Ob.Gyn. News

Kangaroo mother care (KMC), with close skin-to-skin contact between mothers and their low-birthweight newborns, appears to reduce mortality risk by almost one-third, compared with conventional care, according to new research published online in BMJ Global Health.

Starting the contact, which involves mothers carrying the newborn in a sling, within 24 hours of birth and continuing it for at least 8 hours a day both appear to amplify the effect on reducing mortality and infection, the paper states.

Sindhu Sivanandan, MD, with the department of neonatology at Jawaharlal Institute of Postgraduate Medical Education and Research, Puducherry, India, and Mari Jeeva Sankar, MD, in the pediatrics department of the All India Institute of Medical Sciences, New Delhi, looked at existing studies to compare KMC with conventional care and to compare starting the intervention within 24 hours of birth versus a later start.

Mortality risk reduction

Analysis showed that, compared with conventional care, KMC appeared to cut mortality risk by 32% (relative risk, 0.68; 95% confidence interval, 0.53-0.86) during birth hospitalization or by 28 days after birth, while it seemed to reduce the risk of severe infection, such as sepsis, by 15% (RR, 0.85; 95% CI, 0.76-0.96; low-certainty evidence.)

That mortality-risk reduction was found regardless of gestational age or weight of the child at enrolment, time of starting KMC, and whether the intervention was started in a hospital or community.

The studies that had compared early with late-initiated KMC showed a reduction in neonatal mortality of 33%.

Low- and middle-income countries have the highest rates of premature births (gestational age of less than 37 weeks) and low birthweight (less than 2,500 grams). Premature births and low birthweight both are key causes of death and disability.

The World Health Organization recommends KMC as the standard of care among low birthweight infants after clinical stabilization. The American Academy of Pediatrics also promotes immediate KMC.
 

Relevance in the U.S.

Grace Chan, MD, MPH, PhD, an epidemiologist and pediatrician with the Harvard School of Public Health, Boston, said though the practice is promoted by the WHO and AAP, recommendations to families vary widely by providers.

She said the health benefits for KMC are numerous. One of the biggest is that skin-to-skin contact can help transfer heat to newborns who may have trouble regulating their own temperature. That is especially important in cold climates in places where there may be insufficient indoor heat.

She said it’s well-known that preterm babies are at higher risk for apnea, and listening to a mother’s heartbeat may stimulate the child to breathe regularly.

Additionally with KMC, there’s an inherent benefit of a mother or caregiver being able to see any change in a newborn’s color immediately when the baby is held so closely, as opposed to a nurse watching several babies at a time in a neonatal intensive care unit.

This is evidence that starting KMC right away is important, because the risk of death for premature and low-weight newborns is highest in the first 24 hours of life, Dr. Chan noted.
 

Barriers of time

There are some barriers, she noted, in that mothers or other caregivers caring for several young children may not have the time to carry a child in a sling for 8 or more hours at a time.

The authors conclude that their findings have policy implications, particularly for low- and middle-income countries: “KMC should be provided to all low birth weight and preterm infants irrespective of the settings – both health facilities and at home,” they wrote.

The authors caution that, “very low birth weight, extremely preterm neonates, and severely unstable neonates were often excluded from studies. More evidence is needed before extrapolating the study results in these high-risk groups.”

The study authors and Dr. Chan report no relevant financial relationships.

Kangaroo mother care (KMC), with close skin-to-skin contact between mothers and their low-birthweight newborns, appears to reduce mortality risk by almost one-third, compared with conventional care, according to new research published online in BMJ Global Health.

Starting the contact, which involves mothers carrying the newborn in a sling, within 24 hours of birth and continuing it for at least 8 hours a day both appear to amplify the effect on reducing mortality and infection, the paper states.

Sindhu Sivanandan, MD, with the department of neonatology at Jawaharlal Institute of Postgraduate Medical Education and Research, Puducherry, India, and Mari Jeeva Sankar, MD, in the pediatrics department of the All India Institute of Medical Sciences, New Delhi, looked at existing studies to compare KMC with conventional care and to compare starting the intervention within 24 hours of birth versus a later start.

Mortality risk reduction

Analysis showed that, compared with conventional care, KMC appeared to cut mortality risk by 32% (relative risk, 0.68; 95% confidence interval, 0.53-0.86) during birth hospitalization or by 28 days after birth, while it seemed to reduce the risk of severe infection, such as sepsis, by 15% (RR, 0.85; 95% CI, 0.76-0.96; low-certainty evidence.)

That mortality-risk reduction was found regardless of gestational age or weight of the child at enrolment, time of starting KMC, and whether the intervention was started in a hospital or community.

The studies that had compared early with late-initiated KMC showed a reduction in neonatal mortality of 33%.

Low- and middle-income countries have the highest rates of premature births (gestational age of less than 37 weeks) and low birthweight (less than 2,500 grams). Premature births and low birthweight both are key causes of death and disability.

The World Health Organization recommends KMC as the standard of care among low birthweight infants after clinical stabilization. The American Academy of Pediatrics also promotes immediate KMC.
 

Relevance in the U.S.

Grace Chan, MD, MPH, PhD, an epidemiologist and pediatrician with the Harvard School of Public Health, Boston, said though the practice is promoted by the WHO and AAP, recommendations to families vary widely by providers.

She said the health benefits for KMC are numerous. One of the biggest is that skin-to-skin contact can help transfer heat to newborns who may have trouble regulating their own temperature. That is especially important in cold climates in places where there may be insufficient indoor heat.

She said it’s well-known that preterm babies are at higher risk for apnea, and listening to a mother’s heartbeat may stimulate the child to breathe regularly.

Additionally with KMC, there’s an inherent benefit of a mother or caregiver being able to see any change in a newborn’s color immediately when the baby is held so closely, as opposed to a nurse watching several babies at a time in a neonatal intensive care unit.

This is evidence that starting KMC right away is important, because the risk of death for premature and low-weight newborns is highest in the first 24 hours of life, Dr. Chan noted.
 

Barriers of time

There are some barriers, she noted, in that mothers or other caregivers caring for several young children may not have the time to carry a child in a sling for 8 or more hours at a time.

The authors conclude that their findings have policy implications, particularly for low- and middle-income countries: “KMC should be provided to all low birth weight and preterm infants irrespective of the settings – both health facilities and at home,” they wrote.

The authors caution that, “very low birth weight, extremely preterm neonates, and severely unstable neonates were often excluded from studies. More evidence is needed before extrapolating the study results in these high-risk groups.”

The study authors and Dr. Chan report no relevant financial relationships.

Kangaroo mother care (KMC), with close skin-to-skin contact between mothers and their low-birthweight newborns, appears to reduce mortality risk by almost one-third, compared with conventional care, according to new research published online in BMJ Global Health.

Starting the contact, which involves mothers carrying the newborn in a sling, within 24 hours of birth and continuing it for at least 8 hours a day both appear to amplify the effect on reducing mortality and infection, the paper states.

Sindhu Sivanandan, MD, with the department of neonatology at Jawaharlal Institute of Postgraduate Medical Education and Research, Puducherry, India, and Mari Jeeva Sankar, MD, in the pediatrics department of the All India Institute of Medical Sciences, New Delhi, looked at existing studies to compare KMC with conventional care and to compare starting the intervention within 24 hours of birth versus a later start.

Mortality risk reduction

Analysis showed that, compared with conventional care, KMC appeared to cut mortality risk by 32% (relative risk, 0.68; 95% confidence interval, 0.53-0.86) during birth hospitalization or by 28 days after birth, while it seemed to reduce the risk of severe infection, such as sepsis, by 15% (RR, 0.85; 95% CI, 0.76-0.96; low-certainty evidence.)

That mortality-risk reduction was found regardless of gestational age or weight of the child at enrolment, time of starting KMC, and whether the intervention was started in a hospital or community.

The studies that had compared early with late-initiated KMC showed a reduction in neonatal mortality of 33%.

Low- and middle-income countries have the highest rates of premature births (gestational age of less than 37 weeks) and low birthweight (less than 2,500 grams). Premature births and low birthweight both are key causes of death and disability.

The World Health Organization recommends KMC as the standard of care among low birthweight infants after clinical stabilization. The American Academy of Pediatrics also promotes immediate KMC.
 

Relevance in the U.S.

Grace Chan, MD, MPH, PhD, an epidemiologist and pediatrician with the Harvard School of Public Health, Boston, said though the practice is promoted by the WHO and AAP, recommendations to families vary widely by providers.

She said the health benefits for KMC are numerous. One of the biggest is that skin-to-skin contact can help transfer heat to newborns who may have trouble regulating their own temperature. That is especially important in cold climates in places where there may be insufficient indoor heat.

She said it’s well-known that preterm babies are at higher risk for apnea, and listening to a mother’s heartbeat may stimulate the child to breathe regularly.

Additionally with KMC, there’s an inherent benefit of a mother or caregiver being able to see any change in a newborn’s color immediately when the baby is held so closely, as opposed to a nurse watching several babies at a time in a neonatal intensive care unit.

This is evidence that starting KMC right away is important, because the risk of death for premature and low-weight newborns is highest in the first 24 hours of life, Dr. Chan noted.
 

Barriers of time

There are some barriers, she noted, in that mothers or other caregivers caring for several young children may not have the time to carry a child in a sling for 8 or more hours at a time.

The authors conclude that their findings have policy implications, particularly for low- and middle-income countries: “KMC should be provided to all low birth weight and preterm infants irrespective of the settings – both health facilities and at home,” they wrote.

The authors caution that, “very low birth weight, extremely preterm neonates, and severely unstable neonates were often excluded from studies. More evidence is needed before extrapolating the study results in these high-risk groups.”

The study authors and Dr. Chan report no relevant financial relationships.

Researchers may use artificial intelligence (AI) language models such as ChatGPT to write and revise scientific manuscripts, according to a new announcement from the International Committee of Medical Journal Editors. These tools should not be listed as authors, and researchers must denote how AI-assisted technologies were used, the committee said.

These new guidelines are the latest effort for medical journals to define policies for using these large-scale language models (LLMs) in scientific publication. While these AI-assisted tools can help with tasks such as writing, analyzing data, catching mistakes, and much more, they are also prone to errors, noted Casey Greene, PhD, a professor of biomedical informatics at the University of Colorado at Denver, Aurora. It is also not totally clear how information is stored and processed in these kinds of tools, and who has access to that information, he noted.

At the same time, experts argue that these AI tools could have a positive impact on the field by limiting some of the linguistic disparities in scientific publishing as well as alleviating the burden of some monotonous or mechanical tasks that come along with manuscript writing.

What experts can agree on, though, is that the use of AI tools is here to stay. “This is going to become a common tool,” Dr. Greene said. “I don’t think there’s a way out of that at this point.”
 

A change in medical publishing

OpenAI released ChatGPT in November 2022. In its own words, ChatGPT is “a deep learning model that has been trained on a massive amount of text data to understand and generate humanlike text.” Enter a question or a prompt, and it will respond. For example, when asked how the AI tool can be used in scientific publishing, ChatGPT responded:

“ChatGPT can aid scientific publishing by generating ideas, clarifying concepts, conducting preliminary literature reviews, providing proofreading and editing suggestions, and assisting with outlining and organizing scientific papers. However, it should be used as a supportive tool alongside domain expertise and validation from trusted scientific sources.”

Just a few months after ChatGPT became available, researchers began using this tool in their own work. One individual, Som Biswas, MD, a radiologist at the University of Tennessee Health Science Center in Memphis, reportedly used ChatGPT to author 16 scientific articles in just 4 months, according to the Daily Beast. Five of these articles have been published in four different journals. Dr. Biswas declined to be interviewed for this article.

There were also reports of papers with ChatGPT as one of the listed authors, which sparked backlash. In response, JAMA, Nature, and Science all published editorials in January outlining their policies for using ChatGPT and other large language models in the scientific authoring process. Editors from the journals of the American College of Cardiology and the American College of Rheumatology also updated their policies to reflect the influence of AI authoring tools.

The consensus is that AI has no place on the author byline.

“We think that’s not appropriate, because coauthorship means that you are taking responsibility for the analysis and the generation of data that are included in a manuscript. A machine that is dictated by AI can’t take responsibility,” said Daniel Solomon, MD, MPH, a rheumatologist at Brigham and Women’s Hospital, Boston, and the editor in chief of the ACR journal Arthritis & Rheumatology.
 

Issues with AI

One of the big concerns around using AI in writing is that it can generate text that seems plausible but is untrue or not supported by data. For example, Dr. Greene and colleague Milton Pividori, PhD, also of the University of Colorado, were writing a journal article about new software they developed that uses a large language model to revise scientific manuscripts.

“We used the same software to revise that article and at one point, it added a line that noted that the large language model had been fine-tuned on a data set of manuscripts from within the same field. This makes a lot of sense, and is absolutely something you could do, but was not something that we did,” Dr. Greene said. “Without a really careful review of the content, it becomes possible to invent things that were not actually done.”

In another case, ChatGPT falsely stated that a prominent law professor had been accused of sexual assault, citing a Washington Post article that did not exist.

“We live in a society where we are extremely concerned about fake news,” Dr. Pividori added, “and [these kinds of errors] could certainly exacerbate that in the scientific community, which is very concerning because science informs public policy.”

Another issue is the lack of transparency around how large language models like ChatGPT process and store data used to make queries.

“We have no idea how they are recording all the prompts and things that we input into ChatGPT and their systems,” Dr. Pividori said.

OpenAI recently addressed some privacy concerns by allowing users to turn off their chat history with the AI chatbot, so conversations cannot be used to train or improve the company’s models. But Dr. Greene noted that the terms of service “still remain pretty nebulous.”

Dr. Solomon is also concerned with researchers using these AI tools in authoring without knowing how they work. “The thing we are really concerned about is that fact that [LLMs] are a bit of a black box – people don’t really understand the methodologies,” he said.
 

A positive tool?

But despite these concerns, many think that these types of AI-assisted tools could have a positive impact on medical publishing, particularly for researchers for whom English is not their first language, noted Catherine Gao, MD, a pulmonary and critical care instructor at Northwestern University, Chicago. She recently led research comparing scientific abstracts written by ChatGPT and real abstracts and discovered that reviewers found it “surprisingly difficult” to differentiate the two.

“The majority of research is published in English,” she said in an email. “Responsible use of LLMs can potentially reduce the burden of writing for busy scientists and improve equity for those who are not native English speakers.”

Dr. Pividori agreed, adding that as a non-native English speaker, he spends much more time working on the structure and grammar of sentences when authoring a manuscript, compared with people who speak English as a first language. He noted that these tools can also be used to automate some of the more monotonous tasks that come along with writing manuscripts and allow researchers to focus on the more creative aspects.

In the future, “I want to focus more on the things that only a human can do and let these tools do all the rest of it,” he said.
 

New rules

But despite how individual researchers feel about LLMs, they agree that these AI tools are here to stay.

“I think that we should anticipate that they will become part of the medical research establishment over time, when we figure out how to use them appropriately,” Dr. Solomon said.

While the debate of how to best use AI in medical publications will continue, journal editors agree that all authors of a manuscript are solely responsible for content in articles that used AI-assisted technology.

“Authors should carefully review and edit the result because AI can generate authoritative-sounding output that can be incorrect, incomplete, or biased,” the ICMJE guidelines state. “Authors should be able to assert that there is no plagiarism in their paper, including in text and images produced by the AI.” This includes appropriate attribution of all cited materials.

The committee also recommends that authors write in both the cover letter and submitted work how AI was used in the manuscript writing process. Recently updated guidelines from the World Association of Medical Editors recommend that all prompts used to generate new text or analytical work should be provided in submitted work. Dr. Greene also noted that if authors used an AI tool to revise their work, they can include a version of the manuscript untouched by LLMs.

It is similar to a preprint, he said, but rather than publishing a version of a paper prior to peer review, someone is showing a version of a manuscript before it was reviewed and revised by AI. “This type of practice could be a path that lets us benefit from these models,” he said, “without having the drawbacks that many are concerned about.”

Dr. Solomon has financial relationships with AbbVie, Amgen, Janssen, CorEvitas, and Moderna. Both Dr. Greene and Dr. Pividori are inventors in the U.S. Provisional Patent Application No. 63/486,706 that the University of Colorado has filed for the “Publishing Infrastructure For AI-Assisted Academic Authoring” invention with the U.S. Patent and Trademark Office. Dr. Greene and Dr. Pividori also received a grant from the Alfred P. Sloan Foundation to improve their AI-based manuscript revision tool. Dr. Gao reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

Researchers may use artificial intelligence (AI) language models such as ChatGPT to write and revise scientific manuscripts, according to a new announcement from the International Committee of Medical Journal Editors. These tools should not be listed as authors, and researchers must denote how AI-assisted technologies were used, the committee said.

These new guidelines are the latest effort for medical journals to define policies for using these large-scale language models (LLMs) in scientific publication. While these AI-assisted tools can help with tasks such as writing, analyzing data, catching mistakes, and much more, they are also prone to errors, noted Casey Greene, PhD, a professor of biomedical informatics at the University of Colorado at Denver, Aurora. It is also not totally clear how information is stored and processed in these kinds of tools, and who has access to that information, he noted.

At the same time, experts argue that these AI tools could have a positive impact on the field by limiting some of the linguistic disparities in scientific publishing as well as alleviating the burden of some monotonous or mechanical tasks that come along with manuscript writing.

What experts can agree on, though, is that the use of AI tools is here to stay. “This is going to become a common tool,” Dr. Greene said. “I don’t think there’s a way out of that at this point.”
 

A change in medical publishing

OpenAI released ChatGPT in November 2022. In its own words, ChatGPT is “a deep learning model that has been trained on a massive amount of text data to understand and generate humanlike text.” Enter a question or a prompt, and it will respond. For example, when asked how the AI tool can be used in scientific publishing, ChatGPT responded:

“ChatGPT can aid scientific publishing by generating ideas, clarifying concepts, conducting preliminary literature reviews, providing proofreading and editing suggestions, and assisting with outlining and organizing scientific papers. However, it should be used as a supportive tool alongside domain expertise and validation from trusted scientific sources.”

Just a few months after ChatGPT became available, researchers began using this tool in their own work. One individual, Som Biswas, MD, a radiologist at the University of Tennessee Health Science Center in Memphis, reportedly used ChatGPT to author 16 scientific articles in just 4 months, according to the Daily Beast. Five of these articles have been published in four different journals. Dr. Biswas declined to be interviewed for this article.

There were also reports of papers with ChatGPT as one of the listed authors, which sparked backlash. In response, JAMA, Nature, and Science all published editorials in January outlining their policies for using ChatGPT and other large language models in the scientific authoring process. Editors from the journals of the American College of Cardiology and the American College of Rheumatology also updated their policies to reflect the influence of AI authoring tools.

The consensus is that AI has no place on the author byline.

“We think that’s not appropriate, because coauthorship means that you are taking responsibility for the analysis and the generation of data that are included in a manuscript. A machine that is dictated by AI can’t take responsibility,” said Daniel Solomon, MD, MPH, a rheumatologist at Brigham and Women’s Hospital, Boston, and the editor in chief of the ACR journal Arthritis & Rheumatology.
 

Issues with AI

One of the big concerns around using AI in writing is that it can generate text that seems plausible but is untrue or not supported by data. For example, Dr. Greene and colleague Milton Pividori, PhD, also of the University of Colorado, were writing a journal article about new software they developed that uses a large language model to revise scientific manuscripts.

“We used the same software to revise that article and at one point, it added a line that noted that the large language model had been fine-tuned on a data set of manuscripts from within the same field. This makes a lot of sense, and is absolutely something you could do, but was not something that we did,” Dr. Greene said. “Without a really careful review of the content, it becomes possible to invent things that were not actually done.”

In another case, ChatGPT falsely stated that a prominent law professor had been accused of sexual assault, citing a Washington Post article that did not exist.

“We live in a society where we are extremely concerned about fake news,” Dr. Pividori added, “and [these kinds of errors] could certainly exacerbate that in the scientific community, which is very concerning because science informs public policy.”

Another issue is the lack of transparency around how large language models like ChatGPT process and store data used to make queries.

“We have no idea how they are recording all the prompts and things that we input into ChatGPT and their systems,” Dr. Pividori said.

OpenAI recently addressed some privacy concerns by allowing users to turn off their chat history with the AI chatbot, so conversations cannot be used to train or improve the company’s models. But Dr. Greene noted that the terms of service “still remain pretty nebulous.”

Dr. Solomon is also concerned with researchers using these AI tools in authoring without knowing how they work. “The thing we are really concerned about is that fact that [LLMs] are a bit of a black box – people don’t really understand the methodologies,” he said.
 

A positive tool?

But despite these concerns, many think that these types of AI-assisted tools could have a positive impact on medical publishing, particularly for researchers for whom English is not their first language, noted Catherine Gao, MD, a pulmonary and critical care instructor at Northwestern University, Chicago. She recently led research comparing scientific abstracts written by ChatGPT and real abstracts and discovered that reviewers found it “surprisingly difficult” to differentiate the two.

“The majority of research is published in English,” she said in an email. “Responsible use of LLMs can potentially reduce the burden of writing for busy scientists and improve equity for those who are not native English speakers.”

Dr. Pividori agreed, adding that as a non-native English speaker, he spends much more time working on the structure and grammar of sentences when authoring a manuscript, compared with people who speak English as a first language. He noted that these tools can also be used to automate some of the more monotonous tasks that come along with writing manuscripts and allow researchers to focus on the more creative aspects.

In the future, “I want to focus more on the things that only a human can do and let these tools do all the rest of it,” he said.
 

New rules

But despite how individual researchers feel about LLMs, they agree that these AI tools are here to stay.

“I think that we should anticipate that they will become part of the medical research establishment over time, when we figure out how to use them appropriately,” Dr. Solomon said.

While the debate of how to best use AI in medical publications will continue, journal editors agree that all authors of a manuscript are solely responsible for content in articles that used AI-assisted technology.

“Authors should carefully review and edit the result because AI can generate authoritative-sounding output that can be incorrect, incomplete, or biased,” the ICMJE guidelines state. “Authors should be able to assert that there is no plagiarism in their paper, including in text and images produced by the AI.” This includes appropriate attribution of all cited materials.

The committee also recommends that authors write in both the cover letter and submitted work how AI was used in the manuscript writing process. Recently updated guidelines from the World Association of Medical Editors recommend that all prompts used to generate new text or analytical work should be provided in submitted work. Dr. Greene also noted that if authors used an AI tool to revise their work, they can include a version of the manuscript untouched by LLMs.

It is similar to a preprint, he said, but rather than publishing a version of a paper prior to peer review, someone is showing a version of a manuscript before it was reviewed and revised by AI. “This type of practice could be a path that lets us benefit from these models,” he said, “without having the drawbacks that many are concerned about.”

Dr. Solomon has financial relationships with AbbVie, Amgen, Janssen, CorEvitas, and Moderna. Both Dr. Greene and Dr. Pividori are inventors in the U.S. Provisional Patent Application No. 63/486,706 that the University of Colorado has filed for the “Publishing Infrastructure For AI-Assisted Academic Authoring” invention with the U.S. Patent and Trademark Office. Dr. Greene and Dr. Pividori also received a grant from the Alfred P. Sloan Foundation to improve their AI-based manuscript revision tool. Dr. Gao reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

Researchers may use artificial intelligence (AI) language models such as ChatGPT to write and revise scientific manuscripts, according to a new announcement from the International Committee of Medical Journal Editors. These tools should not be listed as authors, and researchers must denote how AI-assisted technologies were used, the committee said.

These new guidelines are the latest effort for medical journals to define policies for using these large-scale language models (LLMs) in scientific publication. While these AI-assisted tools can help with tasks such as writing, analyzing data, catching mistakes, and much more, they are also prone to errors, noted Casey Greene, PhD, a professor of biomedical informatics at the University of Colorado at Denver, Aurora. It is also not totally clear how information is stored and processed in these kinds of tools, and who has access to that information, he noted.

At the same time, experts argue that these AI tools could have a positive impact on the field by limiting some of the linguistic disparities in scientific publishing as well as alleviating the burden of some monotonous or mechanical tasks that come along with manuscript writing.

What experts can agree on, though, is that the use of AI tools is here to stay. “This is going to become a common tool,” Dr. Greene said. “I don’t think there’s a way out of that at this point.”
 

A change in medical publishing

OpenAI released ChatGPT in November 2022. In its own words, ChatGPT is “a deep learning model that has been trained on a massive amount of text data to understand and generate humanlike text.” Enter a question or a prompt, and it will respond. For example, when asked how the AI tool can be used in scientific publishing, ChatGPT responded:

“ChatGPT can aid scientific publishing by generating ideas, clarifying concepts, conducting preliminary literature reviews, providing proofreading and editing suggestions, and assisting with outlining and organizing scientific papers. However, it should be used as a supportive tool alongside domain expertise and validation from trusted scientific sources.”

Just a few months after ChatGPT became available, researchers began using this tool in their own work. One individual, Som Biswas, MD, a radiologist at the University of Tennessee Health Science Center in Memphis, reportedly used ChatGPT to author 16 scientific articles in just 4 months, according to the Daily Beast. Five of these articles have been published in four different journals. Dr. Biswas declined to be interviewed for this article.

There were also reports of papers with ChatGPT as one of the listed authors, which sparked backlash. In response, JAMA, Nature, and Science all published editorials in January outlining their policies for using ChatGPT and other large language models in the scientific authoring process. Editors from the journals of the American College of Cardiology and the American College of Rheumatology also updated their policies to reflect the influence of AI authoring tools.

The consensus is that AI has no place on the author byline.

“We think that’s not appropriate, because coauthorship means that you are taking responsibility for the analysis and the generation of data that are included in a manuscript. A machine that is dictated by AI can’t take responsibility,” said Daniel Solomon, MD, MPH, a rheumatologist at Brigham and Women’s Hospital, Boston, and the editor in chief of the ACR journal Arthritis & Rheumatology.
 

Issues with AI

One of the big concerns around using AI in writing is that it can generate text that seems plausible but is untrue or not supported by data. For example, Dr. Greene and colleague Milton Pividori, PhD, also of the University of Colorado, were writing a journal article about new software they developed that uses a large language model to revise scientific manuscripts.

“We used the same software to revise that article and at one point, it added a line that noted that the large language model had been fine-tuned on a data set of manuscripts from within the same field. This makes a lot of sense, and is absolutely something you could do, but was not something that we did,” Dr. Greene said. “Without a really careful review of the content, it becomes possible to invent things that were not actually done.”

In another case, ChatGPT falsely stated that a prominent law professor had been accused of sexual assault, citing a Washington Post article that did not exist.

“We live in a society where we are extremely concerned about fake news,” Dr. Pividori added, “and [these kinds of errors] could certainly exacerbate that in the scientific community, which is very concerning because science informs public policy.”

Another issue is the lack of transparency around how large language models like ChatGPT process and store data used to make queries.

“We have no idea how they are recording all the prompts and things that we input into ChatGPT and their systems,” Dr. Pividori said.

OpenAI recently addressed some privacy concerns by allowing users to turn off their chat history with the AI chatbot, so conversations cannot be used to train or improve the company’s models. But Dr. Greene noted that the terms of service “still remain pretty nebulous.”

Dr. Solomon is also concerned with researchers using these AI tools in authoring without knowing how they work. “The thing we are really concerned about is that fact that [LLMs] are a bit of a black box – people don’t really understand the methodologies,” he said.
 

A positive tool?

But despite these concerns, many think that these types of AI-assisted tools could have a positive impact on medical publishing, particularly for researchers for whom English is not their first language, noted Catherine Gao, MD, a pulmonary and critical care instructor at Northwestern University, Chicago. She recently led research comparing scientific abstracts written by ChatGPT and real abstracts and discovered that reviewers found it “surprisingly difficult” to differentiate the two.

“The majority of research is published in English,” she said in an email. “Responsible use of LLMs can potentially reduce the burden of writing for busy scientists and improve equity for those who are not native English speakers.”

Dr. Pividori agreed, adding that as a non-native English speaker, he spends much more time working on the structure and grammar of sentences when authoring a manuscript, compared with people who speak English as a first language. He noted that these tools can also be used to automate some of the more monotonous tasks that come along with writing manuscripts and allow researchers to focus on the more creative aspects.

In the future, “I want to focus more on the things that only a human can do and let these tools do all the rest of it,” he said.
 

New rules

But despite how individual researchers feel about LLMs, they agree that these AI tools are here to stay.

“I think that we should anticipate that they will become part of the medical research establishment over time, when we figure out how to use them appropriately,” Dr. Solomon said.

While the debate of how to best use AI in medical publications will continue, journal editors agree that all authors of a manuscript are solely responsible for content in articles that used AI-assisted technology.

“Authors should carefully review and edit the result because AI can generate authoritative-sounding output that can be incorrect, incomplete, or biased,” the ICMJE guidelines state. “Authors should be able to assert that there is no plagiarism in their paper, including in text and images produced by the AI.” This includes appropriate attribution of all cited materials.

The committee also recommends that authors write in both the cover letter and submitted work how AI was used in the manuscript writing process. Recently updated guidelines from the World Association of Medical Editors recommend that all prompts used to generate new text or analytical work should be provided in submitted work. Dr. Greene also noted that if authors used an AI tool to revise their work, they can include a version of the manuscript untouched by LLMs.

It is similar to a preprint, he said, but rather than publishing a version of a paper prior to peer review, someone is showing a version of a manuscript before it was reviewed and revised by AI. “This type of practice could be a path that lets us benefit from these models,” he said, “without having the drawbacks that many are concerned about.”

Dr. Solomon has financial relationships with AbbVie, Amgen, Janssen, CorEvitas, and Moderna. Both Dr. Greene and Dr. Pividori are inventors in the U.S. Provisional Patent Application No. 63/486,706 that the University of Colorado has filed for the “Publishing Infrastructure For AI-Assisted Academic Authoring” invention with the U.S. Patent and Trademark Office. Dr. Greene and Dr. Pividori also received a grant from the Alfred P. Sloan Foundation to improve their AI-based manuscript revision tool. Dr. Gao reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

A new guide offers recommendations for the nonsurgical management of major hemorrhage, which is a challenging clinical problem.

Major hemorrhage is a significant cause of death and can occur in a myriad of clinical settings.

“In Ontario, we’ve been collecting quality metrics on major hemorrhages to try and make sure that a higher percentage of patients gets the best possible care when they are experiencing significant bleeding,” author Jeannie Callum, MD, professor and director of transfusion medicine at Kingston (Ont.) Health Sciences Centre and Queen’s University, also in Kingston, said in an interview. “There were some gaps, so this is our effort to get open, clear information out to the emergency doctors, intensive care unit doctors, the surgeons, and everyone else involved in managing major hemorrhage, to help close these gaps.”

The guide was published in the Canadian Medical Association Journal.
 

Fast care essential

The guide aims to provide answers, based on the latest research, to questions such as when to activate a massive hemorrhage protocol (MHP), which patients should receive tranexamic acid (TXA), which blood products should be transfused before laboratory results are available, how to monitor the effects of blood transfusion, and when fibrinogen concentrate or prothrombin complex concentrate should be given.

Not all recommendations will be followed, Dr. Callum said, especially in rural hospitals with limited resources. But the guide is adaptable, and rural hospitals can create protocols that are customized to their unique circumstances.

Care must be “perfect and fast” in the first hour of major injury, said Dr. Callum. “You need to get a proclotting drug in that first hour if you have a traumatic or postpartum bleed. You have to make sure your clotting factors never fail you throughout your resuscitation. You have to be fast with the transfusion. You have to monitor for the complications of the transfusion, electrolyte disturbances, and the patient’s temperature dropping. It’s a complicated situation that needs a multidisciplinary team.”

Bleeding affects everybody in medicine, from family doctors in smaller institutions who work in emergency departments to obstetricians and surgeons, she added.

“For people under the age of 45, trauma is the most common cause of death. When people die of trauma, they die of bleeding. So many people experience these extreme bleeds. We believe that some of them might be preventable with faster, more standardized, more aggressive care. That’s why we wrote this review,” said Dr. Callum.
 

Administer TXA quickly  

The first recommendation is to ensure that every hospital has a massive hemorrhage protocol. Such a protocol is vital for the emergency department, operating room, and obstetric unit. “Making sure you’ve got a protocol that is updated every 3 years and adjusted to the local hospital context is essential,” said Dr. Callum.

Smaller hospitals will have to adjust their protocols according to the capabilities of their sites. “Some smaller hospitals do not have platelets in stock and get their platelets from another hospital, so you need to adjust your protocol to what you are able to do. Not every hospital can control bleeding in a trauma patient, so your protocol would be to stabilize and call a helicopter. Make sure all of this is detailed so that implementing it becomes automatic,” said Dr. Callum.

An MHP should be activated for patients with uncontrolled hemorrhage who meet the clinical criteria of the local hospital and are expected to need blood product support and red blood cells.

“Lots of people bleed, but not everybody is bleeding enough that they need a code transfusion,” said Dr. Callum. Most patients with gastrointestinal bleeds caused by NSAID use can be managed with uncrossed matched blood from the local blood bank. “But in patients who need the full code transfusion because they are going to need plasma, clotting factor replacement, and many other drugs, that is when the MHP should be activated. Don’t activate it when you don’t need it, because doing so activates the whole hospital and diverts care away from other patients.”

TXA should be administered as soon as possible after onset of hemorrhage in most patients, with the exception of gastrointestinal hemorrhage, where a benefit has not been shown.

TXA has been a major advance in treating massive bleeding, Dr. Callum said. “TXA was invented by a Japanese husband-and-wife research team. We know that it reduces the death rate in trauma and in postpartum hemorrhage, and it reduces the chance of major bleeding with major surgical procedures. We give it routinely in surgical procedures. If a patient gets TXA within 60 minutes of injury, it dramatically reduces the death rate. And it costs $10 per patient. It’s cheap, it’s easy, it has no side effects. It’s just amazing.”

Future research must address several unanswered questions, said Dr. Callum. These questions include whether prehospital transfusion improves patient outcomes, whether whole blood has a role in the early management of major hemorrhage, and what role factor concentrates play in patients with major bleeding.
 

‘Optimal recommendations’

Commenting on the document, Bourke Tillmann, MD, PhD, trauma team leader at Sunnybrook Health Sciences Centre and the Ross Tilley Burn Center in Toronto, said: “Overall, I think it is a good overview of MHPs as an approach to major hemorrhage.”

The review also is timely, since Ontario released its MHP guidelines in 2021, he added. “I would have liked to see more about the treatment aspects than just an overview of an MHP. But if you are the person overseeing the emergency department or running the blood bank, these protocols are incredibly useful and incredibly important.”

“This report is a nice and thoughtful overview of best practices in many areas, especially trauma, and makes recommendations that are optimal, although they are not necessarily practical in all centers,” Eric L. Legome, MD, professor and chair of emergency medicine at Mount Sinai West and Mount Sinai Morningside, New York, said in an interview.

“If you’re in a small rural hospital with one lab technician, trying to do all of these things, it will not be possible. These are optimal recommendations that people can use to the best of their ability, but they are not standard of care, because some places will not be able to provide this level of care,” he added. “This paper provides practical, reasonable advice that should be looked at as you are trying to implement transfusion policies and processes, with the understanding that it is not necessarily applicable or practical for very small hospitals in very rural centers that might not have access to these types of products and tools, but it’s a reasonable and nicely written paper.”

No outside funding for the guideline was reported. Dr. Callum has received research funding from Canadian Blood Services and Octapharma. She sits on the nominating committee with the Association for the Advancement of Blood & Biotherapies and on the data safety monitoring boards for the Tranexamic Acid for Subdural Hematoma trial and the Fibrinogen Replacement in Trauma trial. Dr. Tillmann and Dr. Legome reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

A new guide offers recommendations for the nonsurgical management of major hemorrhage, which is a challenging clinical problem.

Major hemorrhage is a significant cause of death and can occur in a myriad of clinical settings.

“In Ontario, we’ve been collecting quality metrics on major hemorrhages to try and make sure that a higher percentage of patients gets the best possible care when they are experiencing significant bleeding,” author Jeannie Callum, MD, professor and director of transfusion medicine at Kingston (Ont.) Health Sciences Centre and Queen’s University, also in Kingston, said in an interview. “There were some gaps, so this is our effort to get open, clear information out to the emergency doctors, intensive care unit doctors, the surgeons, and everyone else involved in managing major hemorrhage, to help close these gaps.”

The guide was published in the Canadian Medical Association Journal.
 

Fast care essential

The guide aims to provide answers, based on the latest research, to questions such as when to activate a massive hemorrhage protocol (MHP), which patients should receive tranexamic acid (TXA), which blood products should be transfused before laboratory results are available, how to monitor the effects of blood transfusion, and when fibrinogen concentrate or prothrombin complex concentrate should be given.

Not all recommendations will be followed, Dr. Callum said, especially in rural hospitals with limited resources. But the guide is adaptable, and rural hospitals can create protocols that are customized to their unique circumstances.

Care must be “perfect and fast” in the first hour of major injury, said Dr. Callum. “You need to get a proclotting drug in that first hour if you have a traumatic or postpartum bleed. You have to make sure your clotting factors never fail you throughout your resuscitation. You have to be fast with the transfusion. You have to monitor for the complications of the transfusion, electrolyte disturbances, and the patient’s temperature dropping. It’s a complicated situation that needs a multidisciplinary team.”

Bleeding affects everybody in medicine, from family doctors in smaller institutions who work in emergency departments to obstetricians and surgeons, she added.

“For people under the age of 45, trauma is the most common cause of death. When people die of trauma, they die of bleeding. So many people experience these extreme bleeds. We believe that some of them might be preventable with faster, more standardized, more aggressive care. That’s why we wrote this review,” said Dr. Callum.
 

Administer TXA quickly  

The first recommendation is to ensure that every hospital has a massive hemorrhage protocol. Such a protocol is vital for the emergency department, operating room, and obstetric unit. “Making sure you’ve got a protocol that is updated every 3 years and adjusted to the local hospital context is essential,” said Dr. Callum.

Smaller hospitals will have to adjust their protocols according to the capabilities of their sites. “Some smaller hospitals do not have platelets in stock and get their platelets from another hospital, so you need to adjust your protocol to what you are able to do. Not every hospital can control bleeding in a trauma patient, so your protocol would be to stabilize and call a helicopter. Make sure all of this is detailed so that implementing it becomes automatic,” said Dr. Callum.

An MHP should be activated for patients with uncontrolled hemorrhage who meet the clinical criteria of the local hospital and are expected to need blood product support and red blood cells.

“Lots of people bleed, but not everybody is bleeding enough that they need a code transfusion,” said Dr. Callum. Most patients with gastrointestinal bleeds caused by NSAID use can be managed with uncrossed matched blood from the local blood bank. “But in patients who need the full code transfusion because they are going to need plasma, clotting factor replacement, and many other drugs, that is when the MHP should be activated. Don’t activate it when you don’t need it, because doing so activates the whole hospital and diverts care away from other patients.”

TXA should be administered as soon as possible after onset of hemorrhage in most patients, with the exception of gastrointestinal hemorrhage, where a benefit has not been shown.

TXA has been a major advance in treating massive bleeding, Dr. Callum said. “TXA was invented by a Japanese husband-and-wife research team. We know that it reduces the death rate in trauma and in postpartum hemorrhage, and it reduces the chance of major bleeding with major surgical procedures. We give it routinely in surgical procedures. If a patient gets TXA within 60 minutes of injury, it dramatically reduces the death rate. And it costs $10 per patient. It’s cheap, it’s easy, it has no side effects. It’s just amazing.”

Future research must address several unanswered questions, said Dr. Callum. These questions include whether prehospital transfusion improves patient outcomes, whether whole blood has a role in the early management of major hemorrhage, and what role factor concentrates play in patients with major bleeding.
 

‘Optimal recommendations’

Commenting on the document, Bourke Tillmann, MD, PhD, trauma team leader at Sunnybrook Health Sciences Centre and the Ross Tilley Burn Center in Toronto, said: “Overall, I think it is a good overview of MHPs as an approach to major hemorrhage.”

The review also is timely, since Ontario released its MHP guidelines in 2021, he added. “I would have liked to see more about the treatment aspects than just an overview of an MHP. But if you are the person overseeing the emergency department or running the blood bank, these protocols are incredibly useful and incredibly important.”

“This report is a nice and thoughtful overview of best practices in many areas, especially trauma, and makes recommendations that are optimal, although they are not necessarily practical in all centers,” Eric L. Legome, MD, professor and chair of emergency medicine at Mount Sinai West and Mount Sinai Morningside, New York, said in an interview.

“If you’re in a small rural hospital with one lab technician, trying to do all of these things, it will not be possible. These are optimal recommendations that people can use to the best of their ability, but they are not standard of care, because some places will not be able to provide this level of care,” he added. “This paper provides practical, reasonable advice that should be looked at as you are trying to implement transfusion policies and processes, with the understanding that it is not necessarily applicable or practical for very small hospitals in very rural centers that might not have access to these types of products and tools, but it’s a reasonable and nicely written paper.”

No outside funding for the guideline was reported. Dr. Callum has received research funding from Canadian Blood Services and Octapharma. She sits on the nominating committee with the Association for the Advancement of Blood & Biotherapies and on the data safety monitoring boards for the Tranexamic Acid for Subdural Hematoma trial and the Fibrinogen Replacement in Trauma trial. Dr. Tillmann and Dr. Legome reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

A new guide offers recommendations for the nonsurgical management of major hemorrhage, which is a challenging clinical problem.

Major hemorrhage is a significant cause of death and can occur in a myriad of clinical settings.

“In Ontario, we’ve been collecting quality metrics on major hemorrhages to try and make sure that a higher percentage of patients gets the best possible care when they are experiencing significant bleeding,” author Jeannie Callum, MD, professor and director of transfusion medicine at Kingston (Ont.) Health Sciences Centre and Queen’s University, also in Kingston, said in an interview. “There were some gaps, so this is our effort to get open, clear information out to the emergency doctors, intensive care unit doctors, the surgeons, and everyone else involved in managing major hemorrhage, to help close these gaps.”

The guide was published in the Canadian Medical Association Journal.
 

Fast care essential

The guide aims to provide answers, based on the latest research, to questions such as when to activate a massive hemorrhage protocol (MHP), which patients should receive tranexamic acid (TXA), which blood products should be transfused before laboratory results are available, how to monitor the effects of blood transfusion, and when fibrinogen concentrate or prothrombin complex concentrate should be given.

Not all recommendations will be followed, Dr. Callum said, especially in rural hospitals with limited resources. But the guide is adaptable, and rural hospitals can create protocols that are customized to their unique circumstances.

Care must be “perfect and fast” in the first hour of major injury, said Dr. Callum. “You need to get a proclotting drug in that first hour if you have a traumatic or postpartum bleed. You have to make sure your clotting factors never fail you throughout your resuscitation. You have to be fast with the transfusion. You have to monitor for the complications of the transfusion, electrolyte disturbances, and the patient’s temperature dropping. It’s a complicated situation that needs a multidisciplinary team.”

Bleeding affects everybody in medicine, from family doctors in smaller institutions who work in emergency departments to obstetricians and surgeons, she added.

“For people under the age of 45, trauma is the most common cause of death. When people die of trauma, they die of bleeding. So many people experience these extreme bleeds. We believe that some of them might be preventable with faster, more standardized, more aggressive care. That’s why we wrote this review,” said Dr. Callum.
 

Administer TXA quickly  

The first recommendation is to ensure that every hospital has a massive hemorrhage protocol. Such a protocol is vital for the emergency department, operating room, and obstetric unit. “Making sure you’ve got a protocol that is updated every 3 years and adjusted to the local hospital context is essential,” said Dr. Callum.

Smaller hospitals will have to adjust their protocols according to the capabilities of their sites. “Some smaller hospitals do not have platelets in stock and get their platelets from another hospital, so you need to adjust your protocol to what you are able to do. Not every hospital can control bleeding in a trauma patient, so your protocol would be to stabilize and call a helicopter. Make sure all of this is detailed so that implementing it becomes automatic,” said Dr. Callum.

An MHP should be activated for patients with uncontrolled hemorrhage who meet the clinical criteria of the local hospital and are expected to need blood product support and red blood cells.

“Lots of people bleed, but not everybody is bleeding enough that they need a code transfusion,” said Dr. Callum. Most patients with gastrointestinal bleeds caused by NSAID use can be managed with uncrossed matched blood from the local blood bank. “But in patients who need the full code transfusion because they are going to need plasma, clotting factor replacement, and many other drugs, that is when the MHP should be activated. Don’t activate it when you don’t need it, because doing so activates the whole hospital and diverts care away from other patients.”

TXA should be administered as soon as possible after onset of hemorrhage in most patients, with the exception of gastrointestinal hemorrhage, where a benefit has not been shown.

TXA has been a major advance in treating massive bleeding, Dr. Callum said. “TXA was invented by a Japanese husband-and-wife research team. We know that it reduces the death rate in trauma and in postpartum hemorrhage, and it reduces the chance of major bleeding with major surgical procedures. We give it routinely in surgical procedures. If a patient gets TXA within 60 minutes of injury, it dramatically reduces the death rate. And it costs $10 per patient. It’s cheap, it’s easy, it has no side effects. It’s just amazing.”

Future research must address several unanswered questions, said Dr. Callum. These questions include whether prehospital transfusion improves patient outcomes, whether whole blood has a role in the early management of major hemorrhage, and what role factor concentrates play in patients with major bleeding.
 

‘Optimal recommendations’

Commenting on the document, Bourke Tillmann, MD, PhD, trauma team leader at Sunnybrook Health Sciences Centre and the Ross Tilley Burn Center in Toronto, said: “Overall, I think it is a good overview of MHPs as an approach to major hemorrhage.”

The review also is timely, since Ontario released its MHP guidelines in 2021, he added. “I would have liked to see more about the treatment aspects than just an overview of an MHP. But if you are the person overseeing the emergency department or running the blood bank, these protocols are incredibly useful and incredibly important.”

“This report is a nice and thoughtful overview of best practices in many areas, especially trauma, and makes recommendations that are optimal, although they are not necessarily practical in all centers,” Eric L. Legome, MD, professor and chair of emergency medicine at Mount Sinai West and Mount Sinai Morningside, New York, said in an interview.

“If you’re in a small rural hospital with one lab technician, trying to do all of these things, it will not be possible. These are optimal recommendations that people can use to the best of their ability, but they are not standard of care, because some places will not be able to provide this level of care,” he added. “This paper provides practical, reasonable advice that should be looked at as you are trying to implement transfusion policies and processes, with the understanding that it is not necessarily applicable or practical for very small hospitals in very rural centers that might not have access to these types of products and tools, but it’s a reasonable and nicely written paper.”

No outside funding for the guideline was reported. Dr. Callum has received research funding from Canadian Blood Services and Octapharma. She sits on the nominating committee with the Association for the Advancement of Blood & Biotherapies and on the data safety monitoring boards for the Tranexamic Acid for Subdural Hematoma trial and the Fibrinogen Replacement in Trauma trial. Dr. Tillmann and Dr. Legome reported no relevant financial relationships.

A version of this article originally appeared on Medscape.com.

Vasomotor symptoms affect as many as 75% of menopausal women in the United States. Characterized by a sudden onset of flushing, sweating, and chills, symptoms of hot flashes can be managed with hormone therapy, but prolonged use of the treatment poses health risks. In a study published in JAMA Internal Medicine, researchers found that the use of nitroglycerin patches did not result in lasting improvements in the frequency and severity of hot flashes, but there was a short-term benefit.

METHODOLOGY

• The  was a randomized, double-blinded trial involving 134 California women aged 40-62 years.
• Between July 2018 and December 2021, participants self-administered either a nitroglycerin patch at a dosage of 0.2 to 0.6 mg/h or a placebo patch every night.
• Participants were in the late stages of menopause or had already undergone menopause. They reported having seven or more hot flashes per day; at least four were moderate to severe over a 1-week period.
• The primary outcome was a change in the frequency of hot flashes over 5 and 12 weeks.

TAKEAWAY

• Over 5 weeks, the frequency of moderate to severe hot flashes decreased by 3.3 episodes per day in the nitroglycerine group, compared with 2.2 episodes per day in the placebo group (95% CI, −2.2 to 0; P = .05).
• The reduction in overall frequency of hot flashes – either mild, moderate, or severe – over the 5-week period was not statistically significant.
• Over the 12-week period, no statistically significant reductions in hot flashes occurred.
• More than two thirds of participants assigned to the nitroglycerin patches reported having headaches, while three reported chest pain and one had a syncopal episode.

IN PRACTICE

The findings do not support daily use of nitroglycerin patches to treat vasomotor symptoms, the researchers conclude.

“The bottom line is that our study doesn’t allow us to recommend nitroglycerin skin patches as a strategy for consumers to suppress hot flashes in the long term,” Alison Huang, MD, MAS, lead author of the study, said in a press release. “The menopause field is still lacking in effective treatment approaches that don’t involve hormones.”
 

STUDY DETAILS

The study was led by Alison Huang, MD, MAS, a professor of medicine at the University of California, San Francisco. Two of the authors reported grants from the National Institute on Aging.

LIMITATIONS

Almost 20% of women who used the nitroglycerin patches discontinued treatment before the end of the trial because they could not tolerate the medication, experienced an adverse event, or their symptoms did not improve, according to the researchers. In addition, the 1-week period used to screen for severity and frequency of hot flashes may have been too short to confirm that symptoms were prolonged, which could explain the better-than-expected results in the placebo group.

DISCLOSURES

One author served on the medical advisory board of SomaLogic. Another author is an unpaid consultant to Astellas Pharma. Another author reported grants from the National Institutes of Health.

A version of this article first appeared on Medscape.com.

Vasomotor symptoms affect as many as 75% of menopausal women in the United States. Characterized by a sudden onset of flushing, sweating, and chills, symptoms of hot flashes can be managed with hormone therapy, but prolonged use of the treatment poses health risks. In a study published in JAMA Internal Medicine, researchers found that the use of nitroglycerin patches did not result in lasting improvements in the frequency and severity of hot flashes, but there was a short-term benefit.

METHODOLOGY

• The  was a randomized, double-blinded trial involving 134 California women aged 40-62 years.
• Between July 2018 and December 2021, participants self-administered either a nitroglycerin patch at a dosage of 0.2 to 0.6 mg/h or a placebo patch every night.
• Participants were in the late stages of menopause or had already undergone menopause. They reported having seven or more hot flashes per day; at least four were moderate to severe over a 1-week period.
• The primary outcome was a change in the frequency of hot flashes over 5 and 12 weeks.

TAKEAWAY

• Over 5 weeks, the frequency of moderate to severe hot flashes decreased by 3.3 episodes per day in the nitroglycerine group, compared with 2.2 episodes per day in the placebo group (95% CI, −2.2 to 0; P = .05).
• The reduction in overall frequency of hot flashes – either mild, moderate, or severe – over the 5-week period was not statistically significant.
• Over the 12-week period, no statistically significant reductions in hot flashes occurred.
• More than two thirds of participants assigned to the nitroglycerin patches reported having headaches, while three reported chest pain and one had a syncopal episode.

IN PRACTICE

The findings do not support daily use of nitroglycerin patches to treat vasomotor symptoms, the researchers conclude.

“The bottom line is that our study doesn’t allow us to recommend nitroglycerin skin patches as a strategy for consumers to suppress hot flashes in the long term,” Alison Huang, MD, MAS, lead author of the study, said in a press release. “The menopause field is still lacking in effective treatment approaches that don’t involve hormones.”
 

STUDY DETAILS

The study was led by Alison Huang, MD, MAS, a professor of medicine at the University of California, San Francisco. Two of the authors reported grants from the National Institute on Aging.

LIMITATIONS

Almost 20% of women who used the nitroglycerin patches discontinued treatment before the end of the trial because they could not tolerate the medication, experienced an adverse event, or their symptoms did not improve, according to the researchers. In addition, the 1-week period used to screen for severity and frequency of hot flashes may have been too short to confirm that symptoms were prolonged, which could explain the better-than-expected results in the placebo group.

DISCLOSURES

One author served on the medical advisory board of SomaLogic. Another author is an unpaid consultant to Astellas Pharma. Another author reported grants from the National Institutes of Health.

A version of this article first appeared on Medscape.com.

Vasomotor symptoms affect as many as 75% of menopausal women in the United States. Characterized by a sudden onset of flushing, sweating, and chills, symptoms of hot flashes can be managed with hormone therapy, but prolonged use of the treatment poses health risks. In a study published in JAMA Internal Medicine, researchers found that the use of nitroglycerin patches did not result in lasting improvements in the frequency and severity of hot flashes, but there was a short-term benefit.

METHODOLOGY

• The  was a randomized, double-blinded trial involving 134 California women aged 40-62 years.
• Between July 2018 and December 2021, participants self-administered either a nitroglycerin patch at a dosage of 0.2 to 0.6 mg/h or a placebo patch every night.
• Participants were in the late stages of menopause or had already undergone menopause. They reported having seven or more hot flashes per day; at least four were moderate to severe over a 1-week period.
• The primary outcome was a change in the frequency of hot flashes over 5 and 12 weeks.

TAKEAWAY

• Over 5 weeks, the frequency of moderate to severe hot flashes decreased by 3.3 episodes per day in the nitroglycerine group, compared with 2.2 episodes per day in the placebo group (95% CI, −2.2 to 0; P = .05).
• The reduction in overall frequency of hot flashes – either mild, moderate, or severe – over the 5-week period was not statistically significant.
• Over the 12-week period, no statistically significant reductions in hot flashes occurred.
• More than two thirds of participants assigned to the nitroglycerin patches reported having headaches, while three reported chest pain and one had a syncopal episode.

IN PRACTICE

The findings do not support daily use of nitroglycerin patches to treat vasomotor symptoms, the researchers conclude.

“The bottom line is that our study doesn’t allow us to recommend nitroglycerin skin patches as a strategy for consumers to suppress hot flashes in the long term,” Alison Huang, MD, MAS, lead author of the study, said in a press release. “The menopause field is still lacking in effective treatment approaches that don’t involve hormones.”
 

STUDY DETAILS

The study was led by Alison Huang, MD, MAS, a professor of medicine at the University of California, San Francisco. Two of the authors reported grants from the National Institute on Aging.

LIMITATIONS

Almost 20% of women who used the nitroglycerin patches discontinued treatment before the end of the trial because they could not tolerate the medication, experienced an adverse event, or their symptoms did not improve, according to the researchers. In addition, the 1-week period used to screen for severity and frequency of hot flashes may have been too short to confirm that symptoms were prolonged, which could explain the better-than-expected results in the placebo group.

DISCLOSURES

One author served on the medical advisory board of SomaLogic. Another author is an unpaid consultant to Astellas Pharma. Another author reported grants from the National Institutes of Health.

A version of this article first appeared on Medscape.com.

The ways in which artificial intelligence (AI) may transform the future of medicine is making headlines across the globe. But chances are, you’re already using AI in your practice every day – you may just not realize it.
 

And whether you recognize the presence of AI or not, the technology could be putting you in danger of a lawsuit, legal experts say.

The use of AI in your daily practice can come with hidden liabilities, say legal experts, and as hospitals and medical groups deploy AI into more areas of health care, new liability exposures may be on the horizon.

“For physicians, AI has also not yet drastically changed or improved the way care is provided or consumed,” said Michael LeTang, chief nursing informatics officer and vice president of risk management and patient safety at Healthcare Risk Advisors, part of TDC Group. “Consequently, it may seem like AI is not present in their work streams, but in reality, it has been utilized in health care for several years. As AI technologies continue to develop and become more sophisticated, we can expect them to play an increasingly significant role in health care.”

Today, most AI applications in health care use narrow AI, which is designed to complete a single task without human assistance, as opposed to artificial general intelligence (AGI), which pertains to human-level reasoning and problem solving across a broad spectrum. Here are some ways doctors are using AI throughout the day – sometimes being aware of its assistance, and sometimes being unaware:

• Many doctors use electronic health records (EHRs) with integrated AI that include computerized clinical decision support tools designed to reduce the risk of diagnostic error and to integrate decision-making in the medication ordering function.
• Cardiologists, pathologists, and dermatologists use AI in the interpretation of vast amounts of images, tracings, and complex patterns.
• Surgeons are using AI-enhanced surgical robotics for orthopedic surgeries, such as joint replacement and spine surgery.
• A growing number of doctors are using ChatGPT to assist in drafting prior authorization letters for insurers. Experts say more doctors are also experimenting with ChatGPT to support medical decision-making.
• Within oncology, physicians use machine learning techniques in the form of computer-aided detection systems for early breast cancer detection.
• AI algorithms are often used by health systems for workflow, staffing optimization, population management, and care coordination.
• Some systems within EHRs use AI to indicate high-risk patients.
• Physicians are using AI applications for the early recognition of sepsis, including EHR-integrated decision tools, such as the Hospital Corporation of America Healthcare’s Sepsis Prediction and Optimization Therapy and the Sepsis Early Risk Assessment algorithm.
• About 30% of radiologists use AI in their practice to analyze x-rays and CT scans.
• Epic Systems recently announced a partnership with Microsoft to integrate ChatGPT into MyChart, Epic’s patient portal system. Pilot hospitals will utilize ChatGPT to automatically generate responses to patient-generated questions sent via the portal.

The growth of AI in health care has been enormous, and it’s only going to continue, said Ravi B. Parikh, MD, an assistant professor in the department of medical ethics and health policy and medicine at the University of Pennsylvania, Philadelphia.

“What’s really critical is that physicians, clinicians, and nurses using AI are provided with the tools to understand how artificial intelligence works and, most importantly, understand that they are still accountable for making the ultimate decision,” Mr. LeTang said, “The information is not always going to be the right thing to do or the most accurate thing to do. They’re still liable for making a bad decision, even if AI is driving that.”
 

What are the top AI legal dangers of today?

A pressing legal risk is becoming too reliant on the suggestions that AI-based systems provide, which can lead to poor care decisions, said Kenneth Rashbaum, a New York–based cybersecurity attorney with more than 25 years of experience in medical malpractice defense.

This can occur, for example, when using clinical support systems that leverage AI, machine learning, or statistical pattern recognition. Today, clinical support systems are commonly administered through EHRs and other computerized clinical workflows. In general, such systems match a patient’s characteristics to a computerized clinical knowledge base. An assessment or recommendation is then presented to the physician for a decision.

“If the clinician blindly accepts it without considering whether it’s appropriate for this patient at this time with this presentation, the clinician may bear some responsibility if there is an untoward result,” Mr. Rashbaum said.

“A common claim even in the days before the EMR [electronic medical record] and AI, was that the clinician did not take all available information into account in rendering treatment, including history of past and present condition, as reflected in the records, communication with past and other present treating clinicians, lab and radiology results, discussions with the patient, and physical examination findings,” he said. “So, if the clinician relied upon the support prompt to the exclusion of these other sources of information, that could be a very strong argument for the plaintiff.”

Chatbots, such OpenAI’s ChatGPT, are another form of AI raising legal red flags. ChatGPT, trained on a massive set of text data, can carry out conversations, write code, draft emails, and answer any question posed. The chatbot has gained considerable credibility for accurately diagnosing rare conditions in seconds, and it recently passed the U.S. Medical Licensing Examination.

It’s unclear how many doctors are signing onto the ChatGPT website daily, but physicians are actively using the chatbot, particularly for assistance with prior authorization letters and to support decision-making processes in their practices, said Mr. LeTang.

When physicians ask ChatGPT a question, however, they should be mindful that ChatGPT could “hallucinate,” a term that refers to a generated response that sounds plausible but is factually incorrect or is unrelated to the context, explains Harvey Castro, MD, an emergency physician, ChatGPT health care expert, and author of the 2023 book “ChatGPT and Healthcare: Unlocking the Potential of Patient Empowerment.”

Acting on ChatGPT’s response without vetting the information places doctors at serious risk of a lawsuit, he said.

“Sometimes, the response is half true and half false,” he said. “Say, I go outside my specialty of emergency medicine and ask it about a pediatric surgical procedure. It could give me a response that sounds medically correct, but then I ask a pediatric cardiologist, and he says, ‘We don’t even do this. This doesn’t even exist!’ Physicians really have to make sure they are vetting the information provided.”

In response to ChatGPT’s growing usage by health care professionals, hospitals and practices are quickly implementing guidelines, policies, and restrictions that caution physicians about the accuracy of ChatGPT-generated information, adds Mr. LeTang.

Emerging best practices include avoiding the input of patient health information, personally identifiable information, or any data that could be commercially valuable or considered the intellectual property of a hospital or health system, he said.

“Another crucial guideline is not to rely solely on ChatGPT as a definitive source for clinical decision-making; physicians must exercise their professional judgment,” he said. “If best practices are not adhered to, the associated risks are present today. However, these risks may become more significant as AI technologies continue to evolve and become increasingly integrated into health care.”

The potential for misdiagnosis by AI systems and the risk of unnecessary procedures if physicians do not thoroughly evaluate and validate AI predictions are other dangers.

As an example, Mr. LeTang described a case in which a physician documents in the EHR that a patient has presented to the emergency department with chest pains and other signs of a heart attack, and an AI algorithm predicts that the patient is experiencing an active myocardial infarction. If the physician then sends the patient for stenting or an angioplasty without other concrete evidence or tests to confirm the diagnosis, the doctor could later face a misdiagnosis complaint if the costly procedures were unnecessary.

“That’s one of the risks of using artificial intelligence,” he said. “A large percentage of malpractice claims is failure to diagnose, delayed diagnosis, or inaccurate diagnosis. What falls in the category of failure to diagnose is sending a patient for an unnecessary procedure or having an adverse event or bad outcome because of the failure to diagnose.”

So far, no AI lawsuits have been filed, but they may make an appearance soon, said Sue Boisvert, senior patient safety risk manager at The Doctors Company, a national medical liability insurer.

“There are hundreds of AI programs currently in use in health care,” she said. “At some point, a provider will make a decision that is contrary to what the AI recommended. The AI may be wrong, or the provider may be wrong. Either way, the provider will neglect to document their clinical reasoning, a patient will be harmed, and we will have the first AI claim.”
 

Upcoming AI legal risks to watch for

Lawsuits that allege biased patient care by physicians on the basis of algorithmic bias may also be forthcoming, analysts warn.

Much has been written about algorithmic bias that compounds and worsens inequities in socioeconomic status, ethnicity, sexual orientation, and gender in health systems. In 2019, a groundbreaking article in Science shed light on commonly used algorithms that are considered racially biased and how health care professionals often use such information to make medical decisions.

No claims involving AI bias have come down the pipeline yet, but it’s an area to watch, said Ms. Boisvert. She noted a website that highlights complaints and accusations of AI bias, including in health care.

“We need to be sure the training of the AI is appropriate, current, and broad enough so that there is no bias in the AI when it’s participating in the decision-making,” said Ms. Boisvert. “Imagine if the AI is diagnosing based on a dataset that is not local. It doesn’t represent the population at that particular hospital, and it’s providing inaccurate information to the physicians who are then making decisions about treatment.”

In pain management, for example, there are known differences in how patients experience pain, Ms. Boisvert said. If AI was being used to develop an algorithm for how a particular patient’s postoperative pain should be managed, and the algorithm did not include the differences, the pain control for a certain patient could be inappropriate. A poor outcome resulting from the treatment could lead to a claim against the physician or hospital that used the biased AI system, she said.

In the future, as AI becomes more integrated and accepted in medicine, there may be a risk of legal complaints against doctors for not using AI, said Saurabh Jha, MD, an associate professor of radiology at the University of Pennsylvania, Philadelphia, and a scholar of AI in radiology.

“Ultimately, we might get to a place where AI starts helping physicians detect more or reduce the miss of certain conditions, and it becomes the standard of care,” Dr. Jha said. “For example, if it became part of the standard of care for pulmonary embolism [PE] detection, and you didn’t use it for PE detection, and there was a miss. That could put you at legal risk. We’re not at that stage yet, but that is one future possibility.”

Dr. Parikh envisions an even cloudier liability landscape as the potential grows for AI to control patient care decisions. In such a scenario, rather than just issuing an alert or prediction to a physician, the AI system could trigger an action.

For instance, if an algorithm is trained to predict sepsis and, once triggered, the AI could initiate a nurse-led rapid response or a change in patient care outside the clinician’s control, said Dr. Parikh, who coauthored a recent article on AI and medical liability in The Milbank Quarterly.

“That’s still very much the minority of how AI is being used, but as evidence is growing that AI-based diagnostic tools perform equivalent or even superior to physicians, these autonomous workflows are being considered,” Dr. Parikh said. “When the ultimate action upon the patient is more determined by the AI than what the clinician does, then I think the liability picture gets murkier, and we should be thinking about how we can respond to that from a liability framework.”
 

How you can prevent AI-related lawsuits

The first step to preventing an AI-related claim is being aware of when and how you are using AI.

Ensure you’re informed about how the AI was trained, Ms. Boisvert stresses.

“Ask questions!” she said. “Is the AI safe? Are the recommendations accurate? Does the AI perform better than current systems? In what way? What databases were used, and did the programmers consider bias? Do I understand how to use the results?”

Never blindly trust the AI but rather view it as a data point in a medical decision, said Dr. Parikh. Ensure that other sources of medical information are properly accessed and that best practices for your specialty are still being followed.

When using any form of AI, document your usage, adds Mr. Rashbaum. A record that clearly outlines how the physician incorporated the AI is critical if a claim later arises in which the doctor is accused of AI-related malpractice, he said.

“Indicating how the AI tool was used, why it was used, and that it was used in conjunction with available clinical information and the clinician’s best judgment could reduce the risk of being found responsible as a result of AI use in a particular case,” he said.

Use chatbots, such as ChatGPT, the way they were intended, as support tools, rather than definitive diagnostic instruments, adds Dr. Castro.

“Doctors should also be well-trained in interpreting and understanding the suggestions provided by ChatGPT and should use their clinical judgment and experience alongside the AI tool for more accurate decision-making,” he said.

In addition, because no AI insurance product exists on the market, physicians and organizations using AI – particularly for direct health care – should evaluate their current insurance or insurance-like products to determine where a claim involving AI might fall and whether the policy would respond, said Ms. Boisvert. The AI vendor/manufacturer will likely have indemnified themselves in the purchase and sale agreement or contract, she said.

It will also become increasingly important for medical practices, hospitals, and health systems to put in place strong data governance strategies, Mr. LeTang said.

“AI relies on good data,” he said. “A data governance strategy is a key component to making sure we understand where the data is coming from, what is represents, how accurate it is, if it’s reproducible, what controls are in place to ensure the right people have the right access, and that if we’re starting to use it to build algorithms, that it’s deidentified.”

While no malpractice claims associated with the use of AI have yet surfaced, this may change as legal courts catch up on the backlog of malpractice claims that were delayed because of COVID-19, and even more so as AI becomes more prevalent in health care, Mr. LeTang said.

“Similar to the attention that autonomous driving systems, like Tesla, receive when the system fails and accidents occur, we can be assured that media outlets will widely publicize AI-related medical adverse events,” he said. “It is crucial for health care professionals, AI developers, and regulatory authorities to work together to ensure the responsible use of AI in health care, with patient safety as the top priority. By doing so, they can mitigate the risks associated with AI implementation and minimize the potential for legal disputes arising from AI-related medical errors.”

A version of this article first appeared on Medscape.com.

The ways in which artificial intelligence (AI) may transform the future of medicine is making headlines across the globe. But chances are, you’re already using AI in your practice every day – you may just not realize it.
 

And whether you recognize the presence of AI or not, the technology could be putting you in danger of a lawsuit, legal experts say.

The use of AI in your daily practice can come with hidden liabilities, say legal experts, and as hospitals and medical groups deploy AI into more areas of health care, new liability exposures may be on the horizon.

“For physicians, AI has also not yet drastically changed or improved the way care is provided or consumed,” said Michael LeTang, chief nursing informatics officer and vice president of risk management and patient safety at Healthcare Risk Advisors, part of TDC Group. “Consequently, it may seem like AI is not present in their work streams, but in reality, it has been utilized in health care for several years. As AI technologies continue to develop and become more sophisticated, we can expect them to play an increasingly significant role in health care.”

Today, most AI applications in health care use narrow AI, which is designed to complete a single task without human assistance, as opposed to artificial general intelligence (AGI), which pertains to human-level reasoning and problem solving across a broad spectrum. Here are some ways doctors are using AI throughout the day – sometimes being aware of its assistance, and sometimes being unaware:

• Many doctors use electronic health records (EHRs) with integrated AI that include computerized clinical decision support tools designed to reduce the risk of diagnostic error and to integrate decision-making in the medication ordering function.
• Cardiologists, pathologists, and dermatologists use AI in the interpretation of vast amounts of images, tracings, and complex patterns.
• Surgeons are using AI-enhanced surgical robotics for orthopedic surgeries, such as joint replacement and spine surgery.
• A growing number of doctors are using ChatGPT to assist in drafting prior authorization letters for insurers. Experts say more doctors are also experimenting with ChatGPT to support medical decision-making.
• Within oncology, physicians use machine learning techniques in the form of computer-aided detection systems for early breast cancer detection.
• AI algorithms are often used by health systems for workflow, staffing optimization, population management, and care coordination.
• Some systems within EHRs use AI to indicate high-risk patients.
• Physicians are using AI applications for the early recognition of sepsis, including EHR-integrated decision tools, such as the Hospital Corporation of America Healthcare’s Sepsis Prediction and Optimization Therapy and the Sepsis Early Risk Assessment algorithm.
• About 30% of radiologists use AI in their practice to analyze x-rays and CT scans.
• Epic Systems recently announced a partnership with Microsoft to integrate ChatGPT into MyChart, Epic’s patient portal system. Pilot hospitals will utilize ChatGPT to automatically generate responses to patient-generated questions sent via the portal.

The growth of AI in health care has been enormous, and it’s only going to continue, said Ravi B. Parikh, MD, an assistant professor in the department of medical ethics and health policy and medicine at the University of Pennsylvania, Philadelphia.

“What’s really critical is that physicians, clinicians, and nurses using AI are provided with the tools to understand how artificial intelligence works and, most importantly, understand that they are still accountable for making the ultimate decision,” Mr. LeTang said, “The information is not always going to be the right thing to do or the most accurate thing to do. They’re still liable for making a bad decision, even if AI is driving that.”
 

What are the top AI legal dangers of today?

A pressing legal risk is becoming too reliant on the suggestions that AI-based systems provide, which can lead to poor care decisions, said Kenneth Rashbaum, a New York–based cybersecurity attorney with more than 25 years of experience in medical malpractice defense.

This can occur, for example, when using clinical support systems that leverage AI, machine learning, or statistical pattern recognition. Today, clinical support systems are commonly administered through EHRs and other computerized clinical workflows. In general, such systems match a patient’s characteristics to a computerized clinical knowledge base. An assessment or recommendation is then presented to the physician for a decision.

“If the clinician blindly accepts it without considering whether it’s appropriate for this patient at this time with this presentation, the clinician may bear some responsibility if there is an untoward result,” Mr. Rashbaum said.

“A common claim even in the days before the EMR [electronic medical record] and AI, was that the clinician did not take all available information into account in rendering treatment, including history of past and present condition, as reflected in the records, communication with past and other present treating clinicians, lab and radiology results, discussions with the patient, and physical examination findings,” he said. “So, if the clinician relied upon the support prompt to the exclusion of these other sources of information, that could be a very strong argument for the plaintiff.”

Chatbots, such OpenAI’s ChatGPT, are another form of AI raising legal red flags. ChatGPT, trained on a massive set of text data, can carry out conversations, write code, draft emails, and answer any question posed. The chatbot has gained considerable credibility for accurately diagnosing rare conditions in seconds, and it recently passed the U.S. Medical Licensing Examination.

It’s unclear how many doctors are signing onto the ChatGPT website daily, but physicians are actively using the chatbot, particularly for assistance with prior authorization letters and to support decision-making processes in their practices, said Mr. LeTang.

When physicians ask ChatGPT a question, however, they should be mindful that ChatGPT could “hallucinate,” a term that refers to a generated response that sounds plausible but is factually incorrect or is unrelated to the context, explains Harvey Castro, MD, an emergency physician, ChatGPT health care expert, and author of the 2023 book “ChatGPT and Healthcare: Unlocking the Potential of Patient Empowerment.”

Acting on ChatGPT’s response without vetting the information places doctors at serious risk of a lawsuit, he said.

“Sometimes, the response is half true and half false,” he said. “Say, I go outside my specialty of emergency medicine and ask it about a pediatric surgical procedure. It could give me a response that sounds medically correct, but then I ask a pediatric cardiologist, and he says, ‘We don’t even do this. This doesn’t even exist!’ Physicians really have to make sure they are vetting the information provided.”

In response to ChatGPT’s growing usage by health care professionals, hospitals and practices are quickly implementing guidelines, policies, and restrictions that caution physicians about the accuracy of ChatGPT-generated information, adds Mr. LeTang.

Emerging best practices include avoiding the input of patient health information, personally identifiable information, or any data that could be commercially valuable or considered the intellectual property of a hospital or health system, he said.

“Another crucial guideline is not to rely solely on ChatGPT as a definitive source for clinical decision-making; physicians must exercise their professional judgment,” he said. “If best practices are not adhered to, the associated risks are present today. However, these risks may become more significant as AI technologies continue to evolve and become increasingly integrated into health care.”

The potential for misdiagnosis by AI systems and the risk of unnecessary procedures if physicians do not thoroughly evaluate and validate AI predictions are other dangers.

As an example, Mr. LeTang described a case in which a physician documents in the EHR that a patient has presented to the emergency department with chest pains and other signs of a heart attack, and an AI algorithm predicts that the patient is experiencing an active myocardial infarction. If the physician then sends the patient for stenting or an angioplasty without other concrete evidence or tests to confirm the diagnosis, the doctor could later face a misdiagnosis complaint if the costly procedures were unnecessary.

“That’s one of the risks of using artificial intelligence,” he said. “A large percentage of malpractice claims is failure to diagnose, delayed diagnosis, or inaccurate diagnosis. What falls in the category of failure to diagnose is sending a patient for an unnecessary procedure or having an adverse event or bad outcome because of the failure to diagnose.”

So far, no AI lawsuits have been filed, but they may make an appearance soon, said Sue Boisvert, senior patient safety risk manager at The Doctors Company, a national medical liability insurer.

“There are hundreds of AI programs currently in use in health care,” she said. “At some point, a provider will make a decision that is contrary to what the AI recommended. The AI may be wrong, or the provider may be wrong. Either way, the provider will neglect to document their clinical reasoning, a patient will be harmed, and we will have the first AI claim.”
 

Upcoming AI legal risks to watch for

Lawsuits that allege biased patient care by physicians on the basis of algorithmic bias may also be forthcoming, analysts warn.

Much has been written about algorithmic bias that compounds and worsens inequities in socioeconomic status, ethnicity, sexual orientation, and gender in health systems. In 2019, a groundbreaking article in Science shed light on commonly used algorithms that are considered racially biased and how health care professionals often use such information to make medical decisions.

No claims involving AI bias have come down the pipeline yet, but it’s an area to watch, said Ms. Boisvert. She noted a website that highlights complaints and accusations of AI bias, including in health care.

“We need to be sure the training of the AI is appropriate, current, and broad enough so that there is no bias in the AI when it’s participating in the decision-making,” said Ms. Boisvert. “Imagine if the AI is diagnosing based on a dataset that is not local. It doesn’t represent the population at that particular hospital, and it’s providing inaccurate information to the physicians who are then making decisions about treatment.”

In pain management, for example, there are known differences in how patients experience pain, Ms. Boisvert said. If AI was being used to develop an algorithm for how a particular patient’s postoperative pain should be managed, and the algorithm did not include the differences, the pain control for a certain patient could be inappropriate. A poor outcome resulting from the treatment could lead to a claim against the physician or hospital that used the biased AI system, she said.

In the future, as AI becomes more integrated and accepted in medicine, there may be a risk of legal complaints against doctors for not using AI, said Saurabh Jha, MD, an associate professor of radiology at the University of Pennsylvania, Philadelphia, and a scholar of AI in radiology.

“Ultimately, we might get to a place where AI starts helping physicians detect more or reduce the miss of certain conditions, and it becomes the standard of care,” Dr. Jha said. “For example, if it became part of the standard of care for pulmonary embolism [PE] detection, and you didn’t use it for PE detection, and there was a miss. That could put you at legal risk. We’re not at that stage yet, but that is one future possibility.”

Dr. Parikh envisions an even cloudier liability landscape as the potential grows for AI to control patient care decisions. In such a scenario, rather than just issuing an alert or prediction to a physician, the AI system could trigger an action.

For instance, if an algorithm is trained to predict sepsis and, once triggered, the AI could initiate a nurse-led rapid response or a change in patient care outside the clinician’s control, said Dr. Parikh, who coauthored a recent article on AI and medical liability in The Milbank Quarterly.

“That’s still very much the minority of how AI is being used, but as evidence is growing that AI-based diagnostic tools perform equivalent or even superior to physicians, these autonomous workflows are being considered,” Dr. Parikh said. “When the ultimate action upon the patient is more determined by the AI than what the clinician does, then I think the liability picture gets murkier, and we should be thinking about how we can respond to that from a liability framework.”
 

How you can prevent AI-related lawsuits

The first step to preventing an AI-related claim is being aware of when and how you are using AI.

Ensure you’re informed about how the AI was trained, Ms. Boisvert stresses.

“Ask questions!” she said. “Is the AI safe? Are the recommendations accurate? Does the AI perform better than current systems? In what way? What databases were used, and did the programmers consider bias? Do I understand how to use the results?”

Never blindly trust the AI but rather view it as a data point in a medical decision, said Dr. Parikh. Ensure that other sources of medical information are properly accessed and that best practices for your specialty are still being followed.

When using any form of AI, document your usage, adds Mr. Rashbaum. A record that clearly outlines how the physician incorporated the AI is critical if a claim later arises in which the doctor is accused of AI-related malpractice, he said.

“Indicating how the AI tool was used, why it was used, and that it was used in conjunction with available clinical information and the clinician’s best judgment could reduce the risk of being found responsible as a result of AI use in a particular case,” he said.

Use chatbots, such as ChatGPT, the way they were intended, as support tools, rather than definitive diagnostic instruments, adds Dr. Castro.

“Doctors should also be well-trained in interpreting and understanding the suggestions provided by ChatGPT and should use their clinical judgment and experience alongside the AI tool for more accurate decision-making,” he said.

In addition, because no AI insurance product exists on the market, physicians and organizations using AI – particularly for direct health care – should evaluate their current insurance or insurance-like products to determine where a claim involving AI might fall and whether the policy would respond, said Ms. Boisvert. The AI vendor/manufacturer will likely have indemnified themselves in the purchase and sale agreement or contract, she said.

It will also become increasingly important for medical practices, hospitals, and health systems to put in place strong data governance strategies, Mr. LeTang said.

“AI relies on good data,” he said. “A data governance strategy is a key component to making sure we understand where the data is coming from, what is represents, how accurate it is, if it’s reproducible, what controls are in place to ensure the right people have the right access, and that if we’re starting to use it to build algorithms, that it’s deidentified.”

While no malpractice claims associated with the use of AI have yet surfaced, this may change as legal courts catch up on the backlog of malpractice claims that were delayed because of COVID-19, and even more so as AI becomes more prevalent in health care, Mr. LeTang said.

“Similar to the attention that autonomous driving systems, like Tesla, receive when the system fails and accidents occur, we can be assured that media outlets will widely publicize AI-related medical adverse events,” he said. “It is crucial for health care professionals, AI developers, and regulatory authorities to work together to ensure the responsible use of AI in health care, with patient safety as the top priority. By doing so, they can mitigate the risks associated with AI implementation and minimize the potential for legal disputes arising from AI-related medical errors.”

A version of this article first appeared on Medscape.com.

The ways in which artificial intelligence (AI) may transform the future of medicine is making headlines across the globe. But chances are, you’re already using AI in your practice every day – you may just not realize it.
 

And whether you recognize the presence of AI or not, the technology could be putting you in danger of a lawsuit, legal experts say.

The use of AI in your daily practice can come with hidden liabilities, say legal experts, and as hospitals and medical groups deploy AI into more areas of health care, new liability exposures may be on the horizon.

“For physicians, AI has also not yet drastically changed or improved the way care is provided or consumed,” said Michael LeTang, chief nursing informatics officer and vice president of risk management and patient safety at Healthcare Risk Advisors, part of TDC Group. “Consequently, it may seem like AI is not present in their work streams, but in reality, it has been utilized in health care for several years. As AI technologies continue to develop and become more sophisticated, we can expect them to play an increasingly significant role in health care.”

Today, most AI applications in health care use narrow AI, which is designed to complete a single task without human assistance, as opposed to artificial general intelligence (AGI), which pertains to human-level reasoning and problem solving across a broad spectrum. Here are some ways doctors are using AI throughout the day – sometimes being aware of its assistance, and sometimes being unaware:

• Many doctors use electronic health records (EHRs) with integrated AI that include computerized clinical decision support tools designed to reduce the risk of diagnostic error and to integrate decision-making in the medication ordering function.
• Cardiologists, pathologists, and dermatologists use AI in the interpretation of vast amounts of images, tracings, and complex patterns.
• Surgeons are using AI-enhanced surgical robotics for orthopedic surgeries, such as joint replacement and spine surgery.
• A growing number of doctors are using ChatGPT to assist in drafting prior authorization letters for insurers. Experts say more doctors are also experimenting with ChatGPT to support medical decision-making.
• Within oncology, physicians use machine learning techniques in the form of computer-aided detection systems for early breast cancer detection.
• AI algorithms are often used by health systems for workflow, staffing optimization, population management, and care coordination.
• Some systems within EHRs use AI to indicate high-risk patients.
• Physicians are using AI applications for the early recognition of sepsis, including EHR-integrated decision tools, such as the Hospital Corporation of America Healthcare’s Sepsis Prediction and Optimization Therapy and the Sepsis Early Risk Assessment algorithm.
• About 30% of radiologists use AI in their practice to analyze x-rays and CT scans.
• Epic Systems recently announced a partnership with Microsoft to integrate ChatGPT into MyChart, Epic’s patient portal system. Pilot hospitals will utilize ChatGPT to automatically generate responses to patient-generated questions sent via the portal.

The growth of AI in health care has been enormous, and it’s only going to continue, said Ravi B. Parikh, MD, an assistant professor in the department of medical ethics and health policy and medicine at the University of Pennsylvania, Philadelphia.

“What’s really critical is that physicians, clinicians, and nurses using AI are provided with the tools to understand how artificial intelligence works and, most importantly, understand that they are still accountable for making the ultimate decision,” Mr. LeTang said, “The information is not always going to be the right thing to do or the most accurate thing to do. They’re still liable for making a bad decision, even if AI is driving that.”
 

What are the top AI legal dangers of today?

A pressing legal risk is becoming too reliant on the suggestions that AI-based systems provide, which can lead to poor care decisions, said Kenneth Rashbaum, a New York–based cybersecurity attorney with more than 25 years of experience in medical malpractice defense.

This can occur, for example, when using clinical support systems that leverage AI, machine learning, or statistical pattern recognition. Today, clinical support systems are commonly administered through EHRs and other computerized clinical workflows. In general, such systems match a patient’s characteristics to a computerized clinical knowledge base. An assessment or recommendation is then presented to the physician for a decision.

“If the clinician blindly accepts it without considering whether it’s appropriate for this patient at this time with this presentation, the clinician may bear some responsibility if there is an untoward result,” Mr. Rashbaum said.

“A common claim even in the days before the EMR [electronic medical record] and AI, was that the clinician did not take all available information into account in rendering treatment, including history of past and present condition, as reflected in the records, communication with past and other present treating clinicians, lab and radiology results, discussions with the patient, and physical examination findings,” he said. “So, if the clinician relied upon the support prompt to the exclusion of these other sources of information, that could be a very strong argument for the plaintiff.”

Chatbots, such OpenAI’s ChatGPT, are another form of AI raising legal red flags. ChatGPT, trained on a massive set of text data, can carry out conversations, write code, draft emails, and answer any question posed. The chatbot has gained considerable credibility for accurately diagnosing rare conditions in seconds, and it recently passed the U.S. Medical Licensing Examination.

It’s unclear how many doctors are signing onto the ChatGPT website daily, but physicians are actively using the chatbot, particularly for assistance with prior authorization letters and to support decision-making processes in their practices, said Mr. LeTang.

When physicians ask ChatGPT a question, however, they should be mindful that ChatGPT could “hallucinate,” a term that refers to a generated response that sounds plausible but is factually incorrect or is unrelated to the context, explains Harvey Castro, MD, an emergency physician, ChatGPT health care expert, and author of the 2023 book “ChatGPT and Healthcare: Unlocking the Potential of Patient Empowerment.”

Acting on ChatGPT’s response without vetting the information places doctors at serious risk of a lawsuit, he said.

“Sometimes, the response is half true and half false,” he said. “Say, I go outside my specialty of emergency medicine and ask it about a pediatric surgical procedure. It could give me a response that sounds medically correct, but then I ask a pediatric cardiologist, and he says, ‘We don’t even do this. This doesn’t even exist!’ Physicians really have to make sure they are vetting the information provided.”

In response to ChatGPT’s growing usage by health care professionals, hospitals and practices are quickly implementing guidelines, policies, and restrictions that caution physicians about the accuracy of ChatGPT-generated information, adds Mr. LeTang.

Emerging best practices include avoiding the input of patient health information, personally identifiable information, or any data that could be commercially valuable or considered the intellectual property of a hospital or health system, he said.

“Another crucial guideline is not to rely solely on ChatGPT as a definitive source for clinical decision-making; physicians must exercise their professional judgment,” he said. “If best practices are not adhered to, the associated risks are present today. However, these risks may become more significant as AI technologies continue to evolve and become increasingly integrated into health care.”

The potential for misdiagnosis by AI systems and the risk of unnecessary procedures if physicians do not thoroughly evaluate and validate AI predictions are other dangers.

As an example, Mr. LeTang described a case in which a physician documents in the EHR that a patient has presented to the emergency department with chest pains and other signs of a heart attack, and an AI algorithm predicts that the patient is experiencing an active myocardial infarction. If the physician then sends the patient for stenting or an angioplasty without other concrete evidence or tests to confirm the diagnosis, the doctor could later face a misdiagnosis complaint if the costly procedures were unnecessary.

“That’s one of the risks of using artificial intelligence,” he said. “A large percentage of malpractice claims is failure to diagnose, delayed diagnosis, or inaccurate diagnosis. What falls in the category of failure to diagnose is sending a patient for an unnecessary procedure or having an adverse event or bad outcome because of the failure to diagnose.”

So far, no AI lawsuits have been filed, but they may make an appearance soon, said Sue Boisvert, senior patient safety risk manager at The Doctors Company, a national medical liability insurer.

“There are hundreds of AI programs currently in use in health care,” she said. “At some point, a provider will make a decision that is contrary to what the AI recommended. The AI may be wrong, or the provider may be wrong. Either way, the provider will neglect to document their clinical reasoning, a patient will be harmed, and we will have the first AI claim.”
 

Upcoming AI legal risks to watch for

Lawsuits that allege biased patient care by physicians on the basis of algorithmic bias may also be forthcoming, analysts warn.

Much has been written about algorithmic bias that compounds and worsens inequities in socioeconomic status, ethnicity, sexual orientation, and gender in health systems. In 2019, a groundbreaking article in Science shed light on commonly used algorithms that are considered racially biased and how health care professionals often use such information to make medical decisions.

No claims involving AI bias have come down the pipeline yet, but it’s an area to watch, said Ms. Boisvert. She noted a website that highlights complaints and accusations of AI bias, including in health care.

“We need to be sure the training of the AI is appropriate, current, and broad enough so that there is no bias in the AI when it’s participating in the decision-making,” said Ms. Boisvert. “Imagine if the AI is diagnosing based on a dataset that is not local. It doesn’t represent the population at that particular hospital, and it’s providing inaccurate information to the physicians who are then making decisions about treatment.”

In pain management, for example, there are known differences in how patients experience pain, Ms. Boisvert said. If AI was being used to develop an algorithm for how a particular patient’s postoperative pain should be managed, and the algorithm did not include the differences, the pain control for a certain patient could be inappropriate. A poor outcome resulting from the treatment could lead to a claim against the physician or hospital that used the biased AI system, she said.

In the future, as AI becomes more integrated and accepted in medicine, there may be a risk of legal complaints against doctors for not using AI, said Saurabh Jha, MD, an associate professor of radiology at the University of Pennsylvania, Philadelphia, and a scholar of AI in radiology.

“Ultimately, we might get to a place where AI starts helping physicians detect more or reduce the miss of certain conditions, and it becomes the standard of care,” Dr. Jha said. “For example, if it became part of the standard of care for pulmonary embolism [PE] detection, and you didn’t use it for PE detection, and there was a miss. That could put you at legal risk. We’re not at that stage yet, but that is one future possibility.”

Dr. Parikh envisions an even cloudier liability landscape as the potential grows for AI to control patient care decisions. In such a scenario, rather than just issuing an alert or prediction to a physician, the AI system could trigger an action.

For instance, if an algorithm is trained to predict sepsis and, once triggered, the AI could initiate a nurse-led rapid response or a change in patient care outside the clinician’s control, said Dr. Parikh, who coauthored a recent article on AI and medical liability in The Milbank Quarterly.

“That’s still very much the minority of how AI is being used, but as evidence is growing that AI-based diagnostic tools perform equivalent or even superior to physicians, these autonomous workflows are being considered,” Dr. Parikh said. “When the ultimate action upon the patient is more determined by the AI than what the clinician does, then I think the liability picture gets murkier, and we should be thinking about how we can respond to that from a liability framework.”
 

How you can prevent AI-related lawsuits

The first step to preventing an AI-related claim is being aware of when and how you are using AI.

Ensure you’re informed about how the AI was trained, Ms. Boisvert stresses.

“Ask questions!” she said. “Is the AI safe? Are the recommendations accurate? Does the AI perform better than current systems? In what way? What databases were used, and did the programmers consider bias? Do I understand how to use the results?”

Never blindly trust the AI but rather view it as a data point in a medical decision, said Dr. Parikh. Ensure that other sources of medical information are properly accessed and that best practices for your specialty are still being followed.

When using any form of AI, document your usage, adds Mr. Rashbaum. A record that clearly outlines how the physician incorporated the AI is critical if a claim later arises in which the doctor is accused of AI-related malpractice, he said.

“Indicating how the AI tool was used, why it was used, and that it was used in conjunction with available clinical information and the clinician’s best judgment could reduce the risk of being found responsible as a result of AI use in a particular case,” he said.

Use chatbots, such as ChatGPT, the way they were intended, as support tools, rather than definitive diagnostic instruments, adds Dr. Castro.

“Doctors should also be well-trained in interpreting and understanding the suggestions provided by ChatGPT and should use their clinical judgment and experience alongside the AI tool for more accurate decision-making,” he said.

In addition, because no AI insurance product exists on the market, physicians and organizations using AI – particularly for direct health care – should evaluate their current insurance or insurance-like products to determine where a claim involving AI might fall and whether the policy would respond, said Ms. Boisvert. The AI vendor/manufacturer will likely have indemnified themselves in the purchase and sale agreement or contract, she said.

It will also become increasingly important for medical practices, hospitals, and health systems to put in place strong data governance strategies, Mr. LeTang said.

“AI relies on good data,” he said. “A data governance strategy is a key component to making sure we understand where the data is coming from, what is represents, how accurate it is, if it’s reproducible, what controls are in place to ensure the right people have the right access, and that if we’re starting to use it to build algorithms, that it’s deidentified.”

While no malpractice claims associated with the use of AI have yet surfaced, this may change as legal courts catch up on the backlog of malpractice claims that were delayed because of COVID-19, and even more so as AI becomes more prevalent in health care, Mr. LeTang said.

“Similar to the attention that autonomous driving systems, like Tesla, receive when the system fails and accidents occur, we can be assured that media outlets will widely publicize AI-related medical adverse events,” he said. “It is crucial for health care professionals, AI developers, and regulatory authorities to work together to ensure the responsible use of AI in health care, with patient safety as the top priority. By doing so, they can mitigate the risks associated with AI implementation and minimize the potential for legal disputes arising from AI-related medical errors.”

A version of this article first appeared on Medscape.com.

Sowing the seeds of cancer prevention

Are you looking to add to your quality of life, even though pets are not your speed? Might we suggest something with lower maintenance? Something a little greener?

Indoor plants can purify the air that comes from outside. Researchers at the University of Technology Sydney, in partnership with the plantscaping company Ambius, showed that a “green wall” made up of mixed indoor plants was able to suck up 97% of “the most toxic compounds” from the air in just 8 hours. We’re talking about lung-irritating, headache-inducing, cancer risk–boosting compounds from gasoline fumes, including benzene.

Fraser Torpy/University of Technology Sydney

Public health initiatives often strive to reduce cardiovascular and obesity risks, but breathing seems pretty important too. According to the World Health Organization, household air pollution is responsible for about 2.5 million global premature deaths each year. And since 2020 we’ve become accustomed to spending more time inside and at home.

“This new research proves that plants should not just be seen as ‘nice to have,’ but rather a crucial part of every workplace wellness plan,” Ambius General Manager Johan Hodgson said in statement released by the university.

So don’t spend hundreds of dollars on a fancy air filtration system when a wall of plants can do that for next to nothing. Find what works for you and your space and become a plant parent today! Your lungs will thank you.
 

But officer, I had to swerve to miss the duodenal ampulla

Tiny video capsule endoscopes have been around for many years, but they have one big weakness: The ingestible cameras’ journey through the GI tract is passively driven by gravity and the natural movement of the body, so they often miss potential problem areas.

AnX Robotica

Not anymore. That flaw has been addressed by medical technology company AnX Robotica, which has taken endoscopy to the next level by adding that wondrous directional control device of the modern electronic age, a joystick.

The new system “uses an external magnet and hand-held video game style joysticks to move the capsule in three dimensions,” which allows physicians to “remotely drive a miniature video capsule to all regions of the stomach to visualize and photograph potential problem areas,” according to Andrew C. Meltzer, MD, of George Washington University and associates, who conducted a pilot study funded by AnX Robotica.

The video capsule provided a 95% rate of visualization in the stomachs of 40 patients who were examined at a medical office building by an emergency medicine physician who had no previous specialty training in endoscopy. “Capsules were driven by the ER physician and then the study reports were reviewed by an attending gastroenterologist who was physically off site,” the investigators said in a written statement.

The capsule operator did receive some additional training, and development of artificial intelligence to self-drive the capsule is in the works, but for now, we’re talking about a device controlled by a human using a joystick. And we all know that 50-year-olds are not especially known for their joystick skills. For that we need real experts. Yup, we need to put those joystick-controlled capsule endoscopes in the hands of teenage gamers. Who wants to go first?
 

Maybe AI isn’t ready for the big time after all

“How long before some intrepid stockholder says: ‘Hey, instead of paying doctors, why don’t we just use the free robot instead?’ ” Those words appeared on LOTME but a month ago. After all, the AI is supposed to be smarter and more empathetic than a doctor. And did we mention it’s free? Or at least extremely cheap. Cheaper than, say, a group of recently unionized health care workers.

maewjpho/Thinkstock

In early May, the paid employees manning the National Eating Disorders Association emergency hotline voted to unionize, as they felt overwhelmed and underpaid. Apparently, paying six people an extra few thousand a year was too much for NEDA’s leadership, as they decided a few weeks later to fire those workers, fully closing down the hotline. Instead of talking to a real person, people “calling in” for support would be met with Tessa, a wellness chatbot that would hopefully guide them through their crisis. Key word, hopefully.

In perhaps the least surprising twist of the year, NEDA was forced to walk back its decision about a week after its initial announcement. It all started with a viral Instagram post from a woman who called in and received the following advice from Tessa: Lose 1-2 pounds a week, count calories and work for a 500- to 1,000-calorie deficit, weigh herself weekly, and restrict her diet. Unfortunately, all of these suggestions were things that led to the development of the woman’s eating disorder.

Naturally, NEDA responded in good grace, accusing the woman of lying. A NEDA vice president even left some nasty comments on the post, but hastily deleted them a day later when NEDA announced it was shutting down Tessa “until further notice for a complete investigation.” NEDA’s CEO insisted they hadn’t seen that behavior from Tessa before, calling it a “bug” and insisting the bot would only be down temporarily until the triggers causing the bug were fixed.

In the aftermath, several doctors and psychologists chimed in, terming the rush to automate human roles dangerous and risky. After all, much of what makes these hotlines effective is the volunteers speaking from their own experience. An unsupervised bot doesn’t seem to have what it takes to deal with a mental health crisis, but we’re betting that Tessa will be back. As a wise cephalopod once said: Nobody gives a care about the fate of labor as long as they can get their instant gratification.
 

You can’t spell existential without s-t-e-n-t

This week, we’re including a special “bonus” item that, to be honest, has nothing to do with stents. That’s why our editor is making us call this a “bonus” (and making us use quote marks, too): It doesn’t really have anything to do with stents or health care or those who practice health care. Actually, his exact words were, “You can’t just give the readers someone else’s ****ing list and expect to get paid for it.” Did we mention that he looks like Jack Nicklaus but acts like BoJack Horseman?

Anywaaay, we’re pretty sure that the list in question – “America’s Top 10 Most Googled Existential Questions” – says something about the human condition, just not about stents:

1. Why is the sky blue?

2. What do dreams mean?

3. What is the meaning of life?

4. Why am I so tired?

5. Who am I?

6. What is love?

7. Is a hot dog a sandwich?

8. What came first, the chicken or the egg?

9. What should I do?

10. Do animals have souls?

Sowing the seeds of cancer prevention

Are you looking to add to your quality of life, even though pets are not your speed? Might we suggest something with lower maintenance? Something a little greener?

Indoor plants can purify the air that comes from outside. Researchers at the University of Technology Sydney, in partnership with the plantscaping company Ambius, showed that a “green wall” made up of mixed indoor plants was able to suck up 97% of “the most toxic compounds” from the air in just 8 hours. We’re talking about lung-irritating, headache-inducing, cancer risk–boosting compounds from gasoline fumes, including benzene.

Fraser Torpy/University of Technology Sydney

Public health initiatives often strive to reduce cardiovascular and obesity risks, but breathing seems pretty important too. According to the World Health Organization, household air pollution is responsible for about 2.5 million global premature deaths each year. And since 2020 we’ve become accustomed to spending more time inside and at home.

“This new research proves that plants should not just be seen as ‘nice to have,’ but rather a crucial part of every workplace wellness plan,” Ambius General Manager Johan Hodgson said in statement released by the university.

So don’t spend hundreds of dollars on a fancy air filtration system when a wall of plants can do that for next to nothing. Find what works for you and your space and become a plant parent today! Your lungs will thank you.
 

But officer, I had to swerve to miss the duodenal ampulla

Tiny video capsule endoscopes have been around for many years, but they have one big weakness: The ingestible cameras’ journey through the GI tract is passively driven by gravity and the natural movement of the body, so they often miss potential problem areas.

AnX Robotica

Not anymore. That flaw has been addressed by medical technology company AnX Robotica, which has taken endoscopy to the next level by adding that wondrous directional control device of the modern electronic age, a joystick.

The new system “uses an external magnet and hand-held video game style joysticks to move the capsule in three dimensions,” which allows physicians to “remotely drive a miniature video capsule to all regions of the stomach to visualize and photograph potential problem areas,” according to Andrew C. Meltzer, MD, of George Washington University and associates, who conducted a pilot study funded by AnX Robotica.

The video capsule provided a 95% rate of visualization in the stomachs of 40 patients who were examined at a medical office building by an emergency medicine physician who had no previous specialty training in endoscopy. “Capsules were driven by the ER physician and then the study reports were reviewed by an attending gastroenterologist who was physically off site,” the investigators said in a written statement.

The capsule operator did receive some additional training, and development of artificial intelligence to self-drive the capsule is in the works, but for now, we’re talking about a device controlled by a human using a joystick. And we all know that 50-year-olds are not especially known for their joystick skills. For that we need real experts. Yup, we need to put those joystick-controlled capsule endoscopes in the hands of teenage gamers. Who wants to go first?
 

Maybe AI isn’t ready for the big time after all

“How long before some intrepid stockholder says: ‘Hey, instead of paying doctors, why don’t we just use the free robot instead?’ ” Those words appeared on LOTME but a month ago. After all, the AI is supposed to be smarter and more empathetic than a doctor. And did we mention it’s free? Or at least extremely cheap. Cheaper than, say, a group of recently unionized health care workers.

maewjpho/Thinkstock

In early May, the paid employees manning the National Eating Disorders Association emergency hotline voted to unionize, as they felt overwhelmed and underpaid. Apparently, paying six people an extra few thousand a year was too much for NEDA’s leadership, as they decided a few weeks later to fire those workers, fully closing down the hotline. Instead of talking to a real person, people “calling in” for support would be met with Tessa, a wellness chatbot that would hopefully guide them through their crisis. Key word, hopefully.

In perhaps the least surprising twist of the year, NEDA was forced to walk back its decision about a week after its initial announcement. It all started with a viral Instagram post from a woman who called in and received the following advice from Tessa: Lose 1-2 pounds a week, count calories and work for a 500- to 1,000-calorie deficit, weigh herself weekly, and restrict her diet. Unfortunately, all of these suggestions were things that led to the development of the woman’s eating disorder.

Naturally, NEDA responded in good grace, accusing the woman of lying. A NEDA vice president even left some nasty comments on the post, but hastily deleted them a day later when NEDA announced it was shutting down Tessa “until further notice for a complete investigation.” NEDA’s CEO insisted they hadn’t seen that behavior from Tessa before, calling it a “bug” and insisting the bot would only be down temporarily until the triggers causing the bug were fixed.

In the aftermath, several doctors and psychologists chimed in, terming the rush to automate human roles dangerous and risky. After all, much of what makes these hotlines effective is the volunteers speaking from their own experience. An unsupervised bot doesn’t seem to have what it takes to deal with a mental health crisis, but we’re betting that Tessa will be back. As a wise cephalopod once said: Nobody gives a care about the fate of labor as long as they can get their instant gratification.
 

You can’t spell existential without s-t-e-n-t

This week, we’re including a special “bonus” item that, to be honest, has nothing to do with stents. That’s why our editor is making us call this a “bonus” (and making us use quote marks, too): It doesn’t really have anything to do with stents or health care or those who practice health care. Actually, his exact words were, “You can’t just give the readers someone else’s ****ing list and expect to get paid for it.” Did we mention that he looks like Jack Nicklaus but acts like BoJack Horseman?

Anywaaay, we’re pretty sure that the list in question – “America’s Top 10 Most Googled Existential Questions” – says something about the human condition, just not about stents:

1. Why is the sky blue?

2. What do dreams mean?

3. What is the meaning of life?

4. Why am I so tired?

5. Who am I?

6. What is love?

7. Is a hot dog a sandwich?

8. What came first, the chicken or the egg?

9. What should I do?

10. Do animals have souls?

Sowing the seeds of cancer prevention

Are you looking to add to your quality of life, even though pets are not your speed? Might we suggest something with lower maintenance? Something a little greener?

Indoor plants can purify the air that comes from outside. Researchers at the University of Technology Sydney, in partnership with the plantscaping company Ambius, showed that a “green wall” made up of mixed indoor plants was able to suck up 97% of “the most toxic compounds” from the air in just 8 hours. We’re talking about lung-irritating, headache-inducing, cancer risk–boosting compounds from gasoline fumes, including benzene.

Fraser Torpy/University of Technology Sydney

Public health initiatives often strive to reduce cardiovascular and obesity risks, but breathing seems pretty important too. According to the World Health Organization, household air pollution is responsible for about 2.5 million global premature deaths each year. And since 2020 we’ve become accustomed to spending more time inside and at home.

“This new research proves that plants should not just be seen as ‘nice to have,’ but rather a crucial part of every workplace wellness plan,” Ambius General Manager Johan Hodgson said in statement released by the university.

So don’t spend hundreds of dollars on a fancy air filtration system when a wall of plants can do that for next to nothing. Find what works for you and your space and become a plant parent today! Your lungs will thank you.
 

But officer, I had to swerve to miss the duodenal ampulla

Tiny video capsule endoscopes have been around for many years, but they have one big weakness: The ingestible cameras’ journey through the GI tract is passively driven by gravity and the natural movement of the body, so they often miss potential problem areas.

AnX Robotica

Not anymore. That flaw has been addressed by medical technology company AnX Robotica, which has taken endoscopy to the next level by adding that wondrous directional control device of the modern electronic age, a joystick.

The new system “uses an external magnet and hand-held video game style joysticks to move the capsule in three dimensions,” which allows physicians to “remotely drive a miniature video capsule to all regions of the stomach to visualize and photograph potential problem areas,” according to Andrew C. Meltzer, MD, of George Washington University and associates, who conducted a pilot study funded by AnX Robotica.

The video capsule provided a 95% rate of visualization in the stomachs of 40 patients who were examined at a medical office building by an emergency medicine physician who had no previous specialty training in endoscopy. “Capsules were driven by the ER physician and then the study reports were reviewed by an attending gastroenterologist who was physically off site,” the investigators said in a written statement.

The capsule operator did receive some additional training, and development of artificial intelligence to self-drive the capsule is in the works, but for now, we’re talking about a device controlled by a human using a joystick. And we all know that 50-year-olds are not especially known for their joystick skills. For that we need real experts. Yup, we need to put those joystick-controlled capsule endoscopes in the hands of teenage gamers. Who wants to go first?
 

Maybe AI isn’t ready for the big time after all

“How long before some intrepid stockholder says: ‘Hey, instead of paying doctors, why don’t we just use the free robot instead?’ ” Those words appeared on LOTME but a month ago. After all, the AI is supposed to be smarter and more empathetic than a doctor. And did we mention it’s free? Or at least extremely cheap. Cheaper than, say, a group of recently unionized health care workers.

maewjpho/Thinkstock

In early May, the paid employees manning the National Eating Disorders Association emergency hotline voted to unionize, as they felt overwhelmed and underpaid. Apparently, paying six people an extra few thousand a year was too much for NEDA’s leadership, as they decided a few weeks later to fire those workers, fully closing down the hotline. Instead of talking to a real person, people “calling in” for support would be met with Tessa, a wellness chatbot that would hopefully guide them through their crisis. Key word, hopefully.

In perhaps the least surprising twist of the year, NEDA was forced to walk back its decision about a week after its initial announcement. It all started with a viral Instagram post from a woman who called in and received the following advice from Tessa: Lose 1-2 pounds a week, count calories and work for a 500- to 1,000-calorie deficit, weigh herself weekly, and restrict her diet. Unfortunately, all of these suggestions were things that led to the development of the woman’s eating disorder.

Naturally, NEDA responded in good grace, accusing the woman of lying. A NEDA vice president even left some nasty comments on the post, but hastily deleted them a day later when NEDA announced it was shutting down Tessa “until further notice for a complete investigation.” NEDA’s CEO insisted they hadn’t seen that behavior from Tessa before, calling it a “bug” and insisting the bot would only be down temporarily until the triggers causing the bug were fixed.

In the aftermath, several doctors and psychologists chimed in, terming the rush to automate human roles dangerous and risky. After all, much of what makes these hotlines effective is the volunteers speaking from their own experience. An unsupervised bot doesn’t seem to have what it takes to deal with a mental health crisis, but we’re betting that Tessa will be back. As a wise cephalopod once said: Nobody gives a care about the fate of labor as long as they can get their instant gratification.
 

You can’t spell existential without s-t-e-n-t

This week, we’re including a special “bonus” item that, to be honest, has nothing to do with stents. That’s why our editor is making us call this a “bonus” (and making us use quote marks, too): It doesn’t really have anything to do with stents or health care or those who practice health care. Actually, his exact words were, “You can’t just give the readers someone else’s ****ing list and expect to get paid for it.” Did we mention that he looks like Jack Nicklaus but acts like BoJack Horseman?

Anywaaay, we’re pretty sure that the list in question – “America’s Top 10 Most Googled Existential Questions” – says something about the human condition, just not about stents:

1. Why is the sky blue?

2. What do dreams mean?

3. What is the meaning of life?

4. Why am I so tired?

5. Who am I?

6. What is love?

7. Is a hot dog a sandwich?

8. What came first, the chicken or the egg?

9. What should I do?

10. Do animals have souls?

CHICAGO – Patients with advanced estrogen receptor–positive, HER-negative breast cancer receiving first line treatment with CDK4/6 inhibitors (CDK4/6i) experienced no progression-free survival (PFS) or overall survival (OS) advantages over those who received the same therapy as second line treatment.

That was the conclusion of the phase 3 SONIA study, which was presented at the annual meeting of the American Society of Clinical Oncology.

The benefit from first line therapy is not maintained and almost completely disappears when patients in the control arm cross over to receive CDK4/6 inhibition in second line,” said Gabe Sonke, MD, PhD, during his presentation at the meeting.

CDK4/6 inhibitors have shown benefit in both the first-and second-line setting, according to Dr. Sonke, who is a medical oncologist at the Netherlands Cancer Institute, Amsterdam. He added that most guidelines suggest use of CDK4/6 inhibitors in the first line, but there hasn’t been a direct comparison between use in the first and second line.

“Many patients do very well on endocrine therapy alone [in the first line]. Combination treatment leads to a higher risk of the emergence of resistant patterns such as ESR1 mutations, and CDK4/6 inhibitors also come with added costs and toxicities. Given the absence of comparative data between first line and second line, we designed the SONIA trial,” said Dr. Sonke.
 

Study methods and results

The researchers recruited 1,050 pre- and postmenopausal women who were randomized to a nonsteroidal AI in the first line followed by second-line CDK4/6i plus the estrogen receptor antagonist fulvestrant, or a nonsteroidal AI plus a CDK4/6i in the first line and fulvestrant in the second line. The most commonly used CDK4/6i was palbociclib at 91%, followed by ribociclib at 8%, and abemaciclib at 1%.

After a median follow-up of 37.3 months, the median duration of CDK4/6i exposure was 24.6 months in the first-line CDK4/6i group and 8.1 months in the second-line CDK4/6i group.

The median PFS during first-line therapy was 24.7 months in the first-line CDK4/6i group and 16.1 months in the second-line CDK4/6i group (hazard ratio, 0.59; P < .0001), which was consistent with the results seen in CDK4/6i pivotal trials in the first-line setting, according to Dr. Sonke. However, PFS after two lines of therapy was not significantly different between the groups (31.0 months vs. 26.8 months, respectively; HR, 0.87; P =.10).

The safety profile was similar to what had been seen in previous trials with respect to adverse events like bone marrow and liver function abnormalities and fatigue, but there were 42% more grade 3 or higher adverse events in the first-line CDK4/6i group than in the second-line CDK4/6i group. Dr. Sonke estimated that the increase in costs related to adverse events amounted to about $200,000 per patient receiving CDK4/6i as first line.

There were no significant differences between the two groups in quality of life measurement.

Subgroup analyses of patient categories including prior adjuvant or neoadjuvant chemotherapy or endocrine therapy, de novo metastatic disease, visceral disease, bone-only disease, and treatment with palbociclib or ribociclib showed no difference in outcome for first- versus second-line CDK4/6i treatment.
 

Are CDK4/6i costs and side effects worth it?

The findings challenge the need for using CDK4/6 inhibitors as first-line treatment in this population, according to Dr. Sonke, who also raised the following related questions.

“If you were a patient, would you consider a treatment that offers no improvement in quality of life and does not improve overall survival? As a doctor or nurse, would you recommend such a treatment to your patient that nearly doubles the incidence of side effects? And if you were responsible for covering the costs of this treatment, whether as an individual or health care insurance, would you consider it worth $200,000?”

For many patients, particularly in the first line setting where resistance mechanisms are less prevalent, endocrine therapy alone remains an excellent option,” said Dr. Sonke during his presentation.

During the discussion portion of the session, Daniel Stover, MD, who is an associate professor of translational therapeutics at Ohio State University Comprehensive Cancer Center, Columbus, pointed out that the lack of differences in the subanalyses leaves little guidance for physicians.

“We really have a limited signal on who can delay CDK4/6 inhibitors. I think one of the most important outcomes of this study is the focus on the patient, as there were substantially fewer adverse events and of course we need to think about financial toxicity as well,” he said. “I think one of the things that is perhaps most exciting to think about is who are the very good risk patients who can delay CDK4/6 inhibitor [therapy]? I think for the majority of patients, endocrine therapy plus CDK4/6 inhibitor is still the appropriate treatment, but I would argue we need additional biomarkers, be it RNA-based biomarkers, novel PET imaging, or perhaps [circulating tumor] DNA dynamics.”
 

Do cost savings and reduced side effects outweigh first-line PFS benefit?

During the question-and-answer session, William Sikov, MD, spoke up from the audience in support of Dr. Sonke’s conclusions.

“Clearly there are still patients who benefit from that approach, but I think that we have reached an inflection point: I posit that the question has now changed. [We should not ask] why a certain patient should not receive a CDK4/6 inhibitor, but why a certain patient should receive a CDK4/6 inhibitor in the first-line setting,” said Dr. Sikov, who is professor of medicine at Brown University, Providence, R.I.

Dr. Sonke agreed that first-line CDK4/6i is appropriate for some patients, and later echoed the need for biomarkers, but he said that researchers have so far had little luck in identifying any.

“Of course, it’s a shared decision-making between the patient and a doctor, but I think the baseline would be for all of us to consider first line single-agent endocrine therapy,” he said.

Session comoderator Michael Danso, MD, praised the trial but questioned whether the strategy would be adopted in places like the United States, where cost savings is not a major emphasis.

“Progression-free survival is so significant in the first line setting that I can’t imagine that many oncologists in the U.S. will adopt this approach. The other thing is that this was [almost] all palbociclib, so the question remains, would having a different cyclin dependent kinase inhibitor result in the same results? I think the jury’s still out,” said Dr. Danso, who is the research director at Virginia Oncology Associates, Norfolk.

The study was funded by the Dutch government and Dutch Health Insurers. Dr. Sonke has consulted for or advised Biovica, Novartis, and Seagen. He has received research support through his institution from Agendia, AstraZeneca/Merck, Merck Sharp & Dohme, Novartis, Roche, and Seagen. Dr. Sikov has been a speaker for Lilly. Dr. Danso has received honoraria from Amgen and has consulted or advised Immunomedics, Novartis, Pfizer, and Seagen.

CHICAGO – Patients with advanced estrogen receptor–positive, HER-negative breast cancer receiving first line treatment with CDK4/6 inhibitors (CDK4/6i) experienced no progression-free survival (PFS) or overall survival (OS) advantages over those who received the same therapy as second line treatment.

That was the conclusion of the phase 3 SONIA study, which was presented at the annual meeting of the American Society of Clinical Oncology.

The benefit from first line therapy is not maintained and almost completely disappears when patients in the control arm cross over to receive CDK4/6 inhibition in second line,” said Gabe Sonke, MD, PhD, during his presentation at the meeting.

CDK4/6 inhibitors have shown benefit in both the first-and second-line setting, according to Dr. Sonke, who is a medical oncologist at the Netherlands Cancer Institute, Amsterdam. He added that most guidelines suggest use of CDK4/6 inhibitors in the first line, but there hasn’t been a direct comparison between use in the first and second line.

“Many patients do very well on endocrine therapy alone [in the first line]. Combination treatment leads to a higher risk of the emergence of resistant patterns such as ESR1 mutations, and CDK4/6 inhibitors also come with added costs and toxicities. Given the absence of comparative data between first line and second line, we designed the SONIA trial,” said Dr. Sonke.
 

Study methods and results

The researchers recruited 1,050 pre- and postmenopausal women who were randomized to a nonsteroidal AI in the first line followed by second-line CDK4/6i plus the estrogen receptor antagonist fulvestrant, or a nonsteroidal AI plus a CDK4/6i in the first line and fulvestrant in the second line. The most commonly used CDK4/6i was palbociclib at 91%, followed by ribociclib at 8%, and abemaciclib at 1%.

After a median follow-up of 37.3 months, the median duration of CDK4/6i exposure was 24.6 months in the first-line CDK4/6i group and 8.1 months in the second-line CDK4/6i group.

The median PFS during first-line therapy was 24.7 months in the first-line CDK4/6i group and 16.1 months in the second-line CDK4/6i group (hazard ratio, 0.59; P < .0001), which was consistent with the results seen in CDK4/6i pivotal trials in the first-line setting, according to Dr. Sonke. However, PFS after two lines of therapy was not significantly different between the groups (31.0 months vs. 26.8 months, respectively; HR, 0.87; P =.10).

The safety profile was similar to what had been seen in previous trials with respect to adverse events like bone marrow and liver function abnormalities and fatigue, but there were 42% more grade 3 or higher adverse events in the first-line CDK4/6i group than in the second-line CDK4/6i group. Dr. Sonke estimated that the increase in costs related to adverse events amounted to about $200,000 per patient receiving CDK4/6i as first line.

There were no significant differences between the two groups in quality of life measurement.

Subgroup analyses of patient categories including prior adjuvant or neoadjuvant chemotherapy or endocrine therapy, de novo metastatic disease, visceral disease, bone-only disease, and treatment with palbociclib or ribociclib showed no difference in outcome for first- versus second-line CDK4/6i treatment.
 

Are CDK4/6i costs and side effects worth it?

The findings challenge the need for using CDK4/6 inhibitors as first-line treatment in this population, according to Dr. Sonke, who also raised the following related questions.

“If you were a patient, would you consider a treatment that offers no improvement in quality of life and does not improve overall survival? As a doctor or nurse, would you recommend such a treatment to your patient that nearly doubles the incidence of side effects? And if you were responsible for covering the costs of this treatment, whether as an individual or health care insurance, would you consider it worth $200,000?”

For many patients, particularly in the first line setting where resistance mechanisms are less prevalent, endocrine therapy alone remains an excellent option,” said Dr. Sonke during his presentation.

During the discussion portion of the session, Daniel Stover, MD, who is an associate professor of translational therapeutics at Ohio State University Comprehensive Cancer Center, Columbus, pointed out that the lack of differences in the subanalyses leaves little guidance for physicians.

“We really have a limited signal on who can delay CDK4/6 inhibitors. I think one of the most important outcomes of this study is the focus on the patient, as there were substantially fewer adverse events and of course we need to think about financial toxicity as well,” he said. “I think one of the things that is perhaps most exciting to think about is who are the very good risk patients who can delay CDK4/6 inhibitor [therapy]? I think for the majority of patients, endocrine therapy plus CDK4/6 inhibitor is still the appropriate treatment, but I would argue we need additional biomarkers, be it RNA-based biomarkers, novel PET imaging, or perhaps [circulating tumor] DNA dynamics.”
 

Do cost savings and reduced side effects outweigh first-line PFS benefit?

During the question-and-answer session, William Sikov, MD, spoke up from the audience in support of Dr. Sonke’s conclusions.

“Clearly there are still patients who benefit from that approach, but I think that we have reached an inflection point: I posit that the question has now changed. [We should not ask] why a certain patient should not receive a CDK4/6 inhibitor, but why a certain patient should receive a CDK4/6 inhibitor in the first-line setting,” said Dr. Sikov, who is professor of medicine at Brown University, Providence, R.I.

Dr. Sonke agreed that first-line CDK4/6i is appropriate for some patients, and later echoed the need for biomarkers, but he said that researchers have so far had little luck in identifying any.

“Of course, it’s a shared decision-making between the patient and a doctor, but I think the baseline would be for all of us to consider first line single-agent endocrine therapy,” he said.

Session comoderator Michael Danso, MD, praised the trial but questioned whether the strategy would be adopted in places like the United States, where cost savings is not a major emphasis.

“Progression-free survival is so significant in the first line setting that I can’t imagine that many oncologists in the U.S. will adopt this approach. The other thing is that this was [almost] all palbociclib, so the question remains, would having a different cyclin dependent kinase inhibitor result in the same results? I think the jury’s still out,” said Dr. Danso, who is the research director at Virginia Oncology Associates, Norfolk.

The study was funded by the Dutch government and Dutch Health Insurers. Dr. Sonke has consulted for or advised Biovica, Novartis, and Seagen. He has received research support through his institution from Agendia, AstraZeneca/Merck, Merck Sharp & Dohme, Novartis, Roche, and Seagen. Dr. Sikov has been a speaker for Lilly. Dr. Danso has received honoraria from Amgen and has consulted or advised Immunomedics, Novartis, Pfizer, and Seagen.

CHICAGO – Patients with advanced estrogen receptor–positive, HER-negative breast cancer receiving first line treatment with CDK4/6 inhibitors (CDK4/6i) experienced no progression-free survival (PFS) or overall survival (OS) advantages over those who received the same therapy as second line treatment.

That was the conclusion of the phase 3 SONIA study, which was presented at the annual meeting of the American Society of Clinical Oncology.

The benefit from first line therapy is not maintained and almost completely disappears when patients in the control arm cross over to receive CDK4/6 inhibition in second line,” said Gabe Sonke, MD, PhD, during his presentation at the meeting.

CDK4/6 inhibitors have shown benefit in both the first-and second-line setting, according to Dr. Sonke, who is a medical oncologist at the Netherlands Cancer Institute, Amsterdam. He added that most guidelines suggest use of CDK4/6 inhibitors in the first line, but there hasn’t been a direct comparison between use in the first and second line.

“Many patients do very well on endocrine therapy alone [in the first line]. Combination treatment leads to a higher risk of the emergence of resistant patterns such as ESR1 mutations, and CDK4/6 inhibitors also come with added costs and toxicities. Given the absence of comparative data between first line and second line, we designed the SONIA trial,” said Dr. Sonke.
 

Study methods and results

The researchers recruited 1,050 pre- and postmenopausal women who were randomized to a nonsteroidal AI in the first line followed by second-line CDK4/6i plus the estrogen receptor antagonist fulvestrant, or a nonsteroidal AI plus a CDK4/6i in the first line and fulvestrant in the second line. The most commonly used CDK4/6i was palbociclib at 91%, followed by ribociclib at 8%, and abemaciclib at 1%.

After a median follow-up of 37.3 months, the median duration of CDK4/6i exposure was 24.6 months in the first-line CDK4/6i group and 8.1 months in the second-line CDK4/6i group.

The median PFS during first-line therapy was 24.7 months in the first-line CDK4/6i group and 16.1 months in the second-line CDK4/6i group (hazard ratio, 0.59; P < .0001), which was consistent with the results seen in CDK4/6i pivotal trials in the first-line setting, according to Dr. Sonke. However, PFS after two lines of therapy was not significantly different between the groups (31.0 months vs. 26.8 months, respectively; HR, 0.87; P =.10).

The safety profile was similar to what had been seen in previous trials with respect to adverse events like bone marrow and liver function abnormalities and fatigue, but there were 42% more grade 3 or higher adverse events in the first-line CDK4/6i group than in the second-line CDK4/6i group. Dr. Sonke estimated that the increase in costs related to adverse events amounted to about $200,000 per patient receiving CDK4/6i as first line.

There were no significant differences between the two groups in quality of life measurement.

Subgroup analyses of patient categories including prior adjuvant or neoadjuvant chemotherapy or endocrine therapy, de novo metastatic disease, visceral disease, bone-only disease, and treatment with palbociclib or ribociclib showed no difference in outcome for first- versus second-line CDK4/6i treatment.
 

Are CDK4/6i costs and side effects worth it?

The findings challenge the need for using CDK4/6 inhibitors as first-line treatment in this population, according to Dr. Sonke, who also raised the following related questions.

“If you were a patient, would you consider a treatment that offers no improvement in quality of life and does not improve overall survival? As a doctor or nurse, would you recommend such a treatment to your patient that nearly doubles the incidence of side effects? And if you were responsible for covering the costs of this treatment, whether as an individual or health care insurance, would you consider it worth $200,000?”

For many patients, particularly in the first line setting where resistance mechanisms are less prevalent, endocrine therapy alone remains an excellent option,” said Dr. Sonke during his presentation.

During the discussion portion of the session, Daniel Stover, MD, who is an associate professor of translational therapeutics at Ohio State University Comprehensive Cancer Center, Columbus, pointed out that the lack of differences in the subanalyses leaves little guidance for physicians.

“We really have a limited signal on who can delay CDK4/6 inhibitors. I think one of the most important outcomes of this study is the focus on the patient, as there were substantially fewer adverse events and of course we need to think about financial toxicity as well,” he said. “I think one of the things that is perhaps most exciting to think about is who are the very good risk patients who can delay CDK4/6 inhibitor [therapy]? I think for the majority of patients, endocrine therapy plus CDK4/6 inhibitor is still the appropriate treatment, but I would argue we need additional biomarkers, be it RNA-based biomarkers, novel PET imaging, or perhaps [circulating tumor] DNA dynamics.”
 

Do cost savings and reduced side effects outweigh first-line PFS benefit?

During the question-and-answer session, William Sikov, MD, spoke up from the audience in support of Dr. Sonke’s conclusions.

“Clearly there are still patients who benefit from that approach, but I think that we have reached an inflection point: I posit that the question has now changed. [We should not ask] why a certain patient should not receive a CDK4/6 inhibitor, but why a certain patient should receive a CDK4/6 inhibitor in the first-line setting,” said Dr. Sikov, who is professor of medicine at Brown University, Providence, R.I.

Dr. Sonke agreed that first-line CDK4/6i is appropriate for some patients, and later echoed the need for biomarkers, but he said that researchers have so far had little luck in identifying any.

“Of course, it’s a shared decision-making between the patient and a doctor, but I think the baseline would be for all of us to consider first line single-agent endocrine therapy,” he said.

Session comoderator Michael Danso, MD, praised the trial but questioned whether the strategy would be adopted in places like the United States, where cost savings is not a major emphasis.

“Progression-free survival is so significant in the first line setting that I can’t imagine that many oncologists in the U.S. will adopt this approach. The other thing is that this was [almost] all palbociclib, so the question remains, would having a different cyclin dependent kinase inhibitor result in the same results? I think the jury’s still out,” said Dr. Danso, who is the research director at Virginia Oncology Associates, Norfolk.

The study was funded by the Dutch government and Dutch Health Insurers. Dr. Sonke has consulted for or advised Biovica, Novartis, and Seagen. He has received research support through his institution from Agendia, AstraZeneca/Merck, Merck Sharp & Dohme, Novartis, Roche, and Seagen. Dr. Sikov has been a speaker for Lilly. Dr. Danso has received honoraria from Amgen and has consulted or advised Immunomedics, Novartis, Pfizer, and Seagen.

After Canada legalized recreational cannabis in 2018, the rate of acute care for cannabis use during pregnancy in Ontario nearly doubled, data indicate.

A population-based study shows that the rate of cannabis-related acute care use during pregnancy increased from 11 per 100,000 pregnancies before legalization to 20 per 100,000 pregnancies afterward: an increase of 82%. Absolute increases were small, however.

“Our findings are consistent with studies highlighting that cannabis use during pregnancy has been increasing in North America, and this study suggests that cannabis legalization might contribute to and accelerate such trends,” study author Daniel Myran, MD, MPH, a public health and preventive medicine physician at the University of Ottawa in Ontario, said in an interview.

University of Ottawa

The study was published online in the Canadian Medical Association Journal.
 

Risks for newborns  

In a 2019 study, 7% of U.S. women reported using cannabis during pregnancy during 2016-2017, which was double the rate of 3.4% for 2002-2003.

Dr. Myran and colleagues hypothesized that legalizing nonmedical cannabis has affected the drug’s use during pregnancy in Ontario. “We also hypothesized that hospital care for cannabis use would be associated with adverse neonatal outcomes, even after adjusting for other important risk factors that may differ between people with and without cannabis use,” he said.

The researchers’ repeated cross-sectional analysis evaluated changes in the number of pregnant people who received acute care from January 2015 to July 2021 among all patients who were eligible for Ontario’s public health coverage. The final study cohort included 691,242 pregnant patients, of whom 533 had at least one pregnancy with cannabis-related acute care visits. These mothers had a mean age of 24 years vs. 30 for their counterparts with no such visits.

Using segmented regression, the researchers compared changes in the quarterly rate of pregnant people with acute care related to cannabis use (the primary outcome) with those of acute care for mental health conditions or for noncannabis substance use (the control conditions).

“Severe morning sickness was a major risk factor for care in the emergency department or hospital for cannabis use,” said Dr. Myran. “Prior work has found that people who use cannabis during pregnancy often state that it was used to manage challenging symptoms of pregnancy such as morning sickness.”

Most acute care events (72.2%) were emergency department visits. The most common reasons for acute care were harmful cannabis use (57.6%), followed by cannabis dependence or withdrawal (21.5%), and acute cannabis intoxication (12.8%).

Compared with pregnancies without acute care, those with acute care related to cannabis had higher rates of adverse neonatal outcomes such as birth before 37 weeks’ gestational age (16.9% vs. 7.2%), birth weight at or below the bottom fifth percentile after adjustment for gestational age (12.1% vs. 4.4%), and neonatal intensive care unit admission in the first 28 days of life (31.5% vs. 13%).

An adjusted analysis found that patients younger than 35 years and those living in rural settings or the lowest-income neighborhoods had higher odds of acute cannabis-related care during pregnancy. Patients who received acute care for any substance use or schizophrenia before pregnancy or who accessed outpatient mental health services before pregnancy had higher risk for cannabis-related acute care during pregnancy. Mothers receiving acute care for cannabis also had higher risk for acute care for hyperemesis gravidarum during pregnancy (30.9%). 

The rate of acute care for other types of substance use such as alcohol and opioids did not change after cannabis legalization, and acute care for mental health conditions such as anxiety and depression during pregnancy declined by 14%, Dr. Myran noted.

“Physicians who care for pregnant people should consider increasing screening for cannabis use during pregnancy,” said Dr. Myran. “In addition, repeated nonstigmatizing screening and counseling may be indicated for higher-risk groups identified in the study, including pregnancies with severe morning sickness.”
 

The U.S. perspective

Commenting on the study, M. Camille Hoffman, MD, MSc, a maternal-fetal medicine specialist at the University of Colorado in Aurora, said that the findings likely indicate that legalization has made cannabis users less reluctant to come forward for urgent care. “They cannot really claim that this is equivalent to more use, just that more people are willing to present,” she said. Dr. Hoffman was not involved in the study.

The Canadian results do not align perfectly with what is seen in the United States. “It does suggest that there may be more cannabinoid hyperemesis being coded as hyperemesis gravidarum, which is a pregnancy-specific condition vs. a cannabis-dependence-related one,” said Dr. Hoffman.

Literature in the United States often includes tobacco use as a covariate, she added. “This study does not appear to do that,” she said. “Rather, it uses any substance use. Because of this, it is difficult to really know the contribution of cannabis to the adverse pregnancy outcomes vs. the combination of tobacco and cannabis.”

Finally, she pointed out, the proportion of those presenting for acute care for substance use in the 2 years before conception was 22% for acute care visits for cannabis vs 1% for no acute care visits. “This suggests to me that this was a highly vulnerable group before the legalization of cannabis as well. The overall absolute difference is nine in total per 100,000 – hardly enough to draw any real conclusions. Again, maybe those nine were simply more willing to come forth with concerns with cannabis being legal.” 

There is no known safe level of cannabis consumption, and its use by pregnant women has been linked to later neurodevelopmental issues in their offspring. A 2022 U.S. study suggested that cannabis exposure in the womb may leave children later in life at risk for autism, psychiatric disorders, and problematic substance abuse, particularly as they enter peak periods of vulnerability in late adolescence.

As to the impact of legalization in certain U.S. states, a 2022 study found that women perceived legalization to mean greater access to cannabis, increased acceptance of use, and greater trust in cannabis retailers. In line with Dr. Hoffman’s view, this study suggested that legalization made pregnant women more willing to discuss cannabis use during pregnancy honestly with their care providers.

In the United States, prenatal cannabis use is still included in definitions of child abuse or neglect and can lead to termination of parental rights, even in states with full legalization.

“These findings highlight the need for ongoing monitoring of markers of cannabis use during pregnancy after legalization,” said Dr. Myran. He also called for effective policies in regions with legal cannabis, such as increased warning labels on cannabis products.

This study was supported by the Canadian Institutes of Health Research and the University of Ottawa site of ICES, which is funded by an annual grant from the Ontario Ministry of Health and Ministry of Long-Term Care. Dr. Myran reports a speaker fee from McMaster University. Dr. Hoffman reports no relevant financial relationships.
 

A version of this article first appeared on Medscape.com.

After Canada legalized recreational cannabis in 2018, the rate of acute care for cannabis use during pregnancy in Ontario nearly doubled, data indicate.

A population-based study shows that the rate of cannabis-related acute care use during pregnancy increased from 11 per 100,000 pregnancies before legalization to 20 per 100,000 pregnancies afterward: an increase of 82%. Absolute increases were small, however.

“Our findings are consistent with studies highlighting that cannabis use during pregnancy has been increasing in North America, and this study suggests that cannabis legalization might contribute to and accelerate such trends,” study author Daniel Myran, MD, MPH, a public health and preventive medicine physician at the University of Ottawa in Ontario, said in an interview.

University of Ottawa

The study was published online in the Canadian Medical Association Journal.
 

Risks for newborns  

In a 2019 study, 7% of U.S. women reported using cannabis during pregnancy during 2016-2017, which was double the rate of 3.4% for 2002-2003.

Dr. Myran and colleagues hypothesized that legalizing nonmedical cannabis has affected the drug’s use during pregnancy in Ontario. “We also hypothesized that hospital care for cannabis use would be associated with adverse neonatal outcomes, even after adjusting for other important risk factors that may differ between people with and without cannabis use,” he said.

The researchers’ repeated cross-sectional analysis evaluated changes in the number of pregnant people who received acute care from January 2015 to July 2021 among all patients who were eligible for Ontario’s public health coverage. The final study cohort included 691,242 pregnant patients, of whom 533 had at least one pregnancy with cannabis-related acute care visits. These mothers had a mean age of 24 years vs. 30 for their counterparts with no such visits.

Using segmented regression, the researchers compared changes in the quarterly rate of pregnant people with acute care related to cannabis use (the primary outcome) with those of acute care for mental health conditions or for noncannabis substance use (the control conditions).

“Severe morning sickness was a major risk factor for care in the emergency department or hospital for cannabis use,” said Dr. Myran. “Prior work has found that people who use cannabis during pregnancy often state that it was used to manage challenging symptoms of pregnancy such as morning sickness.”

Most acute care events (72.2%) were emergency department visits. The most common reasons for acute care were harmful cannabis use (57.6%), followed by cannabis dependence or withdrawal (21.5%), and acute cannabis intoxication (12.8%).

Compared with pregnancies without acute care, those with acute care related to cannabis had higher rates of adverse neonatal outcomes such as birth before 37 weeks’ gestational age (16.9% vs. 7.2%), birth weight at or below the bottom fifth percentile after adjustment for gestational age (12.1% vs. 4.4%), and neonatal intensive care unit admission in the first 28 days of life (31.5% vs. 13%).

An adjusted analysis found that patients younger than 35 years and those living in rural settings or the lowest-income neighborhoods had higher odds of acute cannabis-related care during pregnancy. Patients who received acute care for any substance use or schizophrenia before pregnancy or who accessed outpatient mental health services before pregnancy had higher risk for cannabis-related acute care during pregnancy. Mothers receiving acute care for cannabis also had higher risk for acute care for hyperemesis gravidarum during pregnancy (30.9%). 

The rate of acute care for other types of substance use such as alcohol and opioids did not change after cannabis legalization, and acute care for mental health conditions such as anxiety and depression during pregnancy declined by 14%, Dr. Myran noted.

“Physicians who care for pregnant people should consider increasing screening for cannabis use during pregnancy,” said Dr. Myran. “In addition, repeated nonstigmatizing screening and counseling may be indicated for higher-risk groups identified in the study, including pregnancies with severe morning sickness.”