ILLUSTRATION: KATTY HUERTAS
SCIENCE science.org 10 DECEMBER 2021 • VOL 374 ISSUE 6573 1309ment for recently diagnosed people. Anyone
in the UPMC system and the nearby region
who tests positive for COVID-19 and is eli-
gible for the antibodies can be referred to get
the therapy, and a UPMC physician explains
that they will receive one of several options.
Those who agree are automatically enrolled
in the trial and their care—a single antibody
infusion—doesn’t change. But behind the
scenes, volunteers are randomly assigned to
one of three antibody therapies, unless the
doctor or patient requests a specific one.
“No one knows if they all work equally well,”
Angus says—an even more pressing question
now, given the rise of the Omicron variant.
About 60% of those approached, nearly
10,000 people so far, have consented to get
monoclonal antibodies and ended up in the
study; typically, 10% to 20% of those eligible
for a trial agree to participate. “The sample
size is staggering,” Angus says, and both
he and Minnier plan to stretch these trial
designs beyond COVID-19. Angus recently
met with pharmacists to discuss comparing
different antibiotics in certain hospitalized
patients. Minnier wants to do the same for
blood pressure medications given in the
ICU, and diabetes drugs for outpatients.
Pragmatic and Bayesian trials aren’t new,
but Angus and Minnier say COVID-19 un-
derscored their appeal and feasibility.
“We may have underestimated our abil-
ity to pivot and change rapidly,” says Peter
Merkel, a rheumatologist at the University ofPennsylvania who leads an international net-
work called the Vasculitis Clinical Research
Consortium. Several years earlier, Merkel
received a grant to experiment with remote
informed consent and patient visits for trials,
but the project struggled because patients
and providers weren’t always accepting of the
technology, and the computer infrastructure
was difficult to navigate.
Pandemic pressure made all the differ-
ence. This time, “We were able to turn what
were pilot ideas to the norm,” Merkel says.
COVID-19 forced his network to collect far
fewer blood samples from volunteers, and
the team is reconsidering the frequency of
future sample collection and physical exams.
“Are there things you can omit without doing
major harm” to the trial, he wonders, such as
imaging studies, tissue samples, and frequent
in-person checkups?
“There’s really no good reason to go
back” to the pre–COVID-19 trial model,
agrees Yana Najjar, an oncologist at UPMC.
Like Merkel, she reoriented her trial test-
ing a drug combination for melanoma, al-
lowing participants to stretch out the time
between intravenous doses from 3 weeks to
6 and shipping medications to sites near her
patients. These changes, Najjar and Merkel
say, not only make life easier for existing
participants, but may help diversify trials,
by attracting people who live in rural areas,
lack transportation, or struggle to find time
and resources to visit a trial site regularly.“We want our trials to represent that popu-
lation with the disease, not just the patients
who live” near study sites, Merkel says.
“This democratizes trials.”
The shift comes with hurdles. “There’s
much more burden on the data science
team,” says Manisha Desai, a biostatistician
at Stanford University. A trial she’s involved
in, on pulmonary arterial hypertension,
assesses how far participants can walk in
6 minutes at home, while researchers
gauge fatigue and shortness of breath via
a video link—a COVID-19–driven change.
Desai notes the data are “so much noisier
than the gold standard,” which is measured
on an in-person test at a clinic. There’s less
uniformity among walking paths, for ex-
ample. Unger is studying to what degree, if
any, shunting parts of studies outside can-
cer centers has compromised quality.
At UPMC, the new normal is a work in
progress. Marroquin doesn’t know yet how
many researchers will be willing to link up
their trials with his vast, new infrastruc-
ture. And he acknowledges that the system
can’t help power trials beyond UPMC. Still,
like others, he sees change enduring. A key
goal, Angus says, is to spur enthusiasm
among more patients by offering easier
to navigate trials that minimize risk and
demand less sacrifice. “The design should
basically ask for as little altruism” as possi-
ble, he says—while tackling as many ques-
tions as it can to speed science along. j