Rheumatology News

TOPLINE: 

In patients with giant cell arteritis (GCA), intravenous methylprednisolone compared with oral glucocorticoids alone does not improve visual acuity and increases the risk for diabetes within the first year. Survival rates do not differ with these two treatments.

METHODOLOGY:

• Researchers conducted a population-based retrospective study at three centers in Sweden to assess the clinical characteristics, treatment-related toxicity, and mortality in patients with GCA who were receiving high-dose intravenous methylprednisolone.
• A total of 419 patients with biopsy-confirmed GCA (mean age at diagnosis, 75 years; 69% women) diagnosed from 2004 to 2019 were included.
• Patients were treated with either intravenous methylprednisolone (n = 111) at a dose of 500-1000 mg per day for 3 consecutive days or oral glucocorticoids alone (n = 308).
• Ischemic visual complications considered to indicate visual involvement were confirmed by an ophthalmologist, and data on visual acuity were collected from ophthalmologic clinic records at initial consultations and follow-up at 3-18 months.

TAKEAWAY:

• Despite a tendency toward improvement, no significant difference in visual acuity was observed with intravenous methylprednisolone compared with oral glucocorticoids.
• Patients treated with intravenous methylprednisolone had a higher risk for newly diagnosed diabetes within a year of GCA diagnosis (odds ratio [OR], 2.59; P = .01).
• The risk for diabetes remained elevated even after adjustment for the cumulative oral glucocorticoid dose at 3 months (adjusted OR, 3.30; P = .01).
• Survival rates did not significantly differ between the treatment groups over a mean follow-up of 6.6 years.

IN PRACTICE:

“In this study on the use of intravenous methylprednisolone treatment in GCA, we found no evidence of a beneficial effect in improving visual acuity or enabling more rapid tapering of the oral glucocorticoid dose,” the authors wrote. “The use of IVMP [intravenous methylprednisolone] was associated with an increased risk of diabetes during the first year compared with oral GC [glucocorticoid], raising questions about the value of IVMP in GCA treatment.”

SOURCE:

The study, led by Hampus Henningson, Department of Clinical Sciences, Rheumatology, Lund University, Lund, Sweden, was published online in Rheumatology.

LIMITATIONS: 

The retrospective nature of the study may have resulted in missing data and difficulty in accurately quantifying the cumulative glucocorticoid doses. The study did not validate the diagnoses of comorbidities but relied solely on diagnostic codes.

DISCLOSURES:

This study was supported by the Swedish Research Council, Swedish Rheumatism Association, Swedish Medical Society, Alfred Österlund’s Foundation, and King Gustaf V’s 80-year foundation. The authors declared no conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

TOPLINE: 

In patients with giant cell arteritis (GCA), intravenous methylprednisolone compared with oral glucocorticoids alone does not improve visual acuity and increases the risk for diabetes within the first year. Survival rates do not differ with these two treatments.

METHODOLOGY:

• Researchers conducted a population-based retrospective study at three centers in Sweden to assess the clinical characteristics, treatment-related toxicity, and mortality in patients with GCA who were receiving high-dose intravenous methylprednisolone.
• A total of 419 patients with biopsy-confirmed GCA (mean age at diagnosis, 75 years; 69% women) diagnosed from 2004 to 2019 were included.
• Patients were treated with either intravenous methylprednisolone (n = 111) at a dose of 500-1000 mg per day for 3 consecutive days or oral glucocorticoids alone (n = 308).
• Ischemic visual complications considered to indicate visual involvement were confirmed by an ophthalmologist, and data on visual acuity were collected from ophthalmologic clinic records at initial consultations and follow-up at 3-18 months.

TAKEAWAY:

• Despite a tendency toward improvement, no significant difference in visual acuity was observed with intravenous methylprednisolone compared with oral glucocorticoids.
• Patients treated with intravenous methylprednisolone had a higher risk for newly diagnosed diabetes within a year of GCA diagnosis (odds ratio [OR], 2.59; P = .01).
• The risk for diabetes remained elevated even after adjustment for the cumulative oral glucocorticoid dose at 3 months (adjusted OR, 3.30; P = .01).
• Survival rates did not significantly differ between the treatment groups over a mean follow-up of 6.6 years.

IN PRACTICE:

“In this study on the use of intravenous methylprednisolone treatment in GCA, we found no evidence of a beneficial effect in improving visual acuity or enabling more rapid tapering of the oral glucocorticoid dose,” the authors wrote. “The use of IVMP [intravenous methylprednisolone] was associated with an increased risk of diabetes during the first year compared with oral GC [glucocorticoid], raising questions about the value of IVMP in GCA treatment.”

SOURCE:

The study, led by Hampus Henningson, Department of Clinical Sciences, Rheumatology, Lund University, Lund, Sweden, was published online in Rheumatology.

LIMITATIONS: 

The retrospective nature of the study may have resulted in missing data and difficulty in accurately quantifying the cumulative glucocorticoid doses. The study did not validate the diagnoses of comorbidities but relied solely on diagnostic codes.

DISCLOSURES:

This study was supported by the Swedish Research Council, Swedish Rheumatism Association, Swedish Medical Society, Alfred Österlund’s Foundation, and King Gustaf V’s 80-year foundation. The authors declared no conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

TOPLINE: 

In patients with giant cell arteritis (GCA), intravenous methylprednisolone compared with oral glucocorticoids alone does not improve visual acuity and increases the risk for diabetes within the first year. Survival rates do not differ with these two treatments.

METHODOLOGY:

• Researchers conducted a population-based retrospective study at three centers in Sweden to assess the clinical characteristics, treatment-related toxicity, and mortality in patients with GCA who were receiving high-dose intravenous methylprednisolone.
• A total of 419 patients with biopsy-confirmed GCA (mean age at diagnosis, 75 years; 69% women) diagnosed from 2004 to 2019 were included.
• Patients were treated with either intravenous methylprednisolone (n = 111) at a dose of 500-1000 mg per day for 3 consecutive days or oral glucocorticoids alone (n = 308).
• Ischemic visual complications considered to indicate visual involvement were confirmed by an ophthalmologist, and data on visual acuity were collected from ophthalmologic clinic records at initial consultations and follow-up at 3-18 months.

TAKEAWAY:

• Despite a tendency toward improvement, no significant difference in visual acuity was observed with intravenous methylprednisolone compared with oral glucocorticoids.
• Patients treated with intravenous methylprednisolone had a higher risk for newly diagnosed diabetes within a year of GCA diagnosis (odds ratio [OR], 2.59; P = .01).
• The risk for diabetes remained elevated even after adjustment for the cumulative oral glucocorticoid dose at 3 months (adjusted OR, 3.30; P = .01).
• Survival rates did not significantly differ between the treatment groups over a mean follow-up of 6.6 years.

IN PRACTICE:

“In this study on the use of intravenous methylprednisolone treatment in GCA, we found no evidence of a beneficial effect in improving visual acuity or enabling more rapid tapering of the oral glucocorticoid dose,” the authors wrote. “The use of IVMP [intravenous methylprednisolone] was associated with an increased risk of diabetes during the first year compared with oral GC [glucocorticoid], raising questions about the value of IVMP in GCA treatment.”

SOURCE:

The study, led by Hampus Henningson, Department of Clinical Sciences, Rheumatology, Lund University, Lund, Sweden, was published online in Rheumatology.

LIMITATIONS: 

The retrospective nature of the study may have resulted in missing data and difficulty in accurately quantifying the cumulative glucocorticoid doses. The study did not validate the diagnoses of comorbidities but relied solely on diagnostic codes.

DISCLOSURES:

This study was supported by the Swedish Research Council, Swedish Rheumatism Association, Swedish Medical Society, Alfred Österlund’s Foundation, and King Gustaf V’s 80-year foundation. The authors declared no conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

Despite some recent progress in compensation equity, women in medicine continue to be paid significantly lower salaries than men.

According to the Female Compensation Report 2024 by Medscape, male doctors of any kind earned an average salary of about $400,000, whereas female doctors earned approximately $309,000 — a 29% gap.

The report analyzed survey data from 7000 practicing physicians who were recruited over a 4-month period starting in October 2023. The respondents comprised roughly 60% women representing over 29 specialties.

In the 2022 report, the pay gap between the genders was 32%. But some women in the field argued substantial headway is still needed.

“You can try and pick apart the data, but I’d say we’re not really making progress,” said Susan T. Hingle, MD, an internist in Illinois and president of the American Medical Women’s Association. “A decline by a couple of percentage points is not significantly addressing this pay gap that over a lifetime is huge, can be millions of dollars.”

The gender gap was narrower among female primary care physicians (PCPs) vs medical specialists. Female PCPs earned around $253,000 per year, whereas male PCPs earned about $295,000 per year. Hingle suggested that female PCPs may enjoy more pay equity because health systems have a harder time filling these positions.

On the other hand, the gap for specialists rose from 27% in 2022 to 31% in 2023. Differences in how aggressively women and men negotiate compensation packages may play a role, said Hingle.

“Taking negotiation out of the equation would be progress to me,” said Hingle.

Pay disparity did not appear to be the result of time spent on the job — female doctors reported an average of 49 work hours per week, whereas their male counterparts reported 50 work hours per week.

Meanwhile, the pay gap progressively worsened over time. Among doctors aged 28-34 years, men earned an average of $53,000 more than women. By ages 46-49, men earned an average of $157,000 more than women.

“I had to take my employer to court to get equal compensation, sad as it is to say,” said a hospitalist in North Carolina.

Nearly 60% of women surveyed felt they were not being paid fairly for their efforts, up from less than half reported in Medscape’s 2021 report. Hingle said that this figure may not only reflect sentiments about the compensation gap, but also less support on the job, including fewer physician assistants (PAs), nurses, and administrative staff.

“At my job, I do the work of multiple people,” said a survey respondent. “Junior resident, senior resident, social worker, nurse practitioner, PA — as well as try to be a teacher, researcher, [and] an excellent doctor and have the time to make patients feel as if they are not in a rush.”

Roughly 30% of women physicians said they would not choose to go into medicine again if given the chance compared with 26% of male physicians.

“Gender inequities in our profession have a direct impact,” said Shikha Jain, MD, an oncologist in Chicago and founder of the Women in Medicine nonprofit. “I think women in general don’t feel valued in the care they’re providing.” 

Jain cited bullying, harassment, and fewer opportunities for leadership and recognition as factors beyond pay that affect female physicians’ feelings of being valued.

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

Despite some recent progress in compensation equity, women in medicine continue to be paid significantly lower salaries than men.

According to the Female Compensation Report 2024 by Medscape, male doctors of any kind earned an average salary of about $400,000, whereas female doctors earned approximately $309,000 — a 29% gap.

The report analyzed survey data from 7000 practicing physicians who were recruited over a 4-month period starting in October 2023. The respondents comprised roughly 60% women representing over 29 specialties.

In the 2022 report, the pay gap between the genders was 32%. But some women in the field argued substantial headway is still needed.

“You can try and pick apart the data, but I’d say we’re not really making progress,” said Susan T. Hingle, MD, an internist in Illinois and president of the American Medical Women’s Association. “A decline by a couple of percentage points is not significantly addressing this pay gap that over a lifetime is huge, can be millions of dollars.”

The gender gap was narrower among female primary care physicians (PCPs) vs medical specialists. Female PCPs earned around $253,000 per year, whereas male PCPs earned about $295,000 per year. Hingle suggested that female PCPs may enjoy more pay equity because health systems have a harder time filling these positions.

On the other hand, the gap for specialists rose from 27% in 2022 to 31% in 2023. Differences in how aggressively women and men negotiate compensation packages may play a role, said Hingle.

“Taking negotiation out of the equation would be progress to me,” said Hingle.

Pay disparity did not appear to be the result of time spent on the job — female doctors reported an average of 49 work hours per week, whereas their male counterparts reported 50 work hours per week.

Meanwhile, the pay gap progressively worsened over time. Among doctors aged 28-34 years, men earned an average of $53,000 more than women. By ages 46-49, men earned an average of $157,000 more than women.

“I had to take my employer to court to get equal compensation, sad as it is to say,” said a hospitalist in North Carolina.

Nearly 60% of women surveyed felt they were not being paid fairly for their efforts, up from less than half reported in Medscape’s 2021 report. Hingle said that this figure may not only reflect sentiments about the compensation gap, but also less support on the job, including fewer physician assistants (PAs), nurses, and administrative staff.

“At my job, I do the work of multiple people,” said a survey respondent. “Junior resident, senior resident, social worker, nurse practitioner, PA — as well as try to be a teacher, researcher, [and] an excellent doctor and have the time to make patients feel as if they are not in a rush.”

Roughly 30% of women physicians said they would not choose to go into medicine again if given the chance compared with 26% of male physicians.

“Gender inequities in our profession have a direct impact,” said Shikha Jain, MD, an oncologist in Chicago and founder of the Women in Medicine nonprofit. “I think women in general don’t feel valued in the care they’re providing.” 

Jain cited bullying, harassment, and fewer opportunities for leadership and recognition as factors beyond pay that affect female physicians’ feelings of being valued.

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

Despite some recent progress in compensation equity, women in medicine continue to be paid significantly lower salaries than men.

According to the Female Compensation Report 2024 by Medscape, male doctors of any kind earned an average salary of about $400,000, whereas female doctors earned approximately $309,000 — a 29% gap.

The report analyzed survey data from 7000 practicing physicians who were recruited over a 4-month period starting in October 2023. The respondents comprised roughly 60% women representing over 29 specialties.

In the 2022 report, the pay gap between the genders was 32%. But some women in the field argued substantial headway is still needed.

“You can try and pick apart the data, but I’d say we’re not really making progress,” said Susan T. Hingle, MD, an internist in Illinois and president of the American Medical Women’s Association. “A decline by a couple of percentage points is not significantly addressing this pay gap that over a lifetime is huge, can be millions of dollars.”

The gender gap was narrower among female primary care physicians (PCPs) vs medical specialists. Female PCPs earned around $253,000 per year, whereas male PCPs earned about $295,000 per year. Hingle suggested that female PCPs may enjoy more pay equity because health systems have a harder time filling these positions.

On the other hand, the gap for specialists rose from 27% in 2022 to 31% in 2023. Differences in how aggressively women and men negotiate compensation packages may play a role, said Hingle.

“Taking negotiation out of the equation would be progress to me,” said Hingle.

Pay disparity did not appear to be the result of time spent on the job — female doctors reported an average of 49 work hours per week, whereas their male counterparts reported 50 work hours per week.

Meanwhile, the pay gap progressively worsened over time. Among doctors aged 28-34 years, men earned an average of $53,000 more than women. By ages 46-49, men earned an average of $157,000 more than women.

“I had to take my employer to court to get equal compensation, sad as it is to say,” said a hospitalist in North Carolina.

Nearly 60% of women surveyed felt they were not being paid fairly for their efforts, up from less than half reported in Medscape’s 2021 report. Hingle said that this figure may not only reflect sentiments about the compensation gap, but also less support on the job, including fewer physician assistants (PAs), nurses, and administrative staff.

“At my job, I do the work of multiple people,” said a survey respondent. “Junior resident, senior resident, social worker, nurse practitioner, PA — as well as try to be a teacher, researcher, [and] an excellent doctor and have the time to make patients feel as if they are not in a rush.”

Roughly 30% of women physicians said they would not choose to go into medicine again if given the chance compared with 26% of male physicians.

“Gender inequities in our profession have a direct impact,” said Shikha Jain, MD, an oncologist in Chicago and founder of the Women in Medicine nonprofit. “I think women in general don’t feel valued in the care they’re providing.” 

Jain cited bullying, harassment, and fewer opportunities for leadership and recognition as factors beyond pay that affect female physicians’ feelings of being valued.

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

Nearly two thirds of patients with rheumatic conditions switched to medical cannabis from medications such as nonsteroidal anti-inflammatory drugs (NSAIDs) and opioids, with the substitution being associated with greater self-reported improvement in symptoms than nonsubstitution.

METHODOLOGY:

• Researchers conducted a secondary analysis of a cross-sectional survey to investigate the prevalence of switching to medical cannabis from traditional medications in patients with rheumatic conditions from the United States and Canada.
• The survey included questions on current and past medical cannabis use, sociodemographic characteristics, medication taken and substituted, substance use, and patient-reported outcomes.
• Of the 1727 patients who completed the survey, 763 patients (mean age, 59 years; 84.1% women) reported current use of cannabis and were included in this analysis.
• Participants were asked if they had substituted any medications with medical cannabis and were sub-grouped accordingly.
• They also reported any changes in symptoms after initiating cannabis, the current and anticipated duration of medical cannabis use, methods of ingestion, cannabinoid content, and frequency of use.

TAKEAWAY:

• Overall, 62.5% reported substituting medical cannabis for certain medications, including NSAIDs (54.7%), opioids (48.6%), sleep aids (29.6%), muscle relaxants (25.2%), benzodiazepines (15.5%), and gabapentinoids (10.5%).
• The most common reasons given for substituting medical cannabis were fewer side effects (39%), better symptom control (27%), and fewer adverse effects (12%).
• Participants who substituted medical cannabis reported significant improvements in symptoms such as pain, sleep, joint stiffness, muscle spasms, and inflammation, and in overall health, compared with those who did not substitute it for medications.
• The substitution group was more likely to use inhalation methods (smoking and vaporizing) than the nonsubstitution group; they also used medical cannabis more frequently and preferred products containing delta-9-tetrahydrocannabinol.

IN PRACTICE:

“The changing legal status of cannabis has allowed a greater openness with more people willing to try cannabis for symptom relief. These encouraging results of medication reduction and favorable effect of [medical cannabis] require confirmation with more rigorous methods. At this time, survey information may be seen as a signal for effect, rather than sound evidence that could be applicable to those with musculoskeletal complaints in general,” the authors wrote. 

SOURCE:

The study was led by Kevin F. Boehnke, PhD, University of Michigan Medical School, Ann Arbor, and was published online in ACR Open Rheumatology.

LIMITATIONS: 

The cross-sectional nature of the study limited the determination of causality between medical cannabis use and symptom improvement. Moreover, the anonymous and self-reported nature of the survey at a single timepoint may have introduced recall bias. The sample predominantly consisted of older, White females, which may have limited the generalizability of the findings to other demographic groups.

DISCLOSURES:

Some authors received grant support from the National Institute on Drug Abuse and the National Institute of Arthritis and Musculoskeletal and Skin Diseases. Some others received payments, honoraria, grant funding, consulting fees, and travel support, and reported other ties with pharmaceutical companies and other institutions.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

Nearly two thirds of patients with rheumatic conditions switched to medical cannabis from medications such as nonsteroidal anti-inflammatory drugs (NSAIDs) and opioids, with the substitution being associated with greater self-reported improvement in symptoms than nonsubstitution.

METHODOLOGY:

• Researchers conducted a secondary analysis of a cross-sectional survey to investigate the prevalence of switching to medical cannabis from traditional medications in patients with rheumatic conditions from the United States and Canada.
• The survey included questions on current and past medical cannabis use, sociodemographic characteristics, medication taken and substituted, substance use, and patient-reported outcomes.
• Of the 1727 patients who completed the survey, 763 patients (mean age, 59 years; 84.1% women) reported current use of cannabis and were included in this analysis.
• Participants were asked if they had substituted any medications with medical cannabis and were sub-grouped accordingly.
• They also reported any changes in symptoms after initiating cannabis, the current and anticipated duration of medical cannabis use, methods of ingestion, cannabinoid content, and frequency of use.

TAKEAWAY:

• Overall, 62.5% reported substituting medical cannabis for certain medications, including NSAIDs (54.7%), opioids (48.6%), sleep aids (29.6%), muscle relaxants (25.2%), benzodiazepines (15.5%), and gabapentinoids (10.5%).
• The most common reasons given for substituting medical cannabis were fewer side effects (39%), better symptom control (27%), and fewer adverse effects (12%).
• Participants who substituted medical cannabis reported significant improvements in symptoms such as pain, sleep, joint stiffness, muscle spasms, and inflammation, and in overall health, compared with those who did not substitute it for medications.
• The substitution group was more likely to use inhalation methods (smoking and vaporizing) than the nonsubstitution group; they also used medical cannabis more frequently and preferred products containing delta-9-tetrahydrocannabinol.

IN PRACTICE:

“The changing legal status of cannabis has allowed a greater openness with more people willing to try cannabis for symptom relief. These encouraging results of medication reduction and favorable effect of [medical cannabis] require confirmation with more rigorous methods. At this time, survey information may be seen as a signal for effect, rather than sound evidence that could be applicable to those with musculoskeletal complaints in general,” the authors wrote. 

SOURCE:

The study was led by Kevin F. Boehnke, PhD, University of Michigan Medical School, Ann Arbor, and was published online in ACR Open Rheumatology.

LIMITATIONS: 

The cross-sectional nature of the study limited the determination of causality between medical cannabis use and symptom improvement. Moreover, the anonymous and self-reported nature of the survey at a single timepoint may have introduced recall bias. The sample predominantly consisted of older, White females, which may have limited the generalizability of the findings to other demographic groups.

DISCLOSURES:

Some authors received grant support from the National Institute on Drug Abuse and the National Institute of Arthritis and Musculoskeletal and Skin Diseases. Some others received payments, honoraria, grant funding, consulting fees, and travel support, and reported other ties with pharmaceutical companies and other institutions.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

Nearly two thirds of patients with rheumatic conditions switched to medical cannabis from medications such as nonsteroidal anti-inflammatory drugs (NSAIDs) and opioids, with the substitution being associated with greater self-reported improvement in symptoms than nonsubstitution.

METHODOLOGY:

• Researchers conducted a secondary analysis of a cross-sectional survey to investigate the prevalence of switching to medical cannabis from traditional medications in patients with rheumatic conditions from the United States and Canada.
• The survey included questions on current and past medical cannabis use, sociodemographic characteristics, medication taken and substituted, substance use, and patient-reported outcomes.
• Of the 1727 patients who completed the survey, 763 patients (mean age, 59 years; 84.1% women) reported current use of cannabis and were included in this analysis.
• Participants were asked if they had substituted any medications with medical cannabis and were sub-grouped accordingly.
• They also reported any changes in symptoms after initiating cannabis, the current and anticipated duration of medical cannabis use, methods of ingestion, cannabinoid content, and frequency of use.

TAKEAWAY:

• Overall, 62.5% reported substituting medical cannabis for certain medications, including NSAIDs (54.7%), opioids (48.6%), sleep aids (29.6%), muscle relaxants (25.2%), benzodiazepines (15.5%), and gabapentinoids (10.5%).
• The most common reasons given for substituting medical cannabis were fewer side effects (39%), better symptom control (27%), and fewer adverse effects (12%).
• Participants who substituted medical cannabis reported significant improvements in symptoms such as pain, sleep, joint stiffness, muscle spasms, and inflammation, and in overall health, compared with those who did not substitute it for medications.
• The substitution group was more likely to use inhalation methods (smoking and vaporizing) than the nonsubstitution group; they also used medical cannabis more frequently and preferred products containing delta-9-tetrahydrocannabinol.

IN PRACTICE:

“The changing legal status of cannabis has allowed a greater openness with more people willing to try cannabis for symptom relief. These encouraging results of medication reduction and favorable effect of [medical cannabis] require confirmation with more rigorous methods. At this time, survey information may be seen as a signal for effect, rather than sound evidence that could be applicable to those with musculoskeletal complaints in general,” the authors wrote. 

SOURCE:

The study was led by Kevin F. Boehnke, PhD, University of Michigan Medical School, Ann Arbor, and was published online in ACR Open Rheumatology.

LIMITATIONS: 

The cross-sectional nature of the study limited the determination of causality between medical cannabis use and symptom improvement. Moreover, the anonymous and self-reported nature of the survey at a single timepoint may have introduced recall bias. The sample predominantly consisted of older, White females, which may have limited the generalizability of the findings to other demographic groups.

DISCLOSURES:

Some authors received grant support from the National Institute on Drug Abuse and the National Institute of Arthritis and Musculoskeletal and Skin Diseases. Some others received payments, honoraria, grant funding, consulting fees, and travel support, and reported other ties with pharmaceutical companies and other institutions.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article first appeared on Medscape.com.

Methadone has been shown to be highly effective for opioid use disorder. So why is it still so difficult to prescribe in the United States and is that about to change?

A recent study from Canada adds to the growing body of evidence supporting methadone’s effectiveness in treating opioid use disorder and bolsters efforts to expand access in the United States by removing restrictive barriers.

This paper included more than 30,000 patients with opioid use disorder and showed those on methadone were almost 60% significantly less likely to stop treatment at 24 months than their peers assigned to buprenorphine/naloxone (adjusted hazard ratio [aHR], 1.58), with no difference in mortality risk (aHR, 0.57).

“In Canada, unlike the US, methadone and buprenorphine/naloxone are both available in office-based settings. Methadone really outperforms buprenorphine/naloxone in being able to retain people in treatment, which is our main goal and comes with a host of benefits,” Bohdan Nosyk, PhD, with Simon Fraser University in Burnaby, British Columbia, Canada, who worked on the study, said in an interview.

In addition, a recent systematic review and meta-analysis of relevant research involving more than 1 million patients with opioid use disorder also showed better treatment retention with methadone than with buprenorphine.

During the COVID-19 pandemic, relaxed methadone regulations, that included take-home medications, did not lead to an increase in overdoses. Instead, these changes improved treatment retention and patient experiences, highlighting the potential benefits of further deregulation.
 

‘Atrocious’ Outdated Policies

However, despite methadone’s proven efficacy and safety for opioid use disorder, it remains vastly underutilized because of outdated US policies restricting its use to opioid treatment programs (OTPs).

“It’s absolutely atrocious that methadone policies have not kept up with the evidence. If you look at other countries that have expanded their access to methadone, their overdose rates have fallen dramatically,” said Leslie Suen, MD, with the University of California, San Francisco, and coauthor of a recent JAMA Viewpoint on this topic.

“Methadone is a very good medication that’s been shown over and over to be very effective and safe,” Alan Leshner, PhD, past director of the National Institute on Drug Abuse, said in an interview.

“There is no reason why it couldn’t be administered through pharmacies or through physicians’ offices as long as it’s done in a controlled and careful way,” said Leshner.

Leshner chaired the committee that produced the 2019 report Medications for Opioid Use Disorder Save Lives.

“We learned during COVID that increasing the amount of take-home methadone and increasing access does not lead to an increase in deaths or an increase in overdose, so it’s hard to find a reason not to do it,” he said.
 

Change Finally on the Horizon?

Several recent and proposed policy changes could revolutionize methadone delivery in the United States.

In March 2022, in response to the pandemic, the Drug Enforcement Administration (DEA) allowed hospitals to dispense up to a 3-day supply of methadone (known as the 72-hour rule) to bridge care transitions without needing OTPs.

In April 2024, the Substance Abuse and Mental Health Services Administration and DEA codified many methadone and buprenorphine delivery flexibilities granted temporarily during the pandemic, including increased use of telehealth assessments and earlier access to take-home methadone doses.

Another contemporary policy change is expansion of the Americans with Disabilities Act mandating that patients taking medications for opioid use disorder, such as methadone, be able to continue treatment when transitioning to settings such as hospitals, jails, and skilled nursing facilities.

At the state level, California Governor Gavin Newsom recently signed a bill, effective immediately, that expands access to methadone treatment in his state.

On the horizon at the federal level is the Modernizing Opioid Treatment Access Act (MOTAA) — the bipartisan and bicameral bill introduced by Sen. Ed Markey (D-MA) and Sen. Rand Paul (R-KY), along with Rep. Donald Norcross (D-NJ) and Rep. Don Bacon, (R-NE) — that would allow methadone to be prescribed by addiction specialists and dispensed in community pharmacies.
 

An Ethical Imperative

“With only about 2000 OTP clinics clustered in urban areas, less than 25% of people who are diagnosed with opioid use disorder are actually able to access methadone,” Caty Simon, with the National Survivors Union, Greensboro, North Carolina, and coauthor of the JAMA Viewpoint, said in an interview.

While MOTAA represents a major step forward, limiting methadone prescribing to addiction specialists may not fully address the treatment gap, particularly in rural and underserved areas, Simon said.

To optimize methadone’s potential, she’d like to see further expansion of prescribing privileges to general healthcare providers.

“As someone with lived and living experience of opioid use and treatment, and somebody who works nationally and locally in organizations of people impacted by drug use, I know people in my area right now — marginalized people of color — who would have much better chances of survival if they were able to access methadone. If MOTAA passed tomorrow, we could save so many lives. There is an ethical imperative to pass it,” Simon said.

Leshner said he is “always very concerned about access, particularly for underserved populations, poor people, people living in rural areas. If you can access the medications you need, you’re in big trouble.”

Is this methadone’s moment? “I’m a little optimistic, but I haven’t seen the progress I would like to see,” Leshner said.

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

Methadone has been shown to be highly effective for opioid use disorder. So why is it still so difficult to prescribe in the United States and is that about to change?

A recent study from Canada adds to the growing body of evidence supporting methadone’s effectiveness in treating opioid use disorder and bolsters efforts to expand access in the United States by removing restrictive barriers.

This paper included more than 30,000 patients with opioid use disorder and showed those on methadone were almost 60% significantly less likely to stop treatment at 24 months than their peers assigned to buprenorphine/naloxone (adjusted hazard ratio [aHR], 1.58), with no difference in mortality risk (aHR, 0.57).

“In Canada, unlike the US, methadone and buprenorphine/naloxone are both available in office-based settings. Methadone really outperforms buprenorphine/naloxone in being able to retain people in treatment, which is our main goal and comes with a host of benefits,” Bohdan Nosyk, PhD, with Simon Fraser University in Burnaby, British Columbia, Canada, who worked on the study, said in an interview.

In addition, a recent systematic review and meta-analysis of relevant research involving more than 1 million patients with opioid use disorder also showed better treatment retention with methadone than with buprenorphine.

During the COVID-19 pandemic, relaxed methadone regulations, that included take-home medications, did not lead to an increase in overdoses. Instead, these changes improved treatment retention and patient experiences, highlighting the potential benefits of further deregulation.
 

‘Atrocious’ Outdated Policies

However, despite methadone’s proven efficacy and safety for opioid use disorder, it remains vastly underutilized because of outdated US policies restricting its use to opioid treatment programs (OTPs).

“It’s absolutely atrocious that methadone policies have not kept up with the evidence. If you look at other countries that have expanded their access to methadone, their overdose rates have fallen dramatically,” said Leslie Suen, MD, with the University of California, San Francisco, and coauthor of a recent JAMA Viewpoint on this topic.

“Methadone is a very good medication that’s been shown over and over to be very effective and safe,” Alan Leshner, PhD, past director of the National Institute on Drug Abuse, said in an interview.

“There is no reason why it couldn’t be administered through pharmacies or through physicians’ offices as long as it’s done in a controlled and careful way,” said Leshner.

Leshner chaired the committee that produced the 2019 report Medications for Opioid Use Disorder Save Lives.

“We learned during COVID that increasing the amount of take-home methadone and increasing access does not lead to an increase in deaths or an increase in overdose, so it’s hard to find a reason not to do it,” he said.
 

Change Finally on the Horizon?

Several recent and proposed policy changes could revolutionize methadone delivery in the United States.

In March 2022, in response to the pandemic, the Drug Enforcement Administration (DEA) allowed hospitals to dispense up to a 3-day supply of methadone (known as the 72-hour rule) to bridge care transitions without needing OTPs.

In April 2024, the Substance Abuse and Mental Health Services Administration and DEA codified many methadone and buprenorphine delivery flexibilities granted temporarily during the pandemic, including increased use of telehealth assessments and earlier access to take-home methadone doses.

Another contemporary policy change is expansion of the Americans with Disabilities Act mandating that patients taking medications for opioid use disorder, such as methadone, be able to continue treatment when transitioning to settings such as hospitals, jails, and skilled nursing facilities.

At the state level, California Governor Gavin Newsom recently signed a bill, effective immediately, that expands access to methadone treatment in his state.

On the horizon at the federal level is the Modernizing Opioid Treatment Access Act (MOTAA) — the bipartisan and bicameral bill introduced by Sen. Ed Markey (D-MA) and Sen. Rand Paul (R-KY), along with Rep. Donald Norcross (D-NJ) and Rep. Don Bacon, (R-NE) — that would allow methadone to be prescribed by addiction specialists and dispensed in community pharmacies.
 

An Ethical Imperative

“With only about 2000 OTP clinics clustered in urban areas, less than 25% of people who are diagnosed with opioid use disorder are actually able to access methadone,” Caty Simon, with the National Survivors Union, Greensboro, North Carolina, and coauthor of the JAMA Viewpoint, said in an interview.

While MOTAA represents a major step forward, limiting methadone prescribing to addiction specialists may not fully address the treatment gap, particularly in rural and underserved areas, Simon said.

To optimize methadone’s potential, she’d like to see further expansion of prescribing privileges to general healthcare providers.

“As someone with lived and living experience of opioid use and treatment, and somebody who works nationally and locally in organizations of people impacted by drug use, I know people in my area right now — marginalized people of color — who would have much better chances of survival if they were able to access methadone. If MOTAA passed tomorrow, we could save so many lives. There is an ethical imperative to pass it,” Simon said.

Leshner said he is “always very concerned about access, particularly for underserved populations, poor people, people living in rural areas. If you can access the medications you need, you’re in big trouble.”

Is this methadone’s moment? “I’m a little optimistic, but I haven’t seen the progress I would like to see,” Leshner said.

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

Methadone has been shown to be highly effective for opioid use disorder. So why is it still so difficult to prescribe in the United States and is that about to change?

A recent study from Canada adds to the growing body of evidence supporting methadone’s effectiveness in treating opioid use disorder and bolsters efforts to expand access in the United States by removing restrictive barriers.

This paper included more than 30,000 patients with opioid use disorder and showed those on methadone were almost 60% significantly less likely to stop treatment at 24 months than their peers assigned to buprenorphine/naloxone (adjusted hazard ratio [aHR], 1.58), with no difference in mortality risk (aHR, 0.57).

“In Canada, unlike the US, methadone and buprenorphine/naloxone are both available in office-based settings. Methadone really outperforms buprenorphine/naloxone in being able to retain people in treatment, which is our main goal and comes with a host of benefits,” Bohdan Nosyk, PhD, with Simon Fraser University in Burnaby, British Columbia, Canada, who worked on the study, said in an interview.

In addition, a recent systematic review and meta-analysis of relevant research involving more than 1 million patients with opioid use disorder also showed better treatment retention with methadone than with buprenorphine.

During the COVID-19 pandemic, relaxed methadone regulations, that included take-home medications, did not lead to an increase in overdoses. Instead, these changes improved treatment retention and patient experiences, highlighting the potential benefits of further deregulation.
 

‘Atrocious’ Outdated Policies

However, despite methadone’s proven efficacy and safety for opioid use disorder, it remains vastly underutilized because of outdated US policies restricting its use to opioid treatment programs (OTPs).

“It’s absolutely atrocious that methadone policies have not kept up with the evidence. If you look at other countries that have expanded their access to methadone, their overdose rates have fallen dramatically,” said Leslie Suen, MD, with the University of California, San Francisco, and coauthor of a recent JAMA Viewpoint on this topic.

“Methadone is a very good medication that’s been shown over and over to be very effective and safe,” Alan Leshner, PhD, past director of the National Institute on Drug Abuse, said in an interview.

“There is no reason why it couldn’t be administered through pharmacies or through physicians’ offices as long as it’s done in a controlled and careful way,” said Leshner.

Leshner chaired the committee that produced the 2019 report Medications for Opioid Use Disorder Save Lives.

“We learned during COVID that increasing the amount of take-home methadone and increasing access does not lead to an increase in deaths or an increase in overdose, so it’s hard to find a reason not to do it,” he said.
 

Change Finally on the Horizon?

Several recent and proposed policy changes could revolutionize methadone delivery in the United States.

In March 2022, in response to the pandemic, the Drug Enforcement Administration (DEA) allowed hospitals to dispense up to a 3-day supply of methadone (known as the 72-hour rule) to bridge care transitions without needing OTPs.

In April 2024, the Substance Abuse and Mental Health Services Administration and DEA codified many methadone and buprenorphine delivery flexibilities granted temporarily during the pandemic, including increased use of telehealth assessments and earlier access to take-home methadone doses.

Another contemporary policy change is expansion of the Americans with Disabilities Act mandating that patients taking medications for opioid use disorder, such as methadone, be able to continue treatment when transitioning to settings such as hospitals, jails, and skilled nursing facilities.

At the state level, California Governor Gavin Newsom recently signed a bill, effective immediately, that expands access to methadone treatment in his state.

On the horizon at the federal level is the Modernizing Opioid Treatment Access Act (MOTAA) — the bipartisan and bicameral bill introduced by Sen. Ed Markey (D-MA) and Sen. Rand Paul (R-KY), along with Rep. Donald Norcross (D-NJ) and Rep. Don Bacon, (R-NE) — that would allow methadone to be prescribed by addiction specialists and dispensed in community pharmacies.
 

An Ethical Imperative

“With only about 2000 OTP clinics clustered in urban areas, less than 25% of people who are diagnosed with opioid use disorder are actually able to access methadone,” Caty Simon, with the National Survivors Union, Greensboro, North Carolina, and coauthor of the JAMA Viewpoint, said in an interview.

While MOTAA represents a major step forward, limiting methadone prescribing to addiction specialists may not fully address the treatment gap, particularly in rural and underserved areas, Simon said.

To optimize methadone’s potential, she’d like to see further expansion of prescribing privileges to general healthcare providers.

“As someone with lived and living experience of opioid use and treatment, and somebody who works nationally and locally in organizations of people impacted by drug use, I know people in my area right now — marginalized people of color — who would have much better chances of survival if they were able to access methadone. If MOTAA passed tomorrow, we could save so many lives. There is an ethical imperative to pass it,” Simon said.

Leshner said he is “always very concerned about access, particularly for underserved populations, poor people, people living in rural areas. If you can access the medications you need, you’re in big trouble.”

Is this methadone’s moment? “I’m a little optimistic, but I haven’t seen the progress I would like to see,” Leshner said.

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

The Food and Drug Administration (FDA) has approved a 10-mg dose of Roxybond (Protega Pharmaceuticals), an opioid analgesic indicated for the management of severe pain in cases where other treatments are not well-tolerated or prove ineffective.

Roxybond, an immediate-release (IR) formulation of oxycodone hydrochloride, is made with Protega’s SentryBond technology, which makes it harder for people to crush, inject, or snort, according to the company.

In a statement from Protega, Paul Howe, the company’s chief commercial officer, said the drug meets an “unmet need for an IR opioid with abuse-deterrent technology that may reduce misuse and abuse while providing pain relief to medically appropriate patients when used as indicated.”

To determine the tablet’s ability to withstand manipulation, more than 2000 in vitro tests were conducted, according to the release. The findings indicate Roxybond reduces — but does not entirely negate — the potential for intranasal and intravenous abuse.

Roxybond was previously approved in 5-, 15-, and 30-mg doses. The 10 mg option provides clinicians with the ability to better modify side effects, manage titration, and provide precision care for patients on opioid therapy, according to Protega.

“For patients, the range of doses can provide better pain control, reduce the risk of side effects, and provide a smoother transition during dosing transitions,” the company stated.

Roxybond is contraindicated in patients with significant respiratory depression, acute or severe bronchial asthma, gastrointestinal obstruction, or hypersensitivity to oxycodone. The drug is not intended for long-term use unless otherwise determined by a clinician. Roxybond also is subject to the FDA’s Risk Evaluation and Mitigation Strategies for opioids.

“The development of Roxybond with SentryBond is a step forward in fighting the national epidemic of prescription opioid overdose,” said Eric Kinzler, PhD, vice president of medical and regulatory affairs for Protega, in a release. “Protega is dedicated to our mission to block the path to abuse and work with healthcare professionals across the continuum of care to reduce misuse and abuse.”

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

The Food and Drug Administration (FDA) has approved a 10-mg dose of Roxybond (Protega Pharmaceuticals), an opioid analgesic indicated for the management of severe pain in cases where other treatments are not well-tolerated or prove ineffective.

Roxybond, an immediate-release (IR) formulation of oxycodone hydrochloride, is made with Protega’s SentryBond technology, which makes it harder for people to crush, inject, or snort, according to the company.

In a statement from Protega, Paul Howe, the company’s chief commercial officer, said the drug meets an “unmet need for an IR opioid with abuse-deterrent technology that may reduce misuse and abuse while providing pain relief to medically appropriate patients when used as indicated.”

To determine the tablet’s ability to withstand manipulation, more than 2000 in vitro tests were conducted, according to the release. The findings indicate Roxybond reduces — but does not entirely negate — the potential for intranasal and intravenous abuse.

Roxybond was previously approved in 5-, 15-, and 30-mg doses. The 10 mg option provides clinicians with the ability to better modify side effects, manage titration, and provide precision care for patients on opioid therapy, according to Protega.

“For patients, the range of doses can provide better pain control, reduce the risk of side effects, and provide a smoother transition during dosing transitions,” the company stated.

Roxybond is contraindicated in patients with significant respiratory depression, acute or severe bronchial asthma, gastrointestinal obstruction, or hypersensitivity to oxycodone. The drug is not intended for long-term use unless otherwise determined by a clinician. Roxybond also is subject to the FDA’s Risk Evaluation and Mitigation Strategies for opioids.

“The development of Roxybond with SentryBond is a step forward in fighting the national epidemic of prescription opioid overdose,” said Eric Kinzler, PhD, vice president of medical and regulatory affairs for Protega, in a release. “Protega is dedicated to our mission to block the path to abuse and work with healthcare professionals across the continuum of care to reduce misuse and abuse.”

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

The Food and Drug Administration (FDA) has approved a 10-mg dose of Roxybond (Protega Pharmaceuticals), an opioid analgesic indicated for the management of severe pain in cases where other treatments are not well-tolerated or prove ineffective.

Roxybond, an immediate-release (IR) formulation of oxycodone hydrochloride, is made with Protega’s SentryBond technology, which makes it harder for people to crush, inject, or snort, according to the company.

In a statement from Protega, Paul Howe, the company’s chief commercial officer, said the drug meets an “unmet need for an IR opioid with abuse-deterrent technology that may reduce misuse and abuse while providing pain relief to medically appropriate patients when used as indicated.”

To determine the tablet’s ability to withstand manipulation, more than 2000 in vitro tests were conducted, according to the release. The findings indicate Roxybond reduces — but does not entirely negate — the potential for intranasal and intravenous abuse.

Roxybond was previously approved in 5-, 15-, and 30-mg doses. The 10 mg option provides clinicians with the ability to better modify side effects, manage titration, and provide precision care for patients on opioid therapy, according to Protega.

“For patients, the range of doses can provide better pain control, reduce the risk of side effects, and provide a smoother transition during dosing transitions,” the company stated.

Roxybond is contraindicated in patients with significant respiratory depression, acute or severe bronchial asthma, gastrointestinal obstruction, or hypersensitivity to oxycodone. The drug is not intended for long-term use unless otherwise determined by a clinician. Roxybond also is subject to the FDA’s Risk Evaluation and Mitigation Strategies for opioids.

“The development of Roxybond with SentryBond is a step forward in fighting the national epidemic of prescription opioid overdose,” said Eric Kinzler, PhD, vice president of medical and regulatory affairs for Protega, in a release. “Protega is dedicated to our mission to block the path to abuse and work with healthcare professionals across the continuum of care to reduce misuse and abuse.”

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

For people with severe symptomatic hip osteoarthritis, total hip replacement (THR) alleviates hip pain and improves function much more effectively than a resistance training program supervised by a physiotherapist, according to the results of a randomized controlled clinical trial. 

In the PROHIP study, the mean increases in Oxford Hip Scores from baseline to 6 months were 15.9 points for THR and 4.5 points for resistance training. The 11.4-point difference in scores was both statistically and clinically significant, the study’s investigators reported in The New England Journal of Medicine. 

“Our results are clear: Surgery is superior to exercise in patients who have hip osteoarthritis and indication for surgery, and now we have finally proven that with the highest level of evidence,” corresponding author Thomas Frydendal, PT, PhD, MSc, told this news organization.

Frydendal, who was involved in the study while working on his PhD at University Hospital of Southern Denmark – Lillebaelt Hospital, Vejle, Denmark, the primary center for the trial, is now a postdoctoral researcher at the Department of Clinical Medicine, Aarhus University, and Department of Orthopedic Surgery, Aarhus University Hospital.

“We believe that our findings are pretty robust,” Frydendal added. “I think if someone in the world conducts a trial similar to ours, they will find fairly close or consistent findings, no matter what type of exercise they choose.”

Charlotte Dahl, Lillebaelt Hospital–University Hospital of Southern Denmark, Vejle Hospital

The PROHIP Study

THR is routinely recommended for the management of severe hip osteoarthritis, but since there are no clinical trial data on the effectiveness of this procedure as compared with first-line treatment such as resistance training, the PROHIP study was conceived. 

The trial was conducted at four Danish orthopedic centers and designed as a superiority study, the hypothesis being that THR would be better at alleviating self-reported hip pain and improving hip function than resistance training. 

Of a possible 1474 individuals with a clinical suspicion of hip osteoarthritis, 791 were deemed eligible for inclusion in the trial. Inclusion criteria were being aged 50 years or older and having an indication for THR based on the presence of hip pain and clinical and radiographic findings.

However, the majority (86%) declined to enter the study, with almost half (43%) deciding to have a THR and enroll in a parallel observational cohort. This meant that only 110 (14%) individuals agreed to participate and underwent randomization, which does limit the study’s generalizability, the PROHIP investigators acknowledged. 
 

Design and Study Population

The change in Oxford Hip Score from baseline to 6 months was selected as the primary outcome measure based on the findings of a prior qualitative study. This 12-item, patient-reported outcome measure gives a score ranging from 0 to 48, with higher scores indicating less hip pain and better hip function. The estimated minimal clinically important difference is a change of 5 points. 

After a baseline assessment, 53 of 109 individuals were randomly assigned to undergo THR and 56 to participate in the resistance training program. Overall, the mean age of participants was 67.6 years, and half were women. The average duration of hip pain was a median of 1.7 years. 

The median time to receipt of the allocated treatment was 2.8 months in the THR group and 0.5 months in the resistance training group. 

Those allocated to the THR group also underwent a “fast track” program that involved patient education, pain management, and early mobilization. 

The resistance training group received 12 weeks of exercise supervised by a physiotherapist and then offered 12 weeks of additional exercise conducted on their own. The physiotherapist-supervised exercise sessions were held twice weekly and lasted for 1 hour. These started with a 10-minute warm-up on a stationary bike, followed by a standard set of resistance-based exercises that included a leg press, hip extension, hip flexion, and hip abduction. 
 

‘Reassuring’ Results

In a comment, consultant orthopedic surgeon Antony Palmer, MA, BMBCh, DPhil, said: “It’s reassuring that patients with advanced symptomatic osteoarthritis do well with hip replacements.”

THR does of course come with the potential risk for complications, but “the rate of these is what you’d expect for that procedure,” Palmer said, who works for the Nuffield Orthopaedic Centre, Oxford University Hospital NHS Foundation Trust, and is a senior clinical research fellow at Oxford University in England.

Dr. Palmer

Resistance Training Role 

A notable finding was that, at 6 months, five (9%) people assigned to the THR arm had not undergone surgery, and 12 (21%) people in the resistance training group had undergone a THR.

This could suggest two things, Palmer suggested in the interview. The first is that there could be a small proportion of people assigned to THR who may not need the operation and do well with exercise therapy. And, conversely, there may be those who would do well having the surgery without first going through the intermediate stage of physical therapy. 

It’s a suggestion that “maybe we’ve got to refine that a bit better and identify the patients that really do benefit from physiotherapy and who might not need hip replacement as a result,” Palmer said.

Or in other words, “should all patients undergo a program of physiotherapy before considering surgery?” he added.
 

Authors’ View

The PROHIP investigators conclude: “These results support current recommendations for the management of hip osteoarthritis and may be used to inform and guide shared decision making in clinical practice.”

Moreover, the results “do not oppose the use of resistance training as initial treatment,” says the authors. 

Frydendal highlighted in his interview that nearly three out of four of the patients reported not to have undertaken any type of supervised exercise before entry into the study, which is a first-line, guideline-recommended option.

“If a patient tells me, ‘I haven’t done any exercise previously,’ I’d recommend starting with completing a 6- to 12-week exercise program that is tailored to your individual needs and evaluate your symptoms afterward,” he said. 

“But we should refer the patient if our first-line treatment does not offer any improvements in the patient’s symptoms, as surgery with total hip replacement is clearly a really good treatment option,” Frydendal said.

The study was funded by the Danish Rheumatism Association, among other independent bodies. Frydendal and Palmer reported no relevant financial relationships. 
 

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

For people with severe symptomatic hip osteoarthritis, total hip replacement (THR) alleviates hip pain and improves function much more effectively than a resistance training program supervised by a physiotherapist, according to the results of a randomized controlled clinical trial. 

In the PROHIP study, the mean increases in Oxford Hip Scores from baseline to 6 months were 15.9 points for THR and 4.5 points for resistance training. The 11.4-point difference in scores was both statistically and clinically significant, the study’s investigators reported in The New England Journal of Medicine. 

“Our results are clear: Surgery is superior to exercise in patients who have hip osteoarthritis and indication for surgery, and now we have finally proven that with the highest level of evidence,” corresponding author Thomas Frydendal, PT, PhD, MSc, told this news organization.

Frydendal, who was involved in the study while working on his PhD at University Hospital of Southern Denmark – Lillebaelt Hospital, Vejle, Denmark, the primary center for the trial, is now a postdoctoral researcher at the Department of Clinical Medicine, Aarhus University, and Department of Orthopedic Surgery, Aarhus University Hospital.

“We believe that our findings are pretty robust,” Frydendal added. “I think if someone in the world conducts a trial similar to ours, they will find fairly close or consistent findings, no matter what type of exercise they choose.”

Charlotte Dahl, Lillebaelt Hospital–University Hospital of Southern Denmark, Vejle Hospital

The PROHIP Study

THR is routinely recommended for the management of severe hip osteoarthritis, but since there are no clinical trial data on the effectiveness of this procedure as compared with first-line treatment such as resistance training, the PROHIP study was conceived. 

The trial was conducted at four Danish orthopedic centers and designed as a superiority study, the hypothesis being that THR would be better at alleviating self-reported hip pain and improving hip function than resistance training. 

Of a possible 1474 individuals with a clinical suspicion of hip osteoarthritis, 791 were deemed eligible for inclusion in the trial. Inclusion criteria were being aged 50 years or older and having an indication for THR based on the presence of hip pain and clinical and radiographic findings.

However, the majority (86%) declined to enter the study, with almost half (43%) deciding to have a THR and enroll in a parallel observational cohort. This meant that only 110 (14%) individuals agreed to participate and underwent randomization, which does limit the study’s generalizability, the PROHIP investigators acknowledged. 
 

Design and Study Population

The change in Oxford Hip Score from baseline to 6 months was selected as the primary outcome measure based on the findings of a prior qualitative study. This 12-item, patient-reported outcome measure gives a score ranging from 0 to 48, with higher scores indicating less hip pain and better hip function. The estimated minimal clinically important difference is a change of 5 points. 

After a baseline assessment, 53 of 109 individuals were randomly assigned to undergo THR and 56 to participate in the resistance training program. Overall, the mean age of participants was 67.6 years, and half were women. The average duration of hip pain was a median of 1.7 years. 

The median time to receipt of the allocated treatment was 2.8 months in the THR group and 0.5 months in the resistance training group. 

Those allocated to the THR group also underwent a “fast track” program that involved patient education, pain management, and early mobilization. 

The resistance training group received 12 weeks of exercise supervised by a physiotherapist and then offered 12 weeks of additional exercise conducted on their own. The physiotherapist-supervised exercise sessions were held twice weekly and lasted for 1 hour. These started with a 10-minute warm-up on a stationary bike, followed by a standard set of resistance-based exercises that included a leg press, hip extension, hip flexion, and hip abduction. 
 

‘Reassuring’ Results

In a comment, consultant orthopedic surgeon Antony Palmer, MA, BMBCh, DPhil, said: “It’s reassuring that patients with advanced symptomatic osteoarthritis do well with hip replacements.”

THR does of course come with the potential risk for complications, but “the rate of these is what you’d expect for that procedure,” Palmer said, who works for the Nuffield Orthopaedic Centre, Oxford University Hospital NHS Foundation Trust, and is a senior clinical research fellow at Oxford University in England.

Dr. Palmer

Resistance Training Role 

A notable finding was that, at 6 months, five (9%) people assigned to the THR arm had not undergone surgery, and 12 (21%) people in the resistance training group had undergone a THR.

This could suggest two things, Palmer suggested in the interview. The first is that there could be a small proportion of people assigned to THR who may not need the operation and do well with exercise therapy. And, conversely, there may be those who would do well having the surgery without first going through the intermediate stage of physical therapy. 

It’s a suggestion that “maybe we’ve got to refine that a bit better and identify the patients that really do benefit from physiotherapy and who might not need hip replacement as a result,” Palmer said.

Or in other words, “should all patients undergo a program of physiotherapy before considering surgery?” he added.
 

Authors’ View

The PROHIP investigators conclude: “These results support current recommendations for the management of hip osteoarthritis and may be used to inform and guide shared decision making in clinical practice.”

Moreover, the results “do not oppose the use of resistance training as initial treatment,” says the authors. 

Frydendal highlighted in his interview that nearly three out of four of the patients reported not to have undertaken any type of supervised exercise before entry into the study, which is a first-line, guideline-recommended option.

“If a patient tells me, ‘I haven’t done any exercise previously,’ I’d recommend starting with completing a 6- to 12-week exercise program that is tailored to your individual needs and evaluate your symptoms afterward,” he said. 

“But we should refer the patient if our first-line treatment does not offer any improvements in the patient’s symptoms, as surgery with total hip replacement is clearly a really good treatment option,” Frydendal said.

The study was funded by the Danish Rheumatism Association, among other independent bodies. Frydendal and Palmer reported no relevant financial relationships. 
 

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

For people with severe symptomatic hip osteoarthritis, total hip replacement (THR) alleviates hip pain and improves function much more effectively than a resistance training program supervised by a physiotherapist, according to the results of a randomized controlled clinical trial. 

In the PROHIP study, the mean increases in Oxford Hip Scores from baseline to 6 months were 15.9 points for THR and 4.5 points for resistance training. The 11.4-point difference in scores was both statistically and clinically significant, the study’s investigators reported in The New England Journal of Medicine. 

“Our results are clear: Surgery is superior to exercise in patients who have hip osteoarthritis and indication for surgery, and now we have finally proven that with the highest level of evidence,” corresponding author Thomas Frydendal, PT, PhD, MSc, told this news organization.

Frydendal, who was involved in the study while working on his PhD at University Hospital of Southern Denmark – Lillebaelt Hospital, Vejle, Denmark, the primary center for the trial, is now a postdoctoral researcher at the Department of Clinical Medicine, Aarhus University, and Department of Orthopedic Surgery, Aarhus University Hospital.

“We believe that our findings are pretty robust,” Frydendal added. “I think if someone in the world conducts a trial similar to ours, they will find fairly close or consistent findings, no matter what type of exercise they choose.”

Charlotte Dahl, Lillebaelt Hospital–University Hospital of Southern Denmark, Vejle Hospital

The PROHIP Study

THR is routinely recommended for the management of severe hip osteoarthritis, but since there are no clinical trial data on the effectiveness of this procedure as compared with first-line treatment such as resistance training, the PROHIP study was conceived. 

The trial was conducted at four Danish orthopedic centers and designed as a superiority study, the hypothesis being that THR would be better at alleviating self-reported hip pain and improving hip function than resistance training. 

Of a possible 1474 individuals with a clinical suspicion of hip osteoarthritis, 791 were deemed eligible for inclusion in the trial. Inclusion criteria were being aged 50 years or older and having an indication for THR based on the presence of hip pain and clinical and radiographic findings.

However, the majority (86%) declined to enter the study, with almost half (43%) deciding to have a THR and enroll in a parallel observational cohort. This meant that only 110 (14%) individuals agreed to participate and underwent randomization, which does limit the study’s generalizability, the PROHIP investigators acknowledged. 
 

Design and Study Population

The change in Oxford Hip Score from baseline to 6 months was selected as the primary outcome measure based on the findings of a prior qualitative study. This 12-item, patient-reported outcome measure gives a score ranging from 0 to 48, with higher scores indicating less hip pain and better hip function. The estimated minimal clinically important difference is a change of 5 points. 

After a baseline assessment, 53 of 109 individuals were randomly assigned to undergo THR and 56 to participate in the resistance training program. Overall, the mean age of participants was 67.6 years, and half were women. The average duration of hip pain was a median of 1.7 years. 

The median time to receipt of the allocated treatment was 2.8 months in the THR group and 0.5 months in the resistance training group. 

Those allocated to the THR group also underwent a “fast track” program that involved patient education, pain management, and early mobilization. 

The resistance training group received 12 weeks of exercise supervised by a physiotherapist and then offered 12 weeks of additional exercise conducted on their own. The physiotherapist-supervised exercise sessions were held twice weekly and lasted for 1 hour. These started with a 10-minute warm-up on a stationary bike, followed by a standard set of resistance-based exercises that included a leg press, hip extension, hip flexion, and hip abduction. 
 

‘Reassuring’ Results

In a comment, consultant orthopedic surgeon Antony Palmer, MA, BMBCh, DPhil, said: “It’s reassuring that patients with advanced symptomatic osteoarthritis do well with hip replacements.”

THR does of course come with the potential risk for complications, but “the rate of these is what you’d expect for that procedure,” Palmer said, who works for the Nuffield Orthopaedic Centre, Oxford University Hospital NHS Foundation Trust, and is a senior clinical research fellow at Oxford University in England.

Dr. Palmer

Resistance Training Role 

A notable finding was that, at 6 months, five (9%) people assigned to the THR arm had not undergone surgery, and 12 (21%) people in the resistance training group had undergone a THR.

This could suggest two things, Palmer suggested in the interview. The first is that there could be a small proportion of people assigned to THR who may not need the operation and do well with exercise therapy. And, conversely, there may be those who would do well having the surgery without first going through the intermediate stage of physical therapy. 

It’s a suggestion that “maybe we’ve got to refine that a bit better and identify the patients that really do benefit from physiotherapy and who might not need hip replacement as a result,” Palmer said.

Or in other words, “should all patients undergo a program of physiotherapy before considering surgery?” he added.
 

Authors’ View

The PROHIP investigators conclude: “These results support current recommendations for the management of hip osteoarthritis and may be used to inform and guide shared decision making in clinical practice.”

Moreover, the results “do not oppose the use of resistance training as initial treatment,” says the authors. 

Frydendal highlighted in his interview that nearly three out of four of the patients reported not to have undertaken any type of supervised exercise before entry into the study, which is a first-line, guideline-recommended option.

“If a patient tells me, ‘I haven’t done any exercise previously,’ I’d recommend starting with completing a 6- to 12-week exercise program that is tailored to your individual needs and evaluate your symptoms afterward,” he said. 

“But we should refer the patient if our first-line treatment does not offer any improvements in the patient’s symptoms, as surgery with total hip replacement is clearly a really good treatment option,” Frydendal said.

The study was funded by the Danish Rheumatism Association, among other independent bodies. Frydendal and Palmer reported no relevant financial relationships. 
 

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

When my normally adorable cat Biscuit bit my ankle in a playful stalking exercise gone wrong, I washed it with soap and some rubbing alcohol, slapped on a Band-Aid, and went about my day.

The next morning, when it was swollen, I told myself it was probably just a hematoma and went about my day.

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

JAMA Network

Dr. Wilson

Dr. Wilson is associate professor of medicine and public health and director of the Clinical and Translational Research Accelerator at Yale University, New Haven, Conn. He has disclosed no relevant financial relationships.

A version of this article appeared on Medscape.com.

When my normally adorable cat Biscuit bit my ankle in a playful stalking exercise gone wrong, I washed it with soap and some rubbing alcohol, slapped on a Band-Aid, and went about my day.

The next morning, when it was swollen, I told myself it was probably just a hematoma and went about my day.

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

JAMA Network

Dr. Wilson

Dr. Wilson is associate professor of medicine and public health and director of the Clinical and Translational Research Accelerator at Yale University, New Haven, Conn. He has disclosed no relevant financial relationships.

A version of this article appeared on Medscape.com.

When my normally adorable cat Biscuit bit my ankle in a playful stalking exercise gone wrong, I washed it with soap and some rubbing alcohol, slapped on a Band-Aid, and went about my day.

The next morning, when it was swollen, I told myself it was probably just a hematoma and went about my day.

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

Dr. Wilson

JAMA Network

Dr. Wilson

Dr. Wilson is associate professor of medicine and public health and director of the Clinical and Translational Research Accelerator at Yale University, New Haven, Conn. He has disclosed no relevant financial relationships.

A version of this article appeared on Medscape.com.

In seconds, Ravi Parikh, MD, an oncologist at the Emory University School of Medicine in Atlanta, had a summary of his patient’s entire medical history. Normally, Parikh skimmed the cumbersome files before seeing a patient. However, the artificial intelligence (AI) tool his institution was testing could list the highlights he needed in a fraction of the time.

“On the whole, I like it ... it saves me time,” Parikh said of the tool. “But I’d be lying if I told you it was perfect all the time. It’s interpreting the [patient] history in some ways that may be inaccurate,” he said.

Within the first week of testing the tool, Parikh started to notice that the large language model (LLM) made a particular mistake in his patients with prostate cancer. If their prostate-specific antigen test results came back slightly elevated — which is part of normal variation — the LLM recorded it as disease progression. Because Parikh reviews all his notes — with or without using an AI tool — after a visit, he easily caught the mistake before it was added to the chart. “The problem, I think, is if these mistakes go under the hood,” he said.

In the data science world, these mistakes are called hallucinations. And a growing body of research suggests they’re happening more frequently than is safe for healthcare. The industry promised LLMs would alleviate administrative burden and reduce physician burnout. But so far, studies show these AI-tool mistakes often create more work for doctors, not less. To truly help physicians and be safe for patients, some experts say healthcare needs to build its own LLMs from the ground up. And all agree that the field desperately needs a way to vet these algorithms more thoroughly.
 

Prone to Error

Right now, “I think the industry is focused on taking existing LLMs and forcing them into usage for healthcare,” said Nigam H. Shah, MBBS, PhD, chief data scientist for Stanford Health. However, the value of deploying general LLMs in the healthcare space is questionable. “People are starting to wonder if we’re using these tools wrong,” he told this news organization.

In 2023, Shah and his colleagues evaluated seven LLMs on their ability to answer electronic health record–based questions. For realistic tasks, the error rate in the best cases was about 35%, he said. “To me, that rate seems a bit high ... to adopt for routine use.”

A study earlier this year by the UC San Diego School of Medicine showed that using LLMs to respond to patient messages increased the time doctors spent on messages. And this summer, a study by the clinical AI firm Mendel found that when GPT-4o or Llama-3 were used to summarize patient medical records, almost every summary contained at least one type of hallucination.

“We’ve seen cases where a patient does have drug allergies, but the system says ‘no known drug allergies’ ” in the medical history summary, said Wael Salloum, PhD, cofounder and chief science officer at Mendel. “That’s a serious hallucination.” And if physicians have to constantly verify what the system is telling them, that “defeats the purpose [of summarization],” he said.
 

A Higher Quality Diet

Part of the trouble with LLMs is that there’s just not enough high-quality information to feed them. The algorithms are insatiable, requiring vast swaths of data for training. GPT-3.5, for instance, was trained on 570 GB of data from the internet, more than 300 billion words. And to train GPT-4o, OpenAI reportedly transcribed more than 1 million hours of YouTube content.

However, the strategies that built these general LLMs don’t always translate well to healthcare. The internet is full of low-quality or misleading health information from wellness sites and supplement advertisements. And even data that are trustworthy, like the millions of clinical studies and the US Food and Drug Administration (FDA) statements, can be outdated, Salloum said. And “an LLM in training can’t distinguish good from bad,” he added.

The good news is that clinicians don’t rely on controversial information in the real world. Medical knowledge is standardized. “Healthcare is a domain rich with explicit knowledge,” Salloum said. So there’s potential to build a more reliable LLM that is guided by robust medical standards and guidelines.

It’s possible that healthcare could use small language models, which are LLM’s pocket-sized cousins, and perform tasks needing only bite-sized datasets requiring fewer resources and easier fine-tuning, according to Microsoft’s website. Shah said training these smaller models on real medical data might be an option, like an LLM meant to respond to patient messages that could be trained with real messages sent by physicians.

Several groups are already working on databases of standardized human medical knowledge or real physician responses. “Perhaps that will work better than using LLMs trained on the general internet. Those studies need to be done,” Shah said.

Jon Tamir, assistant professor of electrical and computer engineering and co-lead of the AI Health Lab at The University of Texas at Austin, said, “The community has recognized that we are entering a new era of AI where the dataset itself is the most important aspect. We need training sets that are highly curated and highly specialized.

“If the dataset is highly specialized, it will definitely help reduce hallucinations,” he said.
 

Cutting Overconfidence

A major problem with LLM mistakes is that they are often hard to detect. Hallucinations can be highly convincing even if they’re highly inaccurate, according to Tamir.

When Shah, for instance, was recently testing an LLM on de-identified patient data, he asked the LLM which blood test the patient last had. The model responded with “complete blood count [CBC].” But when he asked for the results, the model gave him white blood count and other values. “Turns out that record did not have a CBC done at all! The result was entirely made up,” he said.

Making healthcare LLMs safer and more reliable will mean training AI to acknowledge potential mistakes and uncertainty. Existing LLMs are trained to project confidence and produce a lot of answers, even when there isn’t one, Salloum said. They rarely respond with “I don’t know” even when their prediction has low confidence, he added.

Healthcare stands to benefit from a system that highlights uncertainty and potential errors. For instance, if a patient’s history shows they have smoked, stopped smoking, vaped, and started smoking again. The LLM might call them a smoker but flag the comment as uncertain because the chronology is complicated, Salloum said.

Tamir added that this strategy could improve LLM and doctor collaboration by honing in on where human expertise is needed most.
 

Too Little Evaluation

For any improvement strategy to work, LLMs — and all AI-assisted healthcare tools — first need a better evaluation framework. So far, LLMs have “been used in really exciting ways but not really well-vetted ways,” Tamir said.

While some AI-assisted tools, particularly in medical imaging, have undergone rigorous FDA evaluations and earned approval, most haven’t. And because the FDA only regulates algorithms that are considered medical devices, Parikh said that most LLMs used for administrative tasks and efficiency don’t fall under the regulatory agency’s purview.

But these algorithms still have access to patient information and can directly influence patient and doctor decisions. Third-party regulatory agencies are expected to emerge, but it’s still unclear who those will be. Before developers can build a safer and more efficient LLM for healthcare, they’ll need better guidelines and guardrails. “Unless we figure out evaluation, how would we know whether the healthcare-appropriate large language models are better or worse?” Shah asked.
 

A version of this article appeared on Medscape.com.

In seconds, Ravi Parikh, MD, an oncologist at the Emory University School of Medicine in Atlanta, had a summary of his patient’s entire medical history. Normally, Parikh skimmed the cumbersome files before seeing a patient. However, the artificial intelligence (AI) tool his institution was testing could list the highlights he needed in a fraction of the time.

“On the whole, I like it ... it saves me time,” Parikh said of the tool. “But I’d be lying if I told you it was perfect all the time. It’s interpreting the [patient] history in some ways that may be inaccurate,” he said.

Within the first week of testing the tool, Parikh started to notice that the large language model (LLM) made a particular mistake in his patients with prostate cancer. If their prostate-specific antigen test results came back slightly elevated — which is part of normal variation — the LLM recorded it as disease progression. Because Parikh reviews all his notes — with or without using an AI tool — after a visit, he easily caught the mistake before it was added to the chart. “The problem, I think, is if these mistakes go under the hood,” he said.

In the data science world, these mistakes are called hallucinations. And a growing body of research suggests they’re happening more frequently than is safe for healthcare. The industry promised LLMs would alleviate administrative burden and reduce physician burnout. But so far, studies show these AI-tool mistakes often create more work for doctors, not less. To truly help physicians and be safe for patients, some experts say healthcare needs to build its own LLMs from the ground up. And all agree that the field desperately needs a way to vet these algorithms more thoroughly.
 

Prone to Error

Right now, “I think the industry is focused on taking existing LLMs and forcing them into usage for healthcare,” said Nigam H. Shah, MBBS, PhD, chief data scientist for Stanford Health. However, the value of deploying general LLMs in the healthcare space is questionable. “People are starting to wonder if we’re using these tools wrong,” he told this news organization.

In 2023, Shah and his colleagues evaluated seven LLMs on their ability to answer electronic health record–based questions. For realistic tasks, the error rate in the best cases was about 35%, he said. “To me, that rate seems a bit high ... to adopt for routine use.”

A study earlier this year by the UC San Diego School of Medicine showed that using LLMs to respond to patient messages increased the time doctors spent on messages. And this summer, a study by the clinical AI firm Mendel found that when GPT-4o or Llama-3 were used to summarize patient medical records, almost every summary contained at least one type of hallucination.

“We’ve seen cases where a patient does have drug allergies, but the system says ‘no known drug allergies’ ” in the medical history summary, said Wael Salloum, PhD, cofounder and chief science officer at Mendel. “That’s a serious hallucination.” And if physicians have to constantly verify what the system is telling them, that “defeats the purpose [of summarization],” he said.
 

A Higher Quality Diet

Part of the trouble with LLMs is that there’s just not enough high-quality information to feed them. The algorithms are insatiable, requiring vast swaths of data for training. GPT-3.5, for instance, was trained on 570 GB of data from the internet, more than 300 billion words. And to train GPT-4o, OpenAI reportedly transcribed more than 1 million hours of YouTube content.

However, the strategies that built these general LLMs don’t always translate well to healthcare. The internet is full of low-quality or misleading health information from wellness sites and supplement advertisements. And even data that are trustworthy, like the millions of clinical studies and the US Food and Drug Administration (FDA) statements, can be outdated, Salloum said. And “an LLM in training can’t distinguish good from bad,” he added.

The good news is that clinicians don’t rely on controversial information in the real world. Medical knowledge is standardized. “Healthcare is a domain rich with explicit knowledge,” Salloum said. So there’s potential to build a more reliable LLM that is guided by robust medical standards and guidelines.

It’s possible that healthcare could use small language models, which are LLM’s pocket-sized cousins, and perform tasks needing only bite-sized datasets requiring fewer resources and easier fine-tuning, according to Microsoft’s website. Shah said training these smaller models on real medical data might be an option, like an LLM meant to respond to patient messages that could be trained with real messages sent by physicians.

Several groups are already working on databases of standardized human medical knowledge or real physician responses. “Perhaps that will work better than using LLMs trained on the general internet. Those studies need to be done,” Shah said.

Jon Tamir, assistant professor of electrical and computer engineering and co-lead of the AI Health Lab at The University of Texas at Austin, said, “The community has recognized that we are entering a new era of AI where the dataset itself is the most important aspect. We need training sets that are highly curated and highly specialized.

“If the dataset is highly specialized, it will definitely help reduce hallucinations,” he said.
 

Cutting Overconfidence

A major problem with LLM mistakes is that they are often hard to detect. Hallucinations can be highly convincing even if they’re highly inaccurate, according to Tamir.

When Shah, for instance, was recently testing an LLM on de-identified patient data, he asked the LLM which blood test the patient last had. The model responded with “complete blood count [CBC].” But when he asked for the results, the model gave him white blood count and other values. “Turns out that record did not have a CBC done at all! The result was entirely made up,” he said.

Making healthcare LLMs safer and more reliable will mean training AI to acknowledge potential mistakes and uncertainty. Existing LLMs are trained to project confidence and produce a lot of answers, even when there isn’t one, Salloum said. They rarely respond with “I don’t know” even when their prediction has low confidence, he added.

Healthcare stands to benefit from a system that highlights uncertainty and potential errors. For instance, if a patient’s history shows they have smoked, stopped smoking, vaped, and started smoking again. The LLM might call them a smoker but flag the comment as uncertain because the chronology is complicated, Salloum said.

Tamir added that this strategy could improve LLM and doctor collaboration by honing in on where human expertise is needed most.
 

Too Little Evaluation

For any improvement strategy to work, LLMs — and all AI-assisted healthcare tools — first need a better evaluation framework. So far, LLMs have “been used in really exciting ways but not really well-vetted ways,” Tamir said.

While some AI-assisted tools, particularly in medical imaging, have undergone rigorous FDA evaluations and earned approval, most haven’t. And because the FDA only regulates algorithms that are considered medical devices, Parikh said that most LLMs used for administrative tasks and efficiency don’t fall under the regulatory agency’s purview.

But these algorithms still have access to patient information and can directly influence patient and doctor decisions. Third-party regulatory agencies are expected to emerge, but it’s still unclear who those will be. Before developers can build a safer and more efficient LLM for healthcare, they’ll need better guidelines and guardrails. “Unless we figure out evaluation, how would we know whether the healthcare-appropriate large language models are better or worse?” Shah asked.
 

A version of this article appeared on Medscape.com.

In seconds, Ravi Parikh, MD, an oncologist at the Emory University School of Medicine in Atlanta, had a summary of his patient’s entire medical history. Normally, Parikh skimmed the cumbersome files before seeing a patient. However, the artificial intelligence (AI) tool his institution was testing could list the highlights he needed in a fraction of the time.

“On the whole, I like it ... it saves me time,” Parikh said of the tool. “But I’d be lying if I told you it was perfect all the time. It’s interpreting the [patient] history in some ways that may be inaccurate,” he said.

Within the first week of testing the tool, Parikh started to notice that the large language model (LLM) made a particular mistake in his patients with prostate cancer. If their prostate-specific antigen test results came back slightly elevated — which is part of normal variation — the LLM recorded it as disease progression. Because Parikh reviews all his notes — with or without using an AI tool — after a visit, he easily caught the mistake before it was added to the chart. “The problem, I think, is if these mistakes go under the hood,” he said.

In the data science world, these mistakes are called hallucinations. And a growing body of research suggests they’re happening more frequently than is safe for healthcare. The industry promised LLMs would alleviate administrative burden and reduce physician burnout. But so far, studies show these AI-tool mistakes often create more work for doctors, not less. To truly help physicians and be safe for patients, some experts say healthcare needs to build its own LLMs from the ground up. And all agree that the field desperately needs a way to vet these algorithms more thoroughly.
 

Prone to Error

Right now, “I think the industry is focused on taking existing LLMs and forcing them into usage for healthcare,” said Nigam H. Shah, MBBS, PhD, chief data scientist for Stanford Health. However, the value of deploying general LLMs in the healthcare space is questionable. “People are starting to wonder if we’re using these tools wrong,” he told this news organization.

In 2023, Shah and his colleagues evaluated seven LLMs on their ability to answer electronic health record–based questions. For realistic tasks, the error rate in the best cases was about 35%, he said. “To me, that rate seems a bit high ... to adopt for routine use.”

A study earlier this year by the UC San Diego School of Medicine showed that using LLMs to respond to patient messages increased the time doctors spent on messages. And this summer, a study by the clinical AI firm Mendel found that when GPT-4o or Llama-3 were used to summarize patient medical records, almost every summary contained at least one type of hallucination.

“We’ve seen cases where a patient does have drug allergies, but the system says ‘no known drug allergies’ ” in the medical history summary, said Wael Salloum, PhD, cofounder and chief science officer at Mendel. “That’s a serious hallucination.” And if physicians have to constantly verify what the system is telling them, that “defeats the purpose [of summarization],” he said.
 

A Higher Quality Diet

Part of the trouble with LLMs is that there’s just not enough high-quality information to feed them. The algorithms are insatiable, requiring vast swaths of data for training. GPT-3.5, for instance, was trained on 570 GB of data from the internet, more than 300 billion words. And to train GPT-4o, OpenAI reportedly transcribed more than 1 million hours of YouTube content.

However, the strategies that built these general LLMs don’t always translate well to healthcare. The internet is full of low-quality or misleading health information from wellness sites and supplement advertisements. And even data that are trustworthy, like the millions of clinical studies and the US Food and Drug Administration (FDA) statements, can be outdated, Salloum said. And “an LLM in training can’t distinguish good from bad,” he added.

The good news is that clinicians don’t rely on controversial information in the real world. Medical knowledge is standardized. “Healthcare is a domain rich with explicit knowledge,” Salloum said. So there’s potential to build a more reliable LLM that is guided by robust medical standards and guidelines.

It’s possible that healthcare could use small language models, which are LLM’s pocket-sized cousins, and perform tasks needing only bite-sized datasets requiring fewer resources and easier fine-tuning, according to Microsoft’s website. Shah said training these smaller models on real medical data might be an option, like an LLM meant to respond to patient messages that could be trained with real messages sent by physicians.

Several groups are already working on databases of standardized human medical knowledge or real physician responses. “Perhaps that will work better than using LLMs trained on the general internet. Those studies need to be done,” Shah said.

Jon Tamir, assistant professor of electrical and computer engineering and co-lead of the AI Health Lab at The University of Texas at Austin, said, “The community has recognized that we are entering a new era of AI where the dataset itself is the most important aspect. We need training sets that are highly curated and highly specialized.

“If the dataset is highly specialized, it will definitely help reduce hallucinations,” he said.
 

Cutting Overconfidence

A major problem with LLM mistakes is that they are often hard to detect. Hallucinations can be highly convincing even if they’re highly inaccurate, according to Tamir.

When Shah, for instance, was recently testing an LLM on de-identified patient data, he asked the LLM which blood test the patient last had. The model responded with “complete blood count [CBC].” But when he asked for the results, the model gave him white blood count and other values. “Turns out that record did not have a CBC done at all! The result was entirely made up,” he said.

Making healthcare LLMs safer and more reliable will mean training AI to acknowledge potential mistakes and uncertainty. Existing LLMs are trained to project confidence and produce a lot of answers, even when there isn’t one, Salloum said. They rarely respond with “I don’t know” even when their prediction has low confidence, he added.

Healthcare stands to benefit from a system that highlights uncertainty and potential errors. For instance, if a patient’s history shows they have smoked, stopped smoking, vaped, and started smoking again. The LLM might call them a smoker but flag the comment as uncertain because the chronology is complicated, Salloum said.

Tamir added that this strategy could improve LLM and doctor collaboration by honing in on where human expertise is needed most.
 

Too Little Evaluation

For any improvement strategy to work, LLMs — and all AI-assisted healthcare tools — first need a better evaluation framework. So far, LLMs have “been used in really exciting ways but not really well-vetted ways,” Tamir said.

While some AI-assisted tools, particularly in medical imaging, have undergone rigorous FDA evaluations and earned approval, most haven’t. And because the FDA only regulates algorithms that are considered medical devices, Parikh said that most LLMs used for administrative tasks and efficiency don’t fall under the regulatory agency’s purview.

But these algorithms still have access to patient information and can directly influence patient and doctor decisions. Third-party regulatory agencies are expected to emerge, but it’s still unclear who those will be. Before developers can build a safer and more efficient LLM for healthcare, they’ll need better guidelines and guardrails. “Unless we figure out evaluation, how would we know whether the healthcare-appropriate large language models are better or worse?” Shah asked.
 

A version of this article appeared on Medscape.com.

From the largest healthcare companies to solo practices, just every organization in medicine faces a risk for costly cyberattacks. In recent years, hackers have threatened to release the personal information of patients and employees — or paralyze online systems — unless they’re paid a ransom.

Should companies pay? It’s not an easy answer, a pair of experts told colleagues in an American Medical Association (AMA) cybersecurity webinar on October 18. It turns out that each choice — pay or don’t pay — can end up being costly.

This is just one of the new challenges facing the American medical system on the cybersecurity front, the speakers said. Others include the possibility that hackers will manipulate patient data — turning a medical test negative, for example, when it’s actually positive — and take advantage of the powers of artificial intelligence (AI).

The AMA held the webinar to educate physicians about cybersecurity risks and defenses, an especially hot topic in the wake of February’s Change Healthcare hack, which cost UnitedHealth Group an estimated $2.5 billion — so far — and deeply disrupted the American healthcare system.

Cautionary tales abound. Greg Garcia, executive director for cybersecurity of the Health Sector Coordinating Council, a coalition of medical industry organizations, pointed to a Pennsylvania clinic that refused to pay a ransom to prevent the release of hundreds of images of patients with breast cancer undressed from the waist up. Garcia told webinar participants that the ransom was $5 million.
 

Risky Choices

While the Federal Bureau of Investigation recommends against paying a ransom, this can be a risky choice, Garcia said. Hackers released the images, and the center has reportedly agreed to settle a class-action lawsuit for $65 million. “They traded $5 million for $60 million,” Garcia added, slightly misstating the settlement amount.

Health systems have been cagey about whether they’ve paid ransoms to prevent private data from being made public in cyberattacks. If a ransom is demanded, “it’s every organization for itself,” Garcia said.

He highlighted the case of a chain of psychiatry practices in Finland that suffered a ransomware attack in 2020. The hackers “contacted the patients and said: ‘Hey, call your clinic and tell them to pay the ransom. Otherwise, we’re going to release all your psychiatric notes to the public.’ ”

Cyberattacks continue. In October, Boston Children’s Health Physicians announced that it had suffered a “ recent security incident” involving data — possibly including Social Security numbers and treatment information — regarding patients and employees. A hacker group reportedly claimed responsibility and wants the system, which boasts more than 300 clinicians, to pay a ransom or else it will release the stolen information.
 

Should Paying Ransom Be a Crime?

Christian Dameff, MD, MS, an emergency medicine physician and director of the Center for Healthcare Cybersecurity at the University of California (UC), San Diego, noted that there are efforts to turn paying ransom into a crime. “If people aren’t paying ransoms, then ransomware operators will move to something else that makes them money.”

Dameff urged colleagues to understand we no longer live in a world where clinicians only bother to think of technology when they call the IT department to help them reset their password.

New challenges face clinicians, he said. “How do we develop better strategies, downtime procedures, and safe clinical care in an era where our vital technology may be gone, not just for an hour or 2, but as is the case with these ransomware attacks, sometimes weeks to months.”

Garcia said “cybersecurity is everybody’s responsibility, including frontline clinicians. Because you’re touching data, you’re touching technology, you’re touching patients, and all of those things combine to present some vulnerabilities in the digital world.”
 

Next Frontier: Hackers May Manipulate Patient Data

Dameff said future hackers may use AI to manipulate individual patient data in ways that threaten patient health. AI makes this easier to accomplish.

“What if I delete your allergies in your electronic health record, or I manipulate your chest x-ray, or I change your lab values so it looks like you’re in diabetic ketoacidosis when you’re not so a clinician gives you insulin when you don’t need it?”

Garcia highlighted another new threat: Phishing efforts that are harder to ignore thanks to AI.

“One of the most successful way that hackers get in, disrupt systems, and steal data is through email phishing, and it’s only going to get better because of artificial intelligence,” he said. “No longer are you going to have typos in that email written by a hacking group in Nigeria or in China. It’s going to be perfect looking.”

What can practices and healthcare systems do? Garcia highlighted federal health agency efforts to encourage organizations to adopt best practices in cybersecurity.

“If you’ve got a data breach, and you can show to the US Department of Health & Human Services [HHS] you have implemented generally recognized cybersecurity controls over the past year, that you have done your best, you did the right thing, and you still got hit, HHS is directed to essentially take it easy on you,” he said. “That’s a positive incentive.”
 

Ransomware Guide in the Works

Dameff said UC San Diego’s Center for Healthcare Cybersecurity plans to publish a free cybersecurity guide in 2025 that will include specific information about ransomware attacks for medical specialties such as cardiology, trauma surgery, and pediatrics.

“Then, should you ever be ransomed, you can pull out this guide. You’ll know what’s going to kind of happen, and you can better prepare for those effects.”

Will the future president prioritize healthcare cybersecurity? That remains to be seen, but crises do have the capacity to concentrate the mind, experts said.

The nation’s capital “has a very short memory, a short attention span. The policymakers tend to be reactive,” Dameff said. “All it takes is yet another Change Healthcare–like attack that disrupts 30% or more of the nation’s healthcare system for the policymakers to sit up, take notice, and try to come up with solutions.”

In addition, he said, an estimated two data breaches/ransomware attacks are occurring per day. “The fact is that we’re all patients, up to the President of the United States and every member of the Congress is a patient.”

There’s a “very existential, very palpable understanding that cyber safety is patient safety and cyber insecurity is patient insecurity,” Dameff said.

A version of this article appeared on Medscape.com.

From the largest healthcare companies to solo practices, just every organization in medicine faces a risk for costly cyberattacks. In recent years, hackers have threatened to release the personal information of patients and employees — or paralyze online systems — unless they’re paid a ransom.

Should companies pay? It’s not an easy answer, a pair of experts told colleagues in an American Medical Association (AMA) cybersecurity webinar on October 18. It turns out that each choice — pay or don’t pay — can end up being costly.

This is just one of the new challenges facing the American medical system on the cybersecurity front, the speakers said. Others include the possibility that hackers will manipulate patient data — turning a medical test negative, for example, when it’s actually positive — and take advantage of the powers of artificial intelligence (AI).

The AMA held the webinar to educate physicians about cybersecurity risks and defenses, an especially hot topic in the wake of February’s Change Healthcare hack, which cost UnitedHealth Group an estimated $2.5 billion — so far — and deeply disrupted the American healthcare system.

Cautionary tales abound. Greg Garcia, executive director for cybersecurity of the Health Sector Coordinating Council, a coalition of medical industry organizations, pointed to a Pennsylvania clinic that refused to pay a ransom to prevent the release of hundreds of images of patients with breast cancer undressed from the waist up. Garcia told webinar participants that the ransom was $5 million.
 

Risky Choices

While the Federal Bureau of Investigation recommends against paying a ransom, this can be a risky choice, Garcia said. Hackers released the images, and the center has reportedly agreed to settle a class-action lawsuit for $65 million. “They traded $5 million for $60 million,” Garcia added, slightly misstating the settlement amount.

Health systems have been cagey about whether they’ve paid ransoms to prevent private data from being made public in cyberattacks. If a ransom is demanded, “it’s every organization for itself,” Garcia said.

He highlighted the case of a chain of psychiatry practices in Finland that suffered a ransomware attack in 2020. The hackers “contacted the patients and said: ‘Hey, call your clinic and tell them to pay the ransom. Otherwise, we’re going to release all your psychiatric notes to the public.’ ”

Cyberattacks continue. In October, Boston Children’s Health Physicians announced that it had suffered a “ recent security incident” involving data — possibly including Social Security numbers and treatment information — regarding patients and employees. A hacker group reportedly claimed responsibility and wants the system, which boasts more than 300 clinicians, to pay a ransom or else it will release the stolen information.
 

Should Paying Ransom Be a Crime?

Christian Dameff, MD, MS, an emergency medicine physician and director of the Center for Healthcare Cybersecurity at the University of California (UC), San Diego, noted that there are efforts to turn paying ransom into a crime. “If people aren’t paying ransoms, then ransomware operators will move to something else that makes them money.”

Dameff urged colleagues to understand we no longer live in a world where clinicians only bother to think of technology when they call the IT department to help them reset their password.

New challenges face clinicians, he said. “How do we develop better strategies, downtime procedures, and safe clinical care in an era where our vital technology may be gone, not just for an hour or 2, but as is the case with these ransomware attacks, sometimes weeks to months.”

Garcia said “cybersecurity is everybody’s responsibility, including frontline clinicians. Because you’re touching data, you’re touching technology, you’re touching patients, and all of those things combine to present some vulnerabilities in the digital world.”
 

Next Frontier: Hackers May Manipulate Patient Data

Dameff said future hackers may use AI to manipulate individual patient data in ways that threaten patient health. AI makes this easier to accomplish.

“What if I delete your allergies in your electronic health record, or I manipulate your chest x-ray, or I change your lab values so it looks like you’re in diabetic ketoacidosis when you’re not so a clinician gives you insulin when you don’t need it?”

Garcia highlighted another new threat: Phishing efforts that are harder to ignore thanks to AI.

“One of the most successful way that hackers get in, disrupt systems, and steal data is through email phishing, and it’s only going to get better because of artificial intelligence,” he said. “No longer are you going to have typos in that email written by a hacking group in Nigeria or in China. It’s going to be perfect looking.”

What can practices and healthcare systems do? Garcia highlighted federal health agency efforts to encourage organizations to adopt best practices in cybersecurity.

“If you’ve got a data breach, and you can show to the US Department of Health & Human Services [HHS] you have implemented generally recognized cybersecurity controls over the past year, that you have done your best, you did the right thing, and you still got hit, HHS is directed to essentially take it easy on you,” he said. “That’s a positive incentive.”
 

Ransomware Guide in the Works

Dameff said UC San Diego’s Center for Healthcare Cybersecurity plans to publish a free cybersecurity guide in 2025 that will include specific information about ransomware attacks for medical specialties such as cardiology, trauma surgery, and pediatrics.

“Then, should you ever be ransomed, you can pull out this guide. You’ll know what’s going to kind of happen, and you can better prepare for those effects.”

Will the future president prioritize healthcare cybersecurity? That remains to be seen, but crises do have the capacity to concentrate the mind, experts said.

The nation’s capital “has a very short memory, a short attention span. The policymakers tend to be reactive,” Dameff said. “All it takes is yet another Change Healthcare–like attack that disrupts 30% or more of the nation’s healthcare system for the policymakers to sit up, take notice, and try to come up with solutions.”

In addition, he said, an estimated two data breaches/ransomware attacks are occurring per day. “The fact is that we’re all patients, up to the President of the United States and every member of the Congress is a patient.”

There’s a “very existential, very palpable understanding that cyber safety is patient safety and cyber insecurity is patient insecurity,” Dameff said.

A version of this article appeared on Medscape.com.