An Update on Human Factors Usability Testing for Medical Devices

In “The Importance of Usability/Human Factors Testing” (written for MPO in 2020), the central question was simple: When is the work of human factors over, and how does a team know it is ready for summative testing?¹ The answer at that time—human factors engineering (HFE) is never truly finished, only folded into the iterative design cycle the FDA expects—has held up. What has changed, however, is the regulatory scaffolding around that cycle. Six years later, the fundamentals are intact, but the way manufacturers document and justify their human factors work in a marketing submission has been substantially rewritten. This is an update for the engineers who live inside that cycle. 

First, the following outlines what’s to be gained from the effort in human factors. 

• 400% ROI: The potential increase in conversion rates achieved through superior user experience (UX) and intuitive interface design.
• 75% Credibility: The percentage of users who judge a company’s overall credibility based purely on its design.
• 70% Churn: The proportion of consumers who will abandon a brand after experiencing just two bad user experiences.
• 54% Usability Weight: Over half of a user’s overall experience rating is determined by just two factors: how useful a product is and how easy it is to use.

The Fundamentals Still Hold 

Let’s start with what has not changed, because it matters. The case for usability testing rests on the same uncomfortable statistics. A widely cited 2016 Johns Hopkins study estimated that more than 250,000 Americans die each year from medical error, which makes it the third leading cause of death behind heart disease and cancer.² Other work has put the figure as high as 440,000 deaths annually, arguing that error and system failure are chronically under-recorded on death certificates.³ Whatever the exact number, the mechanism is the same: use error is a systematic error, not a random one, and it cannot be inspected out at the end of a program. It is designed out, iteration by iteration.

The workhorse standard, IEC 62366-1:2015 consolidated with Amendment 1:2020, remains the recognized consensus standard for applying usability engineering to medical devices, and the FDA still lists it in its Recognized Consensus Standards database. Amendment 1 corrected inaccuracies without disturbing the underlying usability engineering process, so teams that built their file around the 2015 process did not need to start over. The sample-size heuristics cited in the 2020 article also endure: five to eight participants per formative round will surface the majority of usability problems, and a minimum of 15 participants per distinct user group is expected for a summative validation study. That floor traces to Faulkner’s 2003 study, which collected empirical data from 60 users and found that a sample of 15 uncovered an average 97% (and a minimum of 90%) of all problems, compared with roughly 85% for the old five-user assumption.⁴

It is worth restating the division of labor, because the new guidance sharpens rather than replaces it. Formative testing is diagnostic: run with roughly five to eight users, iterate, and use the results to fix the interface while change is still cheap. Summative testing is confirmatory: a single validation study, run on the final production-equivalent design with representative users under realistic use conditions, whose purpose is to demonstrate that critical tasks can be performed safely and effectively. Skipping formative work to save budget is a false economy. The risks that are not exposed early do not disappear; they simply resurface in the summative, where a failure is far more expensive to remediate.

The tactical advice from the original article is as true today: test formatively and test often; recruit naïve users first to gauge how intuitive the interface is, then enlist experienced users to benchmark against competitive devices; and above all, do not over-train participants before a test and mask the very use errors you are trying to expose.

From product launches to M&A and manufacturing trends, the MPO Source delivers the medtech news you need twice a week. Subscribe here!

What Actually Changed: The Submission Rewrite

The biggest shift is in how human factors information reaches the FDA. For years, the anchor was the 2016 guidance Applying Human Factors and Usability Engineering to Medical Devices, paired with a companion List of Highest Priority Devices for Human Factors Review. That list told manufacturers for which device types the agency expected full human factors data. It has been retired.

In its place, the FDA finalized Content of Human Factors Information in Medical Device Marketing Submissions on May 29, 2026, after an extended draft period that began in 2022. This is not a cosmetic revision. It replaces the highest-priority-device list with a risk-based framework that sorts every submission—510(k), De Novo, PMA, and HDE—into one of three human factors submission categories, driven by whether the device involves critical tasks: tasks that, performed incorrectly or not at all, would cause or contribute to serious harm.

In broad strokes, the highest category applies when a device has critical tasks that require summative validation data in the submission; a middle category applies when there are no critical tasks, or no new or impacted critical tasks for a device modification, provided the manufacturer supplies a documented rationale; and the remaining category covers submissions where the human factors discussion is minimal. The practical consequence for engineers is that the use-related risk assessment (URRA) is no longer a supporting appendix; it is the document that determines what you owe the agency. Your identification and categorization of critical tasks now set the evidentiary bar for the entire submission.

Timing is concrete. Beginning Aug. 1, 2026, the FDA’s revised non-IVD and IVD eSTAR submission templates prompt manufacturers to declare their human factors submission category and provide the corresponding information. The framework is not aspirational; it is built into the submission tooling teams already use. Programs in process should be mapping their existing URRA to the new categories now, rather than revealing the gap at review.

Critical Tasks Are the New Center of Gravity

If there is one concept to internalize from this update, it is the primacy of the critical task. The old mental model asked, “Is my device type on the priority list?” The new model asks, “Does my device have critical tasks, and have I identified, mitigated, and validated them?” That is a better question; device-specific and risk-driven rather than category-driven, but it places more analytical weight on the front end of development.

For engineers, the use-related risk analysis must be rigorous and defensible early. Task analysis, identification of potential use errors, the mapping of those errors to hazards and harms, and the risk controls applied to each critical task all need to be traceable through the design history file. A team that treats the URRA as a late-stage compliance artifact will be unable to justify its submission category, or to explain why a given task was, or was not, deemed critical. It is worth remembering context here: device use errors have repeatedly ranked among the top root causes in FDA recall data, which is precisely the failure mode this front-loaded analysis is meant to catch.

AI, Software, and the Next Frontier

The other development worth flagging is the collision of human factors with AI-enabled software as a medical device (SaMD). The FDA’s January 2025 draft guidance, Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations, signals expectations that go beyond traditional usability: assessment of user cognitive load, transparency about how a model reaches its output, and documentation of how users recognize and override the device. These are human factors problems in a new register. When the user interface is a probabilistic recommendation rather than a fixed physical control, the definition of a use error becomes genuinely harder, and summative study design must account for automation bias and appropriate reliance. Engineers building AI-enabled devices should expect their human factors program and critical-task analysis to expand accordingly.

IRB and the Medi-Vantage Perspective

The institutional review board (IRB) questions raised in the 2020 article have not gone away. Whether a usability study needs IRB oversight remains a case-by-case judgment driven by the risk the study itself introduces to participants—an injection device carrying needlestick risk, or a transdermal patch carrying an allergen exposure. Low-risk studies in which the testing organization has taken appropriate protective measures can still make a defensible case to waive IRB submission, but that decision should be documented, not assumed.

Our perspective at Medi-Vantage is that the 2026 guidance rewards teams that were already doing human factors well and penalizes those who treated it as a documentation exercise. The winners will be programs that build the URRA early, keep critical-task reasoning traceable, and run enough formative testing to enter summative validation with few surprises. The framework has changed; the discipline has not. Human factors is still never “over,” but as of this year, how you prove you did it has a new shape. Engineers who align their design controls to critical tasks now, ahead of the August 2026 template change, will spend far less time reconstructing that story later.

References

1. tinyurl.com/mpo260901
2. tinyurl.com/mpo260902
3. tinyurl.com/mpo260903
4. tinyurl.com/mpo260904

More from this author: Ten Trends and Best Practices with Today’s Medtech M&A

Maria Shepherd has more than 20 years of experience in marketing in small startups and top-tier companies. She founded Medi-Vantage, which provides marketing and business strategy for the medtech industry. She can be reached at [email protected]. Visit her website at www.medi-vantage.com.