More About the FDA's Poor Standards for Anti-Depressant Drug Approval
Prepared by Paula J. Caplan, Ph.D., Harvard University and Emily Cohen, B.A., Yale University

The Food and Drug Administration sets Guidelines for drug companies about the research they must do to get a new drug approved[1]. The Guidelines for anti-depressant drugs, for instance, were created in 1977 with the specification that "a re-review" to keep the Guidelines current would be performed "approximately every 18-24 months." But no update has been done (FDA website), and unfortunately, the Guidelines are riddled with problems. The following are brief descriptions of 23 of the problems with the anti-depressant Guidelines, and these represent only a sample of the their deficiencies. Of course, FDA Guidelines for all drugs should be analyzed in this way.

1. The FDA leaves it up to the drug company to ensure that FDA Guidelines are being followed. The laxity of both the Guidelines and the way they are enforced has made it easy for drug companies to conceal negative results and ethical violations in drug testing.For instance, one Guideline allows drug companies to remove from their studies patients who fail to improve or become worse while on the experimental drug. No one wants patients to keep taking drugs that make them worse, but the problem is this: Many companies simply drop those people from their studies but fail to report having done so; thus, the drug appears to have helped a higher percentage of patients than it really has and to have fewer negative effects than it does, because those with the earliest adverse reactions disappear from the statistical analysis.[2]

2. "More than half of the experts hired to advise [the FDA] on the safety and effectiveness of medicine have financial relationships with the pharmaceutical companies that will be helped or hurt by their decisions... These experts are hired to advise [the FDA] which medicines should be approved for sale, what the warning labels should say, and how studies of drugs should be designed."[3] Furthermore, 54% have "a direct financial interest in the drug or topic they are asked to evaluate. These conflicts include helping a pharmaceutical company develop a medicine, then serving on an FDA advisory committee that judges the drugs... Federal law generally prohibits the FDA from using experts with financial conflicts of interest, but the FDA has waived the restriction more than 800 times since 1998."[4] The decisions that these pharmaceutical experts make affect the health of millions of Americans and billions of dollars in drug sales. The FDA nearly always follows the advice of these committees.[5]

3. The duration of drug trials is usually 4 to 6 weeks. However, drugs may take longer than 6 weeks to achieve maximum effectiveness and/or to produce noticeable negative effects. The pharmacological effects that emerge are highly dependent on the length of the trials, the dose, and the individual person. Serious problems may lie ahead if the study is over before these various effects are observed.[6] As a result: --With respect specifically to "Premenstrual Dysphoric Disorder," for which the FDA approved Prozac, if the duration of a study is only 4-6 weeks, Proza could not be adequately tested, since "PMDD" is defined as diagnosable only after two months of menstrual cycles have been charted for symptoms --Even in many people who are helped, psychotropic drugs simply stop working, even as soon as several months into treatment --It is often hard to find out whether a drug is still helping, or making one worse, because if the person stops taking it and finds that their problems increase, it is extremely difficult to figure out whether that is because the original problem is still there or whether the person is getting worse because of withdrawing too quickly from the drug. There is enormous variability in how slowly the tapering off must be done in order to avoid "drug discontinuation syndrome."[7]

4. One Guideline is that "evidence for safety and efficacy is acquired over the course of many studies carried out over considerable periods of time and at different geographical locations." That is a good Guideline, but:

--drug companies are not required to pool the data from all of their studies, and studies revealing their drug to have no effect or negative ones often fail to see the light of day.[8]

--drug companies have in-house researchers who do many of their studies, and this gives the companies control over what is published

--drug companies that award research funding to, for instance, university researchers, often have the latter sign promises to publish no results without the company's approval

5. The FDA Guidelines specify that an "especially competent individual with experience in clinical pharmacology, psychiatry, or internal medicine" is supposed to "carry out frequent and thorough evaluation of all study subjects," but that "especially competent individual" is handpicked by and often employed by the drug company. Furthermore, the actual contact between the monitors and clinical investigators is quite infrequent and inadequate.[9]

6. According to one Guideline, "Follow up data usually should be obtained whenever possible for at least one week after discontinuation of drug." However, only one week after discontinuation is by no means long enough to determine the effects of stopping the drug. And the phrase "whenever possible" has no place in guidelines for research about whether a new drug should go on the market. The Guidelines do not require long-term safety studies. They include the comment that "The duration of long-term safety studies is usually 3-6 months," but pregnancy, for instance, lasts 9 months. And the drug companies tend to lose interest in long-term studies once a drug gets FDA approval.[10]

7. Related to the previous point, there is a widespread lack of long-term safety studies. To track people over the long term is difficult in and of itself, but in addition, the pharmaceutical companies have tended to lose interest once their drugs are approved. It thus has become the responsibility of the individual consumers to report any adverse effects.

8. It often takes more than two years for the FDA to review its procedures. There are no consistent external checks of the FDA; those responsible for any review are usually associated with either the pharmaceutical companies or within the FDA. The actual FDA does not run the drug tests; they are the responsibility of the respective pharmaceutical companies.[11]

9. The drug companies grossly underreport negative results, and the FDA does not attempt to ensure otherwise. For example, the FDA approved Prozac after incompletely testing its effectiveness in treating depression and without adequate information about the potential adverse effects. Neither the FDA nor Eli Lilly knew for certain how effective the drug would be; it would not necessarily prevent more harm than it might cause. Possible links between fluoxetine and suicide and violence remained concealed. In addition, data were altered to include only those patients with a less severe form of depression, who had the ability to communicate adequately with the investigators and were not taking other medications.[12]

10. The FDA's criteria include the specification that studies may be done on "...Normal volunteers and/or psychiatric patients." The distinction between "and" and "or" can lead to crucial differences in the outcome of the study. There may be a substantial difference between the reactions of "normal" and "psychiatric" participants. It would be quite difficult to see how a drug affects certain symptoms in one who is presumably "normal," and certainly the results of drug tests on "normals" cannot be assumed to predict the effects they will have on "psychiatric patients." Furthermore, the assumption of normality may be premature, since several SSRIs are reported to trigger "manic depression" in someone who has a familial predisposition to mental illness, and that response would not be reliably predictable in some people classified as "normal." In addition, prescribing fluoxetine to people regardless of diagnosis or lack thereof risks adversely altering the serotonin levels in their brains.[13]

11. According to their guidelines, the FDA recognizes that a diagnosis of clinical depression is highly subjective, that it is "impossible to define clear, universally applicable diagnostic criteria..."[14] If there are no universal criteria, and the diagnosis is up to the individual professional's discretion, there can be no standardized way to have "each patient be characterized as to the presence and intensity of various psychopathology..." as the guidelines specify.

12. There is no standard supervision in the clinical trial process. The FDA leaves it up to the sponsoring drug company to ensure that FDA guidelines are followed. It is very likely that because of the substantial interest of the company responsible for the drug testing, some essential steps may be glossed over or distorted in order to achieve the desired results. It has been well-documented that the FDA has close financial ties to the pharmaceutical industry, and according to the Guidelines, the testing process is monitored by those "assigned by the drug company." The "especially competent individual" mentioned in the Guidelines, who is supposed to be the overseer, usually has direct ties with the drug company. On the other hand, the actual contact between the monitors and clinical investigators is quite infrequent and inadequate. There are no FDA specifications about who should do the review, there are no standards to ensure that the information is shared among the individual investigators, and the FDA usually disregards this responsibility. The repercussions if the information is not immediately shared may be great, including excessive expenses and serious negative effects of the drug.[15]

13. Extensive documentation has been produced about drug companies interacting with the FDA to conceal negative results and ethical violations, alter data, and skew information that would favor both organizations. As noted, one built-in source of bias is that the FDA Guidelines mention "...explicit provisions for removing from the study patients whose clinical condition worsens or fails to improve in a reasonable period of time." But removal from the study of those who do not respond to the drugs in the desired manner skews the data in the direction of making the drug's effects seem more positive. In this connection, it is important to note that researchers who work for or are funded by drug companies often fail to mention the dropout rate or the reasons that participants dropped out of the study, or if they do mention them, they often fail to take that information into consideration in drawing their conclusions about the drug's effects. It is of the greatest importance to see who is negatively affected, to what degree they are affected, and when and where in the study the negative effects appeared. Treatment-emergent symptoms warrant careful attention, and more weight should be given to them in research reports than is currently given, and the FDA should insist on this.[16]

14. The guidelines specify that "Evidence for safety and efficacy is acquired over the course of many studies carried out over considerable periods of time and at different geographical locations." However the FDA does not have specifications to ensure that the data are pooled together, and there is no neutral regulator.

15. Drug tests that are done outside the United States are often ignored.

16. The guidelines include the note: "In some institutionalized severely ill female patients, the clinical setting provides sufficient insurance against the risk of pregnancy..." However, mental patients are frequently unprotected in drug testing. They may not be made aware of the potential risks or given much choice in the matter.[17]

17. From the Guidelines: "Women of childbearing potential, children, and individuals with serious diseases [are to be] excluded" from the research. The problem with this is that women of childbearing potential, pregnant women, children, and individuals with serious diseases may be prescribed the drug after it has been approved, although no information would have been gathered about how the medication affects members of those groups.

18. According to the Guidelines, "When treatment emergent symptoms or side effects prohibit further increases, cut back to maximum tolerated dose and continue for at least 14 days... total duration of approximately 6 weeks is usual..." However, the FDA acknowledges that depression can require several months of treatment before a beneficial effect is seen. Furthermore, 14 days is not nearly long enough to determine that it would be safe under the conditions under which most patients would be using the drug. Longer clinical trials might provide data demonstrating greater efficacy and/or more negative effects, depending on the individual and the drug. The general point is that 6 weeks is far too brief a time for testing of a new drug or of an approved drug for a new use.

19. The Guidelines include the specification that "Follow up data usually should be obtained at least weekly during administration and whenever possible for at least one week after discontinuation of drug." But as noted, it may take longer than the duration of the study for negative effects of the drug or withdrawal effects to emerge.

20. There has been no systematic attempt on behalf of the FDA to attain adverse reports from consumers. It has been documented that, when initially approving Prozac, the FDA had limited information about its safety and efficacy, including that it might worsen symptoms of depression, and the true adverse effects of Prozac were not known outside of Eli Lilly at the time of its initial marketing. There were no conclusive data on long-term safety, nor was there a great deal of information about its effects on large populations of patients.

21. The FDA specifies that "It is desirable that a variety of socio-demographic and clinical characteristics of the patients be reported." This is too vague a direction to guarantee that the patients have been randomly selected and therefore can be generalized to the population as a whole or that they have been selected to represent accurately the population for whom the drug will be prescribed. There are no clear specifications regarding what information should be gathered. There is no guarantee that a full history is collected, and this significantly affects the data. For example, one who displays the "clinical characteristics" of depression may be having a normal reaction to the death of a loved one rather than being someone who would otherwise be will depressed. Nor are provisions made for those with extenuating circumstances, such as people who are being abused or sexually harassed, and people who are poverty stricken, who may erroneously be classified as suffering from endogenous depression.

22. The FDA recognizes that defining and diagnosing "depression" is difficult and often subjective. Indeed, the Guidelines leave too much to the judgment of whoever is choosing the research participants, saying that each participant must show "A significant number (4-5) of associated symptoms." As a result, two people could each have 4 symptoms, and the chooser could define one but not the other as depressed. What if a person with 5 symptoms appears less depressed than one with 4? And by excluding from the study people who have 3 severe symptoms, one excludes one category of person for whom the drug, if approved, is quite likely to be prescribed. For these reasons, it is difficult to know how to interpret the research results.

23. It is noteworthy that, given access to all of the empirical research that the FDA could have reviewed (and perhaps did review) before approving Prozac to treat "Premenstrual Dysphoric Disorder," the European Union's Committee for Proprietary Medicinal Products declared that research had not shown PMDD to be a real entity and accordingly forced Eli Lilly in Europe to notify health professionals that they were dropping PMDD as an indication for Prozac. The FDA had approved Prozac to treat PMDD without even considering the question of whether or not PMDD was a real entity.[18]

References and Bibliography

Barbui, C. & Hotopf, M. (2001). Forty years of antidepressant drug trials. Acta Psychiatrica Scandinavica, 104(2), 92-95.

Bennett, D.A. & Jull, A. (2003). FDA: An untapped source of unpublished trials. Lancet, 361(9367), 1402-1403.

Bertucco, M. (2001). FDA under fire. Psychology Today, 34, 10-11.

Bodenheimer, T. (2000). Uneasy alliance: Clinical investigators and the pharmaceutical industry. The New England Journal of Medicine, 342, 1539-1544.

Brimelow, P. & Spencer, L. (1993). Food and drugs and politics. Forbes, 11/23/93.

Cleophas, G.C., & Cleophas, T.J. (2003). Clinical trials in jeopardy. International Journal of Clinical Pharmacology and Therapeutics 41 (2), 51-55.

Friedman, R. A. (2003). What you do know can't hurt you. New YorkTimes, 8/12/03.

Geddes, J.R., Carney, S.M., Davies, C., Toshiaki, A.F., Kupfer, D.J., Frank, E., & Goodwin, G.M. (2003). Relapse prevention with antidepressant drug treatment in depressive disorders: A systematic review. Lancet, 361(9358), 653-661.

Gelenberg, A. J. & Hopkins, H. S. (2001). Treating bipolar disorder: Towards the new Millennium. Psychiatric Times, 18(4),

Glenmullen, J. (2000). Prozac backlash: Overcoming the dangers of Prozac, Zoloft, Paxil, and other antidepressants with safe, effective alternatives. New York: Simon & Schuster, Inc.

Grandinetti, D. (1997). Clinical research: facing down the ethical dilemmas. Medical Economics, 74(25),80.

Kirsch, I. , Scoboria, A., & Moore, T.J. (2002). Antidepressants and placebos: Secrets, revelations, and unanswered questions. Prevention and Treatment, 5, 33.

Marwick, C. (1999). Drug safety takes cooperation. Journal of the American Medical Association, 282(4), 315-316.

Moynihan, Ray. (2004). Controversial disease dropped from Prozac product information. British Medical Journal 328, 365.

Pies, R. (2003). Psychotic depression: An update. Psychiatric Times, 20(8), 24-26.

Stewart, K. A. & Neumann, P. J. (2002). FDA actions against misleading or unsubstantiated economic and quality of life promotional claims: An analysis of letters and notices of violation. Value in Health, 5(5), 390-397.

Stryer, & Bero, (1996). Characteristics of materials distributed by drug companies: An evaluation of appropriateness. Journal of General International Medicine, 11, 575-583.

Sudhop, T., & von Bergmann, K. (2003). Clinical trials, industrial sponsoring, and ethics: A never-ending story. International Journal of Clinical Pharmacology and Therapeutics 41(2), 49-50.

Wolf, S. (2002). Drug safety withdrawals: Who is responsible for notifying patients? Pharmacoepidemiology and Drug safety. 11(8), 641-642.

Wolfe, S., Sasich, L.D., Zieve, A. M., & Vladeck, D.C. (2002). Comments on docket no.02N-0209: FDA enforcement of drug advertising and enforcement laws. Washington, DC: Public Citizen Health Research Group.

Wood, Stein, & Woosley, (1998) New England Journal of Medicine, 339, pp. 1851-1854.

Zimmerman, M., Mattia, J.I., & Posternak, M. A. (2002). Are subjects in pharmacological treatment trials of depression representative of patients in routine clinical practice? American Journal of Psychiatry, 159(3), 469-473.

Zuckerman, D. (2002). Public needs FDA to improve watchdog policies on drugs. DetroitFree Press, 9/2/02.