1 April 1999
For psychiatrists treating depression, the newer selective serotonin reuptake inhibitors (SSRIs) are the drug of choice, since they don't have the serious side effects of the older tricyclic antidepressants (TCAs). Now, a meta-analysis found that SSRIs and TCAs are equally effective, and that the side effects were roughly equal. But many psychiatrists weren't convinced.
The meta-analysis, "Evidence Report on Treatment of Depression--New Pharmacotherapies," was sponsored by the Agency for Health Care Policy and Research (AHCPR) and conducted by the San Antonio Evidence-Based Practice Center at the University of Texas. The report is designed to help clinicians make decisions, and help organizations develop guidelines. It defined 29 drugs, and 3 herbal remedies, as "newer" antidepressants.
"No particular class of drugs is routinely more effective than others," said lead investigator Cynthia D. Mulrow, MD, MSc, professor of medicine and geriatrics at the Center.
"They're wrong," said T. Byram Karasu, MD, Chair of the Work Group on Major Depressive Disorder of the American Psychiatric Association, and chairman of psychiatry at Albert Einstein College of Medicine, New York.
The report defined efficacy as 50% or greater improvement on the Hamilton Depression Rating Scale or equivalent. It examined studies that compared a newer antidepressant to another antidepressant, and it calculated the ratio of effectiveness, or risk ratio, for the 2 drugs. The overall risk ratio between the newer and older drugs was 1.0, with a statistically significant confidence interval of 0.97 to 1.06.
The report calculated dropout rates from adverse effects for 4 pairwise comparisons of drugs. SSRIs, for example, had a dropout rate of 11%. First generation TCAs had a dropout rate of 16%. That's a rate difference of 5%, which is actually 4% after statistical corrections, the report said.
The report's summary, and press release, were posted on the AHCPR's web site <www.ahcpr.gov> on March 18. The full report is scheduled to be posted and published later.
The summary simply said that the SSRIs and TCAs had a "rate difference" in dropouts of 4%, without supporting data.
"That is unfair statistical maneuvering," said Karasu.
The dropout rates of 11% and 16% were edited out of the summary for "space reasons," said Mulrow. "I agree it provides more information."
Independent statisticians agreed that it would have been better to leave it in. "If I were trying to convey this to the public, I would say 11% of the patients on SSRIs and 16% of the patients on TCAs dropped out," said John Allen Paulos, author of A Mathematician Reads The Newspapers, and professor of mathematics at Temple University. To report that the ratio of side effects was 16%/11%, or 1.45, "distorts in one direction," by giving a psychological misperception that the effects are large, he said, and the 4% "distorts in the other direction."
Co-author John D. Williams, Jr., MD, associate professor of medicine, University of Texas, noted that total dropout rates from all causes, including lack of efficacy, was not statistically different for newer and older drugs. So the component due to side effects must be small.
"I have to wonder to what extent this was an effort to do something about the higher cost of these new drugs," said William M. Glazer, MD, associate clinical professor of psychiatry at Harvard Medical School.
And what about psychotherapy? "Traditionally in psychiatry, we use smaller Ns for our clinical trials" than in other specialties, said David J. Kupfer, MD, chair of psychiatry, University of Pittsburgh. "The power is not as great." So these studies "may not be able to detect that kind of signal."