It [medicine] always is directly hurtful; it may sometimes be indirectly beneficial.
Oliver Wendell Holmes Sr
Currents and Counter-Currents in Medical Science
The person who takes medicine must recover twice, once from the disease and once from the medicine.
Attributed to William Osler, MD
Both Holmes and Osler speak truths that apply to patients in whom ADs have been newly prescribed. Not only can patients with depression develop medication-induced adverse effects long before mood improvement can be expected, but they can also experience the treatment as being as noxious as the illness being treated. The proof of these statements' validity is that as many as 70% of primary care patients fail to adhere to either short- or long-term AD treatment, citing adverse effects as the main reason for discontinuing use of their medication.18
Choosing among the many ADs currently on the US market () presents a challenge to primary care physicians, who write more than 75% of the prescriptions for these drugs.22
That choice is complicated by clinical trial data whose applicability to real-life practice is brought into question by unrepresentative patient populations, the fact that all available medications are predicated on an understanding of the underlying mechanism in depression that is outdated, and the impact of vigorous marketing claims that are not borne out by the research.
Pharmaceutical studies that guide physicians' treatment decisions are frequently based on individuals not typical of patients in most practices. Zimmerman et al23
wanted to know what proportion of depressed patients in their Rhode Island outpatient psychiatric practice would be eligible for such studies. They distilled inclusion and exclusion criteria from 31 studies that appeared in 5 major psychiatric journals in the mid-1990s. Ruled ineligible were patients with bipolar depression, recent substance abuse, suicidal ideation, and comorbid Axis I disorders, including anxiety, borderline personality disorder, and depression that had lasted either too long or not long enough. Only 41 (12%) of 346 patients would have made the cut. Prescribers should assume that criterion standard randomized controlled trials are at best general guides for how to medicate actual patients, most of whom will have 1 or more inconvenient comorbid conditions rendering them ineligible for most research trials.
Antidepressant treatment strategies are based nearly exclusively on the half-century-old monoamine hypothesis, which posits that depressive symptoms are driven by noradrenergic, serotonergic, and dopaminergic system abnormalities that will be corrected with drugs that raise brain monoamine levels.24
All ADs in current use are thought to work through monoaminergic manipulation, even as the mechanisms by which this occurs remain incompletely defined.25
Depression is increasingly understood to derive not only from perturbations in monoamine circuits—a so-called chemical imbalance in the brain—but also from “the cumulative impact of genetics, adverse events in childhood and ongoing or recent stress” that affect brain-derived neurotrophic factor levels, hypothalamic-pituitary-adrenal axis function, and glutamate-mediated toxicity in multiple brain regions.25
With only half of depressed patients responding to monoaminergic ADs, Rapaport24
cites evidence for multiple pathophysiologic processes that may contribute to what he calls the complex heterogeneous syndrome of depression
that may not respond to available ADs. “Our most recent consensus suggests that the depressive disorders reflect the interplay between these biological systems and psychosocial experiences, such as early-life trauma and current life stressors,” Rapaport writes.24
Although all AD classes demonstrate comparable efficacy in treating depression,26-28
second-generation ADs have become first-line treatment as a result of vigorous marketing claims that they promote better adherence and are safer in overdose than the cardiotoxic first-generation tricyclic antidepressants (TCAs).29
Although patients have reported tolerating second-generation ADs—so-called modern or atypical ADs—better than first-generation drugs,30
assertions of improved tolerance belie patient actions. An 84-study meta-analysis by Trindade et al31
showed no difference in dropout rates between those taking TCAs and selective serotonin reuptake inhibitors (SSRIs), although the same study found distinct differences between the 2 drug classes in terms of adverse effects that patients considered unacceptable. Although the 2 classes did not differ in the prevalence of headache, tremor, urinary disturbance, or hypotension, TCAs were substantially more likely to induce such anticholinergic symptoms as dry mouth, constipation, dizziness, sweating, and blurred vision, and SSRIs were associated with substantially more nausea, anorexia, diarrhea, insomnia, nervousness, agitation, and anxiety.31
Knowing that all ADs have similar efficacy, the generalist faced with choosing an appropriate AD would do well to base rational treatment decisions on differences in side effects among the available ADs (for prescribing principles, see ). Aggressive side-effect management can improve adherence, enhance comfort and function, and obviate premature discontinuation.40
Side effects can even be harnessed for their therapeutic value, as this review illustrates.
Prescribing Principles for Antidepressant Therapy
Working with side effects means seeking to minimize the noxious and maximize the beneficial while also targeting specific depressive symptoms. In a sense, with divergent side-effect profiles for first- and second-generation drugs, the patient and physician can “pick their poison,” choosing between the potential anticholinergic and antihistaminic effects prominent with TCAs and the serotonergic and noradrenergic effects with SSRIs and such atypical ADs as mirtazapine, venlafaxine, and duloxetine. Drugs within a particular class may also have divergent tolerability. Paroxetine, for example, causes more sedation, constipation, sexual dysfunction, and weight gain than other SSRIs,41
whereas escitalopram and sertraline appear to be the best tolerated in terms of efficacy and acceptability.42
One class of drugs may even be safer than another. For suicidal or impulsive patients, SSRIs are a better choice than TCAs with their cardiac toxicity in overdose.
Side effects are by no means always problematic. For example, a TCA or mirtazapine can provide beneficial sedation to patients in whom insomnia figures prominently, whereas bupropion may energize a patient who is experiencing lassitude. Moreover, any drug may potentially prove a double-edged sword, with consequences intolerable
for one patient, being salutary for another. An example is sertraline, an SSRI, which, like other drugs in its class, can induce a plethora of sexual difficulties when prescribed for depression but can also delay male orgasm in men with premature ejaculation.43
One man's curse becomes another's boon.
Side Effects of ADs.
Among the pharmacological characteristics of the ideal AD, Richelson44
included “no side effects” and “rapid onset of action.” Unfortunately, no available medication meets either criterion. Insult compounds injury when adverse effects dominate the clinical picture well before the AD can be expected to exert therapeutic benefit.
Treatment with ADs occurs in 3 phases: initiation, continuation, and maintenance. During initiation, the goal of triggering remission is achieved through gradual up-titration of the AD dose. Although achievement of remission is far from universal, when it does occur, the AD should be taken throughout this “continuation phase,” defined as a period that is the typical length of an untreated episode. For a first episode, treatment continues for 6 to 12 months. After a second episode, a year or more of treatment is recommended. With 3 episodes or more, lifelong maintenance treatment is recommended in hopes of limiting the intensity and frequency of future episodes or forestalling them altogether.6
Paralleling the 3 stages of AD treatment, side effects occur in 3 overlapping but roughly sequential phases, each requiring active management to maximize chances that patients will continue taking medications long enough to reap benefits. Indeed, a recent background article from the American College of Physicians emphasizes that it is not the specific AD but the length of time an AD is taken that is key. An adequate treatment trial consists of the maximum tolerable dose of a particular medication taken for at least 8 weeks.29
The traditional distinction between short-term and long-term side effects appears to be artificial and misleading.45
A more useful classification includes those effects that arrive early—sometimes within hours—and are self-limiting, such as nausea or headache; those that arrive early and persist, such as sexual dysfunction or fatigue; and those that emerge gradually and accrue, such as weight gain. Early side effects result from AD synaptic effects on neurotransmitter reuptake and receptor blockage, whereas those emerging later stem from such slow-to-develop neuroplastic adaptations as receptor desensitization and down-regulation that are hypothesized to reverse depression.25,27
During the first phase, long before an AD effect is expected, receptor-mediated synaptic side effects dominate the clinical picture. Examples include nausea or dull headache with SSRIs or jitteriness with bupropion. Because these typically diminish or disappear with time, prescribers can reassure patients in whom ADs have been recently prescribed that at least some of the unpleasantness of these early days will dissipate with continued drug exposure. With fluoxetine, activation in the form of nervousness, anxiety, agitation, and insomnia was more likely to wane during the first few weeks than sedation manifesting as daytime somnolence, low energy, and fatigue.46
In a carefully reasoned meta-analysis of AD-benzodiazepine combination therapy, Furukawa et al47
found that fewer patients taking the combination dropped out of treatment than those taking AD alone, although they advised judicious consideration of the potential for developing dependence or precipitating accidents.
“Start low and go slow” is good advice for limiting the noxiousness of these early adverse effects, and patients are more likely to tolerate them if given the expectation that they will be self-limiting. In addition, warn Kelly et al,40
“side effects sometimes exacerbate or masquerade as residual depressive symptoms.” Fatigue, cognitive impairment, apathy, agitation, irritability, sleep disturbances, sexual dysfunction, and gastrointestinal distress are among depressed patients' many concerns that could be attributed to ADs if clinicians are not careful to tease out what was already present before the patients started taking medications. If a patient has out-of-proportion adverse effects in response to low AD doses or no AD response to high doses, cytochrome P450 (CYP) genotypic testing for variations that cause slow or rapid metabolism may be indicated, particularly because many ADs are metabolized through either the CYP2D6 or CYP2C19 subsystems. Physicians should be aware of the potential for drug-drug interactions, particularly with long half-life drugs such as fluoxetine and agents that alter CYP activity.48,49
The second phase is noticeable as the ephemeral activating side effects dissipate and the AD begins to kick in. Classic examples can be found in the sexual realm. Just as individuals whose libido may have flagged as a function of their depression become interested in sex again, their bodies refuse to perform in accustomed ways. This paradox can be so upsetting that patients discontinue use of the ADs just as they are starting to help. Because ultimate treatment success depends on physicians doing whatever they can to ensure adherence, prescribers will do well to anticipate potential sexual side effects with their patients, reassuring them that, if problems emerge, effective strategies and adjunctive therapies are available.50
Weight gain exemplifies the third side effect phase, the late-emerging stage. Through changes in metabolism, eating patterns, exercise, or all 3, weight may stealthily increase until it is impossible to ignore. Again, paradoxically, mood may have greatly improved when weight gain becomes noticeable, and ironically it appears that some individuals—those already at risk for obesity—may be prone to outsized gain.
In an influential study of 401 patients interviewed 2.5 to 3.5 months after initiating AD therapy, Hu et al22
found that 86% reported at least 1 adverse effect and 55% had an adverse effect bothersome enough to challenge their willingness to continue use of their AD. Although dry mouth, drowsiness, and sexual dysfunction were most common, it was the latter 2 plus weight gain that the patients studied by Hu et al judged most bothersome. Following the lead of Hu et al, this review will consider each of these in turn.
Sleep Disturbances and Sedation.
Neurotransmitters associated with wakefulness and cortical arousal include histamine, acetylcholine, norepinephrine, and serotonin, among others. The most prominent neurotransmitter promoting sleep is γ-aminobutyric acid (GABA), and the most prominent neurotransmitter promoting awakeness is histamine. Strategies for inducing sleep thus most frequently focus on either increasing GABA function through allosteric activation of GABA-A receptors or blockage of histamine receptors. Benzodiazepines and the benzodiazepine receptor agonist “Z drugs” (zolpidem, zaleplon, and eszopiclone) work through modulating different components of the GABA-A receptor complex in the hypothalamic sleep-promoting center. Over-the-counter antihistamines, such as diphenhydramine and/or doxylamine, or ADs with strong antihistaminic side effects, such as trazodone and doxepin, block histamine receptors, particularly H1
, in the hypothalamic wake-promoting center.51
Disturbed sleep is a cardinal sign of depression, with difficulty falling asleep, trouble staying asleep, and early-morning wakening each potentially present and problematic. As the AD acts, sleep will ideally normalize. In the first few weeks after starting use of ADs, however, a benzodiazepine or Z-drug may be used to induce sleep while awaiting the AD's salutary effects. Because benzodiazepines induce tolerance, their use for sleep induction is not recommended beyond a month. Although intermediate-acting benzodiazepines may be more likely to facilitate sustained sleep, short- to intermediate-acting benzodiazepines (temazepam, estazolam, and triazolam) are preferred if daytime sleepiness, dizziness, or incoordination becomes a problem. Long-acting benzodiazepines, such as clonazepam, can induce a hangover sensation when used for sleep alone and are best reserved for patients with a strong anxiety component, present not only at bedtime but throughout the day. In these patients, benzodiazepines may slow racing thoughts or reduce muscular tension, making sleep more likely to follow. In elderly patients with reduced hepatic and renal clearance, lower doses of any benzodiazepine are advised to reduce the risk of falls and hip fractures.52
Although data beyond 6 months of use are lacking, the soporific effects of Z-drugs appear less likely to wear off, making them viable options for longer periods.52
Many studies during the past 15 years have shown that Z-drugs do not interfere with AD response. Recently, Fava et al53
have shown in a large placebo-controlled study that eszopiclone prescribed simultaneously with fluoxetine may not only immediately improve sleep but also enhance the speed of onset and magnitude of response of AD effect.
For some patients, ADs, particularly highly serotonergic ones, may relieve all neurovegetative signs of depression except poor sleep. An antihistamine strategy can prove helpful, augmenting ADs with low doses of highly antihistaminic-sedating ADs, such as doxepin or trazodone, at bedtime.52
warns, however, that most commonly used antihistamines are not selective for histamine receptors alone. Over-the-counter antihistamines, for example, are typically highly anticholinergic and can induce confusional states, particularly in elderly patients.
The TCAs with high affinity for histamine receptors work best for sleep induction in doses that are 1 to 2 orders of magnitude less than typical AD doses of 100 mg or more. With an AD dose, a TCA such as doxepin is associated with dry mouth, postural hypotension, cardiac arrhythmia, drowsiness, and weight gain, among other side effects. In a study of doxepin in ultralow doses of 1 to 6 mg, in which it is highly selective only for H1
receptors, anticholinergic effects, memory impairment, and next-day hangover effects occurred no more often than with placebo, thus minimizing myriad other neurotransmitter actions and concomitant adverse effects that occur with full dosing.54
Trazodone causes hypotension, constipation, and rarely priapism.52
Intolerable for most in a full AD dose (150-600 mg), trazodone effectively induces sleep in doses of 25 to 150 mg.51
The dose may need to be titrated down if day-time sedation or a hangover sensation occurs or up if early morning awakening continues to be a problem. Reports published soon after trazodone's introduction more than a quarter century ago associated it with ventricular arrhythmias and idiopathic peripheral edema with substantial weight gain. Janowsky et al55
challenged 20 patients with preexisting cardiac problems or histories of adverse cardiac reactions to trazodone doses ranging from 150 to 600 mg. Two of 20 had increased premature ventricular contraction frequency, and 1 of 20 had a reduction in frequency. Premature ventricular contraction frequency returned to baseline when the trazodone dose was decreased or use of the medication discontinued. Barrnett et al56
reported 10 cases of peripheral edema associated with trazodone in doses ranging from 150 to 600 mg. In all cases, edema resolved with dose reduction or discontinued use of the medication. Of note, patients taking trazodone for its soporific properties rarely require doses within the ranges used in the studies by Janowsky et al and Barrnett et al. Case reports implicate trazodone in acute overdoses (exceeding 1.5 g) as causing QT prolongation and delayed atrioventricular nodal conduction in a first case57
and a potentially lethal ventricular arrhythmia in a second case.58
With its prominent 5-hydroxytryptamine receptor 2 (5-HT2
)–blocking properties, the tetracyclic AD mirtazapine specifically counteracts SSRI-induced nonspecific stimulation of serotonin receptors. Blockade of 5-HT2
is associated with eventual shortened sleep-onset latency, increased total sleep time, and improved sleep efficiency.59
Mirtazapine also has histaminic effects prominent soon after initiation, which dissipate as tolerance to the sedating effect develops.60
By then 5-HT2
sleep-promoting influences predominate, with the added advantage of there being no hangover effect.60,61
A subtherapeutic AD dose of 15 mg or less is recommended for long-term treatment of insomnia.61
As mentioned, TCAs can be sedating, particularly in higher doses, and are therefore usually given in a single dose at bedtime. In contrast, SSRIs are typically considered energizing, possibly through their 5-HT2
–stimulating properties, and are usually taken once daily in the morning. With its greater tendency to cause sedation than other SSRIs, paroxetine is the exception and is often taken at night.41
For unclear reasons, a few patients may experience daytime sedation while taking any of the SSRIs. For them, a first strategy is to switch the morning dose to evening. If this fails to relieve the grogginess, a trial of another AD may be necessary.
First-generation ADs (TCAs) certainly cause sexual dysfunction along with a host of other adverse effects. Introduced in a more sexually repressed age, the 1950s, and causing constipation, urinary retention, blurry vision, and cognitive clouding, among other antihistaminic and anticholinergic effects, these drugs did not draw attention to sexual performance in the way that so-called modern ADs (eg, SSRIs, serotonin and norepinephrine reuptake inhibitors [SNRIs], and bupropion) have. By the late 1980s, Americans were speaking more freely about their sex lives and their dissatisfaction with medications that interfered with their intimate activities. Even so, self-reports of medication-induced sexual dysfunction, at 2% to 16%, have always lagged behind rates elicited by direct questioning,62
which can range up to 70%, even as problems uncovered through deliberate querying may not be clinically important for many experiencing them.63
As a result of substantial serotonin reuptake blockade, second-generation ADs (SSRIs and SNRIs, such as venlafaxine) have been associated with reduced libido in both men and women, as well as genital arousal problems (erection in men and lubrication in women) and delayed or absent orgasm in up to 40% of both men and women.64
The extent of these problems was not recognized until reports began to appear in the medical literature in the 1990s, years after these medications had been hailed as wonder drugs free of the adverse effects that had made TCAs so difficult for many to tolerate.
Although the details of SSRI-induced sexual dysfunction remain unknown, one culprit may be elevated synaptic levels of serotonin, which are theorized to improve mood centrally while indirectly contributing to compromised genital function peripherally. Although details of the specific interaction between serotonin and nitric oxide (NO) remain uncertain, NO is a direct mediator of arousal, with levels increasing in response to sexual stimulation. At least one SSRI (paroxetine) has been shown in the laboratory to interfere with NO synthase, thereby reducing NO levels.65
Ultimately, NO is responsible for triggering production of cyclic guanosine monophosphate, the agent that relaxes smooth muscle, including that found in genital vasculature, a process that allows blood to engorge sexual organs in both men and women. Although varied strategies, such as drug switching, dose reduction, drug holidays, and antidotes, have been tried to reverse sexual dysfunction, only 3 pharmacological strategies have been shown in double-blind trials to have efficacy against these adverse effects.64
To maintain dopamine levels thought necessary for optimal sexual functioning, the first 2 strategies consist of adding to the SSRI regimen either high daily doses of buspirone (60 mg/d), an anxiolytic that prevents serotonin's damping effect on dopamine, or therapeutic daily doses of bupropion (300-450 mg/d), a dopaminergic AD in its own right. Taken only in anticipation of intercourse, the third option, phosphodiesterase 5 (PDE-5) inhibitors (eg, sildenafil, vardenafil, and tadalafil), have become the standard of care.18,40,63
The PDE-5 inhibitors are thought to act through inhibiting the enzyme that inactivates cyclic guanosine monophosphate, thus permitting greater NO effect. In men, these drugs facilitate improved erections and greater orgasmic satisfaction, with nearly 60% of men needing an antidote for SSRI-induced dysfunction reporting “much” or “very much” improved sexual performance in all spheres.18
In women, sildenafil has shown promise for reversing the inadequate lubrication and delayed orgasm induced by SSRIs.66
It is critical to emphasize that PDE-5 inhibitors do not directly enhance libido but only exert their effects in response to sexual stimulation. One second-generation AD in particular appears to spare NO function. Patients taking bupropion alone report one-fourth to one-sixth the rate of sexual dysfunction in all spheres (desire, arousal, and orgasm) compared with patients taking SSRIs or SNRIs, with bupropion rates indistinguishable from placebo.67
This sparing of sexual function is likely the result of bupropion's negligible effects on serotonin levels and appreciable effects on dopamine levels, the neurotransmitter associated with enhancing sexual function.62
Like bupropion, nefazodone has minimal impact on sexual function, but it has carried a Food and Drug Administration black box warning since 2002, the only AD with such a warning for hepatotoxicity. Although most ADs have the potential to cause idiopathic liver injury, nefazodone seems to inflict the most serious damage, although at a rate of only 1 case per 250,000 to 300,000 treatment-years. Of 9 cases of liver failure in the literature up until 2007, 4 resolved on their own, 3 resolved after liver transplant, and 2 resulted in death, with 1 of the deaths occurring after liver transplant.68
Distinguishing AD effects from depressive symptoms can be challenging. Labbate63
emphasizes that sexual dysfunction is frequently both premorbid and multifactorial, resulting from medical factors, such as vascular disease and cardiac medications, and psychological factors, such as marital dissatisfaction. Assessing sexual function before starting use of an AD may prevent false attribution of preexisting dysfunction to the AD. If sexual adverse effects develop after starting use of ADs, however, active intervention is advised rather than waiting for improvement because tolerance to these adverse effects rarely develops.64
Although decreased libido and erectile dysfunction are common in patients with depression, the ejaculatory delay and anorgasmia frequent with SSRIs are not.69
As with many of the side effects attributed to ADs, low libido is for some patients a depressive symptom that may actually be relieved as the AD begins to take effect. Others may experience side effects as changes in familiar body responses that they can readily tolerate, particularly if prescribers have alerted them in advance to these possibilities. Some patients are not particularly concerned about their sex lives. Others may be willing to increase foreplay to overcome reduced libido or take a drug such as a PDE-5 inhibitor to enhance genital engorgement, particularly if their depression is relieved. Like men taking SSRIs for premature ejaculation, some depressed men may actually find delayed orgasm advantageous and desirable. For women, estrogen creams applied directly to the genitals may be helpful. In addition, nonpharmacological means, such as vaginal lubricants, can obviate adding a medication. Physicians and patients should talk about sexual issues before they become problems, preferably introducing the topic when AD therapy is initiated.
Antidepressants differ in their capacity to induce weight gain. Even in low doses, tertiary amine TCAs (eg, amitriptyline, imipramine, and doxepin) are more likely to cause weight increase than secondary amine TCAs (eg, nortriptyline and desipramine). Bupropion is linked to the least weight gain of any AD.40
Among the SSRIs, paroxetine has the greatest potential for inducing weight gain. Mirtazapine carries a reputation for being a particularly potent appetite stimulant, even though it is far from clear that it is any more of a culprit than any other AD. The exact mechanism is disputed, but these drugs' antihistaminic qualities are often implicated. Consistent with their having greater antihistaminic potency than secondary amine TCAs, the greater tendency of tertiary amine TCAs to induce weight gain supports this hypothesis.
Recent publications have challenged simplistic earlier formulations linking weight gain exclusively to histamine receptor antagonism, while also undermining initial hopes that SSRIs would prove weight neutral. “Despite their claimed selectivity,” write Harvey and Bouwer,70
“SSRIs still interact, either directly or indirectly, with various critical neurotransmitter systems.” Norepinephrine, dopamine, serotonin, cholecystokinin, corticotropin-releasing factor, endogenous opioids, neuropeptide Y, and various hormone-like peptides, including leptin, interact in complex ways to regulate appetite and related behaviors. Exquisitely regulated hormonal and metabolic feedback mechanisms involving ingested nutrients, adipose tissue, and hypothalamic nuclei govern energy expenditure, food intake, and ultimately weight gain or loss.71
Specifics of these interactions remain elusive, with many potential sites for pharmaceutical interferences.
Referencing both TCAs and SSRIs, Malone72
emphasizes that all drugs within each class are not created equal: “The tendency to cause weight gain is often related to differential specificity and sensitivity of binding to receptors involved with appetite regulation.” Moreover, a given AD does not identically affect everyone; people may vary not only in their response to any particular AD but also in their genetic predisposition to weight gain and obesity more generally.73
Healthy people gain a kilogram a year during adult life, independent of medication effects.74
Depression is associated with reduced food intake in some patients and increased intake in others, and therefore resolution of depression will be associated with weight gain in the former and weight loss in the latter.
Nonetheless, a frequently confusing and contradictory literature sheds some light on what to expect when starting AD therapy, which ADs are comparatively weight neutral, and what to advise patients about weight in advance of taking these drugs. A study by Sussman et al75
illuminates both the extent and scale of the problem with SSRIs. Observed during 16 to 46 weeks, 13.8% of patients taking SSRIs (1 in 7) gained 7% or more of body weight, although the study did not report possible differences among sertraline, fluoxetine, and paroxetine. In a recent review, Papakostas45
singled out paroxetine over other SSRIs as being particularly problematic. He cited a 26- to 32-week study in which a weight increase of 7% or more occurred in 25.5% (1 in 4) of paroxetine-treated patients vs only 6.8% of fluoxetine-treated patients and 4.2% of sertraline-treated patients. A meta-analysis of studies of escitalopram purports it to be weight neutral.76
In the study by Sussman et al,75
24.5% of imipramine-treated patients gained 7% or more of body weight. A study comparing patients taking imipramine and mirtazapine showed weight gain of 7% or more in 22% of the former but also 13% of the latter patients. Interestingly, average body weight increases overall were only 1.7 kg and 1.4 kg for imipramine and mirtazapine, respectively, suggesting that a few of the patients gained most of the weight. The same phenomenon of most of the weight gain concentrated in a few patients occurred in a 34-week study comparing duloxetine and paroxetine. Although the overall average weight gain was only 1.0 kg and 1.3 kg for the duloxetine- and paroxetine-treated patients, respectively, 10.8% of the duloxetine and 13.8% of the paroxetine groups recorded gains of at least 7% of body weight.77
The only AD on the market that is consistently cited as being weight neutral is bupropion, prescribed not only as Wellbutrin (GlaxoSmithKline, Ann Arbor, MI) for depression but also as Zyban (GlaxoSmithKline, Ann Arbor, MI) for smoking cessation.67,78,79
In summary, it appears that all ADs except bupropion and possibly escitalopram are associated with weight gain of at least a few pounds. In a notably few patients, large weight gain is possible, regardless of the particular AD, and typically occurs gradually over many months. Some ADs are more likely than others to induce marked weight gain, and patients already susceptible to obesity may be more prone to this effect. In a trial with patients with eating disorders, bupropion in its immediate-release form had a 4-fold risk of inducing seizures compared with SSRIs. In the sustained-release preparation, the rate of inducing seizures proved to be the same as for the SSRIs.80
Prudent practice indicates that bupropion should be prescribed warily, if at all, in patients with elevated seizure risk or a history of an eating disorder.