Pragmatic Trials: The Real-World Data Revolution
Insights derived from real-world data are changing every facet of pharma but its impact on clinical trials may prove to be a crowning achievement
If Google, Facebook and other consumer brands have taught us anything, it’s that data has become this century’s most sought-after commodity.
Prized for its unique ability to capture the minutiae of an individual’s lifestyle, it is the ultimate tool for targeted branding, attracting criticism as just the latest strategy in the corporate playbook, pursuing profit at the expense of personal privacy.
As regulatory bodies scramble to contain the real-world data revolution, its impact is already fundamentally changing the nature of how we develop innovative new medicines.
Historically, drug trials have been placed into two camps, explanatory or pragmatic trials, but this distinction is somewhat of a false dichotomy, says 
Jack Sheehan, Real World Evidence Team Leader at Janssen. “Every trial exists on a spectrum between pragmatic and explanatory,” he says.
For example, the more heterogenous the study population, the closer it sits to the pragmatic end of the spectrum, the more homogenous, the closer to the explanatory end, he says.
The logic is clear; the more data available, the greater the likelihood the sample will be more heterogeneous, increasing the probability of finding a drug that can “pragmatically” treat a broader percentage of the population.
Riccardo Perfetti, Senior Medical Officer, Vice President of Global Medical Affairs at Sanofi, points to the opportunities afforded by this proliferation of data. “This new strategy attempts to address two major issues; one is that a randomized controlled study only represents a very small fragment of the potential population and the second is that they are developed in a very controlled setting
where patients and investigators likely behave in a way that is substantially different to real life.
“So, there is a need for an additional set of data that will allow us to understand how drugs behave in a real-world setting,” he adds. “We are all working to improve the methodology that allows for such an assessment in the pragmatic study approach.”
Sheehan cites the GSK Salford Lung study as a watershed moment because it provides a credible case study example of an “embedded” pragmatic trial – the study used insights from a coordinated database of updated electronic health records to monitor patient outcomes while preserving usual care intensity of follow-up. It combined prospective data generation and randomization with usual-care intensity of touchpoints between the patient, study staff and treating clinicians.
This is a much less invasive way to collect data on patient outcomes and provides a more realistic insight into drug adherence than a structured, clinical trial, he says. The frequency of visits and the dosage administered in a structured trial “raises adherence in both arms to artificially high levels.”
If the Salford study is a poster-child for the manifold benefits of a connected database, it also reflects the advantages and disadvantages of implementing this form of pragmatic trial on a wider scale.
It may give Europe an edge over the US, suggests Sheehan. “European countries, particularly the Nordic countries but also the UK’s National Health Service, have a uniform database that the patient population exists in, and they generally exist in that database as long as they live in that country. You have this really great longitudinal data on patients in European countries.”
In the US, it is harder to gather data in such a holistic fashion, he says. “You have patients who go in and out of databases as their insurance changes or as they go from one hospital to another hospital because it's a very fragmented system.”
However, the sheer size and scope of the US means it has the potential to provide a richer, more heterogenous population sample and diversity of treatment practices than individual European countries.
Efforts to circumvent the US’s fragmented market are improving, he says, citing PCORnet, a network that enables interoperability between databases, as a step in the right direction. “The sacrificesyou have to make in choosing to have patients all within one database is diminishing over time.”
For Perfetti, integrating the two areas presents another key obstacle for real-world pragmatic trials. “The overall objective is to give or to describe what happens in real life and one of the concerns is that by injecting too much methodological control and too many of the assumptions that regulate randomized-controlled studies into the real-life data, we may produce something that departs too much from the real-life setting.” He adds. “We may generate an artefact.”
It’s a delicate balancing act, he adds. “The challenges are how to inject rigor and make it vigorously strong from a methodological point of view without altering the natural real-life behavior of both patients, physicians and the healthcare system.”
There are also cost complexities to consider, says Sheehan. “There is this promise that an embedded pragmatic trial could attain great cost savings in the execution. But, if you have a heterogeneous population, you may need to increase the number of patients to find a statistically significant result. Because you've increased the variability in the result, you may need to have a substantial drop in the cost per patient for this approach to actually prove cost-saving in the long run because you will need to recruit more patients.”
However, incorporating data into one centralized database could help address a wider range of researcher questions, he posits. “If you conduct an embedded trial where patients are all in the same database, you can get their data before they enroll in the study to see their natural history and to authenticate certain self-reported patient criteria. You can then track these patients real-world data after the study ends to see if there is a disease-modifying effect. Once the system is in place, there's lots of additional research questions that you can answer without meaningful additional data collection effort.”
Perfetti also highlights the profit yielding potential of this model – for payers. With reams of data demonstrating a drug’s efficacy in a real-world setting and realistic levels of adherence to boot, this will obviously prove an attractive proposition to payers, he says.
Such cost-saving potential comes hand in hand with material benefits for patients, he says. Admitting that pharma has historically focused its energies on advancing the science to the detriment of the patients’ perspective, incorporating adherence insights derived from real-world data into pragmatic trials could bridge this disconnect.
“An application that needs to be administered every hour may provide what you need from a scientific point of view, but the patient will obviously be very annoyed. Changing the dimension of how the patient is able to use certain treatments; this innovation has been neglected and it can no longer be neglected,” says Perfetti.
Jack Sheehan and Riccardo Perfetti will be weighing in on the seismic impact of data at the Data, Evidence, Access USA 2017 event in November.