How many times have you encountered a study — on, say, weight
loss — that trumpeted one fad, only to see another study discrediting it a week
later?
That’s
because many medical studies are junk. It’s an open secret in the research
community, and it even has a name: “the reproducibility crisis.”
For any study
to have legitimacy, it must be replicated, yet only half of medical studies
celebrated in newspapers hold water under serious follow-up scrutiny — and
about two-thirds of the “sexiest” cutting-edge reports, including the discovery
of new genes linked to obesity or mental illness, are later “disconfirmed.”
Though erring
is a key part of the scientific process, this level of failure slows scientific
progress, wastes time and resources and costs taxpayers excesses of $28 billion
a year, writes NPR science correspondent Richard Harris in his book “Rigor Mortis: How Sloppy Science Creates Worthless Cures,
Crushes Hope, and Wastes Billions” (Basic Books).
“When you
read something, take it with a grain of salt,” Harris tells The Post. “Even the
best science can be misleading, and often what you’re reading is not the best
science.”
Take one
particularly enraging example: For many years research on breast cancer was
conducted on misidentified melanoma cells, which means that thousands of papers
published in credible scientific journals were actually studying the wrong
cancer. “It’s impossible to know how much this sloppy use of the wrong cells
has set back research into breast cancer,” writes Harris.
Another study
claimed to have invented a blood test that could detect ovarian cancer — which
would mean much earlier diagnosis. The research was hailed as a major
breakthrough on morning shows and in newspapers. Further scrutiny, though,
revealed the only reason the blood test “worked” was because the researchers
tested the two batches on two separate days — all the women with ovarian cancer
on one day, and without the disease the next. Instead of measuring the
differences in the cancer, the blood test had, in fact, measured the day-to-day
differences in the machine.
So why are so
many tests bogus? Harris has some thoughts.
For one,
science is hard. Everything from unconscious bias — the way researchers see
their data through the rosy lens of their own theses — to the types of beaker
they use or the bedding that they keep mice in can cloud results and derail
reproducibility.
Then there is
the funding issue. During the heyday of the late ’90s and early aughts,
research funding increased until Congress decided to hold funding flat for the
next decade, creating an atmosphere of intense, some would say unhealthy, competition
among research scientists. Now only 17 percent of grants get funded (compared
to a third three decades ago). Add this to the truly terrible job market for
post-docs — only 21 percent land tenure track jobs — and there is a greater
incentive to publish splashy counterintuitive studies, which have a higher
likelihood of being wrong, writes Harris.
One effect of
this “pressure to publish” situation is intentional data manipulation, where
scientists cherry-pick the information that supports a hypothesis while
ignoring the data that doesn’t — an all too common problem in academic
research, writes Harris.
“There’s a
constant scramble for research dollars. Promotions and tenure depend on making
splashy discoveries. There are big rewards for being first, even if the work
ultimately fails the test of time,” writes Harris.
‘Promotions and tenure depend on making
splashy discoveries. There are big rewards for being first, even if the work
ultimately fails the test of time.’
This will
only get worse if funding is cut further — something that seems inevitable
under proposed federal tax cuts. “It only exacerbates the problems. With so
many scientists fighting for a shrinking pool of money, cuts will only make all
of these issues worse,” Harris says.
Luckily,
there is a growing group of people working to expose the ugly side of how
research is done. One of them is Stanford professor John Ioannidis, considered
one of the heroes of the reproducibility movement. He’s written extensively on
the topic, including a scathing paper titled “Why Most Published Scientific
Research Findings Are False.”
He’s found,
for example, out of tens of thousands of papers touting discoveries of specific
genes linked to everything from depression to obesity, only 1.2 percent had
truly positive results. Meanwhile, Dr. Ioannidis followed 49 studies that had
been cited at least a thousand times — of which seven had been “flatly
contradicted” by further research. This included one that claimed estrogen and
progestin benefited women after hysterectomies “when in fact the drug
combination increased the risk of heart disease and breast cancer.”
Other
organizations like Retraction Watch, which tracks discredited studies in real
time, and the Cochrane group, an independent network of researchers that pushes
for evidence-based medicine, act as industry watchdogs. There is also an
internal push for scientists to make their data public so it’s easier to police
bad science.
The public
can play a role, too. “If we curb our enthusiasm a bit,” Harris writes,
“scientists will be less likely to run headlong after dubious ideas.”