The Biden administration has released a sweeping new rule touching on multiple aspects of health information technology in the U.S., including proposals meant to create more transparency and trust in clinical decision support software.
The rule proposed Tuesday would give providers more information in their electronic health record system to assess clinical decision support algorithms and their results.
The changes would promote greater trust in predictive decision support interventions, and help providers assess their applicability for a particular patient population or circumstance, the Office of the National Coordinator for Health IT said.
Artificial intelligence has a potential to cut costs and improve care delivery in healthcare, including in the area of clinical decision support. AIs are trained on datasets to find patterns and make recommendations to clinicians, like alerts for adverse medication interactions or potential cases of sepsis.
Demand for clinical decision support is rising as physicians try to keep up with rapidly shifting medical knowledge and recommendations. In addition, mounting data suggests tools can help fend off adverse outcomes for patients.
However, regulating algorithms has historically been a gray area, until Food and Drug Administration guidance last year broadened the agency’s oversight of clinical decision support software, including those that provide a specific diagnostic or treatment suggestion.
Now, the ONC — which regulates health IT systems that can include the AI tools — is proposing new regulations meant to make more information about the algorithms available in the EHR for providers, ONC head Micky Tripathi said on a Tuesday call with reporters.
The rule proposes a new certification criteria for clinical decision support algorithms, to keep pace with advances in decision-making software.
Along with defining algorithm-based, predictive decision support interventions, the rule requires certified EHRs that enable or interface with the software to allow users to review information about additional source attributes, including around health equity. The rule also requires developers of certified health IT modules to undergo intervention risk management practices for all the predictive decision support interventions they interface with.
Together, the proposals should improve transparency and promote trustworthiness of clinical decision support tools by enabling users, like providers, to better scrutinize them, the ONC said.
New info blocking condition for TEFCA
The rule is also proposing to add new information blocking flexibilities for organizations that participate in the Trusted Exchange Framework and Common Agreement, the government’s nascent framework for nationwide health information exchange.
Since participation is TEFCA is voluntary, there have been questions around industry buy-in since it was finalized over a year ago. Some stakeholders have argued that regulators should exempt organizations that participate in TEFCA from information blocking stipulations, to incentivize companies to join by reducing their compliance burden.
Now, the ONC is proposing a new information blocking condition for parties that participate in TEFCA.
If a TEFCA participant offers to fulfill a data-sharing request from another TEFCA participant through the framework, they wouldn’t be required to offer that data in any alternative manner — even when the data might exceed the minimum data classes and elements required in the Common Agreement.
The proposed manner exception says it’s reasonable for an information blocking actor to prioritize using TEFCA means to send data over other feasible ways.
This approach furthers both the goals of TEFCA adoption and interoperability, according to the final rule.
Other changes
The proposed rule would discontinue year-themed editions of ONC certification criteria for health IT. Instead, it would establish a single set of criteria that would be updated incrementally, with the goal of aligning with standard development cycles and making it easier for health IT developers to maintain their product certificates over time.
The rule also includes provisions meant to help individuals and providers, including those engaged in public health, to securely access electronic health information through broader adoption of standardized application programming interfaces.
It would also standardize electronic case reporting, with the goal of facilitating faster disease tracking and case management, after the lack of a nationwide reporting infrastructure hamstrung the public health response during COVID-19.
Regulators are also proposing that developers adopt the third iteration of the U.S. Core Data for Interoperability dataset by 2025. The dataset would be expanded to include additional data elements related to public health, sexual orientation and social determinants of health.
The proposed rule would also clarify the infeasibility exception to the information blocking rules.
The ONC has yet to finalize a long-awaited regulation creating appropriate disincentives for providers found information blocking. That regulation now has a tentative release date of September.
Regulators also expect to release new standards for EHRs, including expanded uses of APIs around electronic prior authorization and patient engagement, in November.
