If I invented an unprecedented mobile technology for treating tennis elbow, the law would require me to do substantial and expensive clinical trials before bringing it to market. In my last post, I argued that the FDA regulatory system is structurally biased against new technology, including novel software, even if it is low risk. By law, any medical device that is both new and novel—as opposed to an incremental gain-- is put in regulatory purgatory no matter how simple and low risk it might be. To me, that’s an overreaction to the risk of the unknown. And as I point out in this post, it’s a costly one.
Much of the technology now being deployed, or about to be deployed, in mobile health applications is not terribly new. It has been around for years if not decades. But the uses of the technology are sometimes new, even radically new.
In a post a couple of months ago, I mentioned that I really liked the technology behind Uber. In comments on the post, one person laughed at me, and called my observation “a knee slapper.” The person pointed out that any engineering sophomore could probably create the software necessary to make Uber work. Point taken. I was probably showing my age (I still marvel at fax machines). But I think we can all agree the application of the technology to the particular use case is new. Indeed, the Uber use case is disruptive.
The primary determinant of FDA regulation, as I have written before, is the “intended use” of a product. It isn’t so much the technology itself, as the use case that the manufacturer recommends for the technology. FDA bases both its jurisdiction, as well as its decisions on the appropriate classification, on the intended use.
Let’s say I make a video game. As videogames go, there’s nothing especially unique about the computer science behind this particular video game. But from a content perspective, I design the game to be particularly engaging based on cognitive science (which, please assume, I have thoroughly documented). I promote this videogame as a way to improve attention for those with ADHD. ADHD is a recognized and indeed serious condition (my son has it.)
By claiming that the game in effect can treat a disease, I have transformed the videogame into a medical device regulated by FDA. Because there’s never been before a video game that’s been authorized as a treatment for ADHD, it is novel in FDA’s eyes. It is automatically in class III. As a result, the law requires significant regulatory hoop jumping and clinical trials. The technology is old. The use case is new and novel.
“New,” for purposes of this post, simply means that FDA has not had the opportunity to create a regulatory classification for it.
What’s not new?
Roughly 80 percent of what I see coming down the pike in mobile health that qualifies as a medical device does not qualify as new in a regulatory sense. Take, for example, the Biosense product called uChek. UChek got a fair amount of press two years ago because FDA sent an enforcement letter to Biosense. The company had put a product on the market for doing urinalysis. On the one hand, the product is a technological marvel. The brilliant developers were able to use a cell phone to conduct urinalysis instead of the previous machine which was about the size of a cash register. uChek offers great promise in that it reduces the cost of urinalysis and makes it much more widely available.
But the fact is, from an FDA standpoint, it wasn’t terribly “new”, only incrementally so. The basic operating principles for the system were the same as prior chemistry-based urinalysis. A person pees in a cup, puts a test strip in the urine, and then uses the camera on the cell phone to measure the change of color on the strip. The cleverness of the developers is in figuring out how to use a smart phone to do the necessary measuring of the color change. It’s ingenious. But it’s not terribly new in a regulatory sense. Indeed, that’s good news for the company. If the product had actually been new, their pathway to market would’ve required them to go through a class III designation requiring perhaps a million dollars. As it is, all they have to do is demonstrate that their product is substantially equivalent to the cash register type machines to be lawfully on the market.
Many of the new mobile health products I see fall into similar cases of incremental improvements using substantially the same technology but on a different platform. The technique for doing EKG on a cell phone is another clever, incremental improvement. However, the basic technique of measuring EKG is not new, and there is an established FDA classification for it. Viewing x-ray images is nothing new in regulatory speak, even on a cell phone.
Examples of new stuff
But there is, clearly, new mobile health hardware and software coming to market. And it’s the truly novel use cases I’m worried about. Consider the following examples of new software that represent potential game changers:
- A mobile app that uses the camera to analyze moles for potential melanoma using fractal geometry as an aid to physician decision-making.
- A mobile app that uses the microphone to analyze snoring sounds for possible sleep apnea as an aid to physician decision-making.
- A mobile app that uses the phone screen to conduct vision acuity tests so that a consumer can order corrective glasses without the involvement of a health professional.
- A mobile app that uses a video game to improve vision.
- A mobile app that uses the phone speaker to conduct a hearing test so the consumer can then order hearing aids without the involvement of a health professional.
- Software that through a series of questions produces a legally sufficient order to allow a patient access to a prescription drug. Don’t laugh – FDA itself raised this possibility in 2012 as a way to encourage broader use of certain drugs such as those for high cholesterol, hypertension, migraine headaches, and asthma.
- A mobile app that screens for certain psychiatric diseases and if such a disease is suspected, recommends that the user consult a psychiatrist.
- A mobile app that makes decisions about in what order patients will be seen for physician house calls based on information collected from the patients.
- Software that scours electronic health records for diseases that have not been diagnosed based on symptoms that have been observed and data that have been collected, and which offers specific diagnostic and therapeutic recommendations to the physician.
- Software intended for use by elderly patients with multiple chronic conditions that receives data from wearable health sensors, identifies such conditions as tachycardia and signs of respiratory infections (based on established medical knowledge) and communicates this information to caregivers.
- Software that uses a digital recording of breath sounds to diagnose diseases such as asthma, chronic obstructive pulmonary disease, and pneumonia as an aid to physician decision-making.
What exactly makes these things new? Why do they not fit any of the existing FDA classifications? Well, first and foremost it’s because FDA has not created classifications for those specific uses. And there could be many different reasons why FDA would not have created a classification including obviously the fact that a given piece of software might be the first of its kind.
But at a higher level, there are at least three factors that FDA considers in determining whether these fit within an existing classification.
1. Software replacing a professional for the first time.
Professionals make decisions now with professional grade tools, but the idea of creating a software tool to be used by the patient himself without a professional can be quite radical. FDA often treats software intended to replace professionals as new, at least if it has not been done before in that particular functionality. In these cases, FDA would say there is unknown risk. Replacing the mind and judgment of highly-trained medical professionals can produce unintended consequences.
For the record, it does not bother me that folks are hard at work developing similar technologies to replace certain services supplied by attorneys. Innovators are creating software tools to allow people to write simple wills, draft advanced healthcare directives, draft pleadings for small claims court and do other things that frankly do not require three years of law school. Through those efforts, people of lesser means are now, for example, much more likely to have a will, and that’s good for everyone.
As with attorneys, replacing clinicians can be very good or very bad. On the one hand, there are medical equivalents to the simple will. There are aspects of vision testing and hearing testing that have already become highly automated in the professional’s hands, so making them available directly to consumers may not create unacceptable risk.
On the other hand, automating a human process can be tough because it requires a systems-level analysis of all the risks, all the functions, all the processes and all the communication clinicians do that may not be formally documented. It’s in those cases, presumably, that FDA worries about unintended consequences.
At a very practical level, developers of these products will quite likely find resistance from the established medical community. Many of the FDA staff come from that established medical community, and will find it hard to believe that a computer can replace them on anything important.
So, if you come up with a new way to make health professionals obsolete, you can probably expect FDA to treat it as new and put it in class III.
2. A completely new use, like a video game for the treatment of ADHD or the improvement of vision
Here the technology is old, but the use is new. If you read the trade press, you will probably recognize that the two examples I just gave-- a video game for ADHD and a video game for improvement of vision-- were both subject to enforcement actions by the Federal Trade Commission. That agency challenged the marketing of those products on the basis of a lack of medical evidence proving that they were effective.
Put that issue aside. Let’s say I come up with a new way to use an existing technology and indeed I have appropriate medical evidence that shows that it works. The novelty of the use will quite likely lead FDA to make me demonstrate the safety and effectiveness of my product under a class III standard requiring very substantial clinical trials, even if the use of the technology is low risk.
3. A truly new technological capability, like brand-new algorithms that automates what has historically been done manually through a different mode of action
While the great majority of technologies being put into mobile apps now for healthcare are only slight variations of old technologies, there are also some truly new technologies coming down the pike. Software that uses fractal geometry to determine whether a mole is melanoma is new. Obviously this also can fall into the category of taking over health care professionals’ judgment, but even if the technology only aids the healthcare professionals in making the determination, FDA is likely to treat such technologies as new. Among the 1700 FDA classifications that exist, there is not one for technology that diagnoses melanoma through fractal geometry or for that matter anything close.
So what’s the problem?
In my first post, I explained how automatically putting the new and novel stuff in class III means, at a minimum, substantial delays in the marketing of the products, at substantial cost. Even worse are those cases where the entrepreneur gives up in the face of the regulatory obstacles.
This mobile health technology is in fact quite important, and its development needs to be encouraged, not frustrated. If you already agree, please humor me while I make the case that the potential benefits from a societal standpoint outweigh the regulatory policy behind putting every new medical device through a substantial gauntlet.
Of course, each software developer presumably can make a case that their product will truly help patients. But at a higher level, I will focus specifically on the new, cutting-edge mobile health technologies that replace professionals.
Some healthcare professionals will want to shoot me for this, but I believe certain software that “replaces” the healthcare professional has significant, potential societal benefits.
There are good reasons we should want to have software that automates some of what physicians and other health professionals do.
For starters, we have a growing shortage of physicians and other healthcare professionals that make access to healthcare difficult for many, especially the rural poor. It’s no secret that the United States faces a severe shortage of doctors, with some predicting the deficit will grow to as many as 90,000 by the year 2015. No matter whether you think Obamacare is good or bad, Americans will not have access to healthcare if there are not enough physicians to provide it.
While there are ways that we can address the shortage of physicians, it also makes sense to improve the efficiency of the healthcare delivery process. If the use of technology can make that process more efficient and accessible, without sacrificing quality, creating unacceptable risk or increasing net cost, it’s hard to argue against it.
Mobile health strategies can address some of the geographic, cultural, and communications barriers that low-income Americans face. Those barriers include:
- Chronic, debilitating illnesses that limit personal mobility, making face-to-face provider visits impractical;
- Homes isolated from health care facilities and from public transportation;
- Cultural, educational, and communications barriers that limit exposure of disadvantaged Americans to information on programs and opportunities for improving their health.
For example, despite enrollment expansions in Medicaid, individuals covered by this federal-state program tend to work in hourly wage jobs that limit their ability to take personal time during the day, when most clinics and other ambulatory care facilities are open. This forces a choice between visiting a doctor or earning an income – an income these individuals can’t afford to lose. Moreover, Medicaid covers millions of Americans in rural areas, where distances are greater and practical access to physicians, particularly specialists, is even more limited.
These new mobile health technologies can help. For example, these technologies --
- Lie as close at hand as a patient’s pocket or purse, thus bringing care to individuals whose conditions limit mobility or whose residences are too distant to make face-to-face provider visits practical;
- Combat America’s tidal wave of chronic disease, such as diabetes and asthma, by providing treatment-adherence reminders, and other capabilities for managing conditions on a longer-term basis without requiring daily involvement by the physician;
- Present opportunities for driving down the cost of preventative care by making possible screening services, like vision and hearing, that insurance often doesn’t cover;
- Present opportunities for making health care delivery models more efficient and cost-effective by stretching the supervisory span of physicians, matching professional services to the most appropriate skill levels, and expanding opportunities for family members to serve as care givers.
But to help, they need to reach the marketplace.
A group I work with, the CDS Coalition, this fall will conduct a survey and interview CDS developers to get a better sense of the impact that FDA silence – and the resulting ambiguity it creates – is having on innovation in this space. The agency’s silence extends both to the classification issue analyzed in this post, as well as the threshold question of what CDS software FDA will regulate.
Having announced in the spring of 2011 the intention to publish a guidance document explaining the agency’s approach to CDS, the agency has still not done so over four years later. Indeed, the agency had made answering that question a top priority for the fiscal year that just ended on September 30. Still no guidance. Rather than speculate on the impact of that silence, we want to collect the data needed to understand its impact on innovation. Please watch for that survey.
Based on the examples provided in this post, we cannot afford to stymie new medical technology simply because, for example, it does a task previously done by a healthcare professional. While I understand the change in roles can create unforeseen risk, we need to figure out an easier path for new technology that takes on low risk functionalities, the medical equivalent of a simple will.
We need a new paradigm where new technology is quickly evaluated for potential risk, and placed promptly into an appropriate regulatory category. While it is true that anything new inherently carries with it unknown risks, we must not overreact to those risks. Our healthcare system, and low income Americans in particular, are in desperate need of the new. New is not a four letter word.