Payers, drugmakers and regulators are all exploring how real-world evidence can complement randomized clinical trials.
The alignment is notable in an industry with opposing interests and likely points to greater adoption moving forward.
Underlying healthcare trends such as a rising focus on value, as well as legislative changes brought about by the 21st Century Cures Act, will likely serve as further catalysts.
That said, real-world evidence has very real limitations. First and foremost, analyses of data gathered retrospectively from medical or claims records can't imply causation, only association — limiting how useful such evidence is to demonstrate a drug's benefit.
In addition, insurer and provider databases, from which such data is pulled, are only as good as their inputs. Drugmakers need to build analyses from large pools of patient data in order to mitigate potential gaps or errors.
Despite those hurdles, real-world evidence looks set to play a greater role in how drugs are examined after regulatory approval. Here are five reasons why:
Moving from volume to value
A new direction under the 21st Century Cures Act
Proving economics in a cost-conscious world
Reinforcing clinical outcomes
Big data boost
Healthcare's shifting focus from volume-based care to prioritizing paying for value will also serve to promote the use of real-world evidence. Such data helps to prove whether medicines deliver the same patient outcomes in everyday clinical practice as was demonstrated in randomized, controlled trials.
"From the payer side, it is all about this move to a more value-based healthcare landscape," said Alastair Macdonald, VP of real world and late phase at Syneos Health, a pharmaceutical services provider. "How do we demonstrate value for a drug and, therefore, how do we substantiate the price for that particular drug?"
These types of considerations are a principal motivation for experimenting with so-called outcomes-based contracts, where the price paid by an insurer is linked in some way to how the drug performs.
"Some payers are moving towards outcomes-based contracts," explained Bill Dreitlein, director of pharmaceutical policy at the Institute for Clinical and Economic Review. "They are all predicated on real-world evidence."
Currently, publicly announced, outcomes-based contracts only number in the dozens. But most signs point to greater interest moving forward, particularly in areas with expensive or complex medicines such as cell and gene therapies. Greater consensus on how to tap real-world data for those purposes will be a key step.
Enacted into law in 2016, the sweeping legislation known as the 21st Century Cures Act will help solidify a regulatory framework around real-world evidence.
The law directs the Food and Drug Administration to develop guidelines setting out appropriate uses for data collected in the real world. As part of the framework, the agency intends to more clearly articulate its vision for what data sources and methodologies would fulfill its requirements.
As the FDA and others are quick to note, real-world evidence won't supplant information gathered through randomized clinical study. Yet it may eventually become a larger part of how drugs are approved for new uses.
"We're now working on policies to support the use of RWE in the approval of new indications for already marketed drugs," said FDA chief Scott Gottlieb at a workshop in Washington D.C. last September. "This may be especially relevant in settings like rare diseases or other unmet medical needs, where it can be hard to enroll patients in clinical trials."
Along these lines, the FDA last spring used in vitro modeling and real-world safety data to approve Vertex Pharmaceuticals Inc.'s cystic fibrosis drug Kalydeco (ivacaftor) for patients with additional disease-causing mutations.
Real-world evidence also gives drugmakers the opportunity to highlight how pharmaceutical costs are related to associated, medical costs like hospitalization.
Bristol-Myers Squibb & Co. and Pfizer Inc., for example, have used Medicare hospitalization claims to demonstrate how use of their anticoagulant Eliquis is associated with lower bleeding-related medical costs versus warfarin.
While a branded drug will cost more than the cheap generic, if it lowers related medical spending, the switch might be beneficial from a payer's perspective — although not necessarily for the patient.
Particularly in an environment when pharmaceutical pricing is under scrutiny, drugmakers are turning to outcomes-based contracting to help secure coverage. It's no coincidence that many of the publicly announced outcomes-based deals involve drugs in competitive classes looking to unseat familiar alternatives — think PCSK9 inhibitors taking on statins.
Randomized, placebo-controlled studies are rightly extolled as the definitive measure of whether a drug works. Even so, clinical studies still have limitations due to their tightly managed nature and narrowly defined patient populations.
Exclusion criteria can leave out certain groups of patients, leaving physicians unsure of how a drug might perform in an individual with different co-morbidities or parallel treatments. Clinical studies are also by nature somewhat artificial, as investigators more closely manage participant dosing and adherence.
Real-world evidence, then, can help provide a fuller picture of whether a drug's safety and efficacy hold up in patients in less ideal settings.
Drugmakers need to be careful, however. Real-world evidence can't prove a drug works in a different type of patient. But it can reinforce what was previously shown and help convince payers and physicians of the durability of experimental results.
By necessity, real-world studies require large databases from which to pull data of interest.
Since individuals are not prospectively chosen and randomized, investigators need to sift through large pools of data to identify sufficient numbers of target patients.
Johnson & Johnson, for instance, tapped a database of almost 10 million electronic medical records from the Department of Defense system to isolate about 45,000 patients with non-valvular atrial fibrillation who had taken the anticoagulant Xarelto (rivaroxaban).
Drugmakers can use statistical analyses to better sort and divide such large data sets into different groups. A technique called one-to-one propensity score matching allows companies to pair individuals with similar demographic characteristics — a foundation that allows for stronger comparisons to be made between 'treatment A' and 'treatment B.'
"When you have large enough numbers — which is what we are able to do now with large databases — we are able to control things through statistical analyses like one-to-one propensity score matching to mimic a randomized trial without actually having to do the randomized trial," explained Alpesh Amin, professor of medicine at University of California, Irvine.
Amin led a real-world study that compared 20,803 patients who took Eliquis (apixaban) to 20,803 matched individuals receiving warfarin.
Bringing large electronic databases together with more sophisticated statistical analyses has expanded what drugmakers can show with real-world evidence.