Using real-world data (RWD) in medical research has a long history. Frustrated that Royal Navy sailors contracted scurvy on the long trip to India, James Lind, a Scottish physician, decided to find a solution. Using the HMS Salisbury as a testing ground, he divided the crew into groups, and fed them special food and drinks. Some men had cider, while others sucked on oranges. One unfortunate group had to gulp down seawater. Because supplies were so expensive, Lind had to stop his experiment early. But he managed to prove a crucial point: acidic foods like oranges helped fight scurvy, an illness that killed more seamen than French and Spanish broadsides combined.
From these exciting beginnings, RWD has retreated back to the edges of medical life. Meanwhile, traditional clinical trials have ballooned.
Professionals consider them an invaluable part of research, while watchdogs have planted forests of regulations about how they should be conducted. This power is also reflected economically. Advocacy group Research!America found that pharmaceutical and biotech companies spend upwards of $100 billion a year on R&D. According to a 2014 study by Tufts University, meanwhile, it takes around $2.56billion for each new drug to reach market.
Researchers are slowly appreciating that traditional clinical trials have their limitations, however. In an increasingly globalised society, artificial samples fail to accurately predict the effects of new drugs on diverse groups of patients.
The contrived conditions of regular trials hardly help; no matter how strict the rules are in a clinic, no real-world patient ever takes their medication at exactly the right time, every day.
Just as Lind embraced the power of RWD, testing subjects in the field to fix a specific problem, researchers are again using patient data in similar ways.
The result is a major shift in how clinical trials are conducted – a change that also provides huge benefits to patients.
Dr Nilay Shah is in a good position to reflect on these advances. Chair of the division of healthcare policy and research at Mayo Clinic, and an associate professor of health services research at the Mayo Clinic College of Medicine, he has plenty of experience using RWD in practical applications. Through his work as an adviser to the US Food and Drugs Administration (FDA), Shah is also familiar with the regulatory side of clinical research.
Shah’s wealth of knowledge encourages him to take the long view. He highlights that, while RWD has existed for years, it was often too cumbersome to access effectively. “One of the big areas of growth across the past decade – all over the world – has been the availability of deeper and richer seams of electronic medical data,” he explains.
“Before, while the data was available, it wasn’t readily usable. There were no standards, and data from each medical record looked different.”
This problem was only solved after researchers worked to collate data under topic-specific umbrellas. One example is Orpha, a database for conditions such as Gaucher disease, a rare genetic disorder.
Through his work with the Patient- Centred Outcomes Research Institute (PCORI), Shah has helped to develop a similar system, called PCORnet.
“PCORnet aims to create an infrastructure of RWD to be able to create faster evidence especially from clinical trials,” he explains. “Projects like this help harmonise data. They enable us to build models to put all this data together in a standard way which we can then use, something we’re able to do at a scale that was impossible even six to ten years ago.”
Broad technical developments are being dovetailed by a more sensitive approach to the data itself. While electronic health records (EHRs) are crucial, they cannot be used in isolation. By combining EHRs with other metrics, researchers can sketch a detailed picture of a patient’s health.
“We realised early on that using EHR data by itself is not enough,” Shah says. “We need the links to administrative claims and insurance records as well. This is because each healthcare system has its own separate EHR data, but if you track a patient over time you really don’t know the complete mass of data.”
Cheaper and faster
All this is transforming how researchers recruit clinical trials. Unlike traditional tests, RWD allows clinicians to quickly target patients prone to a particular illness. Shah sees these benefits in his own work, on cardiovascular disease. Until recently, the disease was more likely to affect men than women. Medical advances have balanced things out, but clinical trials have yet to catch up. For its part, the American Heart Association uses guidelines based on trials where women account for just 30% of patients.
By using EHRs in conjunction with other data, Shah and his team were able to tweak the recruitment process, and better reflect present conditions. “The rate of advanced cardiovascular cases, with the specific underlying risk group of patients, is now quite different compared with how it was ten or 12 years ago,” he explains.
“We’re examining the differences in the rates of cardiovascular cases, which allows us to design [the clinical trial] very differently than if you were using the data from when the last big trial was published,” he continues.
“Access to more data can also help researchers in other ways, notably by keeping prices down.
“By using all this readily available data, you can conduct more efficient research and decrease the burden on data collection,” Shah says. “The result is that clinical trials are cheaper to organise.”
These changes cut both ways. If RWD helps researchers organise better clinical trials, it can also smooth the process for patients. “In the past, the patient had to come down to multiple trials, or needed to go back to the clinic for additional testing. This was important to meet the clinical trial requirement,” Shah says. “By using RWD, researchers can decrease the burden significantly on the patient.”
Shah highlights how patient-recorded data – using mobile phones to track exercise rates, for example – can help patients save time.
“Researchers are thinking more about how ‘wearable data’ can be incorporated [in clinical trials],” he says. “There are a number of other exciting data sources, involving different collection methods, like online surveys.”
Putting fewer demands on patients mean they’re easier to recruit, a welcome change given 80% of clinical trials don’t have enough participants.
This is especially helpful when investigating conditions that mostly affect older people, like blood clotting. After all, it’s precisely these vulnerable patients who tend to shrink from gruelling clinical trials.
A related change is the spread of so-called ‘pragmatic trials’ whereby RWD is enhanced by studying drugs in the field. Rather than staying in the lab, researchers observe drugs’ effectiveness in hospital wards and surgeries.
Apart from being gentler on patients, pragmatic trials can provide better results, Shah explains. “Because you use more real world populations in pragmatic trials, the applicability of evidence tends to be broader when compared with typical clinical trials,” he says. “Observational research can provide stronger signals as well.”
FDA OK
Despite the clear benefits to trial recruitment and procedure, some clinicians are still reluctant to embrace RWD completely. “Obviously, there are people who feel very strongly that randomised trials are the gold standard, and that [they’re] the only way to generate evidence, especially on the research side,” Shah admits. “I think it’ll require some proof of concept trials before people start using this data more broadly.”
Similar worries exist among regulators. “In recent months, FDA has passed [rules] that require it to consider more real-world evidence,” Shah says, “but while there’s a legislative mandate [to use RWD], this doesn’t mean it’s ready to be used in decision making yet.” Shah’s own uncertainty is reflected by FDA itself. Its new guidelines encourage staff to consider the “characteristics and sources of RWD, and that may be sufficient for use in making various regulatory decisions”, but also emphasises that it doesn’t specifically “mandate [RWD’s] use”.
A particular concern involves protecting patients’ data; a difficult task, given the explosion of new databases and a decentralised recruitment system, but researchers are working hard to balance RWD’s power with patient confidentiality.
“My work involves [giving] patients more power over their own data,” Shah says. “While this means pulling together a lot of meaningful data for research, patients need to own their own data, and choose who they share it with.”
Given all his hard work, Shah is unsurprisingly buoyant about the future of RWD. “I think there’s going to be significant growth of RWD in clinical trials,” he says.
“Specifically, there’s going to be an increase in the use of pragmatic trials, something PCORI has been a big proponent of. [I also think] traditional clinical trials are going to start using a lot more RWD, especially for long-term monitoring and assessment of drugs.”
Through new schemes like ‘Health eHeart’ – which enables patients to submit health data anonymously online – Shah also predicts patient recruitment will keep getting simpler. “I think there might be increasing opportunities to recruit patients more directly than having to make them go through the typical [procedures] to recruit them in the trial,” he says.
“The clinicians will still be engaged, and will have to positively identify patients, but they will no longer be required to constantly come into the clinic. Overall, I suspect there is going to be more efficiency in terms of collecting data over time.”
All this is a long way from the HMS Salisbury, but it’s pleasing to see the pharmaceutical industry coming back to the thoughtful methods first pioneered by James Lind. At least this time, researchers don’t have to worry about running out of oranges halfway through.
