On November 4, 2021, the US Food and Drug Administration (FDA) released new draft guidance for premarket submissions of device software functions. First issued in May 2005, this much-needed update was a long time coming for the MedTech industry. As a result, the new draft intends to reflect the rapidly evolving landscape of digital health and medical technology. Read on for more clarity on the contents of the draft guidance, how it has changed since 2005, and how it will affect your MedTech development and FDA approval process efforts!
New draft guidance for device software FDA premarket submissions
On November 4, 2021, sixteen years after the original “Guidance for the Content of Premarket Submissions for Software Contained in Medical Devices” was issued, the FDA has released an updated draft version of this document. After undergoing quite an overhaul compared to the 2005 version, the draft provides up-to-date guidance for what documents the FDA expects sponsors to provide for premarket submissions. These documents are what enable the FDA to assess how safe and effective device software is.
Initially slated for the end of 2019, this new version has been a long time coming — after all, sixteen years is an eternity when it comes to tech. Medical device development has evolved significantly since 2005, especially when it comes to embedded software. Bakul Patel, director of FDA’s Digital Health Center of Excellence in the Center for Devices and Radiological Health, elaborates:
“As technology continues to advance all facets of health care, software has become an important part of many products and is integrated widely into medical devices. The FDA recognizes this evolving landscape and seeks to provide our latest thinking on regulatory considerations for device software functions that is aligned with current standards and best practices.”
Who and what does the guidance apply to?
This guidance applies to MedTech developers and covers all “device software functions” that align with the device definition from section 201(h) of the Federal Food, Drug, and Cosmetic Act (FD&C 17 Act).
This means that it covers both software in and software as a medical device, known respectively as SiMD and SaMD. SiMD refers to your traditional hardware that comes with embedded software, while SaMD is software that functions independently of any other medical device. Overall, the scope is quite similar to the 2005 guidance in that it is intended for:
- Firmware and other means for software-based control of medical devices
- Stand-alone software applications
- Software intended for installation in general-purpose computers
- Dedicated hardware/software medical devices
- Accessories to medical devices when those accessories contain or are composed of software
It is also worth noting that the FDA emphasizes that this guidance does not apply to automated manufacturing and Quality System software that is not a device.
In terms of premarket submissions, the guidance applies to all types, including:
- Premarket Notification (510(k))
- De Novo Classification Request
- Premarket Approval Application (PMA)
- Investigational Device Exemption (IDE)
- Humanitarian Device Exemption (HDE)
- Biologics License Application (BLA)
What has changed since the 2005 guidance was issued?
The 2021 draft presents a new model for what sponsors should provide in FDA premarket submissions for device software. Although many of the requirements are the same or similar, this version notably establishes a risk-based approach to premarket submissions as well as enhanced documentation levels. Regardless of the level, sponsors are still expected to accurately describe the software and demonstrate what the associated risks are and how they have been mitigated.
Here are some of the other main changes you should be aware of in the 2021 version:
- Software Device Documentation Categories
The 2005 guidance document used three categories to determine what documentation was needed for software, also known as Major, Moderate, or Minor Level of Concern. The new draft guidance shifts to two categories instead; device documentation levels can now either be Basic Documentation Level or Enhanced Documentation Level. The way to determine which category is suitable is by assessing four risk-based factors, i.e. if the device:
- is a constituent part of a combination product
- is intended to test blood donations, or to determine donor and recipient compatibility, or a blood establishment computer software
- is categorized as a Class III device
- or presents a probable risk of death or serious injury upon failure
Then the device is categorized as Enhanced Documentation Level. All in all, the main difference between the two categories is that the Basic Documentation Level is not required to submit the Software Design Specification, while the Enhanced Documentation Level is.
- Risk Management
One of the most obvious changes in the 2021 release of this guidance is the approach taken in regard to risk management. In contrast to the 2005 version, the draft text puts forward a risk-based approach to the type and amount of documents required from sponsors. Previously, sponsors were required to submit what was called a hazard analysis. According to the new draft guidance, sponsors now need to submit a Risk Analysis as well as Risk Management Plan and Risk Management Report.
- Artificial Intelligence / Machine Learning Software Devices
Another prominent change are the updates the FDA has included regarding guidelines for software devices that leverage Artificial Intelligence (AI) and Machine Learning (ML). The 2021 release of the guidance document takes a full lifecycle approach to AI-powered software, encompassing its full journey from conception to deployment. The guideline also states that when you are preparing the description of your AI/ML-powered software, you need to include populations or samples which contributed to the design of your model, outline potential biases and limitations, and describe what steps were taken to deal with them. Since AI and ML are increasingly used in medical devices, this has been hailed as a welcome change by industry players as a whole and is seen as a foundation for Good Machine Learning Practice (GMLP).
In the FDA’s press release, Patel summarized: "We anticipate this draft guidance, which fulfills FDA's commitment in MDUFA IV, will provide clarity, simplicity and harmonization with current best practices and recognized voluntary consensus standards, once finalized”. As stated, this draft guidance for device software in FDA premarket submissions is not the final version — it is being distributed for commenting purposes only. Anyone interested in submitting their feedback to the FDA can do so here.
For up-to-date information on compliance in medical device development in the EU, check out our related white paper authored by independent medical experts: