The only way out of this mess is for us to upend and rebuild the platforms that have failed us. From the top-down, from the bottom up, "end-to-end" -- whichever way works, saves the most lives and causes the least damage in the process. It is societal change, but we can no longer avoid it.
One case in point: Investigative clinical trial research.
It's through clinical trials that vaccines are discovered, better health practices are adopted, lives are saved, and livelihoods are improved. Virtual clinical trials involve conducting experimental research with human volunteers on potentially life-saving treatments, but outside of physical research facilities. Now that a great many more people are working from home, perhaps that same connection could be leveraged to help people live longer while they're at it.
The clinical trial may be considered as an application of experimental scientific method to the validation of therapeutic procedures. The 'trial and error' investigations which presumably have been used for simple problems throughout the history of medicine contain the germ of this experimental approach but are inadequate to solve more complex problems. The need for the development of clinical trials to deal with these has been frequently obscured by undue confidence in tradition and a prior theory.
- John P. Bull
A Study of the History and Principles of
Clinical Therapeutic Trials, 1951
"I think when all of this started to happen in 2020," said Craig Lipset, founder of Clinical Innovation Partners, "there was an initial sentiment of, maybe this is a three-month exercise." Lipset was one of the moderators of the 2nd Annual Virtual Clinical Trials conference, produced by Dynamic Global Events, which this year -- the ironies being obvious -- took place virtually. He continued:
If we can just minimize missed doses, minimize missing data, we'll be able to get through this together. And I think now, there's an increased appreciation that, in particular, when we're engaging with patients with chronic conditions, there is no return to the way things used to be -- at least, not until there's a vaccine, which in reality is another 12 to 18 months of this model
Most every industry relies, to one degree or another, upon a digital platform to secure its information. Investigative clinical studies are about obtaining information through institutionalized testing involving human patients, with careful observation of signals and results throughout the process.
As the name suggests, clinical trials are centered around clinics. More recently, these locales have been dubbed sites -- physical buildings which patients are scheduled to visit, often connected to hospitals or healthcare providers. Something happened this year which made it unsafe for both volunteer patients and clinical workers to visit these places -- everywhere in the world, all at once.
"The question for us," remarked Josh Rose, vice president and global head of clinical trials for platform producer Iqvia (pronounced "eye · cue · via"), also during the Virtual Clinical Trials conference, "is really going to be, what is going to change, and what is going to stick? And how is this situation that we're in right now, going to change things forever?"
In the classic clinical trial, researchers experiment with the application of a therapy -- perhaps a medical one, perhaps physical, on occasion psychological -- on a set of human volunteers. In a sense, this application is a program, albeit not originally the digital variety. In the lexicon of trial practitioners, it's a protocol -- a programmed methodology for applying therapy, to determine whether such a method in general practice would yield a positive outcome. Since the method would be experimental, and thus somewhat different from what doctors would be normally allowed to prescribe, each application of the protocol is called an intervention.
Clinical trials utilize many of the same words we do in the general discussion of technology, although with different meanings. For instance, the clinic or hospital, or wherever the patient interacts with the clinical research associate (CRA), is called the site. Sometimes a multiplicity of sites, often within a given region, is called a site network. When a patient interacts with a site, it involves direct physical contact with the CRA conducting the examination.
"One thing that has changed is what we used to think of as a traditional site," noted Medidata EVP/CTO/CIO Dr. Rama Kondru, in an interview with ZDNet. He continued:
We always had this notion of patients having access to their clinical systems. Now, more and more people are going towards -- thinking about, at least -- a site-less concept, which we kind of introduced a few years ago. We're seeing a lot of demand for our patient-centric tools, that we can connect directly to the patient using apps. That's the key driver, because of COVID.
It is indeed, in at least one sense, a counterpart of the serverless concept: Patients still continue to use sites, and sites still exist, yet the patient doesn't have to make contact with the site directly. The point of care moves from the site to the home.
Yet Dr. Kondru's assessment of the pandemic-altered landscape for clinical trials is not shared by everyone. Pharmaceutical companies continue to take the position that some patients may be more afraid of technology than of travel. Shelley Barnes, who heads global clinical innovations projects for Brussels-based pharmaceutical firm UCB, told conference attendees:
The way that we approach decentralized clinical trials is not always technology-based. We do leverage video telehealth, and we think that is really good for patient engagement. But we also leverage many different things, like wearables and sensors, as well as home nursing. So we look for a good combination of those types of services, as well as technology, to determine what's the best mix, if you will -- the best recipe for that particular study and that patient population.
"We are not a software company," declared Kirsten Engelhardt, Associate Director for Applied Innovation and Process Improvement at pharmaceutical firm Otsuka. It's for that reason, she said, that firms such as hers should find suitable, long-term software vendors. She continued:
The outcome that a clinical trial protocol aims to accomplish, as well as an outcome that indisputably indicates the methodology has failed in its objective, are both called endpoints. Think of a program written in a 1960s or 1970s high-level language, in which a conditional termination is marked with the keyword END. An endpoint of a protocol is essentially that: A point at which a conclusion may be drawn, success or failure may be declared, and the trial stops.
The strict confidentiality of patient data, so that no patient knows the progress of other patients or even of her own, is called endpoint security. In many ways, clinical trial protocols truly are an alternate universe.
As late as 2016, the use of computers in clinical trials has, in many research facilities, been limited to typing records into spreadsheets and writing reports with word processors. That year, Dr. Tomasz Sablinski, the CEO of a protocol design firm called Transparency Life Sciences, declared during a meeting of the New York Academy of Sciences, "The world of clinical trials isn't about gigabytes, and it's not even about cell phones. If you ever go to a clinical research operation conference, you'll hear people talking about reducing the number of pages that are faxed and increasing the security of these documents. The clinical trial industry is operating in the 1980s."
What technological advancements there have been in the field of clinical trials, have largely been limited to the interval between then and now. Today, there exists something called the clinical trial platform. It's the same word we use throughout digital technology. Yet here, in this context, its meaning is more ambiguous than you might expect.
"We want to create a model that's paperless -- that's fully digitalized," declared Bola Oyegunwa, head of hybrid and virtual trials for drug development services firm Covance. He went on:
It's to provide a platform where you capture and input the data, instead of using paper CRFs [case report forms], and that e-source can now be directly connected into the EDC [electronic data capturer], where you can push data into the EDC. So you now eliminate transcription, [and] move towards source data verification, source data review. With e-source platforms for some studies now, you may not even need an EDC.
Source data refers to the information obtained from patients during a study, including what's written on paper. When it gets "captured," it becomes e-source. Electronic Data Capture is the first generation of computerized e-source gathering software for clinical trial facilities. It is also the only generation.
In explaining to prospective clients how EDC works, service provider OpenClinica tells folks that its forms are hosted by a web server, then cites Google.com as an example of another web server that people may have heard of. There are no blocky diagrams with longer rectangles for the foundation layers, shorter ones for the plug-ins, or network addresses floating in space. EDC is a form into whose fields a technician types entries before clicking on Save.
One of the advantages of EDC is that it saves CRAs from having to haul around all those boxes of floppy diskettes, not to mention the shipping costs. "Transcription," to which Oyegunwa referred, is exactly what you're thinking it is: The process of making "source" into "e-source." But as OpenClinica points out, once you save data in one format, it's really hard to make it interchangeable with other programs' formats. The implication there is that, on paper, no one has any doubt as to what data says, or question as to what it means, so there's a big temptation to just leave it there.
"Some of the sites, their EHR [electronic health records] systems or their billing systems don't necessarily integrate with some of the other systems they're being asked to use," remarked Jim Kremidas, Executive Director of the Association of Clinical Research Professionals in Raleigh-Durham, North Carolina. "So if they have a virtual visit in their clinical practice, it needs to somehow automatically interface so that they can bill the patient."
"We're feeling the tech push, fast and furiously," remarked Otsuka's Kirsten Engelhardt. "Everybody from Amazon to Intel to Best Buy to Lyft and Uber are really trying to figure out their role in helping us innovate the space that we're in."
By design as well as necessity, a trial site is an independent entity. Each site is run by a contract research organization (CRO), which typically operates one or more facilities visited by patients who volunteer for trials. Because patients' identities and vital data must be kept private and sacrosanct for the integrity of the studies, there are no other way trials can be managed. This isn't some management-driven silo; it's the way the business must work.
Each study is owned and operated by a sponsor -- often a pharmaceutical company, sometimes a healthcare provider, on occasion the manufacturer of an electronic device used in treatment, that is being prototyped. Though there is no law mandating this, the sponsor is usually the sole source of funding for most trials. Many times, the contract for a clinical trial is awarded to a CRO after a request for bids.
The sponsor sets the guidelines and specifications for the study -- how many or how few patients, whether they may already be hospitalized, how they will be monitored, how often they should be examined. Very often, the sponsor also specifies which software will be used in the study, and how the data will be collected and processed. That's where the partnerships Engelhardt mentioned come into play.
Nevertheless, many sites must contract with visiting, registered nurses for professional patient examination. And they too will bring their own software. Every agent in the clinical trial process could very well be using its own choice of platform, and in the case of cloud-based servers, its choice of cloud provider.
Covance's Oyegunwa told attendees his firm conducted its own study of over 600 active test sites, prior to the pandemic. He told attendees:
One of the challenges that they had is the multiplicity of platforms that they're currently using to support clinical studies. The goal of decentralized trials is not to further complicate, and further add, even more technology platforms that sites need to use. It really needs to be a partnership with sites to understand where the pain points are, and how a modular technology platform potentially can help address those pain points. So they're only using a single platform. We enable platform singularity, or even the ability to actually connect back into the platform that's being provided by CROs or sponsors.
The first commercial software platforms of the 1980s were devised before the era of multitasking. Back then, an accountant using a PC, for instance, didn't have the luxury of looking at accounts receivable and inventory at the same time. Each module in the platform was a separate program, but presumably, all the modules operated on the same database. In not-so-unusual instances, however, some modules actually exported data to each other in interchangeable formats, especially when those modules were produced by separate developers.
Getting data into and out of modules in a platform, so it could be useful elsewhere -- say, in a spreadsheet or a document -- often required more than simple exporting. Businesses often hired consultants (I was one of them) to produce batch files or separate executables that reproduced data from one module's format, in another that a spreadsheet such as Lotus 1-2-3 could import. These businesses devoted so much operating expenses to building these batch files (I know, I reaped the rewards) that they were often unwilling to upgrade or replace their platforms with newer and better versions, for fear of having to start all over again with the batch files.
It was a phenomenon my colleagues and I called platform ossification: Where the perceived value of the upkeep of older systems outweighed the estimated value of replacing them. What would it really cost, I was asked to estimate, to replace WordPerfect 4 for WordPerfect 5, 1-2-3 2.x to 1-2-3 3.x, MS-DOS 3.1 with MS-DOS 5 -- when all the work invested in keeping up the existing systems, failed to amortize?
Clinical studies facilities, up until last February's socio-economic deep freeze, faced a similar situation, albeit almost four decades hence. They had been utilizing modular software platforms. Sponsors used their protocol design modules to program the activities for each study, using some sort of symbolic language and/or flowchart. With such a module, they would determine the endpoints of the study, and which objectives should be met to ascertain whether an endpoint had been reached.
Because the protocol and the endpoints must be executed by the same platform in which they were designed, sites find themselves using all the platforms for all their sponsors, simultaneously -- to borrow Otsuka's Engelhardt's phrase, "bridging the gap." Think of the amount of transcription that must be taking place there, all day long.
Now, keep in mind that sites must maintain very strong security and privacy measures for their patients. And they would prefer to use just one billing system, not three and not 19. Executing billing procedures in one program, when patient activities are being collected from multiple protocols in a variety of platforms, cannot be an easy task. And the types of "integrations" made feasible by services such as Informatica -- where data is transferred to the public cloud and then "transformed" -- may only be legal if sites take precisely the right steps in precisely the prescribed manner. Indeed, such transfers may require direct e-consent from patients -- an explicit grant that requires a whole platform module unto itself, to manage.
Along comes a virus
It has taken ato break up the ossification of EDC in clinical trials. The industry whose structure and practices EDC kept almost inflexible for several decades, within a few days' time was brought to a crashing halt.
Over the past few years, some virtual clinical trial platforms have been testing the waters. Iqvia describes its own using terms that might resonate with the software development community: "Accelerate your timelines," its website suggests, "with patient-centered drug development." It's promoting what it calls a "hybrid model" for deploying virtual study components, splitting responsibility between "on-site" and virtual activities, in a way that mirrors the early "hybrid cloud" models -- giving customers virtual infrastructure in spoonsful. As Iqvia's website explains:
In their simplest form, hybrid models leverage a handful of virtual elements, such as secure online platforms for document exchange, or wearable health monitors for data collection, while maintaining a traditional on-site model for the rest of the trial. More advanced hybrid models may only require a few on-site visits for initial screenings, or for treatments that require a formal healthcare setting. This means investigator sites will continue to be part of the experience – and that they will be leveraged in more meaningful, value-driven ways.
With sponsors deciding the makes and models of devices that trial participants would use -- whether they're wearables, locked-down tablets, open tablets like iPads, or apps on smartphones -- determining the behavior of the new point of care after it moves to the home, has been very difficult. So Iqvia took a bold step, almost a leap of faith: It marketed its Virtual Trials platform directly to sponsors, equipping it with a component no 1980s platform ever attempted: Actual users. Iqvia's service came complete with human beings who serve as virtual trial investigators, taking the place of clinical investigators. Think of it like a managed cloud provider, such as Rackspace ("we handle IT so you don't have to") and you'll get the idea.
Such a scheme could still work today. However, there's now an urgent need for all sites to move to a virtual, or at least "hybrid," model. Admitted Iqvia's Josh Rose very subtly, through the use of a chart:
Where we were pre-COVID, we really thought we were getting towards that point where there was much more awareness, much more interest -- patients were much more receptive to it, and demanding much more patient-centric clinical research that adapts to the way that they live. Our thinking was that, at some point, broad-scale adoption of virtual and decentralized trials would plateau somewhere. But COVID has certainly changed our thinking, very much the way World War II changed the demographics of the workforce in a positive way -- where we have much more balance today than we had before.
In a decentralized model, where telehealth connects CRAs with patients remotely and readings are taken using wearable devices and remote sensors, the point of care moves to the patient's home or safest locale. This gives patients much greater leverage in dictating the terms of their participation, especially with whether they use their own tablets, smartphones, or watches.
In light of this change, Medidata is attempting what would've been called, in an era when more sports were being played, a hail-Mary pass: It has devised a true hybrid cloud-based information complex, where patient data is stored and secured within its own data center, but the mechanism for interfacing with that data is hosted on Amazon AWS. Communications between the public and private cloud components take place using APIs. This way, it can develop a suite of services that behaves like a sponsor platform for sponsors (Medidata Unified Platform), a site management platform for physical and virtual clinics (iMedidata), and a patient portal for volunteers (MyMedidata).
"The entire system can be configured by our sponsors, who are actually setting up these clinical trials," said Medidata's Dr. Kondru. "We can set it up for them, or they can set it up themselves." It's a way for sponsors to continue to lead their protocols and specify their own endpoints, without sites having to throw away their operating manuals with each new trial.
It's a "solution" that begs to be solved using more modern tools. Kondru told us his firm does have an eye toward the cloud-native, container-based model of computing -- a service-driven approach that may be several generations removed from many models the industry has been using up to now. Edge computing infrastructure, involving micro data centers the size of coffee tables housed within sites, could conceivably make way for a truly modern, distributed platform that ensures patient data security, while at the same time maintains optimum service performance levels.
But these container-based, DevOps-friendly systems, from the perspective of the clinical trials industry today, may as well exist in a different universe.
Before the riots began, as the pandemic was wearing on our souls like a corrosive acid, the phrase we began hearing was "the new normal." We now live in a time that is unacceptable as a model of normalcy, for anyone outside of the isolation of an executive bunker. At this moment, our society is unsustainable. The only way out of this situation is to progress forward, but our vision isn't clear enough yet to tell us which way that is.
With absolute certainty -- a rare commodity in recent days -- we can say that our recovery from this pandemic requires a tested, safe, and well-distributed coronavirus vaccine. We may never reach a point as a society that we'd be willing to adopt as "normal," that we'd want to pass off to our children and grandchildren like a well-tended garden, unless and until we make a fully functional, virtual clinical trials ecosystem into a new and sustained reality.