With the growth of health apps, entrepreneurs often need to determine if FDA regulates their particular mobile app. As I’ve explained in other posts and indeed in a book, that question has a number of facets to it, but in this post I want to focus on the most important – the intended use of the app. This concept of intended use is indeed not just the determining factor in deciding whether FDA regulates an app, but for those apps FDA regulates the intended use determines the degree of regulation.
Paraphrasing a bit, FDA regulates any smart phone apps intended for use in the diagnosis or treatment of disease or other conditions. And that intended use is determined mostly by examining the developer’s promotional claims as well as the features of the app.
Let’s focus here on promotional claims. This is where the art comes in. There’s a reason English is taught in liberal arts colleges, and not engineering. Words can be strung together to communicate an endless array of ideas, some of them different only in quite subtle nuances. There is nothing linear about the English language. We can state a concept clearly, or we can state it generally, or we can state it cleverly in ambiguous terms. We can be explicit or we can imply things. Language is an art form.
So if FDA’s jurisdiction is determined by the language we use, and language is an art form with many ways to express closely related but not identical thoughts, how do we know exactly where FDA draws the regulatory line? The most honest answer is we don’t, at least not exactly. There is judgment required.
But that’s not to say that chaos reigns. The system has been working reasonably well for decades because there are ways to discern which intended uses FDA regulates.
Process for Discerning the Scope of FDA Regulation
Here’s the secret sauce:
1. If you are an engineer, don’t try this alone. I love working with engineers; my son is studying to be one at Purdue. But understanding intended use is not an exercise in linear thinking. Instead, it’s a matter of Gestalt. You have to look at the totality of the circumstances-- all of the claims, as well as the design features and the reasons for those features-- and weigh all of that. If that’s not your strong suit, find a liberal arts major. Most bartenders can probably help you.
2. Look at FDA guidance. There are literally hundreds of guidance documents on the agency’s website. You may not be able to stop at just one and get a complete answer. Make sure you scour for clues more completely. For mobile apps, you probably would start with the mobile medical app guidance, but be sure to consider the draft guidances on wellness and the circumstances when a mobile app might be an accessory to a medical device. Also, remember that FDA in effect keeps updating these guidance documents by posting new examples to its website.
3. Look at the FDA classification regulations. There are literally hundreds of them, but by using key terms in your search you can find the ones that are most relevant. Obviously don’t look for medical apps per se, but look for functionality. If FDA has regulated the functionality in other forms, it will regulate the same functionality if done through a mobile app. It gets a little dicey because FDA sometimes leaves outdated regulations in place that are in effect modified by guidance documents under so-called enforcement discretion. Sorry, that’s just the way it is. Don’t neglect the guidance documents.
4. Look at FDA 510(k) clearances. In a practical sense, FDA interprets the classification regulations when deciding which medical devices need to be in class II as it issues these clearances.
5. Look at old clearances. Understanding the scope of class I is more difficult because typically there are not 510(k) clearances (remember, most class I devices do not require FDA clearance to be marketed), but look anyway because device categories that are class I now may have been class II previously in which case the prior clearances give you some clues.
6. Consider enforcement actions. FDA periodically pursues companies that are marketing medical devices if they have not come into compliance beforehand. Stern warning letters are posted on the agency’s website, but the softer enforcement letters are not. So you need to depend on word-of-mouth to hear about the other letters.
7. Consider the risk. Remember, FDA regulation is about protecting the patient. Consider the failure modes, and consider what could happen in unfortunate circumstances. Could someone get hurt by using the product as intended? Be fair and objective; don’t let your emotions cloud your judgment. If there is risk, there is a risk FDA might regulate the app.
8. If it’s still ambiguous, ask someone. You can either ask FDA or a regulatory affairs professional who may have prior experience in getting clarification from FDA on the particular technology. You can ask the agency informally over the phone, via email, or in a meeting, or you can request a definitive decision in the form of a 513(g) letter.
That’s the process; now let’s talk about the substance.
Concept of Intended Use
As I said, the reason this can be tricky is simply the inherently wide range of possible intended uses that may only have subtle differences. But even differences that appear subtle can have a profound impact on the regulatory consequences.
At first glance, the intended use concept may look linear. You could conceivably write on a chalkboard horizontally an array of intended uses: anesthesiology, cardiology, clinical chemistry, dental, gastroenterology, ophthalmology, obstetrics, radiology and the list goes on. Obviously within each of those broad categories, it would be easy to list discrete intended uses. Within radiology, you have x-rays, MRIs, mammography and so forth.
But that’s the easy part. The hard part is assessing the regulatory consequences of the specificity of the intended use (no I don’t mean specific), and any implied claims. You can sell a scalpel claiming that it is useful “in cutting tissue”, or you can sell the exact same scalpel claiming it is useful “in Coronary Artery Bypass Grafting.” FDA looks at those claims differently because even though the second claim is in a sense contained in the first – even in the Coronary Artery Bypass Grafting, the scalpel is only cutting tissue – there is also a claim of suitability for a particular surgery. And that claim carries different risk than the first.
Below, I have attempted to depict conceptually the relation between different levels of general versus specific claims. But this depiction doesn’t capture the nuance of the exact wording of the claim. Consider, for example, a mobile app sold “to use in the diagnosis of congestive heart failure,” versus a mobile app sold “as an aid in the diagnosis of congestive heart failure.” Claims that a mobile app merely aids diagnosis are lower-level claims that get regulated to a lower degree than a pure diagnostic claim. And screening is different from monitoring which is different from diagnosis, even though the technology might in fact be the same.
Magnifying Glass Case Study
To explain that conceptual depiction, I’d like to make this a little bit more concrete. So I will use an example. Let's say we have a cell phone app that basically uses the camera functionality to visualize stuff on a magnified basis. It’s an electronic magnifying glass.
In the example below, it’s important to remember the medical device classification scheme, which divides all regulated medical devices into one of three classes. Class I includes the lowest risk devices, and they typically do not require any sort of FDA review prior to marketing. Class III includes the highest risk devices, and they typically require clinical trials and a very demanding FDA approval process. Class II products are in the middle, and typically require that FDA review a premarket notification and conclude that the device is substantially equivalent to other devices already in that category. For some products that would literally qualify as a medical device but are very low risk, FDA has created a new category that it refers to as “enforcement discretion.” Enforcement discretion basically means that the product meets the literal definition under the statute, but FDA has decided that the risk is so small that it does not make sense to expend its resources enforcing the law on those products. So it basically gives makers of those products a free pass. If it doesn’t even meet the literal definition of a medical device, I refer to it as unregulated.
All of the following claims are consistent, but differ largely in specificity. For each one of them, I identify the regulatory status, and explain how I reasoned my way there.
1. “It works like a general purpose magnifying glass” – unregulated. No even a hint of medical use.
2. “This app magnifies stuff 100x” – unregulated. Simply a tool type claim, but with no clinical use suggestion.
3. “It's a magnifying glass for looking at skin” – unregulated. People look at skin for cosmetic reasons, not just medical.
4. “It's a medical-grade magnifying glass” – enforcement discretion or unregulated. Even though it hints at medical use, I believe FDA would decide that it is very low risk and not worthy of FDA regulation. Indeed, a lawyer would argue that it can't be regulated because there's no claim related to diagnosing or treating a disease. The claim is so general that it includes helping a doctor look at fine print on patient consent forms. That’s not a regulated intended use.
5. “It's a magnifying glass for doctors.” Same as above. This is where FDA has been enlarging enforcement discretion, allowing people to say generally that software has uses within the practice of medicine. This is particularly true for hardware and software used for telemedicine. FDA is allowing companies to make claims at a very high level around their products being used in remote encounters by a doctor without crossing the line into regulated territory. Please understand that literally this claim includes use in diagnostic procedures, but the claim does not explicitly say that.
6. “It's a magnifying glass for looking at moles.” This may have now crossed the line into class I. There is a pretty common understanding of why someone would want to magnify a mole, and it is not for cosmetic purposes. At the same time, FDA could decide that it is in enforcement discretion because it is so simple that the failure modes are not worthy of FDA regulation. This is an evolving area, and FDA is constantly down classifying low risk apps. So this might be a candidate, but it would be one that you would need to talk with FDA about to be sure. This example raises another complication: there needs to be an existing classification regulation for this particular technology to avoid the highest classification. If something is truly novel, it can end up in class III simply because there have been no products like it before. In those cases, the software developer actually has to blaze the trail by getting FDA to create a new, low risk classification through a process called de novo review. And that process is tedious.
7. “It’s a magnifying glass for use by a patient with impaired vision” – class I. Indeed, this is covered explicitly by the classification found at section 886.5540 in title 21 of the Code of Federal Regulations.
8. “It's a magnifying glass for looking at moles at 100x, ideal for seeing evidence of possible melanoma.” From a risk standpoint, this one sits right on the border between class I and class II. On the one hand, there's absolutely nothing special about the software, it’s just a magnifying glass. And it basically still relies entirely on the doctor for the diagnosis. Its failure modes are pretty simple; if it doesn't work it will be obvious to the user. But it makes a claim that somehow the special feature of 100x is ideal for that use, and that may push the app into class II. This is like my scalpel example above. Complicating things, there may not be an established classification regulation for it. Remember too that the claim must be supported by appropriate evidence, but that is a separate issue.
9. “This app magnifies moles to help visualize them, and includes algorithms that help determine whether features of a mole suggest melanoma.” This may be Class II. FDA has said that these are class II when marketed to consumers because consumers are not independently trained in assessing melanoma. FDA may still treat it as class II even if the promotional claims make it clear that the product is only for use by a physician, and the physician is expected to make the diagnosis. To me the question would be whether the doctor is likely to rely on the app, or use her own judgment. This is an issue we hope FDA clarifies in an upcoming guidance document on clinical decision support software.
All of these claims cover the same basic functionality, and arguably just become more specific as you go down the list. But these claims differ profoundly in the regulatory treatment they trigger.
The Best Path
As the manufacturer/developer, you control the intended use of your product. It’s your intent that matters, as evidenced by promotional claims you make and design features you add to the product. So be very deliberate in selecting the intended use for your product. On the one hand, don’t be general about the intended use if general won’t sell. But on the other hand, don’t be more specific than you need to be to achieve your commercial objectives. Above all, don’t be disingenuous. If you limit the labeling on the product to very general claims, but then use back channels like social media or other communication avenues to suggest different uses, FDA will likely nail you. And if you’re an engineer and this article makes you need a drink, just ask the bartender to explain it to you.