Following public disclosure about the causal link of SSRIs to suicidal and violent aggressive behavior—coupled with these drugs’ failure to demonstrate a clinically significant benefit greater than placebo—SSRI prescriptions for adolescents declined by 20% in the U.S. This drop in sales has caused panic attacks not only for the drugs’ manufacturers but also for stakeholders in the psychiatric establishment. Their fear is that when the adult data is fully analyzed sales will take a dive.
Both studies suffer from a fundamental scientific flaw which precludes drawing any valid conclusions. They failed to include a control of any kind—neither placebo, nor a comparative non-drug treatment. NIMH director, Dr. Insel, acknowledged: “Since there was no placebo control group, we do not know how many of these patients would remit without active drug treatment.”
The front page headline in The Washington Post—“Depression Drugs Safe, Beneficial, Studies Say”—is worthy of a tabloid, not a national newspaper that values its credibility. The article by Rob Stein and Marc Kaufman unreservedly endorses grossly misleading, unsubstantiated claims about the tentative findings from two uncontrolled observational studies.
Most other reporters were less swept up by the hype; their news reports reflect somewhat greater caution and uncertainty. The Associated Press noted that the findings are tentative and “the use of antidepressants has proved extremely controversial within recent years as evidence that they on rare occasions worsen suicidal tendencies in children or teenagers.” Indeed, these tendencies have been scientifically proven and replicated by several independent analysts. [See AHRP website for documents and references.] AP notes that patients in the study received “close monitoring and frequent dose adjustments in the first three months — a level of care that few U.S. patients today receive.” So much for the claimed naturalistic “real world” conditions.
The report by Alex Berenson of The New York Times (below) stands out with its sober evaluation, noting the lack of control, and reminders that scientifically valid findings—from controlled tests—show that “the drugs seem to make a small number of patients extremely agitated, a reaction that can lead to violent or suicidal thoughts.” Furthermore, “clinical trials have found that the drugs work only modestly better than placebo pills for most patients.”
Indeed, scientifically valid, replicated findings in controlled studies led the FDA in 2004 to order black box label warnings about the pediatric risk, and warnings that antidepressant users of all ages need be closely monitored for signs of agitation and other symptoms that might signal suicidal behavior during the first weeks of therapy.
Berenson predicts (I think correctly): “The study is likely to increase the controversy over whether depressed teenagers should routinely be prescribed newer antidepressants.” Indeed, a critical examination of these two studies reveals essential flaws in their design and limitations of what the studies actually examined and demonstrated rather than what they report and claim.
1. Neither study was designed to compare antidepressant treatment with other treatments—so neither study can draw any valid inferences about the ill or beneficial effects of antidepressant therapy. Uncontrolled before-after comparison studies are considered “pre-scientific” and not usually published by the better science journals.
2. The “naturalistic” STAR*D efficacy study found a 47% response rate with only a 28% to 33% rate of remission. These results are consistent with placebo controlled trials of SSRIs—which show that those given a placebo do almost as well (in some cases better) than this. So, while treatment may make a difference – it doesn’t necessarily have to be drug treatment.
3. In this study, the rate of "serious" adverse events in patients taking citalopram (Celexa) is reported to be 5%: 1 in 20 persons. Overall, 1 in 40 is reported to have experienced a "serious psychiatric adverse event"—is that mania or other psychotic symptom? More than 1 in 50 persons are reported to have been hospitalized for a psychiatric condition—is that a result of taking Celexa? That’s 20,000 people for every million users. Will that information be disclosed in direct-to-consumer ads for Celexa?
4. Contrary to psychiatry’s claims, depression is NOT like diabetes or any progressively debilitating disease. Like most transient cyclical conditions such as, migraine headaches, colds, premenstrual cramps, depression remits spontaneously over time. (The success of a lot of snake oil rests on a failure to appreciate this fact.)
The fallacy of attributing causal effect to pill (or snake oil) is faulty thinking. Imagine that the study looked at incidence of migraine headache before and after treatment with an antibiotic. If we observe remission of headache after seeing an MD and getting an antibiotic, does it imply antibiotics are either effective or an appropriate remedy for headache? The people who never saw the physician or who went to a doctor but did not receive antibiotic are likely to have had similar or even better recovery rate of headache—simply due to the natural history of the condition.
5. Given the serious risks and uncertain benefit beyond the placebo effect, the question has to be asked: whether the risks of suicide, increased bleeding, birth deformities, and physical dependence, etc. make it really worth using this form of treatment widely?
6. NIMH director, Dr. Thomas Insel, makes several important acknowledgements in his accompanying editorial about STAR*D Phase I results:
a. SSRIs don’t work for 70% of patients. Only 30% of the healthiest, highly educated, currently employed, Caucasian women met the criteria for remission.
b. “Since there was no placebo control group, we do not know how many of these patients would remit without active drug, so even for the 30%, can we be certain of the value of the drug?”
c. “The gulf between research and practice [ ] has led to the unfortunate current state where too many research studies have little immediate relevance to practice, and too little practice is based on research evidence.”
7. In the safety study by Simon et al, ("Suicide Risk During Antidepressant Treatment") the authors say they compared suicide rates before and after starting therapy. They identified 65,103 patients with 82,285 “episodes of antidepressant treatment” between January 1, 1992 and June 30, 2003. Death by suicide was identified by death certificate data, and serious suicide attempt was defined by suicide attempt leading to hospitalization, identified by hospital records so indicating. By these standards some suicides will not be counted because oftentimes suicides are not identified out of consideration for families and because of insurance policy liability exemptions. The Center for Disease Control acknowledges: “Because suicide is particularly subject to inaccurate determination, the incidence of suicide may be underestimated by 10%-50%.” See: CDC. Current Trends Operational Criteria for Determining Suicide: http://www.cdc.gov/mmwr/preview/mmwrhtml/00001318.htm
Simon et al report: “In the 6 months after the index prescription of antidepressant treatment, 31 suicide deaths (40 per 100,000 treatment episodes) and 76 serious suicide attempts (93 per 100,000) were identified in the study group. The risk of suicide attempt was 314 per 100,000 in children and adolescents, compared to 78 per 100,000 in adults. The risk of death by suicide was not significantly higher in the month after starting medication than in the months following.”
However, while these findings are consistent with FDA figures—those trials included a placebo control. And those studies (in FDA’s database) demonstrated that there were significantly lower rates of complete suicide and suicidal acts on placebo compared to an SSRI active drug.
8. Simon’s own previous suicide study based on a similar Pugent Sound database of 35,000 individuals who received treatment for depression during the years, 1992, 1993, and 1994, found: 36 suicides in patients with any secondary mental health service contact which he calculated as 64/100,000 patient years. Simon reported that for patients with a primary care diagnosis of depression—those treated with antidepressants had a suicide rate of 43/100,000 patient years, while such patients who were not treated with antidepressants had a suicide rate of zero.
See: Simon GE, VonKorff M. Suicide mortality among patients treated for depression in an insured population. American J Epidemiology 1998: 147, 155-160.
9. The most obvious major flaw of the study is the failure to compare the outcome of depressed patients treated with an antidepressant and those who did NOT receive drug therapy. What was their outcome?
10. How to explain lower rates of suicidal acts in the week or weeks just after treatment has started?
a. Suicidal acts often are a trigger for seeking treatment—as other studies have shown, often just being seen by a physician is therapeutic–in part placebo effect, in part human contact. But since the studies fail to compare their findings against a control, there is no way to assess whether the suicide rates would be even lower if these people had been treated conservatively rather than with drugs.
b. The authors report that the highest rate of suicide attempts “was primarily attributable to increased risk in the 7 days before the index prescription.” However, irregularities in record keeping (which are acknowledged) raise a can of worms:
“Several limitations should be considered in interpreting these data. First, outpatient prescription records would misclassify the starting date for antidepressant treatment begun during hospitalization. In outpatient prescription records, the period of inpatient treatment (a period of low risk for suicide attempt) would be misclassified as preceding the index prescription rather than following it. The resulting bias would lead to underestimation of risk before treatment was started and overestimation of risk in the first month after starting treatment."
An expert epidemiologist (whom we consulted) had this to say about unreliable record keeping:
“Evaluating the validity of studies like this requires imagining various scenarios. This one suggests that because we do not know about inpatient care in this database, those patients who were hospitalized and started on antidepressants would not be identified as starting until they filled their first outpatient prescription. So the date started would be after the actual date started in hospital. It is a possible scenario along with many others.
The main concern is that drug records are admittedly incomplete, and even when they get a prescription we don’t know how much if any of the drugs were ingested (or when). Also, suicides and attempts are under ascertained in data like this. Both misclassification of exposure and misclassification of endpoints introduce biases that, in this setting, tend to diminish any real association. That is a much more serious concern than the one raised by the authors. The fact that the authors emphasize the other scenario indicates that they are not seriously entertaining the possibility that there could be an adverse effect that they failed to detect (they are trying to prove their point).”
11. Another confounding issue raised about this “real world” study when “free drug samples” are routinely given—is that patients may have actually begun taking an SSRI before the first indicated prescription!
12. The authors’ graphs recording suicide attempt trends over time—3 months before treatment, 6 months after—show that attempt rates are stable. When they rise suddenly, people tend to seek treatment, then the rates decline to their old levels. Since the suicide attempt rate 6 months after treatment is similar to (maybe higher than) the rate several months before treatment, it is unclear what effect antidepressants had. The risk of serious suicide attempt for adolescents remains “four times as high as in adults.”
What’s missing entirely from this report is comparative evidence of suicide attempts among patients who see a physician and are NOT prescribed pills—the authors do not even mention this. Did NO suicide attempts occur in patients Not prescribed a drug—thus replicating Simon’s 1998 study? If so, the authors’ report is biased in an effort to make the drugs look good.
13. The authors’ final sentence seems tailored to an agenda that has nothing to do with the study design or findings. The statement reveals the authors’ primary intention, which is, to roll back the FDA black-box warning on SSRI labels:
“Closer monitoring of antidepressant treatment is clearly needed, but warnings regarding suicide precipitated by antidepressants may do more to discourage effective treatment than to improve the quality of follow-up care.”
This agenda is reinforced by Stein and Kaufman of The Washington Post, who transmit the claim of psychiatry’s stakeholders that these findings have “spurred some psychiatrists to call for the [FDA] to reevaluate its warnings about the drugs, which have been blamed for a decline in their use.”
The Washington Post saw fit to provide a sounding board for spreading unscientific promotional hype such as the following pep talk by Darrel A. Regier, director of research for the American Psychiatric Association: "The take-home message from these studies is that we have treatment that is effective and that the risk from depression is far greater than the risk of treatment. These studies are very important.”
In fact, they illustrate the powerful abiding influence of the post hoc (“after this, therefore because of this”) logical fallacy in thinking. Their motivation is obvious: eliminating strong warnings from SSRI labels because disclosure of risks reduces sales.
Though the Stein and Kaufman were fully apprised about the points of criticism outlined above, they chose to ignore all of them. Bloomberg News reported that NIMH funded the studies, but “Simon, the lead investigator, got a research grant from Lilly and consulting fees from Pfizer.” The Washington Post called the studies “independent.”
This front page article in the Washington Post is an unfortunate example of stenography transmitted as journalism. Uncritical acceptance of un-validated claims by parties with vested interests runs counter to the grand tradition of critical journalism of Lincoln Steffens.
Contact: Vera Hassner Sharav
THE NEW YORK TIMES
Antidepressants Seem to Cut Suicide Risk in Teenagers and Adults, Study Says
By ALEX BERENSON January 1, 2006
Treatment with antidepressant drugs appears to reduce the risk of suicide in depressed teenagers and adults, according to a study of more than 65,000 patients published today in The American Journal of Psychiatry. The study is likely to increase the controversy over whether depressed teenagers should routinely be prescribed newer antidepressants like Paxil and Zoloft.
In October 2004, the Food and Drug Administration ordered drug companies to add strong warnings to the labels of antidepressants after clinical trials suggested that some drugs increased the risk of suicidal thinking and behavior in children and teenagers. The drugs seem to make a small number of patients extremely agitated, a reaction that can lead to violent or suicidal thoughts, psychiatrists say. Since the warning, prescriptions for antidepressants have been flat for adults, according to NDCHealth, which tracks pharmaceutical information, and they have fallen slightly for adolescents.
Now some psychiatrists say they believe that the pendulum has swung unfairly against the drugs and that depressed people are not receiving treatment that could help them. But other doctors continue to say that the drugs should be prescribed cautiously, especially because clinical trials have found that the drugs work only modestly better than placebo pills for most patients. Antidepressants are among the most widely prescribed drugs in the United States, with almost 200 million prescriptions written each year.
A second study, also published today in the psychiatry journal, found that only about 30 percent of patients taking citalopram, a widely used antidepressant sold under the brand name Celexa, recovered fully after 12 weeks of treatment. That figure is comparable to the recovery rate for a placebo, which is generally estimated at 25 percent to 30 percent.
The National Institute of Mental Health financed both studies, and the doctors who conducted them said that drug companies played no part in analyzing the data. The psychiatry journal is published by the American Psychiatric Association, which represents about 35,000 doctors. Dr. Gregory Simon, the psychiatrist who led the study examining suicide risks, said he hoped that it would allay the fears of the parents of teenagers who are considering taking an antidepressant.
"The risk of a serious suicide attempt or a suicide in people taking an antidepressant is quite low, and on average the risk goes down, not up, after people start taking those medicines," Dr. Simon said. Dr. Simon is a researcher at Group Health Cooperative, a nonprofit insurer in the Pacific Northwest that covers about 500,000 people. The study examined suicides and hospitalizations for suicide attempts in the medical records of 65,103 members of Group Health who received antidepressants from 1992 to 2003. The study found that patients were significantly more likely to attempt or commit suicide in the month before they began drug therapy than in the six months after starting it.
On one level, that finding is not surprising, because a serious suicide attempt is likely to prompt psychiatric treatment. But if the newer antidepressants posed a significant suicide risk, suicide attempts would probably rise, not fall, after treatment began, Dr. Simon said. In addition, Prozac and the other newer antidepressants, often called S.S.R.I.’s, for selective serotonin reuptake inhibitors, appeared to be associated with a faster and larger reduction in risk than older classes of antidepressants, which are no longer commonly used. Dr. Robert Freedman, the editor of The American Journal of Psychiatry, said he believed the study "had real public health implications." The study is not perfect, because it did not compare the experiences of patients who were treated with non-drug therapy or no therapy to those who took an antidepressant, Dr. Freedman said.
The F.D.A. generally views the results of randomized clinical trials as more important than those of epidemiological studies like the one conducted by Dr. Simon. In a clinical trial, researchers do examine the effects of two different treatments on two groups of patients who have carefully been selected so they are identical. That way, scientists can assume that any difference in the outcomes of the two groups is produced by the two treatments. But in Dr. Simon’s study, patients were not given another kind of treatment, so it is difficult to know whether their suicide rates fell because of the antidepressants they were given or for another reason. Nonetheless, the findings are striking, Dr. Freedman said. "This was a huge sample of real patients," he said. "It’s the first evidence that the drugs are actually changing suicide rates."
In the second study, researchers tracked 2,876 depressed people who were taking citalopram under the care of psychiatrists or primary care doctors. Based on a widely used measure of depression, about 28 percent of the patients had recovered fully by the end of the study, with 15 percent showing some improvement. Another measure showed slightly better results, with 33 percent achieving remission. Patients who were female, white or married were significantly more likely to recover than those who were black, single or male. Response rates were very similar for people being treated by psychiatrists and for those being treated by primary care physicians.
The trial did not compare people taking citalopram with those taking other antidepressants, other types of therapy or placebos. But the results were consistent with other clinical trials that show that antidepressants are in general only modestly – if at all – more effective than a placebo in treating depression. In fact, patients taking the highest doses of citalopram, 50 milligrams a day or more, were less likely to improve than those taking lower doses. The study is part of a longer trial financed by the National Institute of Mental Health that examined ways to treat depression in people who do not respond to initial drug therapy. Later phases will examine different drugs, multidrug combinations and drugs in combination with therapy, said Dr. Madhukar Trivedi, professor of psychiatry at the University of Texas Southwestern Medical Center in Dallas and the study’s lead author. "The good news is we got a third of patients to remission," he said.
Copyright 2005The New York Times Company
THE WASHINGTON POST
Depression Drugs Safe, Beneficial, Studies Say: Suicide Risk Rejected, But Critics Question Validity of Findings By Rob Stein and Marc Kaufman
Sunday, January 1, 2006; A01
Antidepressants, such as Prozac and similar drugs, help many patients overcome their often disabling psychiatric disease and do not increase the risk for suicide in adults, according to two large studies being published today that counter recent concerns about the popular medications.
The findings from two independent, federally funded studies — the first of their kind — spurred some psychiatrists to call for the Food and Drug Administration to reevaluate its warnings about the drugs, which have been blamed for a decline in their use. "The take-home message from these studies is that we have treatment that is effective and that the risk from depression is far greater than the risk of treatment," said Darrel A. Regier, director of research for the American Psychiatric Association, a group that has been critical of the warnings. "These studies are very important." xxx cut xxxx
FAIR USE NOTICE: This may contain copyrighted (© ) material the use of which has not always been specifically authorized by the copyright owner. Such material is made available for educational purposes, to advance understanding of human rights, democracy, scientific, moral, ethical, and social justice issues, etc. It is believed that this constitutes a ‘fair use’ of any such copyrighted material as provided for in Title 17 U.S.C. section 107 of the US Copyright Law. This material is distributed without profit.