Australia's Therapeutic Goods Administration (TGA) recently posted a draft guidance for regulation of software as a medical device, but the Medical Technology Association of Australia expressed a preference for an international standard for risk classification, in large part because the U.S. FDA has adopted that risk scheme.

The TGA opened the consultation period in February 2019, noting that the scope of the draft was for software as a medical device (SaMD), while software in a medical device is regulated as part of that parent hardware device. The agency said that the transition period for any final regulation would align with the European Union's roll-out of the Medical Device Regulation, which formally goes into force in May 2020 (the related regulations for in vitro diagnostics will not be enforceable until 2022).

One of the concerns expressed by TGA is that SaMD could be imported into Australia "in large volumes" without registration in the Australian Register of Therapeutic Goods (ARTG) by an Australian sponsor. At present, this is allowed only if an Australian citizen receives such items from a supplier or personal acquaintance, including via the Internet, and so long as those items are not distributed to other parties.

IMDRF, EU approaches alike but different

TGA indicated it may use the risk classification principles spelled out in the 2014 guidance by the International Medical Device Regulators Forum, but noted that the European Commission has devised it's own approach, which spans classes IIa, IIb and III (for low-to-medium risk, medium-to-high risk and high risk, respectively). The TGA said the European approach "is in accordance with the IMDRF recommendations," but that the EU approach fails to provide sufficient detail "to capture the different risk categories" for SaMD as found in the IMDRF document.

TGA proposed that for software that processes data to provide information, a class III article would be one that makes a direct diagnosis for a critical situation in which the disease or condition "is fatal or debilitating in a short time frame." The class III designation would also apply if that diagnosis is critical to ward off a risk to public health and when the making of that diagnosis is needed to prevent "a serious deterioration in a person's state of health if not identified." Otherwise, that article would fall into class IIa.

Any screening diagnostic that deals with a condition that is fatal or debilitating in a short time frame would also fall into class III, TGA said, which again applies to a technology that addresses a predicament with public health implications. When the underlying condition could cause a serious deterioration in health, the item would be class IIb, while all other uses of screening diagnostics would fall into class IIa.

No interest in Australian-only approach

The Medical Technology Association of Australia (MTAA) said in response that the association supports an update to regulations for medical device software, but that the group "does not support an Australian-only regulatory framework." MTAA said it prefers the IMDRF approach to risk to the EU approach, adding that the TGA's framework appears to have no direct correlation with either the IMDRF or the EU frameworks.

MTAA said the EU criterion of "monitor physiological processes" could be deemed roughly equivalent to the IMDRF criterion of "drive clinical management," but that the two non-Australian frameworks could lead to divergent risk classifications. One example is when the IMDRF approach provides a risk classification of IIa for products that drive clinical management for serious conditions, the parallel for which under the EU framework would result in a class IIa designation.

MTAA argued further that the EU definitions "will lead to over-regulation of software in the EU and will be out of synch" with the approaches exercised in "the rest of the world, including the U.S." The MTAA letter noted that the FDA is currently chairing the IMDRF work group on SaMD.

Another point raised by MTAA is that while the association generally supports the proposal to remove the regulatory exemption from SaMD imported for personal use, there is a question as to whether TGA will be able to enforce such a requirement. The group said another issue is that the exclusion of import for personal use would preclude users from access if the product's suppliers fail to provide Australia with sufficient volume to meet demand.

The Australasian Society for Ultrasound in Medicine (ASUM) said its concerns revolve in part around the use of software that could be downloaded to a smartphone or computer and used in the absence of professional training and experience. Thus, the ASUM said, it would prefer that such products not be available for personal use.

The ASUM recommended that TGA consider the management of artificial intelligence and machine learning algorithms in ultrasound systems, but pointed to the advent of portable ultrasound systems that connect to smartphones. ASUM further noted that the FDA's approach to clinical decision support (CDS) systems appears to be "more focused on the stimulation of innovative technology and greater competition" than the approaches adopted by other jurisdictions. However, the letter noted that the European Union appears to be "preparing for radical changes led by the development of artificial intelligence," and may reduce regulatory and legal hurdles in order to promote such developments.

MSIA urges TGA to call the whole thing off

The Medical Software Industry Association (MSIA) said the addition of software to the definition of a medical device could have unintended consequences in that it could affect patient safety and the delivery of medical software "without evidence of commensurate benefit."

MSIA said "we respectfully suggest that the first proposed change to the framework should be deleted" in reference to the discussion of risk classification. The association said the FDA had "explicitly excluded" most medical software that supports or provides recommendations about prevention, diagnosis and treatment of diseases because clinicians, rather than developers, "are trained to assess risk and provide treatment." In the absence of a wholesale removal of the risk framework, MSIA said, the TGA should convene with stakeholders to make recommendations in an effort to head off "the likely unintended consequences for all Australians."