On the surface this trial appeared to be just one more negative, or failed,
drug trial funded by industry. However, it found itself published in a
prestigious journal, illustrating the adage, "the devil is in the
Those details, however, are anything but common, according to Steven Roose,
M.D., a professor of clinical psychiatry at Columbia University College of
Physicians and Surgeons and a psychopharmacology researcher at the New York
State Psychiatric Institute.
The trial did not just show that the highly selective serotonin reuptake
inhibitor citalopram (Celexa) did not beat placebo in treating depression in
those over age 75; the trial raised several questions, partially answered
others, and above all, clearly showed that the beast known as a clinical drug
trial is a complex being—one that science is far from clearly defining
Roose was first author on the multisite, randomized, controlled study, on
behalf of the Old-Old Depression Study Group, which includes researchers at
Duke University, the University of Pittsburgh, Cornell Institute of Geriatric
Psychiatry at Weill Medical College, UCLA-Neuropsychiatric Institute, the
University of Pennsylvania, and the Mental Illness Research Education and
Clinical Center at the Forest Research Institute in Philadelphia. The
research, funded by Forest Laboratories, maker of Celexa, appears in the
November American Journal of Psychiatry.
"First of all, to set the stage," Roose told Psychiatric
News, "older people are the most rapidly growing segment of the
population. Second, depression is a very prevalent illness and a very
important illness in this population because it significantly impairs quality
of life, it leads to suicide, and when depression is comorbid with other
medical illnesses common to this population—like cardiovascular
disease—the other illness is often made significantly worse by the
depression. So, we really wanted and need to find safe and effective
treatments for depression in the old-old."
Citalopram seemed a logical choice. It is highly selective for the brain
serotonin reuptake transporter and therefore was thought to be associated with
fewer adverse effects than more general serotonin reuptake inhibitors. It was
also thought to have a lower likelihood of drug-to-drug interactions. Thus the
drug should be relatively safe in the geriatric population.
Forest agreed to the researchers' demand that all control over study design
and methodology would rest with the researchers. In addition, all data had to
be in the researchers' possession, and all analyses had to be completed by the
"That was something that was rather unprecedented," Roose
noted, "and it was an important precedent."
"A lot of people are going to say, `Oh look, this study shows that if
somebody comes in with depression, it doesn't make a bit of difference whether
you give them a drug or don't do anything,'" Roose said. "But we
are saying, `Wait a minute. This is far from doing nothing.'"
The study involved 174 patients at 15 clinical sites. The patients, on
average, were just under 80 years old and had a mean baseline Hamilton
Depression Rating Scale (HamD) score of 24.3. Fifty-eight percent of the
subjects were women.
Patients were randomly assigned to either citalopram (10 mg to 40 mg a day)
or placebo over the course of eight weeks. The primary outcome variable was
improvement over the eight weeks in HamD scores, and remission was defined as
an endpoint HamD score of less than 10.
For those treated with citalopram, 35 percent met remission criteria at
week eight, compared with 33 percent for those on placebo.
"These patients all came in and saw the doctor, and they all had a
medical workup, MRI scan, neuropsychiatric testing, et cetera," said
Roose. "What it all amounts to is that degree of attention apparently
has a therapeutic effect itself." But Roose cautioned, "Even the
attention doesn't work for everybody. If you look at the people with severe
depression, those people didn't do as well on placebo as they did on the
In fact, in the subgroup of patients with severe depression, those on
citalopram averaged a remission rate of 35 percent, compared with 19 percent
for those taking placebo. But that doesn't mean that the drug was more
effective for severe depression than for less-debilitating depression."
What this tells you is that placebo was less effective [for those with
severe depression]," Roose concluded. "That is a very interesting
There are many other factors that affect results in clinical drug trials,
factors Roose characterized as "psychological factors or nonphysiologic
factors." They influence the results, often in ways impossible to
imagine or believe, he noted.
"If you use a drug like citalopram in a trial where you are comparing
it with placebo, the response rate in your study is likely to be about 40
percent," Roose said. "In active comparator trials—where
everybody gets some kind of active drug, even though you don't know which
drug—the response rates average 60 percent to 65 percent," he
added. The same drug, given different methodologies, but given to similar
populations, may have a very different rate of response or remission.
But the clinical-trial field has long struggled with significant placebo
effects in depression treatment trials. Roose questioned just how informative
high placebo response rates are.
"In the clinical setting, we don't give placebos," Roose
observed. "Everybody knows they are getting active drug, so what
response rate should you expect in the clinical setting? It's probably closer
to the response rates in active-comparator trials than those in
placebo-controlled trials." But that does not discount the placebo
Interestingly, Roose noted, the researchers saw a significant effect of
site on remission rates in their study.
"Even though we have very good investigators, and they are all [at]
academic medical center sites, and we demonstrated high interrater reliability
on the Hamilton, where you got treated [in our study] made more difference
than what you got treated with. Site variability is very common, and we just
don't have [a good] understanding of this."
"There are a lot of things that impact a clinical trial that we don't
know about or necessarily understand," Roose suggested.
"If we really want to do meaningful clinical trials, we should really
do the gold-standard, placebo-controlled, randomized trials, but in a way that
sort of mimics the clinical experience." In the clinical setting, the
patient is seen, prescribed a drug, and not followed up for weeks. A clinical
trial should be done where the "attention" factor of the
placebo-effect is mitigated, he said.
"So, [our study was] a negative study then, meaning that the drug
wasn't different from placebo," Roose said. "And it was really a
study that says, `Look, clinical trials are complicated, and there are many
variables that influence outcomes of studies that we don't
However, "it would be wrong to conclude from this study that it
doesn't matter if you give a patient a drug or do nothing," Roose
emphasized. "Frankly, we'd like to believe we understand what influences
our clinical trials, but we really don't. And we have to be humble about
The article, "Antidepressant Pharmacotherapy in the Treatment
of Depression in the Very Old: A Randomized, Placebo-Controlled Trial,"
is posted online at<http://ajp.psychiatryonline.org/cgi/content/full/161/11/2050?>.▪
Am J Psychiatry20041612050