Just as cars are beginning to drive themselves, healthcare decision-making is facing its own automation ramp up.
Soon, patients will be able to enter their current symptoms through a portal, with the help of an intelligent agent, and get an accurate diagnosis or prescription without involving a human doctor. A child with asthma will have body sensors continuously monitoring respiration to direct when the child needs to take medicine.
The technology is moving quickly, which begs the question: Is FDA ready?
Automation and the FDA
When machines take over healthcare decision-making, society will want to ensure they work properly. This will be the FDA’s responsibility, but the organization isn’t prepared for it. The FDA operates under a Congressional statute enacted in 1976, the year IBM launched a two-sided floppy disk and a guy name Steven Jobs founded a new computer company called Apple.
The act was based on a simple premise: New is risky. That’s a problem. Because under the statute, the FDA treats everything new as it would a pacemaker, which requires years of expensive validation. Right now, companies are introducing plenty of technology that is new and surely brings with it some risk. But at face value, a reasonable person would conclude that often new technology does not bear the same risk as a pacemaker.
On September 30, the federal government ended another fiscal year without publishing a much-needed – and supposedly top priority – guidance on clinical decision-support software (CDS). The agency has been working on this guidance for five years, but it still hasn’t published anything. To be clear, the FDA does regulate some high-risk CDS, and it should. But for small companies developing products in the space, the question naturally is: Where is the dividing line? Without guidance, no one knows the answer to that simple question. So how is an entrepreneur to develop a business plan and obtain funding if she can’t answer that simple question for investors?
Five years ago, in response to the FDA’s announced plan to develop a CDS guidance, the CDS Coalition came together to provide the agency with information about these technologies, the industry, and the patients they serve. The Coalition, to which I provide legal counsel, has provided examples of current and future CDS and has suggested principles that the FDA could adopt to discern the difference between high- and low-risk software. Some of those principles are derived from consensus guidelines developed by regulators worldwide. Others stem from the relatively common-sense proposition that the FDA should not regulate software designed to ensure that doctors can independently review the basis for the software’s recommendations and make their own decisions.
The Coalition is focused on the complementary use of software with pharmaceutical products. Software now is used to inform many of the decisions around pharmaceutical care management, including drug selection and dosage. This is an area where FDA rules are very opaque, because the people at the FDA who regulate software and those who regulate drugs are in two entirely different centers and it’s slowing innovation. The Coalition filed a citizen’s petition in August asking for these disparate groups to work together to give the public answers on three hypothetical case studies.
What will keep the FDA relevant?
The FDA needs to adapt. The Coalition is sympathetic to the FDA’s challenges – automation is rapidly changing and growing. It is difficult to write regulatory policy when you don’t know exactly what the technology will look like, or even what it will do.
That’s why the petition tackled not only the issue of getting answers regarding FDA regulation of software used with pharmaceuticals, but the broader question of how the FDA can more effectively provide guidance on fast-moving technology. The Coalition’s proposal is for the FDA to move toward what we refer to as “case study guidance.” In this proposed model of regulating, the FDA would publish guidance not on broad topics where the scope is uncertain, but on specific real-world case studies. In this way, the FDA can give more timely and valuable guidance, without getting hung up on the need to anticipate every twist and turn the technology might take in the future.
The Coalition has also developed a detailed framework for how to regulate new clinical decision support software: It recommends that manufacturers take the responsibility to use the unique capabilities of software to collect and analyze information, in effect creating a better safety net to identify problems and continuously improve new software products. The historical approach with traditional medical devices is to iterate every few years. Modern medical software regulation should not involve rowing the steamship, but rather should embrace what software developers do well, which is continuously collect data and innovate. This approach lessens the need for significant pre-market clinical validation and offers the advantage of gathering real-world evidence to produce greater insights vs. collecting artificial data through clinical trials.
We simply cannot afford for another year to come and go without innovators in the space even knowing the scope of FDA regulation. The cost of FDA compliance is such that this is not an inconsequential detail in a business plan. For example, if the FDA decides that a particular software program requires its pre-clearance, that can easily add hundreds of thousands of dollars to the cost, but more importantly delay market launch by months, if not years, depending on the validation required. These innovations, and the patients they are intended to help, stand waiting.
Bradley Merrill Thompson, a partner in the Washington DC office of law firm Epstein Becker & Green and Chairman of the Board of the firm’s consulting affiliate EBG Advisors, leads a broad coalition of technology companies seeking reform of FDA’s approach to regulating clinical decision support software. To learn more about the CDS Coalition, visit www.cdscoalition.org.