The American Medical Informatics Association said recent guidance from the U.S. Food and Drug Administration for clinical and patient decision support tools is "well-timed to initiate a broad discussion" about the future of CDS and PDS. But AMIA also says there's still a lot to discuss.

Section 3060 of the 21st Century Cures Act calls on FDA to exclude specific functionalities from its definition of medical device – legalese that essentially bars FDA from regulating that software as a device (which would require pre-market review, clinical trials and/or other agency clearances).

Certain of those excluded functionalities describe what is normally understood as clinical decision support and its consumer-facing cousin, patient decision support. But not all CDS/PDS was excluded from the definition of the device by the Cures Act.

[Also: 2017 was a big year for FDA digital health regulations]

FDA's draft guidance attempts to suss out a line between regulated device and unregulated functionality, but in comments submitted to the agency this week, AMIA argues that the proposal is still incomplete – and needs a further discussion around how FDA would regulate those CDS and PDS applications it still considered medical devices.

AMIA makes the point that several trend lines are converging – consumerism and patient engagement, cloud computing, machine learning – that are set to fundamentally change how healthcare knowledge is disseminated for healthcare and wellness.

But the FDA draft guidance has left "lingering confusion among developers and clinicians trying to determine whether specific decision support software is, or is not, considered a device," according to AMIA, since some of the criteria used to determine whether functionality should be excluded is ambiguous.

[Also: How nursing homes can harness clinical decision support systems]

To be useful and practical, AMIA recommended that FDA do three things with its guidance:

First, it should articulate why some types of software are considered exempt and others aren't, using the criteria established by 21st Century Cures.

"This will help industry better understand the rationale of the FDA, and it will ensure that adequate consideration be given to variants of each criterion," wrote AMIA CEO Douglas Fridsma, MD.

Second, FDA should include discussion regarding its intended regulatory controls for CDS software considered a device – even if those intended regulatory controls are still in development: "The omission of regulatory controls breeds uncertainty and anxiety among industry readers," said Fridsma.

Third, it should specify the anticipated literacy levels, and their variance, across intended users of patient decision support software functions.

AMIA noted it has been helping with development, innovation and regulation of decision support for more than 20 years. But in recent years, CDS and PDS innovation have started happening "at a tremendous rate, and we believe several trends are converging to transform the delivery of computable biomedical knowledge at the point of care."

CDS and PDS functionalities now be "based on a trained neural network, multivariate regressions, or fuzzy logic with perhaps millions of data points, enabling a data-driven approach to decision-making at the bedside," officials pointed out.

"Twenty years ago, AMIA led a national dialogue over how to regulate software,” said Fridsma. “Now is the time to revisit old assumptions, and the informatics community is eager to help.”

CDS Coalition, Health IT Now also weigh in


AMIA hasn't been the only industry group to weigh in on the FDA rules. The Clinical Decision Support Coalition, which was founded as an industry group to prevent CDS tools don't become overregulated, released its own set of comments yesterday, criticizing the draft guidance for expanding the scope of FDA regulations without taking a risk-based approach – grouping together many low-risk programs already on the market.

"The resulting over-regulation would be to the detriment of patient care," said CDS Coalition leader Bradley Merrill Thompson, who leads medical device and digital health practices at Epstein Becker & Green, P.C.

And Health IT Now – a Washington, D.C.-based advocacy group whose dozens of healthcare and technology members include everyone from Aetna to Verizon – issued its own feedback to the agency. In its letter to FDA Commissioner Scott Gottlieb, MD, the group noted that it had worked with policymakers to help craft the language in Section 3060 of the Cures Act.

"We believe that the draft guidance is aligned with Congressional intent and appropriately adopts the risk-based framework called for in the law for clinical decision support software," wrote Health IT Now's Executive Director Joel C. White.

He noted concerns with FDA's draft guidance, however, specifically a section that "establishes that a practitioner would be unable to independently evaluate the basis of a recommendation if the recommendation were based on non-public information."

Language in the 21st Century Cures Act "states that health software that enables a health care professional to "independently review the basis for such recommendations that such software presents" shall not be included in the definition of a medical device under the Food, Drug, and Cosmetic Act," said White. "With this language, Congress intended to prohibit the exclusion of so-called 'black box' software that does not provide transparency in the data that goes into software algorithms. They did not intend to establish a bar that the data must be publicly available."

Health IT Now asked that FDA amend its guidance to reflect that that information should be available to healthcare professionals, not necessarily publicly available.

"Similarly, we are concerned with one of the examples of CDS and other software functions for healthcare professionals that remain devices," said White, pointing to a section of the draft guidance that "sets a standard that the algorithm used in the software product must be disclosed to the user.

"Again, we believe that it was Congress’ intent that the data be transparent to the user, not the algorithms into which the data are entered. We request FDA amend this example to reflect transparency of data, not of the algorithm itself."

To a larger point Health IT Now noted that "throughout all of the examples laid out in the document, there is a reliance on information from clinical practice guidelines or other reference information. However, we believe it is important to recognize the growing inclusion of real-world data in medical decision-making. We request that the guidance document be updated to reflect that many CDS tools are, and more will be in the future, incorporating real-world data into their functions."

Twitter: @MikeMiliardHITN
Email the writer: mike.miliard@himssmedia.com