It is difficult to identify a standard trial of CBT because of the remarkable heterogeneity of interventions for anxiety disorders. For example, exposure interventions may include imaginal, in vivo or interoceptive exposure, or the use of virtual reality technology for hard-to-arrange exposures.
Attempts at defining treatment resistance in this context are fraught with difficulty given the wide range of CBT and the diversity of anxiety disorders. Nonetheless, certain key items should be addressed. An adequate course of CBT should include a full protocol of treatment, ideally defined by interventions found to be effective in at least two controlled trials. Treatment protocols often include 12–16 sessions of a combination of information, cognitive restructuring, and exposure interventions, with skill rehearsal and exposure in home-practice assignments (
Barlow, 2001). The use of protocol-driven CBT for anxiety disorders is well tolerated, cost-effective, and efficacious acutely and in the long term (
Foa et al., 2002; McHugh et al., in press;
Otto et al., 2004), with evidence for further gains following acute CBT as patients consolidate and extend learning-based approaches (
Liebowitz et al., 1999).
Supportive interventions, such as brief relaxation treatment or breathing retraining do not constitute an adequate CBT trial as these symptom management procedures alone do not, for example, add to the efficacy of CBT protocols for panic disorder (
Schmidt et al., 2002). Failure to respond to an initial CBT protocol suggests the need for alternative interventions (e.g., greater attention to changing cognitions, refinement of the target or intensity of exposure interventions), other empirically supported treatments, or combined treatment modalities. Relevant recent findings on extinction learning and context that can be applied across the anxiety disorders are reviewed, and next-step strategies for non-response are considered.
EXTINCTION LEARNING AND LIMITING FACTORS
Anxiety disorders are characterized by exaggerated distress and disability in response to phobic cues. As long as an individual can learn and integrate new experiences with phobic stimuli, exaggerated responses to phobic cues should respond to extinction learning, and anxiety, thus diminish. However, this assumes that the correct fears are being targeted and appropriate learning conditions are identified and established by the therapist.
Informational and cognitive interventions often are used to facilitate and enhance initial exposure to fears, which are designed to become increasingly relevant to core fears as therapy progresses. Balance must be maintained between conditions that facilitate exposure to avoided phobic cues and conditions where exposure provides useful learning. Strategies that avert or attenuate anxiety in phobic situations (i.e., “safety behaviors”) may temporarily alleviate anxiety, but reduce long-term efficacy (
Powers et al., 2004;
Salkovskis et al., 1999).
What is learned during extinction in one context (e.g., when safety behaviors or medications are available) may not generalize to a different context (e.g., when no safety behavior or medications are available) (
Bouton, 2002). Stronger findings have been reported for shifts in internal context, such as medication effects (
Mystkowski et al., 2003). Therapeutic approaches for treatment-resistant anxiety disorders should include strategies that maximize fear extinction during exposure and generalize this learning across contexts, seeking to reverse the effects of distraction, safety cues, and context shifts on the longer-term acquisition of a sense of safety when confronted by phobic cues (
Powers et al., 2006).
Attention to shifts in context, specifically internal context provided by pharmacotherapy, is relevant to combining medications with CBT. Context effects may explain the loss of efficacy seen in combination treatment trials when medications are later discontinued (e.g.,
Barlow et al., 2000). However, maintenance and extension of treatment gains are common when CBT is provided during and after medication taper (
Otto et al., 2002).
Context effects introduce additional complexity into the selection of the next-step strategy following partial response to CBT. Should patients be switched to pharmacotherapy, risking the introduction of context effects that may require additional CBT to prevent relapse in the future? If there was no response to CBT, then there is little reason to be concerned about context effects introduced by medications. However, if patients achieved a partial response to CBT, further honing of the targets of CBT interventions and identification of context effects that may be preventing generalization of exposure-based learning should be considered before addition of pharmacologic alternatives.
There is a dearth of evidence addressing whether troubleshooting and persisting with CBT is better than switching treatment modalities (
Kampman et al., 2002). In one small randomized trial, continuation of exposure-based treatment alone was at least as effective as its combination with imipramine or cognitive therapy (
Fava et al., 1997). Further, dropout data from controlled trials of CBT and pharmacotherapy indicate that CBT tends to be more tolerable and acceptable than medication alternatives (
Otto et al., 2005). Nonetheless, non-response to initial CBT does provide a call to therapists to re-evaluate whether core fears are being adequately targeted by cognitive and exposure interventions, with consideration of altering the timing, modality, and context (including the presence of safety cues that may undermine the efficacy of exposure treatments), to try to enhance response to additional CBT (e.g.,
Powers et al., 2004,
2006;
Smits et al., 2006).
In addition, recent research has introduced a new approach to combined treatment, in which pharmacotherapy is used to enhance therapeutic learning rather than attenuate symptoms (
Ressler et al., 2004). The
N-methyl-D-aspartate (NMDA) agonist, D-cycloserine (DCS) augments extinction learning in animal models after a single dose administered immediately before extinction trials (
Walker et al., 2002). In a proof of concept translational RCT, 28 patients with acrophobia were randomized to virtual-reality exposure (VRE) therapy plus one of two doses of DCS or placebo, administered before each of 2 weekly VRE sessions (
Ressler et al., 2004). DCS resulted in significantly less fear of heights compared with placebo during the trial as well as following treatment. In an independent replication of the DCS enhancement effect,
Hofmann et al. (2006) demonstrated better outcome in a sample of 27 outpatients with primarily generalized social anxiety disorder when treated with DCS versus placebo in the context of a five-session CBT protocol combining study pills with the latter four sessions of increasingly challenging public speech exposures. Furthermore, recent evidence also supports the efficacy of combined treatment with DCS for OCD, although there are indications that with ongoing exposure therapy CBT alone can catch up to the boost in efficacy provided by DCS (Kushner et al., in press). If these promising findings are further replicated, this strategy for combined treatment suggests a potentially expeditious way to improve the speed of onset and, perhaps, overall efficacy of CBT for anxiety.