PsychEducation.org (home)

Psychiatric Times Point/Counterpoint

Psychiatric Times no longer offers the original, actual essays. For historical purposes they seemed worth saving. Here are summaries of their views, followed by the two essays.  The pivotal studies from each author, and more recent work, are summarized and linked on another page, Antidepressant Continuation or Not: Relevant Literature. 


Point: Dr. Lori Altshuler

"...the risk of acute switching has been overestimated and more likely occurs in approximately 2% to 14% of patients with BD."

"...retrospective studies and case series have reported that as many as 25% of people treated with antidepressants over the long-term may develop a rapid-cycling course, although the high rate may reflect a selection bias in the patients studied."

"The risks of potential mania-induction and cycle-acceleration must be weighed against the need for and efficacy of antidepressant treatment."

Counterpoint: Dr. Nassir Ghaemi

"The debate is not whether antidepressants should be used in people who have already been adequately treated with mood stabilizers but remain depressed. In the case of severe bipolar depression that does not respond to mood-stabilizer monotherapy, antidepressant use certainly seems justified."

"I want to suggest that depression (whether mild, moderate or severe) can be treated in most patients with BD with mood stabilizers, alone or in combination, in the absence of antidepressants. However, in a minority of cases, antidepressants are needed along with mood stabilizers."

"...when antidepressants are used for acute bipolar depression, there is no justification for continuing them indefinitely in most patients. Again, in a minority of patients (probably about 20%), there may be a need for long-term antidepressant use."



It Is Reasonable To Try and Treat Depression in BD Primarily With Antidepressants
by Lori Altshuler, M.D.

Bipolar disorder (BD) affects approximately 1% of the population and is associated with a high morbidity and mortality (Goodwin and Jamison, 1990). Bipolar disorder is recurrent in almost all cases, and most patients will spend more time in the depressed than the manic phase of their illness over their lifetime. This is true for patients with bipolar I disorder (BD-I) as well as bipolar II disorder (BD-II) (Goodwin and Jamison, 1990; Judd et al., 2002). Suicide attempts and completed suicides are high in this population (Goodwin and Jamison, 1990).

Despite these facts, there have been very few double-blind, placebo-controlled trials comparing the efficacy of different treatments for the depressed pole of BD, and, to date, there is only one treatment for acute bipolar depression approved by the U.S. Food and Drug Administration (olanzapine/fluoxetine [Symbyax]). Thus, clinicians are often faced with the dilemma of needing to treat a phase of the illness for which there are very few empirical studies regarding the optimal treatment and approach. Yet patients with BD continue to present to the treating clinician with depressive symptoms requiring intervention. In some cases these symptoms may be mild, but nonetheless patients will state that these symptoms interfere with their social or occupational functioning, and they seek information about treatment options. In other cases symptoms are moderate-to-severe, and patients come to their clinicians desperate for help and often suicidal.

There is concern in this population that treatment with antidepressants might induce switches into mania or cyclic acceleration. While open case series suggested medication-induced switches into mania occur at high rates, four large, randomized, controlled acute treatment trials of bipolar depression have enhanced our understanding of the natural acute switch rate of the disorder versus the drug-induced switch rate (Calabrese et al., 1999; Nemeroff et al., 2001; Post et al., 2001; Silverstone, 2001). The data from these studies suggested that the risk of acute switching has been overestimated and more likely occurs in approximately 2% to 14% of patients with BD. Switch rates with continued antidepressant use have not been well studied, but preliminary data from one study suggested switch rates of about 14% over one year (Post et al., 2001). Whether this rate is due to antidepressant treatment or is part of the natural course (in which switching to the opposite pole over time is expected) cannot be definitively known, as there was no placebo group in this study.

Another concern of continued antidepressant use is the induction of cycle-acceleration. No randomized, controlled trials, to the author's knowledge, have assessed the risk for cyclic acceleration with antidepressant exposure versus no exposure over time. However, retrospective studies and case series have reported that as many as 25% of people treated with antidepressants over the long-term may develop a rapid-cycling course, although the high rate may reflect a selection bias in the patients studied.

The risks of potential mania-induction and cycle-acceleration must be weighed against the need for and efficacy of antidepressant treatment. Studies done to date have reported an approximately 50% response rate to these agents (Post et al., 2003b). Thus, even with adequate treatment, many patients do not respond to antidepressants. To date, we do not have good predictors of who will have a good antidepressant response. Whether the newer antidepressants (e.g., selective serotonin reuptake inhibitors, bupropion [Wellbutrin], venlafaxine [Effexor]) will have greater response rates than older medications (e.g., tricyclic antidepressants), and whether these newer medications will be less likely to induce switching or cycling, remains to be studied. It is promising that lamotrigine (Lamictal) appears effective in treating some patients with bipolar depression and the olanzapine-fluoxetine combination has recently been approved by the FDA for the treatment of bipolar depression. Preliminary data suggest that olanzapine (Zyprexa) and quetiapine (Seroquel) may also prove effective treatments for bipolar depression and may reduce the need for concomitant antidepressant use, but more studies are needed. To date, however, empirically driven evidence for other antidepressants is limited (Ghaemi et al., 2004).

Some investigators have taken the stance that, because of these concerns, one should use antidepressants sparingly (Ghaemi et al., 2003). Others have suggested that these fears have led clinicians to be too conservative in aggressively managing bipolar depression (Moller and Grunze, 2000). However, a survey on the current standard of care by clinicians in the community suggested that most clinicians appear to use antidepressants routinely for patients with BD (Ghaemi et al., 2000) (see Figure 1 for medications used in a sample population with BD).

Most clinicians add an antidepressant to a mood stabilizer when the mood stabilizer alone is not treating the depression. While there are alternatives to antidepressants, including use of a concomitant mood stabilizer (Young et al., 2000) or nonpharmacologic intervention such as sleep deprivation or light therapy (Szuba et al., 1994), these approaches are even less well studied than antidepressant approaches. Knowing the high rates of functional impairment and risks for suicide in patients with depression (Altshuler et al., 2002; Post et al., 2003a), it is reasonable that clinicians should attempt to treat the depression primarily with antidepressants. In patients with BD-I, antidepressants should never be used without a concomitant mood stabilizer. Whether the concomitant use of mood stabilizers is protective against blocking a switch into mania remains a subject to be further researched. While prior studies suggest a lack of complete protection by a mood stabilizer against the mania-inducing properties of antidepressants (Altshuler et al., 1995; Nemeroff et al., 2001; Post et al., 2001; Quitkin et al., 1981), four studies have suggested a protective prophylactic effect (Bottlender et al., 2001; Jann et al., 1982; Prien et al., 1973; Rouillon et al., 1992). It is possible that mood stabilizers will be variably effective at preventing a manic switch depending on the class of antidepressant agent implicated in the switch; however, this requires more study. One study that combined two mood-stabilizing agents versus adding an antidepressant to a mood stabilizer demonstrated that the combined mood-stabilizer approach is equally as effective at treating depression but is less well tolerated (Young et al., 2000). In patients with BD-II, some clinicians treat only the depressive pole and use antidepressants as monotherapy if the patient's symptoms in the hypomanic phase are not functionally impairing. The benefits of the approach of antidepressant monotherapy versus combined treatment of a mood stabilizer plus an antidepressant are not well studied (Amsterdam and Brunswick, 2003).

In summary, to date, clinicians have very little data to guide their choice regarding the most efficacious approach to the short-term acute treatment of bipolar depression. Data on acute switching suggests that TCAs may be more likely than some of the newer agents to cause a switch into mania. However, as previously mentioned, the acute switch rates may be lower than previously thought. While 50% of patients may not respond to antidepressants acutely, another 50% will, in fact, get acutely better, and the majority of these patients will not switch into mania but may remain well (Post et al., 2001).

If the patient is lucky enough to respond acutely, the question becomes whether long-term antidepressant continuation further renders them vulnerable to course instability and cycling. (Data on this subject again requires further study.) In two studies by my colleagues and I, we found that patients with BD successfully treated with an antidepressant for an acute depressive episode had a reduced risk of depressive relapse if they were continued on antidepressants for at least one year (Altshuler et al., 2003, 2001): 

[Note that these outcomes, in 84 patients total, represent only 15% of the original sample of 549; the rest did not get on an antidepressant and do well for longer than two months, which was required by the study design.  Notice how Dr. Althshuler's presentation of this slide, with which we now continue, does not emphasize this, although she alludes to it in the first sentence -- JP]

These results suggested that there were patients who could tolerate the antidepressants for two months without switching into mania and have a good antidepressant response. In this group of patients, discontinuing antidepressants in those who achieved successful remission put them at significant risk for depressive relapse. In fact, 70% of those who discontinued antidepressants within the first six months after remission (as clinical guidelines would suggest doing) relapsed during the one-year follow-up, whereas only 24% of those who remained on antidepressants throughout the first year after remission experienced a relapse (Altshuler et al., 2003). This follow-up time was not long enough to look at differential risks for cycling. However, the group who remained on antidepressants in both of our studies did not have a higher risk for switching into mania. As switching into mania with continued antidepressant use is one of the major concerns that clinicians have about long-term treatment, perhaps these studies will help revise thinking on this topic.

Four randomized studies exist that directly assess the longitudinal impact of antidepressant medication exposure on the risk for recurrence of bipolar depression in patients with BD who are treated with a mood stabilizer (Johnstone et al., 1990; Kane et al., 1982; Prien et al., 1973; Quitkin et al., 1981). While the studies concluded that a mood stabilizer, lithium (Eskalith, Lithobid), in conjunction with a TCA offered no greater protection against depressive relapse than mood stabilizers alone, these studies did not directly assess the critical question. In three of the studies (Johnstone et al., 1990; Kane et al., 1982; Prien et al., 1973), euthymic patients were followed who had not necessarily been recently depressed, and thus they were not at a point in their illness when they were most vulnerable for risk of relapse.

Our prior studies suggested that if an antidepressant is needed in addition to a mood stabilizer to successfully treat a depressive episode, then the continued combination strategy may be more effective at preventing depressive relapse than continuation of the mood stabilizer alone. If it is found that a patient is sensitive to antidepressants by showing that their course of illness has markedly changed, then the medication can clearly be withdrawn. However, if the patient demonstrates a marked improvement on a medication, continuing this medication for long-term prophylactic efficacy may be a reasonable approach.

There are many unanswered questions in the approach to the patient with bipolar depression. There are issues around acute treatment (response rate versus switch rate) and longitudinal treatment (risk for induction of mania or cycling with continued treatment, risk for relapse back into depression with discontinuation of antidepressant treatment). Clinicians must weigh the morbidity of the patient's acute illness against the risk for switching into mania or inducing cyclic acceleration. Similarly, long-term treatment must take the patient's history of illness into account. Clearly, an attempt to "do no harm" is important, and alternatives to pharmacologic treatment can be considered (psychotherapy, light therapy, sleep deprivation), especially in those individuals with a history of prior antidepressant-induced sensitivities. These considerations notwithstanding, theoretical concern of a risk that may have been inflated (e.g., recent rates for antidepressant-induced switching in outpatient settings are estimated at 2% to 14%) must not interfere with taking appropriate action at treating the depressive pole of BD, which is associated with high rates of morbidity and mortality. Once a person is successfully treated with an antidepressant for an acute depressive episode, the risk for relapse may be increased by the common clinical practice of discontinuing the antidepressant soon after remission. Guidelines more similar to those for maintenance treatment for unipolar depression may ultimately be adopted for some patients with bipolar depression who respond well to an acute antidepressant treatment. However, double-blind, prospective studies are clearly indicated before a definitive answer to this question can be given. At this point in time, withholding a potential treatment because of a theoretical fear and an attempt therefore to "do no harm" may actually be extremely harmful to an individual patient. The risks of switching into mania should be explained to all patients treated for bipolar depression, and careful observation for this change in mood state as a function of treatment should occur over the first eight weeks when treating patients with bipolar depression.

References

Altshuler LL, Gitlin MJ, Mintz J et al. (2002), Subsyndromal depression is associated with functional impairment in patients with bipolar disorder. J Clin Psychiatry 63(9):807-811.

Altshuler L, Kiriakos L, Calcagno J et al. (2001), The impact of antidepressant discontinuation versus antidepressant continuation on 1-year risk for relapse of bipolar depression: a retrospective chart review. J Clin Psychiatry 62(8):612-616.

Altshuler LL, Post RM, Leverich GS et al. (1995), Antidepressant-induced mania and cycle acceleration: a controversy revisited. Am J Psychiatry 152(8):1130-1138 [see comment].

Altshuler L, Suppes T, Black D et al. (2003), Impact of antidepressant discontinuation after acute bipolar depression remission on rates of depressive relapse at 1-year follow-up. Am J Psychiatry 160(7):1252-1262.

Amsterdam JD, Brunswick DJ (2003), Antidepressant monotherapy for bipolar type II major depression. Bipolar Disord 5(6):388-395.

Bottlender R, Rudolf D, Strauss A, Moller HJ (2001), Mood-stabilisers reduce the risk of developing antidepressant-induced maniform states in acute treatment of bipolar I depressed patients. J Affect Disord 63(1-3):79-83.

Calabrese JR, Bowden CL, Sachs GS et al. (1999), A double-blind placebo-controlled study of lamotrigine monotherapy in outpatients with bipolar I depression. Lamictal 602 Study Group. J Clin Psychiatry 60(2):79-88.

Ghaemi SN, Boiman EE, Goodwin FK (2000), Diagnosing bipolar disorder and the effect of antidepressants: a naturalistic study. J Clin Psychiatry 61(10):804-808; quiz 809.

Ghaemi SN, Hsu DJ, Soldani F, Goodwin FK (2003), Antidepressants in bipolar disorder: the case for caution. Bipolar Disord 5(6):421-433.

Ghaemi SN, Rosenquist KJ, Ko JY et al. (2004), Antidepressant treatment in bipolar versus unipolar depression. Am J Psychiatry 161(1):163-165.

Goodwin FK, Jamison KR, eds. (1990), Manic-Depressive Illness. New York: Oxford University Press.

Jann MW, Bitar AH, Rao A (1982), Lithium prophylaxis of tricyclic-antidepressant-induced mania in bipolar patients. Am J Psychiatry 139(5):683-684.

Johnstone EC, Owens DG, Lambert MT et al. (1990), Combination tricyclic antidepressant and lithium maintenance medication in unipolar and bipolar depressed patients. J Affect Disord 20(4):225-233.

Judd LL, Akiskal HS, Schettler PJ et al. (2002), The long-term natural history of the weekly symptomatic status of bipolar I disorder. Arch Gen Psychiatry 59(6):530-537.

Kane JM, Quitkin FM, Rifkin A et al. (1982), Lithium carbonate and imipramine in the prophylaxis of unipolar and bipolar II illness: a prospective, placebo-controlled comparison. Arch Gen Psychiatry 39(9):1065-1069.

Moller HJ, Grunze H (2000), Have some guidelines for the treatment of acute bipolar depression gone too far in the restriction of antidepressants? Eur Arch Psychiatry Clin Neurosci 250(2):57-68.

Nemeroff CB, Evans DL, Gyulai L et al. (2001), Double-blind, placebo-controlled comparison of imipramine and paroxetine in the treatment of bipolar depression. Am J Psychiatry 158(6):906-912.

Post RM, Altshuler LL, Frye MA et al. (2001), Rate of switch in bipolar patients prospectively treated with second-generation antidepressants as augmentation to mood stabilizers. Bipolar Disord 3(5):259-265.

Post RM, Denicoff KD, Leverich GS et al. (2003a), Morbidity in 258 bipolar outpatients followed for 1 year with daily prospective ratings on the NIMH life chart method. J Clin Psychiatry 64(6):680-690; quiz 738-739.

Post RM, Leverich GS, Nolen WA et al. (2003b), A re-evaluation of the role of antidepressants in the treatment of bipolar depression: data from the Stanley Foundation Bipolar Network. Bipolar Disord 5(6):396-406.

Prien RF, Klett CJ, Caffey EM Jr (1973), Lithium carbonate and imipramine in prevention of affective episodes. A comparison of recurrent affective illness. Arch Gen Psychiatry 29(3):420-425.

Quitkin FM, Kane J, Rifkin A et al. (1981), Prophylactic lithium carbonate with and without imipramine for bipolar 1 patients. A double-blind study. Arch Gen Psychiatry 38(8):902-907.

Rouillon F, Lejoyeux M, Filteau MJ (1992), Unwanted effects of long term treatment. In: Long-Term Treatment of Depression (Perspectives in Psychiatry, vol. 3), Montgomery SA, Rouillon F, eds. New York: John Wiley & Sons, pp81-111.

Silverstone T (2001), Moclobemide vs. imipramine in bipolar depression: a multicentre double-blind clinical trial. Acta Psychiatr Scand 104(2):104-109.

Szuba MP, Baxter LR Jr, Altshuler LL et al. (1994), Lithium sustains the acute antidepressant effects of sleep deprivation: preliminary findings from a controlled study. Psychiatry Res 51(3):283-295.

Young LT, Joffe RT, Robb JC et al. (2000), Double-blind comparison of addition of a second mood stabilizer versus an antidepressant to an initial mood stabilizer for treatment of patients with bipolar depression. Am J Psychiatry 157(1):124-126 [see comment].

---


Most Patients With BD Do Not Need, or Would Not Benefit From, Antidepressants
by S. Nassir Ghaemi, M.D.

Voltaire is reputed to have held his contemporary medical colleagues in low regard, saying: "Doctors pour drugs of which they know little, to cure diseases of which they know less, into human beings of whom they know nothing." There is, no doubt, a herd mentality, codified in the "standard-of-care" legal criterion, that physicians share with all of mankind. Progress in medicine depends, however, on the ability to critically examine one's assumptions and a willingness to apply standards of evidence that share at least some aspects of scientific method.

Nowhere more, perhaps, is this critical approach needed than in the controversial clinical problem of antidepressant use in patients with bipolar disorder (BD). While I recognize that others can have different ways of attending to this issue, it seems to me that there are three possible positions: Either everyone should get antidepressants, or no one, or something in between. Most clinicians know that the two extremes are likely to be wrong. Some patients need antidepressants, and some patients never need them. The debate then is about the in-between position, and here it is a matter of emphasis. Should most patients with BD receive antidepressants, or should most patients not receive them? These are the terms of the debate, I think; and my position is that most patients with BD do not need, or would not benefit from, antidepressants. Much of the data behind the statements in this article are referenced and provided in further detail in an article of mine (Ghaemi et al., 2003b).

Lest the terms of this debate be confused, let me be clear on what I am not choosing as the core issue here. The debate is not whether antidepressants should be used in people who have already been adequately treated with mood stabilizers but remain depressed. In the case of severe bipolar depression that does not respond to mood-stabilizer monotherapy, antidepressant use certainly seems justified. In fact, that is one of the conclusions that I wish to reach in this article. I want to suggest that depression (whether mild, moderate or severe) can be treated in most patients with BD with mood stabilizers, alone or in combination, in the absence of antidepressants. However, in a minority of cases, antidepressants are needed along with mood stabilizers. I further will argue, however, that when antidepressants are used for acute bipolar depression, there is no justification for continuing them indefinitely in most patients. Again, in a minority of patients (probably about 20%), there may be a need for long-term antidepressant use. But in the majority of patients with BD, such long-term use is not needed and is not supported by the best-available evidence.

The Evidence

The rationale behind what I call the case for caution is both theoretical and empirical. One needs to begin with a basic philosophy of medical practice. My philosophy is Hippocratic (Ghaemi, 2002). On this perspective, nature heals, and the physician is a handmaiden to nature. The anti-Hippocratic approach (which is the basic philosophy that underlies medical interventionism, whether it be surgery or aggressive pharmacology) sees nature as the enemy; the clinician has to artificially cure the patient: cut the tumor out. There are times when the anti-Hippocratic approach may be justified. Usually, however, the more conservative method leads to more benefits and less harm overall in medical practice. This is the theory behind the famous "first do no harm" maxim. In the case of BD, we know that antidepressants can, at the very least, cause mania. The frequency of this event varies from 5% to 50% depending on the drug and study, but a mean range of 10% to 20% is probably not unreasonable, even with new antidepressants (Ghaemi et al., 2003b). That is the short-term risk. More importantly and usually ignored, is that there is a serious long-term risk. About 20% of people treated with antidepressants long-term appear to develop a rapid-cycling course; that is, they develop more mood episodes over time and a worsened course compared to those not treated with antidepressants. This has been demonstrated in at least one double-blind, randomized study, and it has been suggested by numerous other observational studies with few counterexamples in other observational studies (Ghaemi et al., 2003b). Let me emphasize, because this issue is frequently misunderstood even by experienced researchers, this evidence of worsening of BD in 20% of patients with antidepressants is not based solely on anecdotal, retrospective, observational evidence. It is based on double-blind, randomized, placebo-controlled evidence (Ghaemi, 2002). Furthermore, there are no randomized data that contradict this evidence.

Hence, given that there are risks with antidepressants in treating BD, the Hippocratic approach would demand notable evidence of benefit to outweigh those risks. That evidence does not exist. In short-term studies there are some randomized data suggesting benefit with antidepressants for treating active depression (Ghaemi et al., 2003b). But in long-term studies repeated attempts to demonstrate prevention of depressive episodes with antidepressants have failed (Ghaemi et al., 2001). Most of those long-term studies are with older agents (tricyclic antidepressants), and many assume that newer antidepressants (like selective serotonin reuptake inhibitors) do not possess those risks. This is a major assumption in search of some factual support. Observational evidence is quite limited, and where it exists, such as in our recent report, it appears that SSRIs may not be much different from TCAs in terms of long-term risk of inducing a rapid-cycling course (Ghaemi et al., 2004). Figure 3 illustrates the outcomes found in this report.

Figure 3 (from Ghaemi et al, Am J Psychiatry. 2004 Jan)

Until recently, there were no randomized prophylaxis studies with new antidepressants that have been presented or published. We recently presented the first interim analysis of an ongoing randomized study with modern antidepressants (SSRIs and bupropion [Wellbutrin], mainly) in bipolar depression. Our initial results confirm the previous data with TCAs: There appeared to be no preventive benefit by using antidepressants in combination with mood stabilizers as opposed to mood stabilizers alone in one year follow-up of 35 patients with BD (Ghaemi et al., 2003a).

In summary, my basic conclusion is that there are appreciable risks with antidepressants in BD that are not outweighed by notable evidence of benefit. Hence, the conservative approach of the Hippocratic philosophy would apply, and we would generally avoid using antidepressants except in those specific cases where the risk/benefit analysis for a particular patient suggested otherwise. This contrasts with the current standard of care, where most clinicians appear to give antidepressants routinely to most patients with BD (78% of patients had received antidepressants in one survey we conducted [Ghaemi et al., 2000]) and only avoid antidepressants in those cases where severe problems have resulted. Unfortunately, the long-term worsening of the illness can be difficult to clearly relate to antidepressant use, and thus many patients can be treated for years with antidepressants without benefit before reaching a clinician who might consider simply stopping the antidepressants in favor of more mood-stabilizer use.

Antidepressant Discontinuation

The evidence of possible acute benefit with antidepressants for acute depression would perhaps support their short-term use in individuals with current depression, with the antidepressants being tapered off after recovery from the acute major depressive episode (i.e., in two to six months). This idea of antidepressant discontinuation after short-term use has come under criticism as a result of some observational data from the Stanley Foundation Bipolar Network (SFBN), which found that antidepressant discontinuation led to increased risk of depressive relapse (Figure 2) (Altshuler et al., 2003). These data have widely been accepted at face value, and interpreted as perhaps due to the presumed greater benefit of modern antidepressants (as opposed to TCAs) in BD. However, those data were observational, and perhaps the major problem of observational data is that they are not necessarily fair assessments of a question. The main issue is whether other factors, called confounding variables, might not account for the results (Ghaemi and Soldani, 2003; Rothman, 2002).

For instance, in a study of whether or not a potential toxin was a carcinogen, one would need to know how many people smoked in the sample being studied. If smokers predominated in one group versus another, the finding of increased cancer in that group could well be due to the smoking, rather than the putative toxin. Thus, smoking would be a confounding factor. In the SFBN study, no assessment, control or statistical adjustments were made for potential confounding variables such as rapid cycling, gender, age or bipolar subtype. Thus, the results cannot be taken at face value. The reason why randomized studies are important is because they can remove the problem of confounding bias. In an ongoing randomized study of the same topic, in which all patients received new-generation antidepressants, we have found so far that antidepressant discontinuation does not lead to an increased risk of depressive relapse (Ghaemi et al., 2003a). An older double-blind, randomized study with TCAs also found no such risk (Prien et al., 1984). It is also noteworthy that there appears to be a high risk (nearly 60%) of tolerance with antidepressants in BD, which is about three times the frequency seen in unipolar depression (Ghaemi and Goodwin, 2001). In other words, even in those who have a short-term benefit from antidepressants, long-term continuation appears to lead to eventual depressive relapse.

Hence, perhaps short-term antidepressant use can be justified in patients who are severely depressed, but as a rule, there still is not enough evidence to justify long-term antidepressant use in patients with BD on a routine basis.

Frequency of Antidepressant Use

The question of how frequently one should use antidepressants would seem to need to be resolved in the direction of less frequent use. The status quo is that the vast majority of patients with BD (around 80%) receive antidepressants and remain on them for the long run. In academic centers that specialize in BD, about 50% of patients appear to receive antidepressants. In practices that are conservative about this issue, only about 20% of patients with BD receive antidepressants (Ghaemi and Goodwin, 2001). We have published our experience of the long-term benefit of treating depressive symptoms in patients with BD with mood stabilizers plus low antidepressant usage (Ghaemi and Goodwin, 2001). We have demonstrated that experience in the inpatient setting as well (Ghaemi et al., 1999).

The Antidepressant Era

Still, such a change in common practice will be difficult to implement on a wide scale, because both clinicians and patients are not easily convinced. With regard to patient preferences, this issue is no different than the imperative for primary care clinicians to avoid handing out antibiotics to all patients with the sniffles. Similarly, not all depressed patients seeking antidepressants should receive them.

In the last decade, which Healy (1999) has termed the antidepressant era, clinicians and the public have been attracted to antidepressants for a variety of reasons. On the one hand, the new generation of antidepressants certainly seems to have benefited the large population of people with unipolar depression, and indeed, it is likely that those patients were underdiagnosed and undertreated in the past. But after more than a decade of heavy marketing of antidepressants, with numerous continuing medical education programs, clinicians have become well educated about the signs and symptoms of depression and the pharmacology of antidepressants. However, nowhere near a similar amount of time or money has been spent on educating clinicians about BD and mood stabilizers. In fact, the most widely used mood stabilizer in the world, lithium (Eskalith, Lithobid), has been generic for years, and thus private industry has had no incentive to promote it. Hence, clinicians are not as adept in diagnosing BD as they are in diagnosing depression. We have demonstrated that about 40% of patients with BD are initially misdiagnosed as having unipolar depression and that about a decade elapses before the BD diagnosis is eventually made (Ghaemi et al., 2000, 1999). A recent increase in pharmaceutical marketing of agents for BD has tended not to focus on the complex issue of antidepressant use. Thus, despite some recent increase in use of other agents such as atypical antipsychotics, most patients with BD are still treated, usually unfruitfully, with numerous antidepressants.

Clinicians are also inclined to use antidepressants, even when they make the diagnosis of BD, because of the presumed reduced side-effect burden of those agents compared to mood stabilizers. In so doing, clinicians are confusing clinical and pharmacological safety. Antidepressants are pharmacologically safer than mood stabilizers, but they are clinically less safe in treating BD. Pharmacologically, they have fewer nuisance side effects, less medical risks and are not fatal in overdose. However, clinically, they can cause acute mania or worsen the long-term course of the illness, whereas mood stabilizers either work or do not, but they do not usually worsen BD. Pharmacological safety is irrelevant if a drug worsens the illness being treated. Another factor here is that clinicians who prefer antidepressants because they have fewer side effects than mood stabilizers are reversing the risk/benefit process. They are breaking what I call "Holmes' rule," because historically, the issue is best described by the 19th-century physician Oliver Wendell Holmes (Ghaemi, 2003), who argued that all medications are potentially toxic, and thus none should be used unless there is some evidence supporting their benefit. In applying the Hippocratic philosophy to medicine, "first do no harm" does not mean that we should always start with the medications with the fewest side effects. If this were the case, then we should routinely be using placebo in clinical practice, or perhaps everyone should take vitamin C or manganese for everything. This is the homeopathic approach to medicine. The problem is that none of these agents are likely effective. The risk/benefit calculation starts on the benefit side of the ledger. One should begin with agents for which some rationale for efficacy exists (ideally, real scientific evidence) and only then does one choose among the effective agents by beginning with the ones with the least side effects. Hence, the lower side-effect burden of antidepressants should not lead to their use in treating BD instead of mood stabilizers, because the evidence for efficacy with antidepressants is much more limited than that for mood stabilizers.

This latter point is sometimes questioned. In fact, the evidence for short-term, acute antidepressive efficacy with mood stabilizers is limited, but not much different in extent than that with antidepressants (Ghaemi et al., 2003b). The largest and best-designed study of modern antidepressants failed to find any greater efficacy with the SSRI paroxetine (Paxil) compared to therapeutic lithium levels for acute bipolar depression (Nemeroff et al., 2001). Hence, in terms of short-term data, the evidence for efficacy with antidepressants and mood stabilizers is similar. In terms of long-term data, the mood stabilizers are clearly much more effective. Numerous studies now demonstrate--both with lithium and lamotrigine (Lamictal)--evidence of long-term benefit in prevention of both manic and depressive episodes, with perhaps somewhat more such benefit with lamotrigine. Further, some analyses of randomized data, albeit exploratory, suggest such benefit with the use of divalproex (Depakote) and carbamazepine (Epital, Tegretol) (Ghaemi et al., 2003b). There is much less evidence of benefit, even along such exploratory lines, available with antidepressants.

Undertreatment of Depression

It is sometimes assumed that excessive caution in using antidepressants to treat bipolar depression will lead to undertreatment of depression and increased risk of suicide. This is certainly an important clinical point, and I do not want to leave an impression that we should not be aggressively trying to treat depressive symptoms in people with BD. My main point is that, in the long run, aggressive antidepressant use does not appear to be a solution to this problem. Again, the evidence strongly points to less depressive burden in the long term with effective mood stabilizers (like lithium and lamotrigine), while the evidence with antidepressants suggests lack of efficacy. Further, the single psychotropic agent that rather clearly has demonstrated ability to reduce the occurrence of completed suicide, and to increase life expectancy, is lithium. Modern antidepressants have not been shown to demonstrate that benefit, and, perhaps, may even be associated with suicide in some cases (Ghaemi et al., 2003b). I realize that this last point is controversial, and I cannot do justice to the topic here, but it is important to realize that we do not yet have convincing evidence that antidepressants, especially SSRIs, decrease completed suicide rates in unipolar or bipolar depression. The U.S. Food and Drug Administration database suggests, if anything, a slightly increased rate of suicidality with SSRIs compared to placebo in the randomized clinical trials of unipolar depression (Khan et al., 2000). While other factors may be relevant, it at least seems clear that antidepressants have not been proved to clearly reduce suicide in bipolar depression, and, in fact, if suicide risk in people with BD is a major concern, the most evidence-based approach is to emphasize lithium treatment.

Conclusion

In an old tale, a medical student on graduation day looked at her friend and said, "Now we will never be wrong. It will always be a difference of opinion." Hopefully, we can progress beyond mere opinion. To do so, we need to highlight and examine our underlying assumptions and values about medical practice. We also need to agree upon some basic principles of scientific evidence. Only then might we hope to withstand Voltaire's judgment.

References

Altshuler L, Suppes T, Black D et al. (2003), Impact of antidepressant discontinuation after acute bipolar depression remission on rates of depressive relapse at 1-year follow-up. Am J Psychiatry 160(7):1252-1262.

Ghaemi SN, ed. (2002), Polypharmacy in Psychiatry. New York: Marcel Dekker.

Ghaemi SN (2003), Mood Disorders: A Practical Guide. Philadelphia: Lippincott Williams & Wilkins.

Ghaemi SN, Boiman EE, Goodwin FK (2000), Diagnosing bipolar disorder and the effect of antidepressants: a naturalistic study. J Clin Psychiatry 61(10):804-808.

Ghaemi SN, El-Mallakh RS, Baldassano CF et al. (2003a), Antidepressant treatment in bipolar depression: long-term outcome. No. 113. Presented at the 42nd Annual Meeting of the American College of Neuropsychopharmacology. San Juan, Puerto Rico; Dec. 9.

Ghaemi SN, Goodwin FK (2001), Long-term naturalistic treatment of depressive symptoms in bipolar illness with divalproex vs. lithium in the setting of minimal antidepressant use. J Affect Disord 65(3):281-287.

Ghaemi SN, Hsu DJ, Soldani F, Goodwin FK (2003b), Antidepressants in bipolar disorder: the case for caution. Bipolar Disord 5(6):421-433.

Ghaemi SN, Lenox MS, Baldessarini RJ (2001), Effectiveness and safety of long-term antidepressant treatment in bipolar disorder. J Clin Psychiatry 62(7):565-569.

Ghaemi SN, Rosenquist KJ, Ko JY et al. (2004), Antidepressant treatment in bipolar versus unipolar depression. Am J Psychiatry 161(1):163-165.

Ghaemi SN, Sachs GS, Chiou AM et al. (1999), Is bipolar disorder still underdiagnosed? Are antidepressants overutilized? J Affect Disord 52(1-3):135-144.

Ghaemi SN, Soldani F (2003), Meta-analysis of observational studies: the case of rapid-cycling bipolar disorder. Acta Psychiatr Scand 108(1):1-3.

Healy D (1999), The Antidepressant Era. Cambridge, Mass.: Harvard University Press.

Khan A, Warner HA, Brown WA (2000), Symptom reduction and suicide risk in patients treated with placebo in anti-depressant clinical trials: an analysis of the Food and Drug Administration database. Arch Gen Psychiatry 57(4):311-317 [see comments].

Nemeroff CB, Evans DL, Gyulai L et al. (2001), Double-blind, placebo-controlled comparison of imipramine and paroxetine in the treatment of bipolar depression. Am J Psychiatry 158(6):906-912.

Prien RF, Kupfer DJ, Mansky PA et al. (1984), Drug therapy in the prevention of recurrences in unipolar and bipolar affective disorders. Report of the NIMH Collaborative Study Group comparing lithium carbonate, imipramine, and a lithium carbonate-imipramine combination. Arch Gen Psychiatry 41(11):1096-1104.

Rothman KJ (2002), Epidemiology: An Introduction. New York: Oxford University Press.

Figures

Figure 1

 

Figure 3

   (from Ghaemi et al, Am J Psychiatry. 2004 Jan)