ReviewThe effects of safety behaviors during exposure therapy for anxiety: Critical analysis from an inhibitory learning perspective
Introduction
Anxiety, broadly defined, is a natural reaction to perceived threat and is manifested cognitively (e.g., racing thoughts), physiologically (e.g., autonomic arousal), and behaviorally (e.g., escape). Although anxiety is evolutionarily adaptive, those with pathological anxiety (e.g., DSM-5 defined anxiety disorders) experience anxiety in the absence of real threat. That is, if “normal anxiety” serves as an alarm system, the 18% of adults and 25% of children in the United States with anxiety disorders experience frequent false alarms that cause substantial distress and functional impairment (American Psychiatric Association [APA], 2013, Kessler et al., 2005, Merikangas et al., 2010).
In its general form, exposure-based cognitive-behavioral therapy (CBT) for clinical anxiety entails the guided, systematic, and repeated confrontation with feared stimuli (e.g., situations, objects, thoughts). Exposure has demonstrated substantial transdiagnostic efficacy and effectiveness in previous research (Abramowitz, Deacon, & Whiteside, 2011). Accordingly, exposure is considered the first-line intervention for anxiety disorders by international health care bodies (e.g., American Psychiatric Association [APA], 2013, NICE, 2005).
Safety behaviors are overt or covert actions performed to prevent, escape, or minimize a feared catastrophe and/or associated distress (Telch & Lancaster, 2012). Safety behaviors are functionally related to anxious beliefs and are logical, if unnecessary. To illustrate, a man with a fear of germs might wear gloves when using public transportation (i.e., prevent contamination), exit a bus after a child sneezes (i.e., escape contamination), or look out the window and tell himself “relax” when on a crowded flight (i.e., minimize his anxiety associated with possibly becoming contaminated). Although topographically similar, safety behaviors are functionally distinct from adaptive coping (e.g., telling oneself “it's okay if I get germs on me”) or non-pathological safety maneuvers (e.g., washing hands after handling raw meat; Thwaites & Freeston, 2005). That is, whereas attempts to remain safe when faced with actual threat ensure survival, performing such behaviors in the absence of real threat is unnecessary and even generates and maintains distress (see Helbig-Lang & Petermann, 2010). Other examples of situational safety behaviors commonly endorsed by anxious patients are presented in Table 1.
Research consistently implicates safety behaviors in the maintenance of anxiety disorders; accordingly, safety behaviors are traditionally eliminated from anxious patients' behavioral repertoire over the course of exposure therapy (e.g., Abramowitz et al., 2011, Barlow et al., 2011). Yet recent debate as to whether safety behaviors are unconditionally harmful during exposure has challenged this notion. Although substantial evidence—as well as clinical convention—advocates the elimination of safety behaviors during exposure, Rachman, Radomsky, and Shafran (2008) called for a reconsideration of this axiom. Consequently, the role of safety behaviors during exposure has garnered renewed research attention. Results from these studies, however, are mixed and controversial. For example, in a recent meta-analysis of the effects of safety behaviors on exposure, Meulders, van Daele, Volders, and Vlaeyen (2016) concluded that the aggregate data “was inconclusive and could not provide strong evidence supporting either the removal or addition of [safety behaviors] during exposure” (p. 151).
Meta-analytic studies carry the benefit of pooling data across multiple studies to increase statistical power when testing a specific hypothesis (e.g., “do safety behaviors interfere with exposure therapy on specific outcomes?”). However, if—as in the present paper—the aim is to go beyond testing a discrete statistical hypothesis and instead conduct a rigorous conceptual examination of a specific topic, systematic qualitative reviews are important alongside meta-analyses (e.g., Albarracín, 2015, Garg et al., 2008). Indeed, the latter allow for more in-depth discussion of theoretical mechanisms underlying improvement (i.e., therapeutic change processes) while still adhering to rigorous review criteria and presenting opposing perspectives in a balanced manner. In light of inconsistent results in the extant literature on safety behaviors, a qualitative systematic review of the literature on the effects of safety behaviors would be helpful for clinicians and researchers working with anxious individuals.
The judicious use of safety behaviors is a controversial thesis; furthermore, clinicians are left without clear direction, given that inconsistent study findings carry contradictory clinical implications. If safety behaviors are not as detrimental as previously assumed, perhaps judiciously incorporating them into exposure therapy will improve treatment retention and outcome (e.g., Rachman et al., 2008). Alternatively, if safety behaviors are deleterious in the long-term, then encouraging anxious patients to rely on these strategies might be iatrogenic. There are also theoretical implications of a systematic review of the safety behavior research. As discussed below, prevailing models of exposure therapy have enhanced our understanding of the treatment of clinical anxiety (e.g., Craske et al., 2008), yet these approaches are limited and fail to address all aspects of long-term treatment gains or failure (i.e., relapse). Therefore, it is important to bridge the gap between advances in theoretical models of exposure therapy and the empirical literature base related to safety behavior use during exposure. In sum, given the growing popularity of newer models of exposure therapy (e.g., inhibitory learning theory) and the possibility for the judicious use of safety behaviors to either augment or diminish exposure's efficacy, a theory-based analysis of this topic is greatly needed. The current review aims to critically examine the extant literature addressing the role of safety behaviors in exposure therapy from an inhibitory learning perspective. Because the effects of distraction have been reviewed elsewhere (e.g., Parrish et al., 2008, Podină et al., 2013), the present paper will focus on other situational safety behaviors. First, we will explicate current evidence-based theories of the therapeutic mechanisms underlying exposure, emphasizing recent developments in inhibitory learning theory (Craske et al., 2014, Craske et al., 2008). Second, we critically evaluate the theoretical and empirical evidence for the deleterious effect of safety behaviors during exposure therapy. Third, we examine the theoretical and empirical evidence for the proposed advantages of incorporating safety behaviors during exposure. Fourth, we highlight research limitations and future directions. Finally, we offer clinical recommendations based on the aggregated available research.
Section snippets
Literature search strategy and inclusion criteria
Studies included in the present review were identified through an electronic literature search (via the PsycINFO and PubMed databases), supplemented by checking reference lists of published studies (Horsley, Dingwall, & Sampson, 2011). Articles were included if they (a) were published or in press before October 2016,1
How does exposure work?
Traditionally, the dominant explanatory model for exposure's effectiveness has been emotional processing theory (EPT), initially proposed by Rachman (1980), outlined by Foa and Kozak (1986), and revised by Foa and colleagues (Foa et al., 2006, Foa and McNally, 1996). EPT posits that fear extinction during exposure results from the activation of a fear structure (a fear-based association between a stimulus and its significance; e.g., “dog” and the fear of being attacked by the dog) paired with
Safety behaviors interfere with exposure therapy
As shown in Table 2, safety behaviors have been associated with poorer outcomes in clinical and experimental work. Specifically, research shows that exposure in which safety behaviors are encouraged are associated with poorer treatment outcome relative to exposure in which safety behaviors are systematically prevented. Several mechanisms have been offered to explain this effect, yet inhibitory learning theory proposes additional pathways through which safety behaviors may be deleterious.
Safety behaviors cause misattributions of safety
Acknowledging the apparent failure for anxious patients to improve from naturalistic disconfirmation of fearful beliefs (e.g., a patient fails to overcome panic disorder despite experiencing 25 non-lethal panic attacks), Salkovskis (1991) considered the undermining effect of safety behaviors. In his misattribution of safety hypothesis, Salkovskis proposed that anxious individuals whose feared catastrophes did not occur in the context of performing safety behaviors concluded not that the feared
Inhibitory learning theories of how safety behaviors interfere with exposure
Inhibitory learning theory suggests additional mechanisms through which safety behaviors might interfere with exposure. Specifically, safety behaviors may prevent the (a) maximal violation of negative expectancies, (b) generalization of inhibitory associations across contexts, and (c) development of distress tolerance critical for exposure therapy. Each of these possibilities is discussed next.
Summary
In summary, theory suggests that safety behaviors interfere with exposure by promoting safety misattributions, disrupting therapeutic information processing, attenuating negative expectancy violation, contextualizing inhibitory learning, and dampening distress tolerance. Although clinical and analogue research supports the notion that safety behaviors should be eliminated during exposure, studies are methodologically limited and results are mixed. Indeed, occasional null findings lend merit to
Safety behaviors do not necessarily interfere with exposure therapy
As mentioned, not all research supports recommendations to eliminate safety behaviors at the start of exposure (see Table 2). Rachman et al. (2008) recently proposed the “judicious use” of safety behaviors: the careful and strategic implementation of safety behaviors in the early and/or most challenging stages of treatment. Advocates of this method highlight positive consequences, such as enhanced treatment acceptability and approach behavior (e.g., Levy & Radomsky, 2014). Research on the
Safety behaviors enhance exposure's acceptability and tolerability
Citing high treatment refusal and dropout rates in exposure therapy, Radomsky and colleagues have called for research exploring how this treatment may be modified to enhance its tolerability and acceptability without diminishing its efficacy. Safety behavior proponents have argued that exposure might be perceived as more tolerable and acceptable if safety behaviors are strategically incorporated. That is, if patients feel safer knowing they may perform safety behaviors, they might be more
Inhibitory learning theories of how safety behaviors enhance exposure
There is a dearth of research on the potential for safety behaviors to optimize exposure therapy through mechanisms critical to inhibitory learning theory, yet it is possible that safety behaviors optimize inhibitory learning processes. Specifically, the inclusion of safety behaviors during exposure may allow for the (a) violation of negative expectances, (b) generalization of inhibitory associations to other contexts, and (c) development of distress tolerance. These theoretical advantages of
Summary
Proponents of incorporating safety behaviors into exposure highlight positive consequences such as enhanced treatment acceptability and tolerability, approach behavior, and self-efficacy. Consistent with the theory of judicious use of safety behaviors (Rachman et al., 2008), some research shows that safety behaviors are benign or beneficial during exposure. Findings, however, are inconsistent—especially at follow-up. Moreover, evidence for the advantages of safety behaviors on inhibitory
Limitations and methodological considerations
A number of limitations associated with the literature reviewed here dampen the generalizability of study findings and qualify the conclusions reached by investigators. One notable shortcoming concerns the ecological validity of how exposure was delivered in these studies. Specifically, many of the experiments discussed in this review utilized brief (3- to 30-min) exposures, whereas most exposure therapy manuals used in clinical settings recommend longer durations (e.g., 30 to 90 min; Abramowitz
Conceptual issues and future research directions
Rachman et al. (2008) stated that although safety behaviors may be useful early in treatment, they should ultimately be eliminated from patients' behavioral repertoire. Yet it is unclear what the “judicious use” of safety behaviors looks like. Which safety behaviors should be strategically implemented, and at what dose? At what point and rate should safety behaviors be incrementally eliminated? Should safety behaviors be differentially eliminated as a function of the exposure context (e.g.,
Conclusions and clinical recommendations
Our review of the literature indicates that although safety behaviors are not unconditionally deleterious, they tend to interfere with exposure outcomes, possibly by promoting safety misattributions, preventing therapeutic information processing, or interfering with other mechanisms central to inhibitory learning theory. Therefore, clinicians are recommended to eliminate safety behaviors as quickly as anxious patients are willing, as there is not sufficient empirical support to recommend the
References (106)
- et al.
Strategies for improving long-term outcomes in cognitive-behavioral therapy for obsessive-compulsive disorder: Insights from learning theory
Cognitive and Behavioral Practice
(2014) - et al.
The relevance of analogue studies for understanding obsessions and compulsions
Clinical Psychology Review
(2014) - et al.
Addressing relapse in cognitive behavioral therapy for panic disorder: Methods for optimizing long-term treatment outcomes
Cognitive and Behavioral Practice
(2011) - et al.
“If I feel anxious, there must be danger”: Ex-consequentia reasoning in inferring danger in anxiety disorders
Behaviour Research and Therapy
(1995) - et al.
Does habituation matter? Emotional processing theory and exposure therapy for acrophobia
Behaviour Research and Therapy
(2010) Context, ambiguity, and unlearning: Sources of relapse after behavioral extinction
Biological Psychiatry
(2002)- et al.
Mediators of exposure therapy for youth obsessive-compulsive disorder: Specificity and temporal sequence of client and treatment factors
Behavior Therapy
(2015) - et al.
The Driving Behavior Survey: Scale construction and validation
Journal of Anxiety Disorders
(2011) - et al.
Memory biases in the anxiety disorders: Current status
Clinical Psychology Review
(2002) - et al.
Optimizing inhibitory learning during exposure therapy
Behaviour Research and Therapy
(2008)