In Medicine, Evidence can be Confusing
By Rita Rubin
USA TODAY
Medicine is littered with missteps:
•For years, doctors urged post-menopausal women to take
estrogen to protect their hearts. But in 2002, a study of
16,000 women found that those on estrogen plus progestin actually
had more heart attacks than those on a placebo. In 2004, a
similar study found no heart benefit from estrogen alone.
•In the 1980s, doctors enthusiastically prescribed two
new drugs that seemed to eliminate one type of potentially
dangerous irregular heart beat. But a 1989 study found that,
compared to a placebo, the drugs actually raised the risk
of sudden death.
•A 1968 report about one comatose patient concluded that
dexamethasone, a corticosteroid, could save the lives of patients
with deadly cerebral malaria. But a 1982 study of 100 comatose
patients with cerebral malaria found that, compared to a placebo,
dexamethasone actually prolonged comas.
In each example, a randomized controlled trial, in which patients
were randomly assigned to receive either the treatment or
a placebo, dispelled long-held beliefs. Such trials are the
cornerstone of "evidence- based medicine," which
has galvanized doctors perhaps more than any other subject.
Critics condemn evidence-based medicine as "cookbook
medicine" that devalues the doctor's experience and the
patient's preference.
Proponents argue that evidence from randomized controlled
trials has stanched the flow of private and public dollars
for useless or even harmful treatments. More important, they
say, the information has saved countless lives. Both sides
agree on one point: Keeping up with the latest evidence is
virtually impossible. Consider that there are 25,000 medical
journals worldwide, says obstetrician/gynecologist David Grimes,
vice president of biomedical affairs at Family Health International
in Research Triangle Park, N.C. "We fall behind on the
first day of medical school, and we never catch up,"
Grimes says.
Compared to, say, astronomy, medicine "is a fairly young
science, and there are lots of steps and missteps," Grimes
notes. Medical science dates to the French Revolution in the
late 1700s, he says, and the first randomized controlled trial,
a comparison of the antibiotic streptomycin to a placebo for
treating tuberculosis, wasn't published until 1948. There
have been a million since then, Grimes says.
Doctors have always had to juggle what they learned in their
training, what they're seeing in their own patients, and what
they read in medical journals, says oncologist Paul Wallace,
senior adviser to the Kaiser Permanente Care Management Institute
in Oakland. But in the old days, Wallace notes, delving into
the literature simply meant flipping open a textbook. Today,
though, an average of 82 randomized controlled trials are
published each day in the medical literature, says Paul Keckley,
executive director of the Vanderbilt Center for Evidence-Based
Medicine in Nashville. An internist who doesn't read and remember
19 of them is behind the times, he says. And it shows in the
wide range of medical opinions patients might encounter, Keckley
says. "If you took your car in to get it fixed, and you
got five mechanics and you got five different plans, you'd
be concerned," he says. "The same thing is happening
in health care."
To help doctors keep up, several medical journals scan the
scientific literature, select the most significant randomized
controlled trials for their readers, and then summarize them.
Brian Haynes edits two such journals: Evidence-Based Medicine
and its sister publication, the ACP (American College
of Physicians) Journal Club. Evidence-Based Medicine
has a staff of seven research associates who read about
120 medical journals. About one in 50 articles makes the first
cut, says Haynes, chair of clinical epidemiology and biostatistics
at McMaster University in Hamilton, Ont. They are then sent
out to physician reviewers in the relevant specialty. Their
charge? Haynes says: "Tell us how different it is from
what they already know."
Randomized controlled trials differ from so-called observational
studies, in which patients who chose a particular treatment
are compared with those who didn't. The estrogen story illustrates
a weakness of observational studies. Study after study of
postmenopausal women who chose to take estrogen found they
had far fewer heart attacks than other women. But they also
tended to be better-educated and have healthier lifestyles.
Apparently, those differences, not estrogen, deserved credit
for their healthier hearts. Evidence-Based Medicine's
October issue illustrates the wide variety of randomized trials
covered in each issue: Telephone counseling helps smokers
quit, adding the drug salmeterol to the usual asthma medications
may increase respiratory-related deaths, and there is little
evidence that lifestyle changes can prevent heartburn. Still,
only about 50% to 60% of the main conditions leading to hospitalization
of internists' patients have been studied in well-done randomized
controlled trials, Haynes says.
Slow to adjust
Even when the evidence is clear, doctors and hospitals can
be slow to change their practices. Alfred Hallstrom, a biostatistician
at the University of Washington, oversaw the 1980s randomized
controlled trial that found two drugs used to treat a type
of irregular heartbeat actually hiked the risk of dying. When
the trial found that encainide and flecainide were harmful,
"there were a lot of people who suspected there was something
wrong with the study, rather than the drugs," Hallstrom
recalls. "The non-randomized published reports were just
glowing. They were rapidly becoming the drugs of choice."
Doctors did finally stop prescribing them because of the trial.
Says Hallstrom: "I don't think it halted as quickly as
you might expect. The prejudice about the benefit of these
drugs was pretty strong."
Sometimes, doctors are slow to adopt proven therapies. Take
blood clots, the top preventable cause of sickness and death
in hospitals. "We know from many trials how to prevent
that from happening," Haynes says, citing compression
boots and heparin, a blood-thinning medication. "The
studies in hospital settings show less than a third of patients
get these interventions. Then they get something they didn't
bargain for — a clot."
In early 2004, the record for protecting at-risk patients
against blood clots at northeast Philadelphia's Jeanes Hospital
was only a little better than a third, says Patrick McDonnell,
who spearheaded an effort to improve that statistic. McDonnell,
associate professor of clinical pharmacy at Temple University,
arranged a talk for doctors on the subject, and George Miller,
a fellow pharmacist at the 200-bed hospital, developed a one-page
form to help doctors assess whether newly admitted patients
are at risk for a blood clot. Risk factors include cancer
and heart failure as well as obesity and age over 40.
Family practitioner John Woodward, chair of Jeanes' pharmacy
and therapeutics committee, says doctors are sometimes too
busy with patients' urgent problems to consider something
that might never happen, such as a clot. "Your first
task is to get them breathing," Woodward says. For him,
the campaign to prevent blood clots is personal. Five years
ago, his mother-in-law died suddenly hours after arriving
home following gallbladder surgery. "We think she dropped
dead because of a blood clot in her lungs," Woodward
says. "She was older, she had had surgery. We didn't
prophylax her. It was one of those preventable deaths, we
think."
Late last month at Jeanes, Beth Torrito's age and the fact
that she was having major surgery for a ruptured ovarian cyst
made her high-risk for a blood clot. While hospitalized, she
received three heparin shots a day in her abdomen. Unlike
many, Torrito, 45, was familiar with blood clot prophylaxis.
When her late mother was recovering at home from surgery for
a broken hip five years ago, "I used to give her heparin
shots, so I knew exactly what they were," says Torrito,
of Elkins Park, Pa. The shots left her bruised, she says,
but she knows they were for her own good. By raising doctors'
awareness about blood clots, Jeanes has seen an increase in
the use of preventive measures and a decline in the problem,
McDonnell says. Since the program's inception, the percentage
of patients who experience blood clots has dropped from nearly
3% to below 1%, McDonnell says.
Wisdom has its place
Evidence-based medicine has some prominent critics, including
former National Institutes of Health director Bernadine Healy,
who created the Women's Health Initiative, the landmark randomized
controlled study that shot down the popular notion that postmenopausal
women could protect their hearts by taking hormones. "If
it hadn't been done then ... I think we'd have estrogen in
the drinking water right now," Healy says. And yet, she
says, "the implication is that the medicine that was
practiced up until 1990 was not evidence-based, was not science-based."
While "passionate" about randomized trials, Healy
says, "I'm just saying there are multiple dimensions
of evidence. There has to be a certain amount of wisdom, and
a certain amount of clinical understanding." McMaster
University physician Gordon Guyatt, a professor of clinical
epidemiology and biostatistics who coined the term "evidence-based
medicine" in 1990, says doctors should have a "healthy
respect" for the evidence, but should not be tyrannized
by it. Even some researchers involved in the study of drugs
for irregular heartbeats thought they might still prescribe
the drugs to older patients, Hallstrom says. For those patients,
the doctors thought, improving their quality of life by eliminating
arrhythmias might be worth increasing their risk of death.
As Haynes, the editor of Evidence-Based Medicine, says,
"evidence doesn't make decisions. Evidence gives you
some information about the possibilities of what will happen.
The people who make decisions about care are the patient and
the practitioner."
October 16, 2006 |
