How does one manage a patient with panic disorder who initially sought treatment within a family practice setting but now requires referral to a psychiatrist because neither of the first two pharmacologic interventions worked? The first intervention was lorazepam, 0.5 mg as needed, which helped briefly with the acute attacks. However the attacks continued and worsened over time. The internist tried fluoxetine, 20 mg/day but the patient became more anxious and discontinued the medication.

Effective treatments for panic disorder are relatively straightforward and are supported by a substantial evidence base. Both medications and cognitive behavior treatment (CBT) have been studied in multiple randomized controlled trials and found to be effective alone and in combination. Combination treatment has a slight advantage over monotherapy only in the short-term (3—6 months) time window after which effects are comparable (1). Assessing outcome in panic disorder is more complex than in depression because one has to account for changes in attack frequency and intensity, concomitant phobic avoidance, sensitivity to internal bodily sensations, and generalized anticipatory anxiety, in addition to changes in social or work function. Thus, assessing whether a treatment is working requires measurement of multiple domains [nicely provided in the Panic Disorder Severity Rating Scale (2)] rather than a single domain as in depressive disorders.

Selective serotonin reuptake inhibitors (SSRIs) are the pharmacological treatment of choice for panic disorder because of their reasonable tolerability and broad spectrum of action (i.e., they are also effective against all of the other comorbid anxiety disorders as well as comorbid depression, all of which occur at high rates in patients with panic disorder). Treatment must be initiated at lower doses because some patients with panic disorder experience overstimulation with the standard starting doses used in depression. This may be more common with an SSRI such as fluoxetine, which is a bit more stimulating than other SSRIs. There are no differences in efficacy among any of the SSRIs so choice of agent is best determined by avoidance of specific side effects that might mirror anxiety symptoms of a given patient or might be less tolerable because of comorbid medical illness or concomitant medications. Both clinical trials and naturalistic studies suggest that about one-quarter to one-third of patients taking SSRIs for panic disorder experience substantial improvement in their condition, akin to remission, whereas another one-third experience notable improvement, akin to response. For some patients not responding to an initial trial of an SSRI, switching to a different SSRI or to a serotonin-norepinephrine reuptake inhibitor is likely to be effective.

Benzodiazepines may be required in panic disorder, often as adjunctive treatment to antidepressants, to move the responsive patient closer to remission. Benzodiazepines may also be used as monotherapy in patients either unresponsive to or intolerant of antidepressants. Concomitant benzodiazepines can be used from the outset of antidepressant treatment to achieve a more rapid response (3), although this entails a subsequent tapering of the benzodiazepine, and this is more time consuming for both patient and doctor.

For maximum treatment benefit, benzodiazepines must be given on a regular schedule to attain relatively constant blood levels. Although the use of benzodiazepines on an as-needed basis is attractive to patients who may want to take a medication only when they are symptomatic, and to physicians concerned about "addiction" or "dependence," this practice not only is ineffective but may also be harmful for two reasons. First, patients may experience withdrawal effects when the medication wears off, which will be hard to distinguish from endogenous symptoms of anxiety. Second, patients usually take as-needed benzodiazepines in anticipation of facing feared situations, and this impairs their ability to desensitize themselves to these situations, consistent with recent studies that patients taking as-needed benzodiazpines have worse outcomes with CBT than those taking regularly scheduled benzodiazepines (4). Hence, taking benzodiazepines once or twice daily (e.g., clonazepam once or twice a day) will separate the cue of pill taking from any anxiolytic effect in stressful situations during the day and optimize the patient's ability to improve self-management of anxiety. Available evidence shows that individuals with panic disorder taking benzodiazepines chronically generally lower rather than increase their dose over time (i.e., there is no dose escalation or risk of addiction) (5), and there is no evidence that use of these agents increases the risk of abuse of or dependence on other substances in individuals who do not already have a substance use problem.

Cognitive behavior treatment is shown by increasing evidence to be not only as effective as medications for panic disorder but may also have longer lasting effects after discontinuation. Although medications are more widely available and easier to use correctly. The limited data available suggest that between one-third and one-half of patients with panic disorder will have a relapse to some degree after discontinuation of medication in the course of 6—12 months (6). Symptom reemergence after the end of CBT sessions appears to be somewhat less, although the evidence in this regard is not uniform (7). Although CBT requires specialized training, recent studies have explored the use of masters level clinicians to deliver CBT in primary care settings and found that the treatment is quite effective, although probably not at the quality of delivery by specialists with doctoral level training (8). Computer-assisted training programs are also actively being developed to facilitate delivery by clinicians less familiar with standard CBT protocols and to provide patients with video and other graphic illustrations that improve delivery of CBT concepts and possibly retention of CBT material (9).

These considerations will inform the treatment of the patient presented here. Initial considerations must focus on the bad experience this patient has had with two separate medications. The first was ineffective, and the second caused adverse effects that aggravated the underlying panic syndrome. As many patients with panic disorder are unusually sensitive to bodily sensations and apprehensive about taking medication, much time will have to be spent engaging this patient, understanding what treatment he or she might prefer, determining whether he or she is still willing to try medication, after understanding the reason that these initial trials failed, and offering the option of CBT if that is something he or she would prefer and if resources are available to deliver it.

Psychoeducation is the initial core phase of CBT and can be performed well by a knowledgeable clinician who is able to articulate a model of panic disorder that emphasizes a genetic vulnerability to experience increased anxiety and have the "fight or flight alarm" go awry in the face of stress. This clinician must also be able to present a prescriptive solution that validates either the use of medication or CBT as a way to respectively either block the panic symptoms or learn to reprogram the brain so that it processes initial fear signals and bodily sensations differently, resulting in lowered vulnerability to panic, even when stressed. This patient illustrates the major public health challenge for panic disorder (10): many more patients fail to improve because of failure in delivery of effective evidence-based treatments rather than because of resistance to evidence-based treatments.