A “Gold Standard” for Population Health? Revisiting the FDA’s Relationship with the Hippocratic Oath

This commentary was first published on April 13, 2018, on Health Affairs Blog.
Copyright ©2018 Health Affairs by Project HOPE — The People-to-People Health Foundation, Inc.

Perhaps the greatest turning point in the history of medicine occurred sometime in the 20th century, when medical practice started doing more good than harm. While “First, do no harm” is always sound advice, it may not lead to optimal outcomes. Today’s regulators have succeeded fabulously in their charge by minimizing the risks of large-scale catastrophes such as the tragedies caused by the administering of sulphanilamide elixir, (an improperly prepared sulfanilamide medicine, which led to the 1938 Food, Drug, and Cosmetic Act) or thalidomide, but that has not saved us from the opioid epidemic. Meanwhile, critics complain that the Food and Drug Administration’s (FDA’s) deliberate processes delay approvals at the expense of patients. Is there an alternative that can better serve public health? “Do no harm” cannot be the default, especially if it limits the FDA from using its expertise to efficiently approve useful products. We suggest enabling the FDA to use a population health context to meet 21st century expectations for pharmaceutical risk.

A population health paradigm can provide faster and better access to innovation—and better guidance for consumers. Population health programs seek to improve the health outcomes broadly of populations as opposed to focusing on individuals. Conditional or limited duration approvals of promising innovations can speed patient access. Innovators, who will avoid the expense of multiple Phase 3trials, can provide outcomes guarantees; meanwhile, patients, who gain faster access, can make meaningful commitments to reporting and compliance. The progress of routine, real-world data collection can extend the FDA’s reach and power as well as better protect the public’s health—and protect individual FDA reviewers from criticism. As importantly, a population health approach can bring flexibility to the current, all-or-nothing nature of FDA approvals that favors highly capitalized companies, justifies high drug prices, and slows progress.

US drug regulators adopted a “never again” mindset in response to profoundly unsafe products, but regulations required efficacy much later than safety—only added in 1964. In the 5 decades since, the FDA has adopted a Precautionary Principle approach in which the fear of doing harm comes first—even before benefit for patients. The resulting mindset considers the death of a patient waiting for a treatment to be an “Act of God,” but all risks, however minor, must be scrutinized through repeated Phase 3 trials. The creation of the FDA Breakthrough Designation attempts to address the perception of slow approvals. Policy tweaks, such as greatly simplifying a physician’s Compassionate Use application, may pacify individuals who have no time to wait. A recent example occurred when highly publicized parental demands for the use of an antiviral ultimately did not save the child, and the company behind the antiviral found itself in a no-win media situation. Understandably, there is a push for “right-to-try” initiatives in the states. But none of these mechanisms acknowledges the broader question of appropriate access for patients or addresses the potentially large non drug costs associated with experimental therapies. These current and proposed approaches do not encourage rational, let alone fair, improvements to public health.

While some call for reducing the FDA’s power, we propose expanding it to recognize the wealth of population health approaches. The goal is to meet 21st-century expectations for faster progress as a critical contribution to improve US health writ large. Population health is defined by the outcomes of a population, capturing the determinants of health and identifying actions that improve outcomes. This proposal shifts the focus away from the health of individuals as assessed in traditional randomized controlled trials (RCTs) on small, restricted, “simplified” cohorts of patients. Instead, emphasis is on the health of populations as measured directly—populations comprised of many more people, with more individuals with varying characteristics and with multiple comorbidities and compliance histories. The new power would include the ability to act on real-world observations. All this would shift risk away from “do no harm” and into the faster assessment of real-world factors that can positively (or negatively) impact the population’s health.

FDA decisions have always affected population health, partly because its decisions govern which medicines reach the population. However, headlines about individuals facing dire healthcare situations put reviewers and clinicians in difficult positions. Patients do not want medicines that are neither safe nor effective, and we believe meaningful explanations of the risks of developmental drugs would be very helpful to them, but such information is rarely available. Meanwhile, risk adversity makes the FDA reviewers’ jobs harder, and the process ultimately frustrates advocates, patients, and innovators too.

At every step, the FDA can play a critical role in collecting and interpreting the data, offering specific and broad guidance to healthcare professionals, and responding to patient demands, especially in helping all stakeholders understand disease states and their progression. As a consequence, the agency can continue to play a leading role in protecting and promoting the public health (the FDA’s jurisdiction encompasses about 20% of the US economy). Indeed, we can think of no one who could do it better.

How the Insurance Industry Deals with Risk

Most regulators approach uncertainty as a problem, for example, a 95% confidence interval for the frequency of harms that is (-0.10, 0.05) includes 0.0 (no harms), but it also includes positive values for harms (such as 0.04). By contrast, the multitrillion dollar insurance industry (much larger than the pharmaceutical industry) is all about managing uncertainty instead of trying to avoid it. Of course, the FDA is not about to insure your car, but we think that the way insurers frame risk is valuable for the FDA to consider.

Insurers consider risk in a broader context. After a fender bender, a Lamborghini may be more expensive to fix than a Dodge Dart, and that justifies a higher auto insurance premium for the owner of the fine Italian car. But, the auto policy on your Camry covers damage whether you hit a pickup truck, a bus, or a Tesla. Insurers have the data to analyze the risks—and when an insurer discovers they’ve insured a bad driver, the premiums for that individual’s car will go up. But insurers, regulators, and consumers generally prefer to not microstratify the risks—it is easier and less hassle to sell and administer broad-brush protection for a larger population instead of attempting to customize policies for each individual. And, all else aside, the details and circumstances matter; higher risk means that insurance costs more in some states. Five important lessons from insurers are:

• Choose a practical balance between no risk and too much risk
• Accidents happen (similarly recognized in the National Vaccine Injury Compensation Program)
• It is OK to change your mind about risk when you get new information
• Understand the spectrum of risks: rare to common, mild to catastrophic
• Context and behavior affect risk

As with autos, there is ever-present population risk in healthcare. For example, medical errors kill 250,000 to 400,000 people each year. Despite the high real-world prevalence, RCTs rarely report medical errors. Indeed, the trial sponsors’ oversight of patients likely offers significant protection against medical errors. So, how much do RCTs reports of harms and errors help the FDA understand the real-world risk context in which compliance is worse and comorbidities are greater? Perhaps real-world risks overwhelm a product’s harms or its benefits, or the use of a novel agent, despite its risks, actually reduces underlying risks in some ways. This is just one of the real-world risk contexts that suggests a different kind of population-health consideration by the FDA could be valuable.

Ways the FDA Can Embrace Population Health

We are not suggesting changing the “gold standard” of the FDA, but, rather, the agency can use a broader methodology and a more comprehensive array of available evidence to make better decisions sooner. Twenty-first–century data collection and analytics are easily a good enough reason for change. Here are some ways to make it happen:

1. Be practical and balanced: For some developmental products, where the evidence is unclear but lives are at immediate risk, instead of requesting another Phase 3 trial, the FDA could offer “conditional approval” for a limited market to generate real-world evidence. This could work like Medicare’s Coverage with evidence development, which allows temporary Medicare coverage while outcomes are tabulated. It would also create incentives for innovators to avoid other background medical errors—not necessarily a bad consequence. In exchange for the acceleration, the innovator would provide financial guarantees to patients, payers, and regulators in case the drug doesn’t work.
2. Accidents happen: The FDA could require real-world information in addition to the current data review process. As part of an Innovator's New Drug Application, the FDA already requires access to all available evidence on the candidate medicine, irrespective of where or how that data are generated, so this need not be a significant additional burden. But attention to real-world data will help determine the background risks not necessarily associated with the drug.
3. It is okay to change your mind: Use readily available data sources for on going monitoring. Instead of sinking more investment into noncompatible electronic medical records, we suggest using smartphone applications or claims monitoring to detect ongoing population health issues associated with products. Innovators could be required to guarantee outcomes through a performance bond, even with final approval. Using this as context, clawing back drugs or drug approvals for certain indications would be seen as successes, not failures.
4. The spectrum of risks: The application of the FDA’s risk/benefit expertise would include judging when the population health approach would be most useful. That would involve considering issues such as limited versus large populations affected, whether the condition is severe, how life-improving the treatment might be, the side effects of the treatment, the duration of observation needed—and the safety underlying it all.

21st Century Versus 19th Century

In our diverse nation with its variable patterns of care, does it still make sense to limit decisions solely to whether the patient with symptoms fitting the description on the label can expect the drug to be safe and effective when he or she uses it for a specific medical condition under the guidance of his or her physician? It seems to us that the current mission of the agency matches 21st-century needs, but individual FDA reviewers need confidence in their authority to consider what is both unknown and not yet knowable. We need to create a truly science-based decision-making process that uses the most up-to-date approaches now available and allows for further improvements in the future–indeed invites them.

We question the value of right-to-try and compassionate use policies. These approaches are unlikely to provide data that can benefit future patients and could promote “cures” that are, probably at best, placebo but may be much worse. Removing the FDA’s expert consideration of data, however incomplete those data are; would take us back a century to the risks and harms of “snake oil.” Even an individual treatment Investigational New Drug Application carries a policy hazard because it could be misinterpreted as an FDA endorsement. The FDA’s past successes are a foundation to best address or redefine patient needs and regulators’ dilemmas; accelerate some approvals; and set different accountability on innovator sponsors.

The consequences of an intervention, both good and bad, must be measured in the reality of US healthcare, itself offering opportunities for considerable improvement, and include a fair estimation of the consequences of no intervention at all. The legal system will also need some adjustment, but we do not attempt to encompass those elements here.

How It Could Work

1. A hypothetical drug for pulmonary fibrosis (a disease with high mortality) has gone through Phase 3 trials. While the drug is very effective, there are concerns with kidney toxicity, so the drug is at risk of being disapproved.
2. Instead, the drug is given conditional approval for a specified real-world trial period and in a limited population during which:
1. Adverse events can result in the drug being pulled from the market
2. The manufacturer must enroll patients in a registry that collects data continuously and reports to the FDA monthly
3. The manufacturer promises to refund payers the full purchase price if the drug is withdrawn during the trial period, as well as the cost of treating adverse events
4. Trial prices are set at some fraction of comparable drugs
5. Patients are enrolled in a compensation plan and opt out of medical malpractice.
3. Within the trial period, based on performance, the drug may be granted approval for some range of indications or certain patient characteristics.

A Refined Gold Standard

An FDA gold standard based on population health will deliver the same assurance of safety and efficacy for which the agency is justifiably proud. However, the refined gold standard can accommodate greater sensitivity to the patient who is waiting as well as recognize the vulnerabilities of the sponsor of a new medicine. There is immense value in new, population-based sources of evidence, including clinical experience in less than ideal populations or in less controlled study settings. “Good enough” real-world data is now far too valuable to ignore. It is far better to empower the FDA to make more flexible decisions while maintaining its critical oversight instead of sidelining those who are most competent to judge risks and benefits. For this to happen, the FDA will need to lead.