13 Things That Don't Make Sense (23 page)

Eventually, they found the source of this much-quoted, never-questioned statistic: Henry Knowles Beecher. In
The Powerful Placebo
, published in the
Journal of the American Medical Association
in 1955, Beecher made the first loud call for the use of double-blind, placebo-controlled trials in assessing medical treatments.
The paper documents his analysis of a dozen studies, an analysis that produced the magical 35 percent figure.

It wasn’t enough to convince Hróbjartsson and Gøtzsche, so they carried out a meta-analysis. This is what scientists do when
they are faced with a long series of conflicting answers to a question; essentially, it is a formalized way of analyzing all
previous attempts to answer the question. They examine the quality of each one: its experimental methods, its biases, its
statistical analyses. The idea is to get a flavor of each set of results and then put them together in a way that reflects
how much weight should be given to their stated results. In the end, such a study makes some pronouncement about the overall
weight of evidence for or against a hypothesis.

Hróbjartsson and Gøtzsche’s meta-analysis of the placebo effect took the data from 114 clinical trials that had compared placebo-treated
patients with untreated patients. Overall, there were around 7,500 patients suffering from about forty different conditions
ranging from alcohol dependence to Parkinson’s disease. Over this wide spectrum of complaints, they found no evidence that
placebo treatments had significant effects on health. The only place there was possibly some effect was in the trials that
involved pain relief, but even here it was hard to be sure. Pain is a subjective measure, and patients like to please their
doctors, Hróbjartsson points out; they may well have reported less pain than they actually felt. Certainly the objective measures,
such as blood pressure and cholesterol levels, showed no placebo response. The researchers called for doctors to stop using
placebos in clinical situations. “The use of placebo outside the aegis of a controlled, properly designed clinical trial cannot
be recommended,” they said.

In 2003 Hróbjartsson and Gøtzsche revisited the analysis, this time with data from 156 trials and 11,737 patients. Their results,
published in the
Journal of Internal Medicine
, were unchanged. They “found no evidence that placebo interventions in general have large clinical effects, and no reliable
evidence that they have clinically useful effects.” Placebo, they conclude, is a far from proven phenomenon; the only possible
exception is in pain relief, and even here the placebo response was not clearly above what they would expect to see in doctor-pleasing-biased
subjective reporting. “Most patients are polite and prone to please the investigators by reporting improvement, even when
no improvement was felt … we suspect reporting bias occurred,” the researchers write.

Hróbjartsson and Gøtzsche’s work is well respected and has contributed a significant amount to the controversy over our handling
of placebo. Nevertheless, we have significant evidence from equally well-respected researchers that the placebo effect is
real. Brain imaging has shown the pathways involved in the brain, for example. In 2005 researchers from the University of
Michigan published their work with a
positron emission tomography
(
PET
) scanner, showing the endorphin system in the hypothalamus activating when patients received an injection they had been told
was a pain medication. Reporting bias seems unlikely given that these trial patients were being deliberately hurt (by a saline
injection in the jaw) as part of the experiment; they had no reason to report less pain in order to please the researchers
carrying out the experiment.

An editorial accompanying Hróbjartsson and Gøtzsche’s original paper seems to sum up the general feeling. Though the author,
John Bailar of the University of Chicago, admitted to little more justification than a “pesky, utterly unscientific feeling
that some things just ought to be true,” he suggested their conclusions were “too sweeping.” Things that happen in research
labs “may obscure a real effect of placebo that would be evident in nonresearch settings.” The solution to this problem is
unforthcoming, however; “it is not clear how one could study and compare the effects of placebo in research and nonresearch
settings, since that would of course require a research study.”

Perhaps an informal visit to Turin would help Bailar. It certainly cured me of any doubt about the reality of the placebo
effect.

WHEN
I asked Fabrizio Benedetti if I could experience a placebo response for myself, he was far from convinced it would work. Normally,
his team won’t tell their trial volunteers what kind of experiment they are carrying out; such knowledge might skew the results.
It didn’t in my case. In a windowless basement room below Turin’s towering San Giovanni Battista hospital, I repeatedly subjected
myself to pain. And, against all my expectations and with my full knowledge of what they were doing, the doctors present were
able to reduce it with nothing more than a lie.

My first experiment measured the effects of caffeine on muscle performance, following a routine that involves exercising before
and after a small cup of cold, rather unpleasant coffee. While I was drinking the coffee, the white-coated Dr. Antonella Pollo,
one of Benedetti’s colleagues, filled my head with stories about how caffeine is a banned substance in athletics. Her sister,
she said, does archery. She is always told not to drink anything containing caffeine before an event; apparently; it enhances
muscle performance and gives an unfair advantage. I knew there was a lie somewhere—perhaps there was no caffeine in the coffee,
maybe caffeine has no effect on muscle performance, or maybe Pollo simply reduced the resistant weights for the exercise session
after the coffee break—but I was definitely able to do more after the coffee than before.

When the experiment came to an end, Pollo came clean. There was no caffeine in the coffee. Nevertheless, I had been sufficiently
convinced of my increased powers to perform much better the second time around. She looked rather pleased. The experiment
was far from rigorous and—in my quick and dirty clinical trial, at least—had many flaws. What’s more, she hadn’t expected
it to work at all on someone who knew what was going on.

The next test came from another white-coated doctor, Luana Colloca, who entered the room holding what looked like a couple
of button cell batteries on a plastic strip. They were electrodes. “Do you mind receiving an electric shock?” she said.

When I consented, she strapped the electrodes to my forearm. Then she wired me up to a computer programmed to manipulate the
mind as it gives a series of electric shocks.

The computer screen told me—via a red or a green light—whether the shock I was about to get would be mild or severe. The deception
here comes from a conditioning, where the brain learns to associate a color with an anticipation of a particular level of
pain. The screen shows a color, and about five seconds later the computer gives a shock. Green for severe (something like
an electric fence jolt) and red for mild (no more painful than a light touch on the arm). But once the conditioning is established,
playing with the color can play with the brain’s perception of pain.

It worked. After around fifteen minutes of conditioning, the last run of shocks all felt mild, like a touch on the arm, whether
introduced under a red or a green light. But they were all severe, Colloca told me afterward. By rights, every one of them
should have felt like touching an electric fence.

In some ways, I shouldn’t be surprised. The brain is an astonishing organ, a supremely complex collection of molecules that
process signals—both chemical and electrical—to give us our sense of who we are and how we experience the world around us.
With careful control of the signals going in, why shouldn’t that sense be open to manipulation?

We know there are many ways to alter the state of a human brain and the body it oversees. The most obvious are the five senses:
the smell of cut grass evokes a particular memory state; the taste of chocolate releases serotonin; the touch of a lover and
the sight of a big-eyed puppy both release the oxytocin molecules that bond us to our partners or our children (or our dogs);
the sound of a scream sends a rush of adrenaline through us, making us ready for fight or flight.

Electrical signals can bypass conscious bodily control too. Sufferers of Parkinson’s disease, for example, can have their
tremors stopped with a microchip implanted in the hypothalamus. Benedetti, an experienced neurosurgeon, performs such implantations;
not only can they help a Parkinson’s patient’s motor control, but they also provide a tool for investigating the neural mechanisms
of the placebo effect. Tell patients that their implant’s settings have been altered so that it will be harder to control
their movements, and they respond by doing everything at a snail’s pace. Tell them the opposite—that the electrodes are now
set for optimum mobility—and suddenly the movements become normal. In neither case does anyone need to touch the electrode
settings to achieve the effect: expectation of a significant improvement—or degradation—in the motor control of Parkinson’s
patients gives them just that; tell them they’re going to be impaired, and they will be. It’s not just about positive thinking:
it’s about the chemical or electrical signals that positive thinking produce.

Benedetti has shown this explicitly. The classic Parkinsonian symptoms of muscle stiffness and tremors are caused by explosive
bursts of signals coming out of a specific region of the brain: the subthalamic nucleus. Injections of the drug apomorphine
reduce this hyperactivity to near-normal levels and take away the associated stiffness and tremors. Benedetti’s team took
a group of sufferers who had had electrodes implanted in the subthalamic nucleus, and gave them apomorphine injections for
a few days. They then covertly switched the injection to saline—still telling the patients that the injection would relieve
their symptoms. It did, and measurements through the implanted electrodes showed reduced activity in the neurons of the subthalamic
nucleus. Placebo, it seems, is all in the brain—and it is real.

IT
is here that the placebo effect turns into something like medicine’s equivalent of dark energy: a repeatable, measurable phenomenon
that could still turn out to be an illusion. A broad analysis of the best clinical data says it might not exist—at least not
in significant amounts. But even with full knowledge of what was going on, I found myself powerless to resist the placebo
effect. It is not simply about deception, a sugar pill being perceived as an efficacious cure. We can create it with mind
tricks, brain implants, or chemical cocktails, and we can see it working on brain scans. Though there is scientific evidence
that the placebo effect is a myth, or that we have misled ourselves about what is going on, there is perhaps more evidence
pointing the other way.

Clinical studies show you can cut morphine use by half—over the long term—if you just make sure the patient knows you’re giving
it. Telling patients they are being injected with a painkiller—while injecting them with saline—is as effective as injecting
6–8 mg of morphine. Studies at the U.S. National Institutes of Health found that cocaine abusers in a recovery clinic can
get by on half doses too—as long as they know they’re getting something. Expectation is a powerful thing.

In fact, we’re back at diazepam. On its own—administered covertly—it does nothing. It’s about diazepam
plus
the expectation chemicals that anticipation of a dose produces; the expectation chemicals are quite good by themselves, but
with diazepam added to the mix, you’re really in for a treat.

These expectation chemicals have a dark side too, though. Benedetti and Colloca have already started to put warnings out that
placebo research could be exploited for questionable purposes. We are only wading in the shallows of the science of placebo,
and it’s already clear that this, like genetics, could be a murky pond. “There are … potentially negative outcomes of
placebo research,” they wrote in a
Nature Reviews
article in 2005. “If future research leads to a full understanding of the mechanisms of suggestibility of the human mind,
an ethical debate will then be required.”

That is especially true in light of the
nocebo effect
, where deliberately inducing anxiety can make pain worse. Benedetti is one of the few people who have been able to study
this phenomenon; if researching placebo poses an ethical dilemma for doctors, nocebo doubles it.

Nocebo
means “I shall harm.” In a nocebo study, the harmless medicine is delivered with a phrase such as, “This really will make
you feel much worse.” It could prove an extremely valuable tool, and Benedetti is already using his nocebo experience to overcome
the limitations of current painkillers, but what kind of ethics committee gives approval to a scheme designed to make patients
more uncomfortable through lying to them? None. Which is why Benedetti has to rely on paid volunteers who are willing to suffer.

It started in 1997, when he and his colleagues were testing the idea that anxiety makes pain worse. They injected a group
of patients who were recovering from painful surgery with proglumide, a chemical that blocks the action of cholecystokinin
(CCK), a neurotransmitter chemical associated with anxiety. When they gave these patients an inert pill and told them it would
make them feel worse, it simply didn’t. It was impossible to induce the nocebo effect when CCK was blocked.

It was a good result, but scientifically lacking—there was no control group that
didn’t
get the CCK-blocking proglumide and thus
did
feel the additional discomfort that anxiety can bring. Unfortunately (for Benedetti, if not for the patients), there was no
ethical approval for a control group.

It took Benedetti nearly ten years to get approval and volunteers for a follow-up study. At the end of 2006 his team published
a paper showing that we—or rather our neurotransmitters—can turn anxiety into pain. The volunteers underwent a routine involving
a tourniquet, some injections, and a verbal warning that their pain would increase while Benedetti’s team took blood samples
and asked them how they rated their pain. The blood samples gave the researchers what they were looking for: proof that proglumide
stops us from turning chemical signals of anxiety into exaggerated pain. Proglumide is the only CCK blocker licensed for human
use, but it is not particularly effective. When researchers manage to develop something better, they will have a drug that
can be mixed with narcotics to alleviate physiological and psychological pain simultaneously. Though nocebo seems somewhat
dark—one can imagine it being exploited to produce extra anxiety and thus pain in interrogations, for example—at least it
has positive applications too.