A vision for clinical trial abundance

What does clinical trial abundance look like? And what still stands in our way?

This post is part of a series on clinical trials - why they are so important, why they are so inefficient, and how we can improve them. (You need not have read the previous posts to follow this one). In my last post, I shared, in painstaking detail, exactly how inefficient today’s clinical trials are. Wasteful activities cost huge amounts of money and, despite efforts at reform, remain widespread. Today, I’ll dive deeper into what’s being done about the problem and what might be missing from these solutions.

A vision for clinical trial abundance. (Generated by ChatGPT, of course.)

Envisioning Clinical Trial Abundance

Before diving into solutions, I’d like to paint a picture of what we’re trying to achieve. When IFP published its report, “The Case for Clinical Trial Abundance”, it was the first time I had seen the word “abundance” being used in reference to clinical trials. I hope this framing becomes more popular. I’ve worked on many efforts to reform and modernize clinical trials, and too many of these reforms have suffered from a lack of focus. It’s not uncommon to see lots of different goals get cited: improving patient access to trials, improving evidence quality, lowering drug development costs, improving trial equity, obtaining better clinical data throughout the “drug lifecycle”, and more. Along with the proliferation of goals has come a proliferation of terminology: pragmatic trials, point-of-care trials, practical trials. Each of these takes the idea of trial abundance and bundles it with lots of other concepts that, while potentially valuable, may also distract us from the goal of abundance.

Above all else, achieving clinical trial abundance means making clinical trials cheaper. Abundance itself, along with almost every other goal we want to achieve with trials, - whether it’s clinical knowledge creation, new medicines, or greater health equity - is best served by lowering the cost of conducting trials. And fortunately, making trials dramatically cheaper seems like a realistic goal. In discussions I’ve been involved in, leaders in the clinical trial industry have said that it ought to be possible to lower clinical trial costs by 10 times. IFP notes that the RECOVERY COVID trial, a model for how we might approach clinical trials in the future, cost 1/80^th of what we spend on traditional trials. In practice, clinical trial reform might not involve rapid cost-cutting, but rather “bending the cost curve”; lowering the rate of increase in trial costs such that over the course of 10-20 years, clinical trials become dramatically cheaper in real terms.

Cheaper clinical trials would be transformative. Suddenly, lots of pharmaceutical investments that might not have penciled out financially become feasible, leading to the development of more new treatments. Government and private philanthropists can make more cost-effective investments in improving our medical knowledge - for instance, coming up with new uses for existing drugs that are no longer under patent. It also becomes possible to expand clinical trials, allowing more patients to benefit from experimental treatments and the evidence that trials generate. And finally, cheaper trials will enable our clinical research system to keep pace with the growing demands placed on it by modern drug discovery methods, which could lead to a dramatic increase in promising new medicines that need clinical testing.

Streamlined trials are coming.

What exactly would a faster, cheaper clinical trial look like? Fortunately, we have some examples, and a picture is starting to emerge. These trials would recruit patients faster by automatically scanning patients’ health records to find eligible participants. Once patients are recruited, the trial would be far simpler: Instead of using a separate, paper-based process for data collection, investigators would collect most of the data they need directly through the electronic health record. The trials would have fewer endpoints, less data collection, and would be more focused on obtaining actionable evidence. From a patient’s perspective, participating in one of these modernized trials would not seem much different from their routine care, except that they may need to share extra information and check in with their doctor more frequently.

This would all be made possible by deep integration of trials into electronic health records and other software used by clinical trial investigators. To start, a digital, interoperable clinical trial protocol would be sent to healthcare providers, kicking off a whole stream of digital workflows that would make trial operations seamless - an idea that I like to refer to as a “plug and play” trial. Once the investigator receives the protocol, they would be able to enroll patients into a trial almost as easily as they order drugs and procedures. And once patients are enrolled, investigators would be automatically prompted by their systems to schedule study visits and collect needed data. Reporting would be automatic too, with researchers receiving routine and real-time updates and monitoring the incoming data for issues.

This future is starting to take shape. The technologies we need to execute trials seamlessly already exist. Industry has designed data standards that define how trial protocols and data can be exchanged, and vendors sell software and tools to enable many of the digital workflows I discussed. Even Epic, the largest EHR vendor, is joining in, launching a suite of clinical trials tools with features like automated trial matching and integrated data collection and exchange. To the extent there are gaps in functionality, they’re nothing that a team of experienced, motivated developers and clinical researchers couldn’t address.

Scaling in medicine is hard. Scaling in clinical trials is even harder.

With these advancements in technology, should we be optimistic that clinical trials are going to get dramatically faster and cheaper in the near future? History gives us reason to be skeptical. In 2010, the National Academy of Medicine convened leaders from FDA, the pharmaceutical industry, academia, and elsewhere to discuss how to transform clinical research in the US. Leaders highlighted multiple challenges: overly complex studies, limited availability of trials to patients, excessive bureaucracy, lack of standardized data collection, and more. Ten years later, in 2020, the Academy again convened clinical research leaders to discuss how to transform trials. They highlighted many of the exact same problems and added to the list a failure to adopt some of the new technologies – particularly IT – that had been transforming health care and the entire economy over the prior ten years. I wouldn’t be surprised if, in 2030, the Academy issued yet another report - this one criticizing our clinical trials enterprise for continuing to rely on paper while failing to take advantage of AI.

If we are ultimately unable to make clinical trials more efficient, it won’t be because we lacked the technical ability to do so. It will come from an inability to scale. Those who have worked in engineering or business probably hear the verb “scale” a lot. It reflects a hard-won insight: Accomplishing a task once is just the first step; real success requires finding ways to scale up that success to serve an entire sector or market. To scale a product – whether it’s a clinical trial, a car, or a chain restaurant – you need to apply an engineering mindset to it. You need to break down the product’s development and delivery into smaller, repeatable components. For each component, you need to standardize, automate, iterate, and streamline its production. And you need to figure out how to configure those components to meet the needs of many different customers.

Scaling is hard in any industry, but it’s especially hard in health care, and even harder in clinical trials. Scaling requires standardized, repeatable, reproducible components. Yet in the health care system where clinical trials are administered, health care providers and settings are extremely diverse, as are the conditions being treated. Our health care systems are fragmented and resist one-size-fits-all solutions, and rules and regulations make building any scalable technology extremely complex. Clinical trials start with these same challenges and add a host of others: each trial has a distinct protocol designed to answer a distinct research question, sets up a distinct infrastructure to execute the trial, and goes through a distinct regulatory review process. This makes standardization and automation difficult. Even the culture of clinical trials – built around hypothesis-driven scientific questions and run by a team of part-time clinician-investigators – is not amenable to scalable, repeatable processes.

So while I am optimistic about new technology’s ability to transform trials, I think making clinical trials more efficient will require us to face clinical trials’ scaling problem head-on. Our goal should be to move the clinical trials industry from pilots and proof-of-concept projects to real systemic change. My next post in this series will talk about how we might achieve this.