In 2023, the FDA identified 62 medical device recalls with 42 identified to date in 2024. Most recently, Tandem Diabetes Care recalled its Apple iOS app after over 200 reported injuries. While the issues behind the 2024 recalls are diverse, software problems are an important theme. Software, while transformative for healthcare and other safety-critical sectors, poses significant risks when it fails — as in the high-profile cases of the Boeing 737 Max 8 and Tesla. For medical device manufacturers, detecting and addressing software issues is vital to ensure patient safety — especially as devices increasingly depend on software and AI for diagnostics and treatments.
Let’s take Smiths Medical as a recent example of why software issues are so difficult to troubleshoot. Smiths recently recalled approximately 86,000 syringe pumps, specifically the Medfusion 3500 and 4000 models. These two models have been recalled every other year since 2020 due to software problems. A critical FDA warning letter sent to Smiths in late 2021 highlighted systemic issues with their existing design validation procedures and practices.
Validation is paramount in ensuring trust in safety-critical devices. It confirms a device’s design follows the approved design plan for medical devices in the US, and conforms to FDA Quality System Regulations (QSR: 21 CFR part 820). Smith’s failure to provide sufficient evidence that each device fully complied with this section highlights a significant quality issue and perhaps a broader industry problem. Understanding the difficulty of meeting these rigorous standards is key to preventing quality problems in the future.
Using Informed Awareness to Transform Care Coordination and Improve the Clinical and Patient Experience
This eBook, in collaboration with Care Logistics, details how hospitals and health systems can facilitate more effective decision-making by operationalizing elevated awareness.
Without knowing Smith’s specifics, the difficulty in meeting these standards most likely stems from software’s growing complexity and the inability of current safety critical infrastructure to keep pace. The infrastructure and quality systems deployed in most medical device companies were designed around 1990s hardware systems, which never anticipated today’s software complexity.
Smiths Medical, part of ICU Medical, is a major player in the infusion therapy market with $2.5 billion in sales. Despite their deep pockets and commitment to quality, they face common industry hurdles: documentation, traceability, and reliability. Their struggle to grasp how systems are interconnected and the ripple effect of changes highlights a broader industry challenge.
Today’s software landscape is fundamentally different than it was just 20 years ago. It’s now cloud-based, incorporates artificial intelligence and machine learning (AI/ML), and relies on open-source components that can harbor unknown vulnerabilities. While leveraging open-source libraries and off-the-shelf components can accelerate time to market, they often lack the stringent controls of regulated software built from scratch. Unidentified vulnerabilities in medical devices connected to the Internet pose security risks that can result in malfunctions, service interruptions, or malicious hacker attacks that could endanger patients or entire hospital networks. The usage of AI – and the extensive data required – introduce additional complexity.
According to a McKinsey study, between 2006 and 2016, the average design complexity and total effort for medical software surged by over 30% CAGR while productivity remained flat, rising just 2% annually. McKinsey attributes the trend to software differentiation (i.e. functionality) as well as the growing adoption of cloud-based solutions.
At ViVE 2024, Panelists Share Prior Authorization Progress and Frustration in Payer Insights Program
At the Payer Insights sessions on Day 1 of ViVE 2024, a panel on prior authorization offered compelling insights from speakers who shared the positive developments in this area after years of mounting frustration. Speakers also shared challenges as they work with providers to figure out how policy developments and technology will work in practice.
Historically, quality and development teams have used disparate tools and methodologies that are not integrated because of varying priorities – quality emphasizes compliance while development prioritizes building the product. This misaligned approach often results in retroactive attempts to piece together traceability matrices and design documentation by manually copying and pasting from different systems to assemble their Design History File (DHF). This after-the-fact method presents two key problems. First, the resulting DHF is often incomplete because of the manual transfer of data between systems. This makes it challenging to identify the root causes of device complaints when it is in the market and prolongs resolution times. Second, this process is highly inefficient and error-prone, consuming precious time that could be better spent on improving the product.
Relying on outdated tools such as Excel for traceability documentation has become particularly problematic because of version confusion and human errors when manually transferring data. Tracking changes in Excel is especially challenging with a single feature tied to multiple users and thousands of interdependencies. Today’s systems, which can comprise tens of thousands of features, simply overwhelm Excel’s capacity. Despite their widespread use, these tools can’t support the volume of data in modern technologies and the fast iteration cycles development teams prefer, resulting in version control issues and data inaccuracies which can compromise patient safety.
How can we fix this issue? Let’s use technology and the bright and dedicated people working in MedTech according to the following principles:
- Utilize machines for process enhancement. Computers excel at following processes, a key component of regulatory compliance. Machines efficiently track complex relationships and manage vast amounts of data that humans can’t – such as the interaction between a traceability matrix and the entire contents of a DHF. Computers can also automate testing protocols to identify and fix problems earlier in the development cycle.
- Manage the risk, not the paperwork. In contrast to machines, humans are strong at thinking. By empowering teams to spend their time evaluating, and mitigating risks, rather than managing documents, better patient outcomes will follow.
- Adopt modern tools to create modern products. If MedTechs can upgrade their infrastructure to utilize today’s development tools while maintaining compliance, medical innovation will soar. By reducing the compliance burdens and leveling the playing field for emerging companies, developers and innovators can swiftly bring new products to market, making high-quality healthcare more accessible to a broader audience.
The sophistication of software-driven applications in healthcare will only increase over time. Cutting-edge technology empowers developers to proactively detect and mitigate issues before they cause problems, reducing recalls and fostering trust in the industry. By embracing modern tools, manufacturers pave the way for a culture of continuous improvement and enhanced patient safety.
Photo: metamorworks, Getty Images
Erez Kaminski is the CEO and Founder of Ketryx Corporation, the first and only connected lifecycle management software for MedTech that unifies quality and R&D efforts to deliver safer, more innovative software 3x faster through the tools developers know and love. Over the last decade, he worked in industries including computational mathematics, biotech, and energy, helping build monitoring systems for pharmaceutical equipment and AI for medication management. Erez is deeply concerned with improving patient care and health outcomes with software solutions.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
