We know how frequently our patients complain of anxiety. Anxiety disorders are common, chronic conditions. They also increase the risk for mood and substance disorders, and complaints of anxiety are found in a wide range of other psychiatric and medical conditions, as well.

Pharmacologically, the two pillars of anxiety treatment for several decades have been the benzodiazepines and antidepressants (MAOIs, TCAs, SSRIs, and SNRIs), but new medications — particularly the atypical antipsychotics and anticonvulsants — have emerged in recent years to expand our repertoire.

Atypical Antipsychotics

Atypical antipsychotics (AAPs) are prescribed widely — sometimes with data to support their use, sometimes not. As of September 2013, no AAP has been approved for use in anxiety, although its not uncommon to see one used when a patient is refractory to other treatments.

The mechanism of action of AAPs in anxiety is unclear. Some, like aripiprazole (Abilify) have serotonin-1A partial agonist properties, similar to buspirone (BuSpar), while others, like quetiapine (Seroquel), have strong antihistamine properties, similar to hydroxyzine (Vistaril, Atarax). No common mechanism has been determined.

As an important historical footnote, two first-generation antipsychotics have been approved for anxiety: trifluoperazine (Stelazine) for short-term treatment of generalized anxiety, and the combination of perphenazine and amitriptyline (formerly marketed as Triavil) for depression and anxiety (Pies R, Psychiatry (Edgemont) 2009;6(6):2937). But these drugs rarely appear on psychiatrists radar screens these days.

Generalized Anxiety Disorder

So how is the evidence? For generalized anxiety disorder (GAD), the best data are for quetiapine (Seroquel), particularly the XR form. In three industry-funded, placebo-controlled trials enrolling more than 2,600 subjects, subjects responded better to quetiapine XR (50 or 150 mg/day, but not 300 mg/ day) than to placebo, as measured by a ?50% decrease in the Hamilton Anxiety Scale (HAM-A) over eight weeks. One study also found quetiapine XR to be superior to escitalopram (Lexapro) 10 mg/day while another showed equivalence to paroxetine (Paxil) 20 mg/ day. Remission was significantly more common with the 150 mg dose than with placebo (Gao K et al, Expert Rev Neurother 2009;9(8):11471158).

Despite these impressive numbers, quetiapine XR has not earned FDA approval for GAD, most likely because of the potential for widespread and prolonged use of this agentwhich has well-known metabolic side effects and requires close monitoringwhen safer alternatives are available. Its also possible that its short-acting (and cheaper) cousin quetiapine may do just as well as the XR form, but the two have not been studied head-to-head.

Randomized controlled trials of other AAPs in GAD have been unconvincing. Risperidone (Risperdal) was no more effective than placebo in a large (N=417) trial of patients with GAD refractory to anxiolytics (Pandina GJ et al, Psychopharmacol Bull 2007;40(3):4157) even though a smaller study (N=40) was positive (Browman-Mintzer O et al, J Clin Psychiatry 2005;66:13211325). Olanzapine (Zyprexa) was effective in a very small study (N=46) as an adjunctive agent with fluoxetine (Prozac), but subjects experienced significant weight gain (Pollack MH et al, Biol Psychiatry 2006;59(3):211225). Several smaller, open-label trials have shown some benefit for other AAPs (reviewed in Gao K, op.cit) but, other than the ones discussed here, larger placebo-controlled studies have been equivocal.

Other Anxiety Disorders

What about other anxiety disorders? For OCD, a pooled analysis of three studies of risperidone (0.5 to 2.25 mg/ day) found risperidone to be slightly better than placebo, but the authors of the analysis suggested that these studies may have been affected by publication bias, given the variation in effect sizes (Maher AR et al, JAMA 2011;306(12):13591369).

PTSD is a complex disorder in which AAPs are frequently used, and small studies of olanzapine (15 mg/ day, N=19) (Stein MB et al, Am J Psychiatry 2002;159:17771779) and risperidone (Bartzokis G et al, Biol Psychiatry 2005;57(5):474479) as adjunctive treatment for combat-related PTSD have shown some promise, but other published trials, including a more recent larger PTSD trial (Krystal JH et al, JAMA 2011;306(5):493-502), have been negative.

Because most trials have been small, and negative trials have been as numerous as positive onesnot to mention the lack of head-to-head trials of these agentsits difficult to make a solid recommendation for any particular AAP in the treatment of anxiety. The existing meta-analyses of these agents for specific anxiety disorders argue for further study (Fineberg NA, FOCUS 2007;5(3):354360) and larger trials. Of course, what were treating may also vary in significant ways, a point well return to later.


Newer on the anti-anxiety scene are the anticonvulsants. All anticonvulsants work via some combination of sodium- or calcium-channel blockade, GABA potentiation, or glutamate inhibition, but individual agents vary in their precise mechanisms. Because anxious symptoms are thought to result from activation of fear circuits, primarily involving the amygdala, hippocampus, and periaqueductal gray, and because anticonvulsants are designed to specifically prevent excessive neuronal activation, their use in anxiety seems rational. Do the data support this?

Unfortunately, despite more than a dozen anticonvulsants approved for human use, only one anticonvulsant (other than the benzodiazepines and barbiturates, which wont be discussed here) shows a benefit for anxiety in several randomized clinical trials, and thats pregabalin (Lyrica), for GAD.

Pregabalin is a GABA analogue but its primary effect appears to be blockade of the alpha-2-delta subunit of the N-type calcium channel, preventing neuronal excitation and neurotransmitter release. (This is also one mechanism of action of gabapentin [Neurontin], a close relative.)

Generalized Anxiety Disorder

Several controlled trials, all funded by the drugs manufacturer, have shown that pregabalin, at doses ranging from 300 to 600 mg/day, can reduce symptoms of generalized anxiety as measured by the HAM-A. Three of these studies also found pregabalins effect to be similar to that of lorazepam (Ativan), alprazolam (Xanax), and venlafaxine (Effexor), respectively. A later meta-analysis of placebo-controlled anxiety trials (with no pharmaceutical industry funding) found pregabalin to have a higher effect size (0.5) in reduction of HAM-A scores than the benzodiazepines (0.38) and SSRIs (0.36) for GAD (Hidalgo RB et al, J Psychopharm 2007;21(8):864872).

Despite its apparent efficacy, pregabalin is also associated with an elevated, dose-dependent risk of dizziness, somnolence, and weight gain (Strawn JR and Geracioti TD, Neuropsych Dis Treat 2007;3(2):237243). It is likely that these adverse effects explain why pregabalin was rejected by the FDA as a treatment for generalized anxiety disorder back in 2004, and again in 2009, even though it was approved in Europe in 2006 for this indication.

Other Anxiety Disorders

Other than pregabalin, placebo-controlled clinical trials reveal few other bright spots for anticonvulsants in anxiety disorders. For the treatment of panic disorder, gabapentin, at doses as high as 3600 mg/day, has been shown in an open-label study to be more effective than placebo. Several open-label studies in PTSD show some benefit of topiramate (median 50 mg/day) and lamotrigine (500 mg/day but N=10 only), while social phobia may benefit from pregabalin (600 mg/day) and gabapentin (9003600 mg/ day). Anecdotal reports of improvement in OCD can be found for just about every anticonvulsant, but the only one with several such reports is topiramate (Topamax) (mean dose 253 mg/day), particularly in augmentation with SSRIs (for a review, see Mula M et al, J Clin Psychopharm 2007;27(3):263272). As always, open-label studies need to be interpreted with caution, as those that are negative are unlikely to be published.

Why the Mixed Results?

A casual read of the data, not to mention abundant case reports and anecdotal evidence, suggests that many anticonvulsants and atypical antipsychotics could work for anxiety disorders, but in controlled trials, most show little or no effect compared to placebo. Why the discrepancy? A very likely answer is because of the heterogeneity of anxiety disorders themselves. Not only are the typical presentations of OCD, PTSD, and social phobia likely to be very dissimilar to each other (see the Expert Q&A with Dr Pine in this issue), but even within a given diagnosis, anxiety can manifest very differently.

Moreover, comorbidity is very high in anxiety disorders. Fear disorders like phobia, panic, and OCD are commonly seen together, as are the distress or misery disorders like GAD and PTSD. All of the above are highly comorbid with mood disorders and substance abuse or dependence (Bienvenu OJ et al, Curr Top Behav Neurosci 2010;2:319), not to mention medical illnesses.

The way we describe and measure anxiety itself creates tremendous variability. There are distinct differences, for instance, between criteria for GAD in the DSM (used in most American research), and in the ICD-10 (used primarily in Europe). ICD-10, for instance, requires autonomic arousal while the DSM does not; and the DSM criteria for GAD require significant distress or impairment, unlike ICD-10. Similarly, the most commonly used symptom rating scale, the HAM-A, contains some items that pertain to somatic anxiety, and others addressing psychic anxiety. Medications may target somatic and psychic symptoms differently (Lydiard RB et al, Int J Neuropsychopharmacol 2010;13(2):229 241).

And then theres the consideration of what we call anxiety in the first place. Weve shed the vague psychoanalytic label of neurosis, and since DSM-III weve described these conditions as anxiety disorders, but the boundaries have continued to shift. DSM-5, for instance, includes two new categories of Obsessive-Compulsive Disorders (which includes OCD, body dysmorphic disorder, and others) and Trauma and Stressor-Related Disorders (which includes PTSD and adjustment disorders), reflecting differences in neurobiology and treatment relative to other anxiety disorders. Some even argue that anxiety, in many cases, is simply the brain using its own fear circuitry in an adaptive way, in which case, nothing is dysfunctional at all (Horowitz AV and Wakefield JG, All We Have To Fear. New York: Oxford University Press; 2012; see also Kendler KS, Am J Psychiatry 2013;170(1):124125).

So when it comes to medication management, asking whether a given medication is useful for anxiety is like asking whether a turkey sandwich is a good lunchtime meal: for some people, it hits the spot, but for others (like vegetarians) it should be avoided. A better understanding of the neurobiology of different anxiety disorders, the response of individual symptoms to particular medications, and the role of other drugsand psychotherapiesin their management, will help us to optimize, and individualize, outcomes for our anxious patients.

TCPR’S VERDICT: Atypical antipsychotics and anticonvulsants may have a role in the treatment of anxiety disorders. The lack of FDA approval or strong evidence supporting any individual treatmentwith a few exceptionsmay speak more to the problems of diagnosis and clinical trial methodology than to the failures of medications themselves.