Human Factors Validation Testing: Do You Actually Need It?

Engineer observing a participant use a handheld medical device in a usability lab from behind glass: human factors validation testing
In this post
  1. What the FDA finalized in May 2026
  2. Category 1: the modification that touches nothing
  3. Category 2 and the rationale that has to hold up
  4. When human factors validation testing is not optional
  5. What a defensible rationale actually contains
  6. The date that mattered was August 1
  7. Where Nopmark fits
  8. FAQ

There are 3 human factors submission categories in the FDA's final guidance, and which one your device falls into decides whether you owe a full usability study or a written rationale. Two of the three categories ask for a rationale rather than human factors validation testing, and one of those two is open to devices that do have critical tasks. Teams that scope a study before working out which category they are in are frequently paying for evidence the submission never asked them to produce.

What the FDA finalized in May 2026

The guidance is called Content of Human Factors Information in Medical Device Marketing Submissions, issued as Guidance for Industry and Food and Drug Administration Staff on 29 May 2026 under docket FDA-2015-D-4599, from the Center for Devices and Radiological Health. It is final, and it closes out a draft that had been sitting since 9 December 2022. It applies to premarket notifications (510(k)s), De Novo requests, premarket approval applications (PMAs), and humanitarian device exemption (HDE) applications.

You reach your category by working through 4 decision points, and it is worth naming them up front because most summaries of this guidance skip straight to the categories and lose the structure that produces them. Decision Point A: is this a modification to an existing device? Decision Point B: is there a change to the user interface, the intended device users, the intended device uses, the intended use environments, the training, or the labeling? Decision Point C: based on your use-related risk analysis, are there critical tasks, or new or impacted critical tasks? Decision Point D: given the history of use, the complexity of the user interface, and the adequacy of existing risk control measures, should human factors validation test data be submitted?

It does not replace the 2016 guidance on applying human factors and usability engineering to medical devices. It sits alongside it as a companion. So the older document still governs how you do the engineering, and this one governs what you hand over about it. Your usability engineering process carries on as before. Your submission content changes.

What FDA has done here is write down a decision structure where reviewers and manufacturers used to negotiate case by case. A structure you can read in advance lets you scope the work before you spend the budget, rather than after a reviewer asks for something nobody planned for.

Category 1: the modification that touches nothing

Category 1 is what you get when you answer Yes at Decision Point A and No at Decision Point B. So the device already holds marketing authorization, and the change in front of you touches none of the 6 items on that list.

What you submit is a conclusion and high-level summary: a statement justifying that the device modifications do not affect the human factors considerations of the modified device. If you are leaning on earlier HFE or UE evaluations, describe those as part of the summary. No usability study.

One nuance from Decision Point A that gets missed. You can answer Yes to it for a genuinely new device, provided that device has the same or a similar user interface as one of your own legally marketed devices and you are leveraging the human factors information from that device. So a new product is not automatically pushed down the new-device branch.

Every item on that list carries weight. So take two firmware changes and ask where each one lands. Does the change alter a screen, an alarm, a workflow, or a set of instructions? A firmware fix that touches none of them is a Category 1 change. A firmware change that adjusts an alarm threshold is not, because the alarm belongs to the user interface and the clinician's response to it is a use-related consideration. Most of the judgment in this guidance lives in the gap between those two examples.

You still have to write that summary and you still have to make it persuasive. A study, though, is not on the list.

Category 2 and the rationale that has to hold up

Here is where most write-ups of this guidance, including several that rank well for it, get the shape wrong. Category 2 is not one route. It is two, and they arrive from different decision points.

The first route is a No at Decision Point C: a new device with no critical tasks, or a modified device with no new critical tasks introduced and no existing critical tasks impacted. You then submit a rationale describing the basis of that determination.

The second route is a No at Decision Point D, and this is the one worth knowing about. Your device does have critical tasks, you reached Decision Point D because of it, and you can still land in Category 2 by submitting a rationale for why human factors validation data are not needed in this submission. FDA lists what that rationale weighs: the user interface history of use for the intended use, users and environments; the complexity of the user interface; and the adequacy of existing risk control measures.

So a device with critical tasks is not automatically a validation study. Teams that read Category 2 as the no-critical-tasks box and stop reading have skipped the route that most established products actually qualify for.

Everything still turns on the term critical task, and the guidance defines it tightly: a user task which, if performed incorrectly or not performed at all, would or could cause serious harm to the patient or user, where harm is defined to include compromised medical care. That last clause does real work. Compromised care counts as harm, so delayed therapy or a missed alarm belongs in the analysis even where nobody is physically injured.

Your determination comes out of the use-related risk analysis, and for a modified device FDA asks you to run the URRA on the final finished device rather than only on the change, so that critical tasks upstream and downstream of the modification are caught. Thin analysis tends to produce thin rationale. No amount of careful writing in the submission will repair it afterwards.

When human factors validation testing is not optional

Category 3 is a Yes at Decision Point D. What you submit is a comprehensive HFE or UE report containing every element the guidance lists in its report section, including the human factors validation testing itself.

FDA also names circumstances where it may decide validation data are needed even when that device type has not historically submitted them: a significant difference from similar legally marketed devices that affects use, such as a novel technological feature, a new indication, a changed use environment or a new user group; new information such as recalls, adverse events, problem reports or complaints identifying a safety signal attributed to use error; or an increase in the severity of possible harm from use error.

Read that list as a warning about assuming your predicate's treatment carries over. For a genuinely new device with a meaningful user interface and a pathway to serious harm through use error, Category 3 is where you should expect to be, and the guidance is not trying to talk you out of it. The 3-category structure exists to set rigor by risk, so that the lowest-risk changes stop carrying the same documentation load as the highest-risk ones.

What a defensible rationale actually contains

Which rationale you owe depends on which No got you here, so write to the decision point you actually answered.

  • For a Decision Point C rationale, describe the basis of your determination that there are no critical tasks, or no new and no impacted ones. Name the URRA it derives from and reference it rather than summarizing it, because a reviewer needs to trace your conclusion back to the analysis that produced it. State which tasks you examined and why each falls short of the critical-task definition. You have to show the negative finding rather than assert it.
  • For a Decision Point D rationale, work the 3 considerations FDA actually lists: history of use for this intended use, these users and these environments; the complexity of the user interface; and the adequacy of your existing risk control measures. A rationale that argues only "we have always done it this way" engages none of them.
  • For a modification, a comparison against the prior version that walks the same 6 Decision Point B dimensions and shows exactly what moved, and for cumulative changes, whether they interact.
  • The known use-related problems review. Complaints, MDRs, recalls and problem reports for your device and for similar devices are the evidence FDA itself points to when it decides validation data are needed after all, so a rationale that ignores that history is arguing against the reviewer's own source.

FDA also publishes sample report outlines for each category in the appendices, and it recommends a Pre-Submission if you are unsure whether a rationale will fly for your specific submission. Both are cheaper than an information request.

None of this is exotic. Most of it already exists inside a competent usability engineering file. What changes is that it now has to be assembled into an argument aimed at a reviewer rather than left as evidence sitting in the design history file waiting to be asked for.

The date that mattered was August 1

FDA said it recognized that the agency and industry might need a minimum of 60 days to operationalize the policies in the guidance. For submissions already pending after publication, and for those received before August 1, 2026, FDA generally did not anticipate that manufacturers would be ready to include the newly recommended information. It added that it intends to review such information if submitted at any time.

That date has now passed. It was never a compliance cliff and nothing becomes unlawful on the other side of it, so treat it as a readiness marker rather than a deadline: submissions going out from here should be built to the 3-category structure, because the agency has now said it expects people to have had time to absorb it.

Where Nopmark fits

Human factors is the medical device discipline where the gap between doing the work and documenting the work does the most damage. Teams run a usability study, learn something real from it, and then present it in a form that answers a question the reviewer did not ask.

Nopmark's Human Factors and Usability Engineering course covers the engineering side and the submission side together: use-related risk analysis, critical task identification, and how validation evidence gets structured for a marketing submission. Critical task determination in particular is the hinge this entire guidance turns on, and it is the hardest thing to pick up from reading the standard on its own.

Ready to get your critical task analysis onto firmer ground? The Human Factors and Usability Engineering course from Nopmark covers exactly this, taught by accredited instructors: https://www.nopmark.com/courses/hfe

FAQ

Is the guidance final or draft?

Final, issued 29 May 2026 as guidance for industry and FDA staff under docket FDA-2015-D-4599. It finalizes a draft that had been open since 9 December 2022.

Does this replace the 2016 human factors guidance?

No. It complements it. The 2016 guidance covers applying human factors and usability engineering to your device. This one covers what human factors information goes into the marketing submission.

What happens to a submission filed before August 1, 2026?

FDA said it generally did not anticipate that manufacturers would be ready to include the newly recommended information in submissions received before that date, and that it intends to review such information if submitted at any time.

Can I really omit validation testing under Category 2?

Yes, by either of two routes. If your device has no critical tasks (or a modification introduces and impacts none), you submit a rationale for that determination. If it does have critical tasks, you can still submit a rationale for why validation data are not needed, weighed against user interface history of use, user interface complexity, and the adequacy of existing risk control measures. The rationale carries the burden either way.

My device has critical tasks. Am I automatically in Category 3?

No. Critical tasks take you to Decision Point D, not straight to a validation study. Category 3 is a Yes at that decision point.

Which submission types does it apply to?

510(k) premarket notifications, De Novo requests, PMAs, and HDE applications.

How does this interact with IEC 62366?

The guidance governs FDA submission content. IEC 62366-1 governs the usability engineering process itself. Running a compliant process under the standard does not by itself tell you which of the 3 categories your submission falls into, and that determination is what the guidance is for.

Keep reading