AMIA seeks policy refinements in its comments to NIH and FDA

The American Medical Informatics Association is offering guidance to two federal agencies, responding in separate letters that seek changes in potential public policies.

In a letter submitted on Monday, AMIA warned that National Institutes of Health that a data management and sharing policy it’s proposing would be ineffective and potentially increase burdens on researchers.

Late last week, the organization, which represents healthcare informaticists, wrote the Food and Drug Administration to ask it to provide greater clarity on how it will define and regulate clinical decision support software.

Monday’s letter to the NIH was in response to the agency’s call in November for a request for public comment on its proposed data management and information sharing policies. AMIA says the rules, as currently proposed, do not fix challenges in existing policies.

For example, under existing NIH policy, grant applications are not scored on the quality of their data sharing plans. “This has led to suboptimal and incomplete sharing plans and likely has contributed to data silos and increased scrutiny over whether and how NIH-supported research data is made available to researchers and the broader public,” the association explained.



The NIH’s policy, a second take that reflects changes sought by the industry to an earlier version, still falls short, AMIA contends. “The policy continues to undervalue the importance of data management and sharing as an essential part of modern research,” the association noted in its letter signed by Patricia Dykes, chair of the AMIA board and director of research at the Brigham and Women’s Center for patient safety, research and practice. “It envisions a two-part plan developed after a project has been chosen to receive funding, perpetuating a check-the-box exercise that will only make the task of managing ever-growing volumes of data more difficult.”

The NIH must view scientific data “as the principle result of scientific research—not the conference presentation or journal publication describing the data,” Dykes said. The association recommended that NIH revise its draft policy to achieve three goals—optimize scientific data once generated; incentivize improvements in data management and sharing practices; and coordinate disparate Institute, Center and Office policies.

The complete comments are available here.

AMIA’s guidance to the FDA is in response to help it sought in understanding the types of clinical decision support software currently being used or that are being planned for development, as stakeholders struggle to understand what types of decision support software is subject to FDA oversight.

There presently is confusion among clinicians and developers of what types of decision support can be used or not used. For example, there is a provision in the statute that says, “the software developer should describe the underlying data used to develop the algorithm and should include plain language descriptions of the logic or rational used by an algorithm to render a recommendation.”

While informaticists working with FDA said this was an improvement upon initial language, the provision still raised follow-on questions about where in the workflow information should be available and how the information should be presented. AMIA further anticipates there will be concerns over the difference between software that informs and software that drives clinical management, as well as the concept of independent review.

The organization also wants FDA to take steps to understand what types of clinical decision support are being used today and other types being planned for development in healthcare delivery systems, and then establish regulatory parameters based on the findings.

The association’s complete comments are available here.