Anxiety sensitivity: selective review of promising research and future directions

Anxiety sensitivity (AS) is the fear of anxiety and arousal-related sensations [1]. AS, or the fear of fear, has been conceptualized as a trait-like cognitive predisposition. When anxious, individuals high in AS become acutely fearful owing to the belief that these anxiety sensations have harmful physical, psychological or social consequences. Therefore, AS theoretically amplifies pre-existing states of anxiety in the short term and theoretically increases the risk of panic psychopathology and perhaps other anxiety disorders in the long term [2,3]. The purpose of this editorial is to selectively describe key aspects of work on AS and highlight future directions that may have translational and public health significance. Thus, rather than offer a comprehensive review of extant work, we call attention to some of what we believe is the most scientifically promising and exciting work in the field on this topic.

The vast majority of scientific work on AS has sought to explore how, and to what extent, it relates to increased risk of anxiety symptoms and problems. The preponderance of this work has focused on panic psychopathology. Overall, a diverse array of studies, ranging from laboratory to field approaches, indicate that AS is concurrently, incrementally and prospectively associated with panic psychopathology. For example, AS-measured pre-exposure to biological challenge (a provocation methodology) is a significant predictor of post challenge anxiety symptoms [4]. These effects have been observed above and beyond the tendency to experience negative emotional symptoms (e.g., trait anxiety and negative affectivity), are relatively specific to self-reported distress (e.g., psychophysiological responding) and are apparent from adolescence through to adulthood [4,5]. Similar associations have been observed in real-world settings and across a range of cultural groups[6].

Prospective studies similarly suggest that AS predicts the future onset of unexpected panic attacks for adolescents [7,8] and young adults [9–12]. Such longitudinal studies also suggest that these effects are applicable to anxiety symptoms more generally [13], and therefore, are not necessarily limited solely to panic psychopathology. Although less well studied, there are two reports suggesting AS is related to the future development of anxiety psychopathology, with some degree of explanatory specificity for panic attacks and panic disorder relative to other anxiety conditions [14,15]. These studies should be viewed as preliminary at this (early) stage of research development. “Other recent investigations have begun to link AS to post-traumatic stress [16] and other forms of psychopathology (e.g., substance use disorders[17,18]).” These emerging areas of study suggest that fears of anxious arousal and its consequences may be relevant to multiple aspects of mental health. Although not yet definitively established, this work has benefited by linking specific, lower-order facets of AS (i.e., physical, psychological and social concerns) to psychopathology. Studies have identified correspondence of specific components of AS that match or map onto a core defining feature of a given disorder (e.g., physical concerns and panic, cognitive dyscontrol and depression [19]).

Owing to the theoretical promise of targeting AS using intervention strategies, other research has begun to test whether, and to what extent, this construct can change or be changed. Theoretical models of AS predict that this construct can be altered via intervention [2]. Although originally assumed to be largely stable, meaningful variability in the construct has been observed over time [20]. Theoretically, learning to alter cognitive processes and beliefs regarding the perceived negative consequences of anxiety symptoms can be achieved through cognitive (e.g., cognitive restructuring), behavioral (e.g., interoceptive exposure) and pharmacological strategies [21]. There is empirical evidence that AS is indeed malleable [22]. For example, emerging data from prevention studies focused on AS suggest that this cognitive factor can, in fact, be targeted for systematic change among nonclinical individuals and, by extension, may be a useful therapeutic target for health promotion in addition to treatment [23].

Other research has focused on delineating the latent structure of the construct. The premise of this work is that by explicating its latent structure, theoretical models of AS can be refined leading to advances in other areas of research and clinical intervention. To date, there have been two principal approaches to understanding the latent structure of AS. The first type of latent structural research has involved factor analytic approaches. Research on adults across diverse populations suggests that AS, as indexed by the 16-item Anxiety Sensitivity Index (ASI [24]), the 36-item ASI-Revised (ASI-R [25]) and, more recently, the ASI-III [26], is multidimensional and hierarchical. Specifically, AS is composed of a global, higher-order factor with a number of lower-order facets among adults and youths, including physical, psychological and social concerns [27].

Taxometrics is the second approach more recently applied to the study of AS latent structure. Taxometrics entails a set of statistical procedures used to determine whether the latent structure of a construct is continuous or categorical (i.e., taxonic [28]). Such research has found that AS is taxonic with a relatively low base-rate taxon (approximately 9–15%). AS taxonicity has been observed across multiple measures and manifest indicators of the latent construct, and among multiple populations, including military cadets [29], adults from North America [30], adults from different regions of the world [31,32], youths [33] and across gender [34]. Initial studies have also demonstrated evidence of the incremental, predictive, discriminant and construct validity of the AS taxon with respect to anxiety problems (e.g., panic attacks and post-traumatic stress) and other panic-related vulnerability processes [16,29,30,31,35]. These taxometric studies indicate that AS may be composed of: a high base-rate form of AS putatively conceptualized as low risk or normative or adaptive; and a low base-rate AS taxon putatively conceptualized as high risk or maladaptive.

Taxonicity does not necessarily imply the absence of latent within-class quantitative gradations or hierarchical structure [36]. To help ensure that further theoretical development of AS validly reflects its latent structure and so is maximally clinically meaningful, both latent qualitative variability observed in taxometric research and quantitative variability observed in factor analytical research should be empirically and theoretically considered [32,36]. Two recent efforts have attempted to do so by sequentially applying taxometric and then confirmatory and exploratory factor analyses to the same data among youths [37] and adults [38]. In summary, these studies found that: AS is taxonic; the well-established hierarchical, multidimensional factor structure of AS fits the low-risk form of AS but not the AS taxon or high-risk form; and the AS taxon demonstrates a unique, previously undocumented latent continuous factor structure. Thus, the latent structure of AS may be more accurately conceptualized as taxonic latent class structure composed of two types or forms of AS, each of which is characterized by its own unique latent dimensional structure.

Future directions

Anxiety sensitivity & psychopathology vulnerability

Future work is needed to more definitively evaluate whether or not AS is related to the onset of panic disorder and agoraphobia, and how specific such effects are to these conditions relative to other clinical conditions. Another promising line of study within this broad vulnerability context will be to explore the relations between AS and post-traumatic stress specifically, as well as between AS, panic and post-traumatic stress more broadly. As noted, emerging theoretical and empirical work has documented associations between anxiety vulnerability, anxiety disorders and substance use problems [39]. It will thus be important for future research to delineate the relations between AS and substance use problems, as well as to explore the role(s) of AS in the treatment and prevention of substance use problems (e.g., tolerating withdrawal processes) and substance–anxiety comorbidity (e.g., nicotine dependence and panic disorder). Beyond linking AS to anxiety and other clinical conditions, clarifying the relations between this cognitive factor and other emotional vulnerability factors (e.g., self-regulation processes) will probably be formative in facilitating an understanding of underlying mechanisms of risk and resiliency.

Multifactorial risk models

The majority of studies focusing on AS and anxiety outcomes have involved tests of main effects. There is some limited evidence that in the presence of a number of other risk factors, AS moderates risk for anxiety symptoms and panic attacks [10,40]. However, this work is largely restricted to cross-sectional tests. Furthermore, there are currently few data on mediational processes related to AS and anxiety problems, although there is at least one study that suggests AS may mediate treatment outcome for panic disorder [41]. Thus, one important future direction will involve the increased sophistication of risk models involving AS and anxiety outcomes.

We believe that there are a number of important directions to pursue within this domain. First, future study could prospectively evaluate the moderating role of AS in relation to psychopathology processes and outcomes. Furthermore, research evaluating additive, interactive and moderational processes linking AS to other risk factors (e.g., emotional regulation factors and substance use problems) is likely to be important for increasing the construct validity of AS-related vulnerability process models.

Future study also may involve prospective evaluation of mediational processes that further explicate connection(s) between AS and panic, such as context insensitive and inflexible avoidance. Other research could importantly examine the mediating role(s) of AS in terms of the linkages between other risk factors and anxiety psychopathology.

Although theoretically and clinically promising, it is noteworthy that limited research has explored shared or common risk factors or processes linked to both AS and panic psychopathology.

In such efforts, research related to learning history, genetic heritability and gene–environment interactions may be particularly promising. In addition, efforts to explore AS as a common risk factor underlying co-occurrence and relations between panic, post-traumatic stress and substance-use problems may be another promising area for future study. Future work explicating these types of risk mechanisms will probably be central to the development of efficacious preventive interventions [42].

Latent structure

In light of emerging latent structural findings, two key directions may be useful for advancing latent structural research. Future research efforts that explore the latent structure of AS from the integrative conceptual perspective that concurrently considers multiple forms of variability (taxonic and dimensional) may be especially useful. Such efforts are likely to result in increased construct valid inferences with respect to the latent structure of AS and improved efforts to match the measurement model of AS to its latent structure. In doing so, it may be helpful for latent structural research to step beyond sequential application of coherent cut kinetic taxometric and factor analytical techniques and, instead, employ emerging structural equation mixture models that permit latent class and latent continuous variable parameter estimation simultaneously [43]. In addition, we believe that increased attention to the translation of knowledge about AS latent structure to clinical advances (e.g., assessment and targeted prevention) is a critical next step in this line of work [44].

Malleability

Empirical evidence from descriptive and intervention studies suggests that AS is malleable [22]. This work is important as it suggests that AS is, at least, a variable risk factor for certain anxiety and perhaps other types of clinical conditions and, moreover, that it can be usefully targeted for therapeutic change. However, evidence that indicates changing this factor will alter the risk for panic and related psychopathology from a preventative standpoint is lacking. Thus, it is not clear if AS represents a variable marker or variable causal risk factor for panic psychopathology [45]. To make this distinction, it is important for future research to examine changes in the construct prospectively following randomized, experimental (prevention oriented) manipulation. This type of research is of use as it may help establish an empirical foundation for using prevention strategies to target AS in order to prevent anxiety problems [42].

We also believe that there is promise in using latent structural findings to facilitate understanding of AS change processes. For example, it may be the case that categorical variability associated with AS taxonicity, as well as within-class continuous variability, are related to differential forms and/or levels of risk for anxiety outcomes. It therefore would be informative to consider taxonic and within-class dimensional properties of AS over the course of clinical prevention and intervention in an effort to better understand the nature of AS change processes. Such knowledge is clinically important for a number of reasons. For instance, such knowledge may guide the nonarbitrary operationalization of clinically significant change in AS in response to developmental events across the lifespan, as well as to prevention or intervention strategies.

Developmental psychopathology perspective

Future work may benefit from a more concerted effort to study AS within a developmental psychopathology conceptual and methodological framework. Research illuminating the putatively bidirectional relations between AS and anxiety problems over time may be particularly important for understanding etiology, maintenance and extinction of clinical levels of AS and anxiety psychopathology more generally. As one illustrative example, it will be important that future research evaluate whether or not, and how, AS shapes panic psychopathology and how the onset of panic problems reciprocally shape AS over time.

Affect tolerance & sensitivity

Finally, extant theory and empirical study has focused on identifying and studying AS as a unique variable and affect sensitivity process. Extant work on AS has not yet tested or developed theories regarding its likely relations with other affect tolerance and/or sensitivity processes. However, we believe that such knowledge may be essential for a comprehensive, contextualized understanding of AS within a more construct valid nomological network of emotional tolerance and sensitivity(ies). Moreover, little theoretical or empirical study has explicitly evaluated the relations between affect sensitivity and tolerance more generally. Such a study may have important clinical implications as these processes are at the heart of many third-wave behavior therapies and may have unique theoretical implications for integrating the study of AS with emotion regulation theory [46].