Back when I was growing up in the early 1960s, there was a popular song out by Bobby Vee, called “Devil or Angel”. I believe it contained lyrics along the lines of, “Dear, whichever you are, I need you.” The title of the song might also be a good summation of the way psychotropic drugs are portrayed in the popular press and other media. And, sad to say, even some of my colleagues in the mental health profession fall into one of two armed camps, when it comes to the role of medications for mood and behavior. This dichotomy parallels the schism described in Tanya Luhrmann’s influential study of psychiatry, aptly entitled, Of Two Minds. Very roughly, Luhrmann argued that the field of psychiatry is still divided between those who see mental illness as a psychological problem amenable to psychosocial therapies; and those who see it as a problem of abnormal brain chemistry, best treated by pharmacotherapy. Despite many attempts to bridge this conceptual chasm — Dr. George Engel’s “biopsychosocial model” is one example — the schism persists to this day.

And this is truly a shame. The “Angel or Devil” dichotomy does nobody any favors, and certainly does not help patients with serious emotional disturbances. In truth, the human brain is the crucible in which all the elements of our experience and sensation are transformed into thought, feeling, and action. We can affect the function and structure of the brain directly, by altering its chemical constituents; or we can affect its function and structure indirectly, by pouring helpful words into the ear of the patient. Speech, music, poetry, art, and a myriad of other “inputs” are all transduced into neuronal connections and electrochemical processes in the brain.

This does not mean that we ought to greet our patients by asking, “How are your serotonin molecules this morning, Mrs. Jones?” Part of our shared behavior as human beings is the use of language that speaks to our felt experience, not our neurons. But this does not mean that our experience is ultimately something over and above the workings of our brains. Moreover, far from being “cosmetic” in nature, many psychotropic medications work at the most fundamental level of the gene, actually increasing the production of nerve growth factors.

These are all reasons why we should not dismiss psychotropic medications out of hand. They are neither agents of the devil, as some extremist factions argue; nor are they angels of redemption, as one might conclude from the “rainbow and butterfly” ads put out by some pharmaceutical companies. Psychotropic medications, as I tell my patients, are neither a crutch nor a magic wand; they are a bridge between feeling bad and feeling better. The patient must still walk — sometimes painfully — across that bridge. This means doing the hard work of changing thoughts, feelings, and behaviors. Medications can often aid that process, and are sometimes needed to get the patient’s work in therapy moving. For example, some patients with very severe depression are so lethargic and cognitively impaired that they can’t fully engage in psychotherapy. After three or four weeks of antidepressant treatment, many of them are able to benefit from “talk therapy”, which then may provide long-term protection against depressive relapse. Some evidence suggests that initial antidepressant treatment can help “set up” the patient for subsequent long-term psychotherapy. As a recent review by Dr. Timothy J. Petersen [1] concluded,

“…sequential use of psychotherapy after induction of remission with acute antidepressant drug therapy may confer a better long-term prognosis in terms of preventing relapse or recurrence and, for some patients, may be a viable alternative to maintenance medication therapy.”

Other evidence indicates that talk therapy and medication work synergistically — one reinforcing the other. Medications may help more with “somatic” aspects of depression, such as impaired sleep and appetite; psychotherapy, more with cognitive aspects, such as guilt or hopelessness. Evidence from brain imaging studies suggest that each intervention may work through overlapping but somewhat different mechanisms: antidepressant medication seems to work “from the bottom up”, arousing lower brain centers associated with emotion. Psychotherapy appears to work from “the top down” by changing neural patterns in higher brain centers, such as the prefrontal cortex.

Given the huge literature on psychotropic medications, I am focusing on antidepressants in this essay — a diverse group of agents that has been the focus of tremendous controversy. In recent years, for example, questions have been raised regarding both the efficacy and safety of antidepressants. There is a voluminous literature on these topics, but here is my best professional synopsis. Antidepressants seem to “show their stuff” more robustly in cases of severe depression, but this may be partly an artifact of how most studies are designed and analyzed. For example, the most recent review from Kirsch and colleagues [2] suggests that in mild-to-moderate depression, antidepressants do not work better than a sugar pill (placebo). In very severe depression, Kirsch et al found, the newer antidepressants outperform placebo, though their benefits are not as robust as in earlier studies (1960s-70s) of the “old” tricyclic antidepressants.

However, we need to put these recent findings in perspective. Numerous posts on the internet have declared, based on the Kirsch et al study, that “Antidepressants Don’t Work!” But this is not what the study showed. Rather, it lumped together results from 47 antidepressant trials and found that the active drug showed a clinically significant “separation” from placebo only in the most severe cases of depression. This is actually much better than finding that antidepressants work only for very mild depression! That said, the Kirsch study attributed the apparent benefit of antidepressants in the most severely ill patients to reduced responsiveness to placebo rather than to increased effectiveness of the drug.

There are a number of problems with the Kirsch study, many of which are nicely discussed in Dr. Grohol’s recent blog (2/26/08) on this website. For one thing, the entire Kirsch study turns on whether a 2-point improvement in a single depression rating scale (the Hamilton Rating Scale for Depression, or HAM-D) amounts to a “clinically significant” (not just statistically significant) change. That is, of course, a matter of judgment. Second, the Kirsch study looked only at antidepressant trials in the FDA data base done prior to 1999; an analysis of more recent trials might have produced different results. Third, the kind of “number crunching” that goes on in any meta-analysis (basically, a study of studies) can obscure not only individual differences, but also subgroup differences. That is, a given patient with certain depressive symptoms—or a subgroup with certain features—may do quite well on an antidepressant, but the results are “submerged” in the overall mediocre success rate in the study as a whole.

There are many other reasons why studies of antidepressants may be yielding less than spectacular results in more recent decades, and the interested reader can find details in an editorial by Kobak and colleagues, in the February 2007 Journal of Clinical Psychopharmacology. These authors point out, among other things, that if the interviews producing HAM-D depression scores are not performed skillfully, the results of the study may be distorted. Kobak and colleagues pointed to several instances in which poor interviewing technique led to outcomes showing little difference between the antidepressant and placebo; conversely, good interviewing technique led to a more robust improvement rate (“effect size”) for the antidepressant. It is not clear how many such “junk interview” studies were included in the Kirsch et al meta-analysis.

Part of the relatively weak showing of antidepressants in recent studies (compared with those done in the 1960s and 70s) may be due to the increasingly “good show” put on by the placebos. What might account for this? My colleague David Osser MD, Associate Professor of Psychiatry at Harvard Medical School, observes that placebo response rates have actually been rising in recent years, as confirmed by Dr. B. Timothy Walsh and colleagues (JAMA Vol. 287 No. 14, April 10, 2002 ). Dr. Osser thinks it likely that this “placebo inflation” is due, in part, to recruitment of less severely ill subjects for study. The less ill the subjects, the more likely a “sugar pill” is going to work for them. Dr. Osser points out (as suggested by Walsh et al) that subjects in modern studies are often recruited from ads in magazines, rather than from samples of “real” patients, who are often much sicker.

There is a larger point to be made about the kind of analysis Kirsch et al have done. Basically, it involved crunching numbers on trials in which, usually, a single antidepressant was tested over a period of a few weeks. But when psychiatrists use a “full court press” and treat depressed patients over many months, using various combination and augmentation strategies, we often see better results with medication. For example, a recent series of carefully-controlled, multi-stage studies known as STAR*D, sponsored by the National Institute of Mental Health, looked at remission rates in patients with resistant major depression. These patients had gone through several levels of intensive antidepressant treatment, without full recovery. After the fourth and final “hoop” was jumped through, the cumulative rate of remission (few or no symptoms) was about 67% [3]. The nature of the STAR-D study precluded use of a placebo group. However, the cumulative remission rate of 67% is certainly much higher than generally reported rates of remission with placebo, which average around 30%.

To be sure, non-specific interventions, such as talking to a friend, taking up a hobby, joining a club, etc. might work as well as an antidepressant for many patients with mild depressive symptoms. (Many individuals with “normal sadness”, of course, will feel better simply by waiting a few weeks). But for those with the most severe types of depression — and certainly for those with psychotic depression — medication is often required, at least in the early stages of treatment. Patients with depression due to bipolar disorder (“manic-depressive illness”) will require special treatment using a “mood stabilizer”, and may actually become agitated or manic if treated with an antidepressant. It is critically important that the patient with depression is carefully evaluated to rule out a bipolar disorder [4].

With regard to safety, there is probably a very small subgroup of depressed patients who will worsen with an antidepressant. Data from the U.S. Food & Drug Administration (FDA) suggest that a small minority of children and adolescents may develop suicidal thoughts or behaviors (“suicidality”) when treated in the short-term with an antidepressant. About 4 in 100 taking an antidepressant may develop these thoughts or behaviors, versus about 2 in 100 taking a placebo [5]. No actual suicides occurred, in the studies reviewed by the FDA.

Indeed, other lines of evidence from other countries call into question the association between antidepressants and suicidal behavior. For example, several studies from the Netherlands and other European countries suggest that as prescriptions for serotonergic antidpressants (“SSRIs”, such as Prozac and Zoloft) declined from 1998-2005, suicide rates actually rose in children and adolescents. Conversely, increased prescription of SSRIs is associated with decreased suicide rates in several European countries [6]. Moreover, results of a comprehensive review of pediatric trials conducted between 1988 and 2006 suggest that the benefits of antidepressant medications greatly outweigh their risks to children and adolescents with major depression and anxiety disorders [7]. Another study of over 226,000 depressed veterans found that SSRIs actually had a protective effect against suicide attempts, in all adult age groups [8].

In my own experience over the past 25 years, antidepressant treatment — usually in combination with talk therapy — may literally be life-saving for seriously depressed adult patients. I have also found that in many cases of “paradoxical” or adverse reactions to antidepressants, the patient actually suffers from an undiagnosed bipolar disorder. Although the use of antidepressants in bipolar disorder is controversial, I try to avoid it whenever possible.

So– “devil or angel”? Asking this of psychotropic medication is a bit like asking, “Will fire burn down my house, or will it warm it in the winter?” In this piece, I have focused almost entirely on antidepressant medication. If I were to go on at even greater length — discussing mood stabilizers, antipsychotics, and anti-anxiety agents — we would see that pharmacotherapy is neither devil nor angel. It is merely one instrument in service of helping the patient. As such, it may do good or ill, depending on the skill of the physician, the constitution of the patient, and the nature of the illness. Medication may be over-sold and “hyped”, as it is by many in the pharmaceutical industry; or it may be vilified and disparaged, as it has been by some vociferous anti-psychiatry groups in this country. In the end, as physician and educator Alfred Stille (1813-1900) observed: “It is quite as necessary for the physician to know when to abstain from the use of medicine as it is…[to know] when medication is necessary…”

* * *

Acknowledgment: I would like to thank Dr. John Grohol for inviting this piece, and Dr. Dave Osser for his helpful comments.

The author is Professor of Psychiatry and Lecturer on Bioethics & Humanities at SUNY Upstate Medical University in Syracuse, NY; and Clinical Professor of Psychiatry at Tufts University School of Medicine in Boston. He is the author of several textbooks in psychiatry, as well as the forthcoming book, Everything Has Two Handles: The Stoic’s Guide to the Art of Living. Dr. Pies reports no conflicts of interest with respect to the material in this piece.


1. Petersen TJ: Enhancing the efficacy of antidepressants with psychotherapy Journal of Psychopharmacology, Vol. 20, No. 3 suppl, 19-28 (2006)

2. Kirsch I, Deacon BJ, Huedo-Medina TB et al: Initial Severity and Antidepressant Benefits: A Meta-Analysis of Data Submitted to the Food and Drug Administration. PLoS Medicine. Accessed at:

3. Rush AJ, Trivedi MH, Wisniewski SR et al: Acute and longer-term outcomes in depressed outpatients requiring one or several treatment setps: a STAR*D report. Am J Psychiatry 2006;163:1905-17.

4. Ghaemi, SN, Miller CJ, Berv DA, Klugman J, Rosenquist KJ, Pies R: Sensitivity and Specificity of a New Bipolar Spectrum Diagnostic Scale. Journal of Affective Disorders 2005; 84:273-77.

5. FDA

6. Gibbons RD, Brown CH, Hur K et al: Early Evidence on the Effects of Regulators’ Suicidality Warnings on SSRI Prescriptions and Suicide in Children and Adolescents. Am J Psychiatry 2007; 164:1356-1363.

7. Bridge JA, Iyengar S, Salary CB et al: Clinical response and risk for reported suicidal ideation and suicide attempts in pediatric antidepressant treatment: a meta-analysis of randomized controlled trials. JAMA. 2007; 297:1683-96.

8. Gibbons RD, Brown CH, Hur K et al: Relationship between antidepressants and suicide attempts: an analysis of the Veterans Health Administration Data Sets. Am J Psychiatry 2007;164:1044-49.