An article about clinical trials. Good weekend reading:
Trial and error
Clinical trials have proved a valuable medical innovation. But they still
have their weaknesses
OCTOBER 30th marks the 50th anniversary of one of the main tools of
scientific medicine: the modern clinical trial. Although the idea of testing drugs
on a few people before inflicting them on everyone else was not original,
Austin Bradford Hill's publication of his trial of an antibiotic called
streptomycin on patients suffering from tuberculosis set a new standard. It was,
in the jargon, a randomised controlled trial: “controlled” because only half the
volunteers received the drug (the others remained untreated, acting as a
benchmark, or “control” against which the effectiveness of the antibiotic could
be assessed); and “randomised” because those who took the drug were
chosen, as statistical sampling theory requires, at random.
Doctors are toasting this landmark, and rightly so. Clinical trials have improved
both medical practice and patients' prospects enormously. For example, the
fact that cholesterol-lowering drugs are now given routinely to reduce the risk
of cardiovascular disease, or that American women over the age of 50 are
encouraged to have yearly mammograms, is largely due to clinical trials. But
the power of trials is such that it is all too easy to be blind to their faults.
First, although randomised trials work well for drugs, it is harder to test other
sorts of treatment with them. Second, their expense means that private firms
are often the only organisations that can afford to finance them. Such firms are
usually interested in things that they can patent, and their scientific detachment
may sometimes be coloured by the desire to turn a quick buck. Third, the
controlled conditions of a trial may be so unlike the outside world that
something which passes in trials will fail in the clinic. And fourth, the question a
trial sets out to answer may be of great significance to the researcher, but of
little personal relevance to the ultimate consumer: the patient.
The burden of proof
One of the main reasons why few medical interventions, apart from new drugs,
have been put through their paces in clinical trials is that it is rarely regarded as
sufficient to compare a new treatment with no treatment at all. If there is no
pre-existing treatment to act as a control, a “placebo” is often created to give
the appearance that the patient is being treated with the product under trial.
(This is because even the impression of treating a disease sometimes has a
positive benefit for the patient, a phenomenon known as the placebo effect.)
But while it is easy to compare the efficacy of, say, Viagra, with that of a
dummy pill, it is harder to come up with a placebo to prove the worth of such
things as minor surgery. Other medical procedures, too, are difficult to
benchmark. What, for example, is the appropriate control when assessing the
benefits of psychotherapy? Time with a gibbering amateur compared with a
session with a trained professional? Despite this systematic difficulty, however,
some doctors are trying to extend clinical trials to areas outside
pharmaceuticals.
Baruch Brody, a bioethicist at Baylor College of Medicine in Houston, is
among them. He is helping to run a trial designed to assess arthroscopic
surgery—a controversial approach to treating damaged knees through small
“keyhole” incisions.
Some critics feel that this operation is no better than letting nature take its
course. To find out, the 200 patients in the Houston trial are being given an
amnesia-inducing drug to blot out the memory of the operation before they are
whisked into the operating room. But only half are actually undergoing the
operation. The rest have to make do with a quick swipe across the knee with a
scalpel, to mimic a surgery scar. That, it is hoped, means that no patient will
know whether he has been operated on. Preliminary results from the trial are
expected early next year.
The difficulty of running trials for non-drug procedures is inherent in the nature
of the trials themselves. But the failure to test promising drug treatments is
more often a failure of the laws of property, for the drugs in question are
usually either established, out-of-patent medicines whose therapeutic range
might be extended (aspirin, for example, has been found to help prevent and
treat heart attacks), or herbal preparations that, being natural products, do not
count as “inventions” for patenting purposes. Useful therapies are thus being
missed because firms cannot profit from them.
This is, of course, a legitimate area for government intervention. And
governments (and charities) do, indeed, intervene. But it would be better if
ways to use private rather than taxpayers' money could be found. Sometimes
they can. Rory Collins, co-director of Oxford University's clinical-trials service
unit (which demonstrated the efficacy of aspirin against heart disease) has
managed to piggy-back on a commercial trial in order to test the notion that
folic acid, a food-supplement commonly given to pregnant women, may also
prevent heart attacks.
The trial in question has been paid for by Merck, a large American drug
company, to test one of its own cholesterol-lowering drugs. But since folic acid
is suspected of enhancing the efficacy of the Merck product, including it
caused few qualms.
Financial interests may also operate in other, darker, ways. Trials are
sometimes terminated prematurely, and “inconvenient” data suppressed. But
there are grey areas here. It is not always wrong to stop a trial when no benefit
is accruing to those on the experimental treatment. Indeed, it would frequently
be wrong not to stop it. Killing off unpromising trials and then burying the
results, while publicising the data from more successful experiments, is,
however, a much more dubious practice.
One example of this is a dispute between researchers at Toronto's Hospital for
Sick Children and a Canadian drug company called Apotek. This firm is
accused of pulling the plug on a trial when a drug intended to treat a blood
disorder was found to be floundering—and then trying to muzzle the scientist
who spoke out. And that is not the only instance. In 1996 it emerged that a
researcher at the University of California, San Francisco was forced by Boots
Pharmaceuticals (a company subsequently bought by BASF) to withdraw a
paper which showed that the firm's expensive drug for thyroid disease was no
more effective than cheaper treatments that already existed.
The third problem, that the results of clinical trials may be irrelevant to the real
world, has several components. One which has been recognised and largely
dealt with over the past few years is inappropriate sampling. For a long time,
most drugs were tested only on men—usually white men, even in countries
such as America, which have large non-white populations.
Changes in the rules for drug approval, combined with more sensible attitudes
by researchers, mean that this happens more rarely now. But another difficulty
is less easy to eradicate. Sometimes the treatment in the trial—meted out by
experts—is so much better than what is normally on offer to patients that the
results exaggerate its real benefit. The ENRICHD trial, under way in America, is
testing the role of professional psychotherapy in improving the physical health
of patients who have had a heart attack and are depressed. But, according to
Dr Brody, the quality of treatment in the trial may far exceed what is likely to
be available outside it.
The gap between trial and reality becomes even more glaring when studies are
conducted in poor countries, where the compounds under test (such as AZT,
an anti-HIV drug) are beyond the purse of most of those who might benefit
from them. But the idea of benefit often seems to get lost even in trials in the
rich world. The advantages of life-saving drugs such as streptomycin are
obvious. Many trials, however, are looking for subtler effects—without, in the
view of observers such as Hilda Bastian, the head of the Consumers' Health
Forum of Australia, assessing whether the treatment really improves patients'
lives, as opposed to changing a few physiological indicators.
There are data to back up Ms Bastian's view. According to Caroline Sanders
and her colleagues at the University of Bristol, fewer than 5% of the trials
published between 1980 and 1997 measured the emotional well-being and
“social function” of their participants.
Moreover, even when such measures are incorporated into trials, translating
the results into routine practice is a tricky business. International organisations
such as the Cochrane Collaboration are working on better ways to analyse,
package and present the results of clinical trials to both doctors and patients,
so that those results can be put into practice. But it is an uphill struggle. After
all, 50 years after the first clinical trial to cure tuberculosis, the disease still kills
almost 3m people a year.
economist.com |
