobehavioral issues.
Artificial Intelligence
The responses were so similar that I began to wonder if they’d all accessed Chat GPT, Co-Pilot or Grok and essentially let them write the piece. One report explicitly says it did this.
Besides the similarities in points made, there was a scripted, A.I. flavor, to the writing – something that Meghan O’Rourke in the New York Times picked out as a feature of A.I. in a recent New York Times article. (Thanks to Grace Jackson for sending this gift link).
It was hard to avoid my A.I. is driving all this impression for another reason. I ended up in a conversation with 4 very sensible, professionals in their early 30s, who are fans of A.I – as almost everyone I know of all ages is and Meghan was for a while.
They’d had reason to ‘do their own research’ on the wisdom of getting maternal vaccines in pregnancy, including RSV vaccines either maternally or in the form of nirsevimab – Beyfortus – the monoclonal antibody (drug not vaccine, which is often ‘sold’ to us as a vaccine) which is given just after birth.
My sensible professionals oral responses uncannily echoed the SSRIs in pregnancy written media coverage, in downplaying any risks with phrases like earlier smaller studies hinted at a slight increase in risks but bigger more recent and better conducted studies along with approval by regulators and widespread uptake have cast doubt on this so that nothing is firmly established whereas the fact that RSV is a dangerous virus and leads to hospitalizations with potentially fatal outcomes is firmly established.
American Chat GPTs don’t make anyone aware that the RSV Vaccine and Beyfortus Drug trials have shown increased rates of pre-eclampsia in the mother along with preterm births and significant jaundice and neonatal deaths in the infants – in order to manage a chesty condition that up till this almost never ever causes deaths in infants from developed countries unless they already have serious additional health problems.
No-one is likely to get any hint that suppressing RSV might lead to MPV infections – which likely killed Pope Francis, who had lots of doctors and widgets at his beck and call. See Who Knew Antibodies could have Antibodies.
Nor are any hints ever likely to come from A.I. that the big players in the field – in this case pharma – have a host of outfits working to them like Sense about Science, who trot out experts to face the media and rubbish anything that attempts to outline possible risks linked to treatments, or Chemrisk who infiltrate pharma partners, in this case APA or ACOG or the Society of Maternal and Fetal medicine, and get statements out that make it look like these organizations (stakeholders in pharma’s distribution channel) all support the fact that antidepressants or vaccines save lives.
I was left feeling that trying to stop young professionals today taking SSRIs or RSV vaccines in pregnancy would be like Canute trying to stop an approaching Tsunami. A.I. is going to lock the door on any possibility of change and make anyone who might demur sound like an anti-vaxxer.
The response of these impressive people could have been scripted by A.I. as Shawn Johnson and Andrew East’s responses about Beyfortus seem to have been.
Shawn:
When they can’t breathe, it’s the scariest thing a parent will ever go through
We definitely did a lot of research as parents having a baby about Beyfortus – whether we should get it or not. Beyfortus helps prevent a serious lung disease caused by RSV for babies under 1
I talked about the common side effects with my babies doctor, my Ob-Gyn, my pediatrician, every on-call pediatrician, they all highly encouraged it
Andrew
Even with food or the sleep-schedule, are you going to do sleep training or not, everyone has a lot of opinions but for us, the people we look up to, and the research we did, it [Beyfortus] was such an easy decision.
Chat GPT etc list the risks of Beyfortus as sore arms, with a hint that other things could happen and if they do consult a doctor. A doctor who is not going to have a clue as to what might be happening you – see Religion and Science.
A Time of Miracles
Back in the 1990s and 2000s we got the internet which caused a buzz quite like today’s A.I. buzz. It was slower to develop as this really was the start of a new era that had to lead through social media before we got to the point we are at now.
People began to do their own research and formed groups like Paxil Progress or followed Charles Medawar’s Social Audit. It looked for all the world as though the internet was going to democratize medicine. It didn’t.
Just as it was new for patients, so too, it was new for companies, who took a little bit more time to find their footing on this new ground. But find their footing they did. When we searched the web for information on the drug we were on, we landed on company sites first, which sounded very patient friendly, even encouraging us to report any side effects to FDA.
These sites peddled the mantra of RCTs being Gold Standard Science. Not a hint that company studies are designed to hide the problems – a Gold Standard Way to Hide Adverse Events. No hint that neither you nor your doctor, nor the publishing journal, nor regulators have access to the data from the studies or can establish the patients even existed. A possible hint at the bottom in small print that the study was ghostwritten – disappearing fast now as A.I. moves in and is putting ghostwriters out of business.
Even I briefly figured that A.I. might be a good thing – it might let people coming to see me or other doctors do their own research and ask more penetrating questions and get decent information. As Adam Urato just 3 weeks ago said – ‘Information is Key’.
Rather than telling people to “do your own research” it’s beginning to look like it might be much better to tell them not to do their research. ‘Information is not Key’.
Religion or Science
It’s terribly difficult to know what to make of the supposed miraculous transformation of water into wine at the Wedding Feast at Cana – some 2000 years ago. This rabbit out of a hat trick stands almost completely at odds with everything Xt said.
Two thousand years later, in 1979, a important medical case unfolded. Raphael Osheroff, a successful nephrologist became depressed and was persuaded by his doctor and others to get admitted to Chestnut Lodge Hospital in Maryland. Over several months, he deteriorated badly, lost several stone in weight, became unkempt and disheveled with feet that were bleeding from agitated pacing. His family rescued him and brought him to Silver Hill Hospital.
Chestnut Lodge were true believers in Psychoanalysis for whom antidepressants might be a sticking plaster but were not a cure. The Lodge figured a true cure might take years.
After a few weeks in Silver Hill, where a much more pragmatic psychoanalytically trained doctor started Osheroff on amitriptyline, he was clearly recovering.
Osheroff took a legal action against Chestnut Lodge. This set up a celebrated but misleading debate. The debate lost sight of Osheroff and centered on what, if any, was the true science behind psychiatry. The leading lights of a biological psychiatry, then emerging in the US, were pitched against leading psychoanalysts. On the biomedical side Gerald Klerman framed the debate in terms of Evidence Based Medicine and the right of patients to have treatments proven by science to work.
Antidepressants had been proven to work using gold-standard scientific methods. On the other side, psychoanalysis had not been proven to work and didn’t look like it could ever be proven to work using scientific methods.
Klerman’s argument was the first iteration of the later Pfizer pandemic slogans – Follow the Science, along with Science will Win. This is as much a religious position as the Freudian position was, making biological psychiatry along with psychoanalysis the disease that each claims to be able to treat.
(With thanks for Eric Caplan for all his work on the Osheroff Case).
Following received truths is Religion not Science.
Science starts when the experiment to test a received truth starts. Science starts after you start a drug or a therapy. It aims at achieving a consensus between a doctor or doctors and you and your family or friends on the observables that appear after the experiment starts. Achieving consensus might in the process require a consensus testing of any interim views we have. When a beta-blocker causes a heart rate to increase rather than slow, stopping it might be a better test than doubling the dose.
This should have been easy to do in Raphael Osheroff’s case – a doctor talking to doctors – but the Chestnut Lodge ideologues systematically stripped away Ray’s identity as a doctor. Pharma – ably assisted by Evidence Based Medicine believers – do essentially the same thing for business rather than ideological reasons – by relegating any views we or our doctors have to the level of anecdotes or misinformation.
Science is about achieving a consensus between us on an event in front of us. It is always context dependent. It is not about applying or forcing acceptance of context free truths. ‘Information is not Key’.
The heart of clinical science lies closer to something else Adam Urato said:
“I’ve been taking care of pregnant women in my hometown for the past 20 years. I take care of my patients as if they were my neighbors, because they are my neighbors… A big part of care is giving patients the proper information about risks and benefits of treatment and then supporting their choices”.
It’s right that its her choice but not right if this is seen in terms of an autonomy that makes her liable for whatever goes wrong. Its more about shared consequences and continuity of care.
A.I. and the internet are fundamentally unscientific. They strip away context – they strip away our voice and our doctor’s voice also. Our doctors are increasingly becoming Follow The Science Badge wearing Puppets.
A.I. is a tool that we can access, just as nuclear energy, guns, and medicines are tools. The Magic, or the Miracles lie in us and our use of these tools – not in the tools.
To adapt a phrase from Mahatma Gandhi:
How can those who think they possess absolute truth be fraternal scientific?
The great risk is the Miracle of A.I. will turn the very best Wine into Water.
Commenting on my own post is not a great look but just in case the message is not obvious – this is the Shipwreck of the Singular message
The post is about care but a care that is completely science based rather than just warm feelings. It ain’t easy for a doctor to really see and hear the person in front of us especially when something we have done has caused a problem. This is science – remaining alert when a new experiment is happening rather than being in autopilot mode – it’s also caring.
We need to reclaim the meaning of doing science, being scientific and link Caring to that. Science is a passionate thing not dispassionate.
A second point is that some of us have been finding it harder and harder to get articles covering adverse events – the bread and butter of RxISK – published. Articles increasingly get rejected without review
Is it possible or is this just me being paranoid but it makes perfect business sense to have A.I. screen all articles when they come in – to reduce the workload for busy editors or you could say support busy editors –
and a minor tweak or ambiguous wording in the filter could lead all such articles being rejected. For instance putting a premium on Gold Standard RCTs rather than what would be viewed essentially as Case Reports. Such a filter would also make perfect business sense
D ,
Historically, Editors may not be up for the Challenge but the way AI involves everyone’s space, they may be perceived as dinosaurs if they don’t embrace what everyone else is using to save time and energy and to make it a fairer way to publish. That is, if they want a fairer way to publish.
Does AI interfere with Science?
AI doesn’t so much interfere with science as it reshapes it—sometimes radically, sometimes subtly. It’s a tool, a collaborator, and occasionally a disruptor. Let’s break it down:
How AI Enhances Science
Data Analysis at Scale: AI can sift through mountains of data—genomic sequences, astronomical observations, climate models—at speeds no human could match.
Pattern Recognition: It spots correlations and anomalies that might elude traditional methods, leading to new hypotheses or discoveries.
Simulation & Modeling: AI helps simulate complex systems like protein folding, neural networks, or planetary dynamics with astonishing accuracy.
Automation of Routine Tasks: From lab robotics to literature reviews, AI frees up scientists to focus on creative and strategic thinking.
Where AI Might Interfere or Raise Concerns
Bias in Algorithms: If trained on flawed or incomplete data, AI can reinforce existing biases—especially in fields like medicine or social sciences.
Opacity of Decision-Making: Some AI models (especially deep learning ones) are “black boxes,” making it hard to understand why they reach certain conclusions.
Overreliance: There’s a risk of scientists leaning too heavily on AI-generated results without sufficient scrutiny or validation.
Ethical Dilemmas: In areas like gene editing or surveillance, AI’s capabilities can outpace the ethical frameworks meant to govern them.
Real-World Impact
In drug discovery, AI has accelerated the identification of promising compounds.
In climate science, it’s helping refine predictions and model interventions.
In physics, AI is used to analyze data from particle accelerators and even assist in the search for dark matter.
So, rather than interfering, AI is more like a powerful microscope—it doesn’t change the laws of science, but it changes how we see and interact with them. The key is ensuring that human judgment, ethics, and curiosity remain at the helm.
Would you say AI is helping or hindering your own view of progress?
A little dissection
Bias in Algorithms: If trained on flawed or incomplete data, AI can reinforce existing biases—especially in fields like medicine or social sciences.
Automation of Routine Tasks: From lab robotics to literature reviews, AI frees up scientists to focus on creative and strategic thinking.
The key is ensuring that human judgment, ethics, and curiosity remain at the helm.
GSK, in particular, has waded through many Scandals, in China, with Drugs, with Paroxetine Birth Defects, and so on.
And so human judgment, ethics and ‘curiosity’ let themselves down, as did ‘creative and strategic thinking’ but if they want to be nice and not blame the corporates for the relentless screw-ups then they can do that (and inserting a bit of humour) ‘they remain at the helm’
‘AI can reinforce existing biases’
That is quite the admission relevant to the ‘miracle’
As clinical science has been framed in this post, A.I. at present is far more likely to harm our chances of getting good clinical care or justice or even a fair trial after an event than not harm it.
This is true for a few reasons. There are now good RxISK related examples of articles saying helpful things about an SSRI preventing certain visual problems sailing into print but articles covering adverse events being serially rejected without review. This speaks to a filter and will compromise the ability of a doctor to help us if we present with something not in the respectable medical literature.
An A.I. filter linked to the BMJ, NEJM, JAMA, Nature, Science risks killing and injuring us. As an editor I’ve made aware of the various filters being put in place to make my job less onerous.
Second A.I. facilitates marketing or what might be better called propaganda. Peter Selley just emailed over a video worth watching. https://www.youtube.com/watch?v=BnpKnTlWZaI
This would have any woman racing out to get vaccinated up to the hilt in the morning. It comes endorsed by the Society of Maternal and Fetal Medicine, supported by a grant from Pfizer. SMFM are a bunch of innocents who don’t even begin to understand how much they have been penetrated by and played by companies. Their oh so professionally video is supported by an educational grant from Pfizer – which likely leaves them feeling we are in control of what’s going on. Our ethics are shaping the conversation – are at the helm – not Pfizer’s. The naivety is breathtaking.
D
Staying slightly from the AI theme, from this side of The Pond it is incredible to see the degree of Pharma Infiltration into august bodies like the Society for Maternal-Fetal Medicine.
https://www.youtube.com/watch?v=aZ6lo1ZWKFQ
“Many patients have no side effects at all” from Pfizer’s Abrysvo RSV vaccine when pregnant.
I read this and thought – AI is HOMOGENISING our thoughts. Feeling marginally pleased with this insight, I looked it up. Others are there well before me – have already coined the term and are researching what AI does to our thinking. All pretty obvious stuff – but disturbing when you take it out of the lab and into our lives.
‘SAT prompts are designed to be broad enough to elicit a multiplicity of responses, but the use of A.I. had a HOMOGENIZING effect. “The output was very, very similar for all of these different people, coming in on different days, talking about high-level personal, societal topics, and it was skewed in some specific directions,” Kosmyna said. For the question about what makes us “truly happy,” the L.L.M. users were much more likely than the other groups to use phrases related to career and personal success. In response to a question about philanthropy (“Should people who are more fortunate than others have more of a moral obligation to help those who are less fortunate?”), the ChatGPT group uniformly argued in favor, whereas essays from the other groups included critiques of philanthropy. With the L.L.M. “you have no divergent opinions being generated,” Kosmyna said. She continued, “Average everything everywhere all at once—that’s kind of what we’re looking at here.”
A.I. is a technology of averages: large language models are trained to spot patterns across vast tracts of data; the answers they produce tend toward consensus, both in the quality of the writing, which is often riddled with clichés and banalities, and in the calibre of the ideas.’
https://www.newyorker.com/culture/infinite-scroll/ai-is-homogenizing-our-thoughts
It ties in directly with something you were saying the other day about AI enabling a suffocating weight of evidence.
‘This will add to the weight of evidence. And the Groks of this world will dig out their prior work and keep citing it until it becomes almost impossible to view anything else as being sensible’.
Ofc the problem is industry owns the volume of information, weight of ‘evidence’, that smothers risks – exactly the battle Adam is fighting now on X.
Here is a Chat GPT, Co-Pilot, Grok answer to – Will A.I. create a Tsunami?
Maybe. But it is just another platform/media: Google search, Twitter, Facebook, TikTok, Youtube, etc.
But pharmaceutical companies used to employ ghost writers as paper mills then pay prestigious journals to publish, pay media and journalists who learned creative writing to produce their propaganda. These types of services may be disappearing, replaced by AI.
As Google search could be both an enemy and supporter for patients, AI may be the same. AI is also a tool and many tricks(bias) are hidden. The users have to be clever.