Changes in the way data are being collected are on track to
revolutionize the use of real-world data in drug approvals, Janet
Woodcock MD, the longtime director of the FDA’s Center for Drug
Evaluation and Research, said here.
“Traditional drug development programs … don’t usually get much
insight into who to use a drug in and who not to use a drug in,”
Woodcock said Thursday at an event on the future of health data
sponsored by Datavant, a company that helps healthcare systems with
patient identification. “Are there ways we can capture and utilize the
data collected … to make the clinical trial enterprise more effective
and efficient?”
The widespread use of electronic health records has drastically
improved the availability of health data, she said. With paper charts,
individual physicians “had to do archaeological expeditions — literally —
to find out what was going on. The same is true of FDA — we used to get
tractor-trailers’ worth of paper, literally [to approve drugs] … Now we
get data sets that are standardized.” The other big factor “is the
digital revolution in general — people have wearables, a lot of people
have iPhones, we have telehealth that’s developing. All these things are
coming together to provide rich data sources that we can turn into
evidence. Our task is to take all the data out there and turn it into
something that’s actionable.”
Real-World Data in Use
FDA already has approved numerous supplements and many drugs based on
real-world evidence, Woodcock noted. “These are usually for rare
conditions, such as male breast cancer, which doesn’t happen very often
but is just as devastating as any other cancer. We were able to use
evidence generated from use to add that indication to the label.” Using
this type of evidence plays into “the long-term vision, articulated
quite some time ago by the [National Academies of Sciences, Engineering,
and Medicine] as the ‘learning healthcare system’ — that we’re able to
make all the information we collect actionable, to assimilate that
learning from treating patients and rapidly transform it back into best
practices … We’re very, very far from that, but that’s the long-term
vision.”
“In the shorter term, what we’re going to see is in terms of drug
development and the intervention space and the best practices of
delivering healthcare,” she continued. “We’re already generating lots of
insights about what’s the best way to do this or that … That will
continue to grow.” In addition, data gleaned from wearables and other
new technologies will become part of healthcare delivery.
However, as both the industry and regulators know, “the data out
there are not necessarily fit for this purpose right now,” said
Woodcock. Take for example, the medical chart of a cancer patient. “What
does it say in the chart about cancer? Do you have the right cancer
written down? Sometimes, the answer is no. Are the biomarkers in there?
Are they entered correctly? That’s important information for the
patients … It ought to be right the first time.”
Resisting Standardization
One problem with trying to make data more actionable is that
“everyone resists standardization,” she said. “Doctors, especially, want
to do things their own way.” She related a story from Peter Bach, MD,
director of the Center for Health Policy and Outcomes at Memorial Sloan
Kettering Cancer Center, in New York City, who wanted to have lymphoma
care and treatment standardized, so researchers could compare outcomes
from different treatments. But some physicians documented their cases
differently than the others, so he asked them why and one of them
replied, “I don’t agree with the consensus definition of stages of
lymphoma. I have my own stages.”
“We need to sell people on the return of that evidence — that if you
put something down in a standard way, you’ll get something back,”
Woodcock said. However, “it’s an uphill battle. There is much discussion
in healthcare circles about what the electronic record is doing to
clinicians and to the doctor-patient relationship. It doesn’t have to be
that way. That can be fixed, but we’re really going to have to focus on
that.”
One problem that crops up with real-world data is that it’s hard to
screen out the confounders. “But something we’re exploring is, can you
randomize people in your practice and collect real-world outcomes data …
without a very elaborate clinical trial apparatus, and can you still
get a valid result that’s actionable?” she said.
“We’re collaborating with a lot of groups trying to sort this out —
different groups are trying to raise the reliability of that
[observational] evidence … Where randomized clinical trials [RCTs] are
ongoing, we’re doing a companion observational study, and we’ll see what
the results are. We hope to have 20 retrospective ones done by next
year, and seven prospective ones. Of course, the real-world study will
get done a lot quicker than the RCT … but it will be very interesting to
see how they compare and what kind of answers they give.”
Privacy Tradeoffs
The privacy tradeoffs involved in collecting data also are important
to think about, said former FDA Commissioner Robert Califf, MD. “In what
circumstances do we trade off privacy for the benefit of getting better
healthcare? Where are we when health systems advertise the nice doctor
in the white coat to cure your disease, when for the most part the
doctor is guessing? The guesses are not uneducated guesses, but the fact
is whenever we do prospective studies, we find out we’re wrong about
half the time, but don’t know which half we’re wrong [or right] about.”
For example, take the recent red meat controversy. “Is it OK to eat red
meat or not? The data are contradictory because good studies haven’t
been done.”
The use of aspirin to prevent heart attacks is another example, added
Califf, who will soon be leaving his position as professor of
cardiology at Duke University to work for the Alphabet health subsidiary
Verily Life Sciences. Although aspirin has been shown to be effective
for secondary prevention, “it’s always bothered me that we don’t know
the right dose of aspirin,” he said. “Now we have enrolled over 17,000
people in a randomized trial of baby aspirin versus full-strength
aspirin, and we’ll soon know the answer. But this is 140 years after
aspirin first went on the market … It’s possible to know a lot more than
we know if we just have the discipline to admit uncertainty.”
https://www.medpagetoday.com/publichealthpolicy/fdageneral/82711
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.