ECRI recently released its annual list of the "Top 10 Health Technology Hazards," which identified risks with artificial intelligence (AI)-enabled health technologies as the top threat to healthcare organizations in the coming year.
According to ECRI, the report focuses on "generic hazards," or those that can occur based on risks inherent to the use of certain types or combinations of medical technologies. The report does not address risks or problems associated with specific models or suppliers.
To create the list, ECRI engineers, scientists, clinicians, and other patient safety analysts nominated topics for consideration based on their own expertise and insight by:
After the nomination phase, professionals from ECRI's different program areas, as well as external advisors, reviewed the topics to select the top 10. The final list weighed several factors, including severity, frequency, breadth, insidiousness, public profile, and preventability. Although any of the criteria on their own could warrant including a topic, all hazards on the list must be, to some degree, preventable.
According to ECRI, biases present in the data used to train AI models, or mismatches between the training data and the target patient population, can result in inappropriate responses or disparate health outcomes. AI systems can also create false or misleading responses or experience a change in their performance over time due to data drift.
AI technology can also lead to disappointing results if organizations have unrealistic expectations, fail to properly define their goals, don't have sufficient governance and oversight, or don't adequately prepare their data for AI use.
"The promise of artificial intelligence's capabilities must not distract us from its risks or its ability to harm patients and providers," said ECRI president and CEO Marcus Schabacker. "Balancing innovation in AI with privacy and safety will be one of the most difficult, and most defining, endeavors of modern medicine."
Devices like ventilators, dialysis machines, and infusion pumps have typically been used in acute care settings but are now seeing increased at-home use. However, patients often don't have the technological support they need to use these devices safely and effectively at home.
According to ECRI, there have been several examples of patient harm due to improper setup or lack of familiarity with medical devices used in the home. Errors with the devices may go undetected or readings may be misinterpreted, leading to care delays or other harm.
"Minimizing the risk of harm requires providing home users with the support they need to operate, maintain, and troubleshoot the device successfully," ECRI writes.
Many essential tools used by healthcare organizations, such as scheduling and billing services and EHRs, are often provided by third-party vendors. However, a healthcare provider's operations can be severely impacted if a cybersecurity event incapacitates or degrades operations at a third-party vendor.
Cyberattacks on third-party vendors can leave healthcare providers unable to access critical services, reliable data, or effective communication channels. This can then delay, prevent, or degrade care, leading to patient harm.
ECRI recommends that healthcare organizations thoroughly vet their vendors before they acquire their services, build in redundancies, conduct incident response tests, and develop recovery procedures. Government bodies, regulatory agencies, and others in the industry should also move toward a collective approach to protecting organizations from cybercrime and vendor risk.
Supply chain challenges have made it easier for substandard or sometimes fraudulent medical devices and supplies to reach the U.S. market and care settings. For example, FDA has warned that plastic syringes made in China could be prone to failure, an action that could affect over 1 billion products in U.S. healthcare facilities.
Because these substandard or fraudulent products may be more likely to malfunction or fail, there is a greater risk of misdiagnoses or injuries. Disruptions and recalls of these products could also stress the supply chain, leading to product shortages and costing organizations time and money.
To minimize risks from these products, ECRI recommends organizations carefully vet all their suppliers, test out new products before purchasing, establish processes for investigating and reporting questionable products, and create response plans for potential supply chain challenges.
According to ECRI, the three elements needed for a fire (an oxidizer, a fuel, and an ignition source) can be present wherever oxygen is delivered. Although surgical staff in acute care facilities are likely aware of the fire risks in the OR, ECRI said that more attention is needed to prevent fires when supplemental oxygen is used in other patient care areas, including the home.
"Clinicians, caregivers, and even patients need to understand the risks associated with each element [needed for a fire to start] and take proper precautions," ECRI writes. People present in areas where supplemental oxygen is being used should also know how to respond if a fire occurs.
According to ECRI, setting the default alarm limits on anesthesia units at zero or similarly unsafe lower limits can cause dangerous situations to go undetected. This can then lead to patient harm, including patient awareness during surgery, brain damage, or even death.
ECRI recommends vendors configure their anesthesia units with safe lower default limits for critical alarms. Organizations should also confirm that the default settings for their units are set to appropriate values.
A lack of carefully vetted workflows can lead to mistakes in temporary holds of medication orders. Failure to hold a medication or not restarting a held medication as needed can result in patient harm.
According to the Institute for Safe Medication Practices, errors with holds on orders are often due to uncertainty about what a hold order means, how it should be communicated, or what process should be followed.
Organizations should have a well-defined and well-understood process for holding and resuming medications, ECRI writes. Providers should also be able to clearly see a hold order and all its related information in the EHR.
Improper line management can cause infections if pathogens enter the connector and infusion line, or environmental contaminants contact the insertion site. Healthcare-associated infections can lead longer hospital stays, increased patient morbidity, or even death.
To minimize infection risks, healthcare workers should follow best practices for infection control, such as using aseptic technique when inserting, accessing, or maintaining lines.
If infusion lines are allowed to lie on the floor, they can also create a tripping hazard for both staff and patients, potentially leading to falls. Staff should move any infusion lines away from the floor.
Inappropriate adhesive choices can lead to skin tears, blisters, or adverse reactions like contact dermatitis. Factors that could contribute to skin injuries from medical adhesives include a patient's age or medical conditions, the strength of an adhesive, or errors when applying or removing a product.
ECRI recommends organizations purchase and provide staff with easy access to an appropriate selection of adhesive products. Clinicians should also assess patients for potential risk of medical-adhesive-related skin injuries and choose a product with the lowest strength that can still meet the requirements.
Investigations of infusion systems can be challenging, and organizations that don't have the expertise or resources to conduct a thorough investigation will find it difficult to prevent future infusion-related medication errors or other incidents.
"The process of identifying root causes and implementing corrective measures starts with creating a culture in which incidents are immediately reported," ECRI writes. Staff should also be educated on how to respond immediately after an incident and on the steps they should take to support an investigation.
(ECRI's Top 10 Health Technology Hazards for 2025 list, 12/4; ECRI news release, 12/4; Beavins, Fierce Healthcare, 12/4)
The rapid development of AI in healthcare calls for thoughtful governance, but there is no one-size-fits-all approach. These three insights from health system leaders will help you build effective AI governance structures that align with your organization's unique needs and culture.
Create your free account to access 1 resource, including the latest research and webinars.
You have 1 free members-only resource remaining this month.
1 free members-only resources remaining
1 free members-only resources remaining
Never miss out on the latest innovative health care content tailored to you.