Behavioural and Cognitive Psychotherapy, 2005, 33, 177–188
Printed in the United Kingdom doi:10.1017/S1352465804001985
Safety-Seeking Behaviours: Fact or Function? How Can We
Clinically Differentiate Between Safety Behaviours
and Adaptive Coping Strategies Across Anxiety Disorders?
North Cumbria Mental Health and Learning Disabilities Trust, Carlisle, UK
Mark H. Freeston
Newcastle Cognitive and Behavioural Therapies Centre, UK
Abstract. Safety-seeking behaviours are seen as playing a key role in the maintenance
of various anxiety disorders. This article examines their role in panic disorder and social
phobia and suggests that, whilst there are clear theoretical differences between safety-seeking
behaviours and adaptive coping strategies, the difﬁcult issue in clinical practice is being able
to distinguish between the two. It builds on previous work by Salkovskis and colleagues
and provides a detailed discussion of the problems in distinguishing between safety-seeking
behaviours (direct avoidance, escape and subtle avoidance) and adaptive coping strategies in
clinical practice. The suggestion is made that topology can only be a guide to categorizing
the two types of responses and they can only be fully distinguished by taking into account the
intention of the individual and their perceived function to that individual in the speciﬁc context.
It is suggested that further analysis of the use of safety-seeking behaviours aimed at avoiding
a variety of outcomes at differing levels of catastrophe may provide useful information that
would clarify our understanding of the role of such behaviours in maintaining anxiety disorders.
Keywords: Safety behaviours, safety-seeking behaviours, coping strategies, anxiety disorders,
panic disorder, social phobia.
Our anxious patients often arrive at the ﬁrst session and describe disabling anxiety permeating
throughout their lives. Whole lifestyles can be constructed around speciﬁc fears. A recent
clinical example was a 29-year-old woman who had avoided virtually any activity without
a trusted adult present for over 13 years due to fears of having a heart attack. She had
never experienced her feared outcome throughout the history of her disorder. However, at
assessment she reported that at times during the previous week she had believed more strongly
than ever that she had to keep performing speciﬁc behaviours in order to prevent a heart attack.
Current cognitive models of anxiety disorders emphasize the role of such behaviours, labelled
Reprint requests to Richard Thwaites, Department of Psychological Services, 13 Portland Square, Carlisle, Cumbria,
CA1 1PT, UK. E-mail: email@example.com
© 2005 British Association for Behavioural and Cognitive Psychotherapies
178 R. Thwaites and M. H. Freeston
“safety behaviours” or “safety-seeking behaviours” (Salkovskis, 1991), in reducing short-term
anxiety, but more importantly in preventing longer-term cognitive change (Gelder, 1997). In
other words, these models propose that people continue to believe, for example, that they may
have a heart attack because they have prevented themselves from having the opportunity to
disconﬁrm their worst fears (Clark and Ehlers, 1993).
Safety-seeking behaviours are central in our understanding of the maintenance of anxiety
disorders and, as such, many current treatments concentrate on gradually dismantling these
behaviours and helping patients design opportunities to test out their beliefs (Clark, 1999).
This is in contrast to earlier treatments, which emphasized the teaching of “appropriate” skills
and strategies to patients to control their anxiety. Indeed some effective current treatments for
panic disorder include elements (e.g. breathing control) that could, under some circumstances,
function as safety behaviours for some individuals (e.g. Craske and Barlow, 2001). Salkovskis
(1996) has noted that some speciﬁc responses to anxiety (e.g. distraction) have been viewed
both as helpful components of behavioural anxiety management programs and also as problem
behaviours that have to be removed in order to allow effective exposure or belief testing to
Many patients also have idiosyncratic strategies that they have learned from either mental
health professionals (e.g. distraction techniques, breathing control) or self-help materials, or
have developed themselves. This is not surprising, given that the majority of us have our own
coping strategies that we deploy to reduce our anxieties in different situations. For example, one
can imagine an inexperienced public speaker who experiences some anxiety might adaptively
spend considerable time preparing for a presentation as a way of coping. In a socially phobic
individual, the same behaviour could function as a safety-seeking behaviour.
How can we recognize when a given behaviour is a safety behaviour and when it is an
adaptive coping strategy? Rather than considering these as dichotomous behaviours, however,
perhaps the same behaviour could function, for any given person, both as an adaptive coping
strategy and as a safety behaviour, but to differing degrees and in different contexts. With
repeated, excessive, or situationally inappropriate use, it is possible that the behaviour shifts
along a continuum from adaptive coping strategy to safety behaviour, depending upon the
intention, actual function and objective beneﬁt to the individual.
Given the proposed role of safety behaviours in maintaining various anxiety disorders, it
becomes important to distinguish between safety-seeking behaviours (that would need to be
gradually dropped) and adaptive coping strategies (that would continue to help the individual
with no long term cost). This article examines current theoretical understanding of the two and
highlights ways that aid in their discrimination and may increase the effectiveness of CBT at
an individual level.
Before attempting to distinguish between coping strategies and safety behaviours, clear
deﬁnitions are required.
Safety behaviours or safety-seeking behaviours
Early behavioural attempts to discuss avoidance concentrated on not entering or escaping from
situations (e.g. Marks, 1987). Safety signal theory was extensively developed and applied to
Safety seeking behaviours 179
agoraphobia by Rachman during the 1980s (e.g. Rachman, 1984a, 1984b) and can be viewed
as a precursor of the current concept of safety behaviours in cognitive models developed in
the late 1980s and early1990s. By 1999, the Oxford Group described safety behaviours in
panic as those that: “are intended to avoid disaster, and these responses have the secondary
effect of preventing the disconﬁrmation that would otherwise take place” (Salkovskis, Clark,
Hackmann, Wells and Gelder, 1999, p. 573).
Note that safety behaviours are here deﬁned with reference to both their “intended purpose
and their consequences”. Salkovskis, Clark and Gelder (1996) identiﬁed three main types
of safety behaviour in panic (though these classiﬁcations can also be applied to other
anxiety disorders): “direct avoidance” of situations; “escape” from situations; and, “subtle
avoidance” within the anxiety-provoking situation. Clark (1999) highlights that, although
labelled “behaviours”, there are numerous examples where safety behaviours are internal
processes or cognitive strategies (e.g. using distraction when panicking, cognitive rehearsals
of conversation in social phobia).
There have been several attempts to create typologies of safety behaviours using factor
analytic approaches (e.g. Kamphuis and Telch, 1998; Hughes, Budd and Greenaway, 1999).
The derived factors can be approximately mapped onto the smaller number of categories
described by Salkovskis et al. (1996) but appear to be based on a combination of function
and topology and, as such, may be clinically useful but less valid at a theoretical level. For
example, relaxation as described by Kamphuis and Telch (1998) could represent both subtle
avoidance or an adaptive coping strategy, depending on its idiosyncratic function for the
individual. Furthermore, without detailed information about an individual, would avoiding
caffeine (avoiding somatic arousal) be an adaptive behaviour or a safety behaviour? This
illustrates the difﬁculty in trying to categorize responses to anxiety based upon a topological
description of the behaviour rather than upon an idiosyncratic understanding of its function
for the individual.
Clinically, it has long been observed that it is the “availability” of safety behaviours rather
than their “usage” that limits new learning and maintains threat beliefs (e.g. carrying anxiolytic
medication without using it). It has been suggested that patients misattribute the lack of
a feared catastrophe to the safety behaviour (Salkovskis, 1991). Recently, however, it has
been suggested that the availability of safety behaviours for patients with claustrophobia
“interferes with treatment by redirecting patients’ attentional resources away from the
threat, thereby reducing the processing of threat-relevant information” (Powers, Smits and
Telch, 2004, p. 448). This additional mechanism requires further experimental investigation
across anxiety disorders. Finally, recent thinking suggests that positive cognitions in panic
mediate the relationship between bodily sensations and anxiety (Casey, Oei and Newcombe
2004). Research is needed to investigate whether safety-seeking behaviours inﬂuence positive
cognitions (e.g. perceived self-efﬁcacy) as well as threat-related cognitions.
Adaptive coping strategies
Although there appears to be widespread agreement on what constitutes a safety behaviour at
a theoretical level, it is more difﬁcult to deﬁne an adaptive coping strategy. There is a notion
that it is something that individuals do in order to reduce anxiety, and which does not maintain
or worsen future responses to the same stimulus or stimuli. In other words, in the same way
that safety behaviours may be deﬁned, at least to some extent, by their consequences, coping
180 R. Thwaites and M. H. Freeston
behaviours are at least partially deﬁned by their longer term impact or, more precisely, their
lack of negative impact. There is a danger that the deﬁnitions may become dependent upon
each other or tautological.
Davey, Burgess and Rashes (1995) have distinguished between avoidance-based coping
(which would be included within safety-seeking behaviours), coping by excessive monitoring
for information (which appears similar to some of the safety behaviours observed in anxiety
disorders), and coping strategies that allow the individual to change the meaning of a threat.
In a similar manner, Hughes et al. (1999) included a category labelled Effective Coping in
response to anxiety and panic. This included such strategies as “I tell myself that I can cope
with the anxiety” and “I try not to think about how I am feeling”, which appears to combine
elements of “effective coping” and subtle avoidance.
To summarize, safety behaviours seek to “prevent or minimize a feared catastrophe” (Clark,
1999, p. 7), whereas adaptive coping strategies seek to reduce anxiety but do not seek to
prevent an “imagined” catastrophe and therefore do not prevent disconﬁrmation of unhelpful
Whilst the above distinction could be seen as somewhat arbitrary to the non-cognitive
therapist, it is of critical clinical signiﬁcance. Initial evidence suggests that reducing safety-
seeking behaviours can increase the effectiveness of therapeutic interventions (e.g. Morgan
and Rafﬂe, 1999; Salkovskis et al., 1999). However, it may raise ethical issues to encourage
patients to reduce helpful coping safety behaviours that are not having deleterious effects in the
long term. Thus, from a clinical perspective, it is essential to be able to discriminate between
safety behaviours and coping. This paper examines exactly how to distinguish between them
in the context of therapy for panic disorder and social phobia.
Distinguishing between safety behaviours and helpful coping strategies
There are three key dimensions. First, the “topology” of the behaviour; what does the behaviour
look like? Is there something inherent in some behaviours that cause them to operate as safety
behaviours rather than helpful coping strategies? Does this vary according to context? Second,
what is the “intention” behind the behaviour? Does the behaviour become a safety behaviour
due to the individual’s purpose behind the act? Third, what are the “consequences of the
behaviour”? Do the consequences allow us to distinguish between them?
A cognitive model of panic
Despite Rachman’s call for experimental analyses of the strength of safety signal during the
1980s (Rachman, 1984a), the ﬁrst studies of the current cognitive concept of safety behaviours
in panic were not carried out until the 1990s. One of the ﬁrst such studies found correlations
between subtle avoidance behaviours and speciﬁc panic cognitions (Salkovskis et al., 1996).
For example, cognitions related to fainting were associated with holding onto both people
and objects. Cognitions about losing control and acting foolishly were associated with efforts
to keep control, moving slowly and looking for an escape route. Supporting this, Kamphuis
and Telch (1998) similarly found speciﬁc associations between anxiety cognitions and safety
Safety seeking behaviours 181
To date, there is limited evidence supporting the hypothesis that safety behaviours maintain
anxiety cognitions. Importantly, Salkovskis et al. (1999) found that even a 15-minute period
of exposure during which the panic patient attempted not to perform safety behaviours was
associated with signiﬁcantly greater reductions in anxiety and anxiety cognitions than a similar
15-minute period in which safety behaviours were continued. Although the sample size was
small, this study provides support for the central hypothesis that safety behaviours maintain
erroneous anxiety beliefs by preventing their disconﬁrmation. Finally, a recent study found that
the severity of panic attack was better predicted by fearful cognitions than by beliefs about
self-efﬁcacy and coping ability (Richards and Richardson, 2002). The authors concluded
that enhancing coping strategies would be less effective than reducing anxious cognitions,
consistent with the notion that some attempts to cope or to control panic are at best ineffective
and at worst unhelpful.
Direct avoidance behaviours
Some of the most obvious safety behaviours in panic are those that involve direct avoidance of
a situation or stimulus and there is an extensive literature examining the relationship between
panic and direct avoidance (see Craske and Barlow, 1988; Clum and Knowles, 1991). For
example, we commonly hear patients talk of avoiding particular places (e.g. city centre) or
of staying in particular places (e.g. at home in a “safety zone”). In addition, patients talk
about avoiding situations that have particular characteristics, including those that are busy
(e.g. city centre on Saturdays), hot (e.g. shopping centre) or those in which a rapid exit may
be impeded (e.g. cinema, theatre). Earlier accounts of avoidance stressed the role of factors
such as social demand and secondary gain (Craske and Barlow, 1988) as well as gender
and co-morbidity (Clum and Knowles, 1991). However, recent accounts have emphasized
misinterpreted stimuli and idiosyncratic catastrophic beliefs (Salkovskis et al., 1996). For
example, an individual who ﬁrst notices a sense of increased body temperature, which then
leads to a fear of passing out, is more likely to avoid hot environments than an individual
who notices dizziness and depersonalization, which then leads to fears of going mad. The
latter is perhaps more likely to avoid situations in which people may witness him going mad
or losing control. In panic disorder, it appears relatively unlikely that direct avoidance could
be an adaptive coping strategy. Complete avoidance of a situation to prevent a perceived
catastrophe will, at best, maintain the anxiety cognition or, at worst, strengthen it. Both the
intention behind the behaviour (prevent feared catastrophe from occurring) and the outcome
(maintain or strengthen belief) mark this type of avoidance as a safety behaviour. However,
to use the suggested criteria of Salkovskis et al. (1996), if the direct avoidance behaviour had
been intended to avoid the anxiety rather than the feared outcome, would this be classed as a
coping behaviour? Might an individual want to avoid the unpleasant effects of panic without an
associated catastrophic belief? Salkovskis et al. (1996) have suggested that responses intended
to avoid anxiety alone are potentially adaptive coping strategies. This clearly would not apply
to individuals who believed that anxiety itself could be harmful (Salkovskis, 1991).
Escape behaviour in panic disorder is similar to direct avoidance and is extremely common
amongst individuals who experience recurrent panic attacks. A clear example was exhibited
182 R. Thwaites and M. H. Freeston
by a recent patient recovering from severe panic disorder, who expressed surprise that having
reduced direct avoidance, he was still extremely anxious leaving his house, despite no recent
panic attacks. Investigation revealed that whilst in the city centre, he would begin to notice an
increased heart rate, which would cause him to feel even more anxious due to fears of having a
panic attack and the associated depersonalization and derealization. His catastrophic thoughts
were based on how “weird” he felt, and revolved around other people being able to see him
looking like a “weirdo” and then being vulnerable to assault. One safety behaviour was to
leave the situation and get home as quickly as possible. He was aware that he had no idea
what would actually happen if he stayed in town rather than heading for his “comfort zone”.
Clearly, examples such as ﬂeeing a situation fulﬁl the deﬁnition of a safety behaviour in that
it prevents a catastrophic outcome and prevents disconﬁrmation.
Depending upon the feared outcome, escape behaviour in panic can include leaving a
particular room, leaving a building or area (e.g. shopping centre) or having to return to a
perceived place of safety. It is difﬁcult to imagine an escape behaviour for an individual with
panic disorder that is not intended to avoid a perceived catastrophe, although there may be a
plausible alternative rationale that has social acceptability or face validity, such as discomfort
due to heat.
Distinguishing between adaptive coping strategies and safety behaviours can be more difﬁcult
in this category. The behaviours are often idiosyncratic, and so it becomes harder to identify
them and increasingly difﬁcult to determine their function. These behaviours include both
those that are planned in advance (e.g. carrying a bottle of water or anxiolytic medication) and
those performed once the anxiety is experienced (e.g. leaning against the wall, sitting down).
In theory, there is nothing inherent in the topology of any of these acts that automatically
identiﬁes them as safety behaviours. However, in practice, certain examples can be more
easily identiﬁed as safety behaviours based on the frequency with which they are reported by
panic patients. Examples of these behaviours include sitting near exits in pubs and restaurants,
only going into certain situations with a companion, and sitting down when legs feel weak.
However, most people are likely to have purposefully sat near an exit in a pub or restaurant
at some point in their lives, with a “commonsense” logic to this behaviour (e.g. waiting for
friends, preferring quiet), rather than to prevent a catastrophe such as getting trapped, having
a panic attack and dying. It is clear that “surface motivation” cannot be used to distinguish
safety behaviours and coping strategies. Two individuals may both choose to sit near the door
to avoid the heat and to get more fresh air. One may do this because he does not want to
get too hot or smell of cigarette smoke, whereas the other may perform the same behaviour
due to a wish to avoid breathlessness and choking to death. The complication arises when an
individual presents both of the previous rationales for sitting by the door. Sometimes patients
will justify their behaviours by reporting “rational” reasons for their actions (especially when
out of the situation and so less anxious). Safety behaviour can also become so engrained
for some individuals that they have difﬁculty identifying them (Kamphuis and Telch, 1998).
Again, the key questions are of the type “What was the worst thing that could have happened if
you had sat in the middle of the pub?” To access the anxiety-related rationale for the behaviour,
heightened levels of emotion may need to be evoked via cognitive exposure or actual exposure
to the stimulus.
Safety seeking behaviours 183
The context of the safety behaviour, whilst clearly associated with the intention, deserves
speciﬁc discussion. The same behaviour by the same individual across different contexts
could be both a safety behaviour and an adaptive coping strategy. For example, sitting down in
response to a novel physical sensation and associated fear of collapsing may be adaptive if the
stimulus is sufﬁciently different and/or is preventing a feared outcome that differs in important
ways. Therefore, in addition to the intention of the individual in performing the behaviour,
one must also consider whether the feared outcome is imagined, exaggerated or perhaps even
objectively realistic in this situation.
Finally, what then for an individual who has learned about the panic cycle (e.g. from self-
help material or from health professionals) and then visualizes the panic cycle in order to help
stay in an anxiety provoking situation? At what stage could visualization become a safety
behaviour? It may be argued that if the individual continues to do this on every such occasion,
it has indeed moved from coping to safety behaviour.
The cognitive model of social phobia
The model developed by Clark and colleagues (e.g. Clark, 1997; Wells, 1997) suggests that
an individual with social phobia holds dysfunctional beliefs about him/herself and, more
speciﬁcally, about him/herself in relation to social situations (Clark, 2001). As in other anxiety
disorders, safety-seeking behaviours are hypothesized to play a central role in maintaining
social phobia. People often believe that the feared social catastrophe would have occurred
had they not performed the safety behaviour. Again, the behaviour reduces the anxiety in the
short term but actually maintains the anxiogenic beliefs in the long term (Heimberg, 2002).
In social phobia, patients often engage in a multiplicity of safety behaviours attempting to
prevent feared outcomes at several levels. For example, an individual could perform safety
behaviours to prevent the feared outcomes of, ﬁrst, the face going red (e.g. keep cool, avoid
eye contact), second, other people noticing (e.g. wear high necks and make up, put hand over
face) and third, people who have noticed the redness thinking badly of her (e.g. provide an
alternative explanation of red face) (Clark, 2001).
As in other anxiety disorders, safety behaviours can lead to an accentuation of feared symp-
toms or “contaminate” the situation. For example, wearing a jacket to hide underarm sweating
can cause increased sweating (Clark, 2001), while excessive self-monitoring and attempting
to memorize what has been said can cause the individual to appear cold or disinterested. In
this case, an unfriendly or critical response from others may result (Clark and McManus,
2002). Certain safety behaviours can thus make the feared outcome objectively more likely.
Research has begun to empirically evaluate the role of safety behaviours in the maintenance
of social phobia. In a case series of eight socially phobic patients, Wells et al. (1995) found
that, as predicted, exposure plus decreased safety behaviours was signiﬁcantly better than
exposure alone in reducing anxiety and catastrophic beliefs. In a small sample of 14 patients,
Morgan and Rafﬂe (1999) found that patients in a CBT group treatment for social phobia
beneﬁted signiﬁcantly more when instructed to drop safety behaviours in addition to the
In a recent article discussing aspects of the Clark and Wells model of social anxiety, Hughes
(2002) suggested that “while ‘counter-productive’ safety behaviours should indeed, be given
184 R. Thwaites and M. H. Freeston
up, calming tactics (e.g. breathing control, postural practice) might have a valuable role and
need to be distinguished from undesirable safety behaviours” (p. 428). Although Hughes does
not discuss the practical difﬁculties of attempting to distinguish between the “calming tactics”
or the coping strategies that he advocates and safety behaviours, his suggestion is consistent
with previous theoretical discussions relating to panic (Salkovskis et al., 1996). The following
section will discuss various types of safety behaviours in social phobia and then suggest
that further work is required to guide clinicians in the differentiation between helpful coping
strategies (e.g. calming tactics) and subtle safety behaviours.
Direct avoidance behaviours
Direct avoidance has been hypothesized to play an important role in the maintenance of social
phobia. Wells (1997) refers to the limited “bandwidths” within which social phobics commonly
operate. For example, there may be long-standing patterns of avoiding speciﬁc “unsafe”
situations (e.g. talking in groups, eating in front of others) or certain “threatening” behaviours
(e.g. disagreeing with people, making a complaint). There is likely to be nothing inherent in the
topology of such behaviours that causes them to function as safety behaviours, but clinicians
can usually discriminate between “rational” avoidance (which may avoid objectively negative
outcomes and/or have no impact on unrealistic fears) and direct avoidance (which maintains
an unrealistic fear). Standard questions will usually sufﬁce (e.g. What do you believe/fear may
happen? How likely is this to occur? What if you hadn’t [performed behaviour]?).
However, it is likely that there are situations with some ambiguity and where the line
between rational avoidance of situations and direct but subtle avoidance becomes blurred.
Consider the recent example of a socially phobic individual who avoided phone calls using
a normal telephone at home and instead made phone calls from a speakerphone in his car
when driving. This made the call easier for him in that pauses were more acceptable. It is
further hypothesized that the distraction of concurrently driving also reduced his self-focus
and consequent anxiety. Indeed, the strategy of decreasing self-focus has been reported as
useful in reducing anxiety (Wells and Papageorgiou, 1998). The patient viewed this phone call
strategy as helpful, yet it may also be considered direct avoidance, a subtle safety behaviour,
and adaptive coping strategy to differing extents at the same time. Whilst, this allowed him to
actually make important phone calls, it was probably maintaining a number of negative beliefs
that may need addressing before the end of therapy.
Like direct avoidance of situations, escape behaviours are commonly reported by socially
phobic patients, and again they are generally unlikely to function as adaptive coping strategies.
Leaving a situation when both anxiety and the belief in a feared outcome is high, is likely to
cause the incident to be perceived as a near miss (e.g. “If I hadn’t got out of the that room then
I would have [x] and people would have thought [y]”). The topology of such acts does not
inherently identify them as safety behaviours, but it does provide an indication of their likely
function. Further, escaping from a situation is less ambiguous than some subtle behaviours.
Once again, the intention behind the behaviour and the function are crucial.
The majority of subtle safety behaviours in social phobia can be distinguished from adaptive
coping strategies on the basis of outcome. For instance, behaviours that actually contaminate
Safety seeking behaviours 185
the social situation and are obviously counter-productive are unlikely to be adaptive. For
example, for individuals who fear drinking in public due to shaking, gripping a glass tightly
is usually a counter-productive strategy. What about behaviours, however, that may have an
adaptive function in a speciﬁc situation but are used over-frequently or in too many situations?
For example, for an important phone call, a list of points that need to be discussed may be
extremely helpful. However, if the same behaviour is applied to all phone calls or conversations,
at what point does this move from adaptive strategy to subtle safety behaviour? The same
situation would apply for an individual that repeatedly practised a presentation. When would
adequate preparation become a safety behaviour? Salkovskis et al. (1996) have suggested that
this would depend on the intention of the individual (i.e. “what the person believes they are
avoiding” (p. 458). However, the related context, function and cost of the behaviour are also
important. At what point does the additional improvement to actual performance diminish to
the extent that the behaviour is now serving as reassurance rather leading to improvement?
For example, practising a presentation 10 times may be adaptive for someone who rarely
presents but less so for an experienced public speaker. Also, the novelty of the situation is
an important factor: if an individual were to over prepare for every public speaking situation
(e.g. regular teaching), this is more likely to be functioning as a safety behaviour. However,
if this strategy is used selectively, such as for new situations or when the “objective” stakes
are high (e.g. ﬁrst keynote address), this would suggest that it may be an adaptive coping
strategy. It is likely that through discussion, patient and therapist could reach agreement not
only on the intention of the behaviour, but also on the novelty of the situation, the objective
risk or likelihood of feared outcome, the cost and actual function and outcome, and thus
determine when an adaptive coping strategy would become a safety behaviour. What then
for the suggestion that patients should be taught “calming techniques” (e.g. Hughes, 2002)?
Whilst this is not inconsistent with the cognitive model, the therapist would need to be clear
how this would ﬁt an individual formulation to avoid teaching safety behaviours. It may be
difﬁcult to do within the group format as described by Hughes. What is the intention? Does
it aim to prevent the physical symptoms, to prevent others from noticing, or to change their
interpretation? Although most of the calming strategies advocated would aim to reduce anxiety
(adaptive), they may also prevent the individuals from disconﬁrming beliefs predicated on the
existence of these anxiety symptoms. This is a complex issue and further research would be
invaluable in determining how coping strategies actually inﬂuence the degree of belief in the
various idiosyncratic catastrophic outcomes.
This article has discussed the role of safety-seeking behaviours across two anxiety disorders
and attempted to clarify the difference between them and adaptive coping strategies. Although
at a theoretical level there are clear differences between them, it may be less clear in clinical
practice. The division by Salkovskis et al. (1996) of safety behaviours into direct avoidance,
escape behaviour and subtle avoidance has proved a useful framework in the examination
of safety behaviours across panic disorder and social phobia and has been applied to OCD
elsewhere (Thwaites and Freeston, in preparation).
Although each disorder may be characterized by an increased likelihood of speciﬁc types of
safety behaviours, there are a number of key clinical principles that apply across disorders. First,
the topology of a behaviour does not automatically distinguish between a safety behaviour and
186 R. Thwaites and M. H. Freeston
coping strategy. For experienced clinicians, topology may be used as a guide, but ultimately
each set of behaviours should be approached with an open mind.
Second, it is likely that safety behaviours and adaptive coping strategies can only be
distinguished on the basis of the intention of the individual, their perceived function to
that individual in the speciﬁc context, and the subsequent impact on positive and negative
cognitions. The clinician needs to engage the patient in a detailed discussion to identify
exactly what the patient is trying to avoid or prevent. If the behaviours have become habitual
or the patient is unable to access relevant cognitions, an appropriate task exposing to the
situation and/or the affect may be necessary. In the theoretically possible, but clinically rare,
case when the individual is trying to avoid anxiety alone without any avoidance of further
consequences, this behaviour could be considered adaptive (Salkovskis et al., 1996). However,
even in these cases the reduction in anxiety may be detrimental to change in that it can prevent
dysfunctional beliefs from being accessed and subsequently tested if access and level of belief
are functions of anxiety. This paper suggests that the distinction is made more complicated by
the fact that some behaviours can concurrently function as both adaptive coping strategies and
safety behaviours with respect to different feared consequences.
Third, Clark (2001) has emphasized that subtle avoidance behaviours in social phobia can
function at a variety of levels, all of which are aimed at avoiding a perceived catastrophic
outcome or outcomes, but at different stages in the process. This same detailed analysis could
be applied to behaviours in other disorders where there may be a range of possible catastrophes,
albeit some more “catastrophic” than others. An individual with panic disorder could perform
a variety of safety behaviours in order to prevent a feared outcome. For example, she could
use direct avoidance of busy shops to avoid the initial symptoms of anxiety (that she fears
could lead to going mad) but also to avoid the people who would see her losing control.
This is similar to panic patients who fear dying, but if the panic attack did not kill them they
would “die of embarrassment”. Although most deﬁnitions of safety behaviours specify that
the behaviour is to avoid a feared outcome rather than just the initial anxiety, it is possible
to categorize safety behaviours further, based on the inference chain leading to the perceived
catastrophic outcome. For example, safety behaviours in panic could be intended to avoid
the initial stimuli, the physical response to the initial stimuli, perceived consequences of the
physical response, the environmental reaction to physical response and so on. It is possible
that this type of categorization could provide clinically useful detail in planning experiments
but may be highly idiosyncratic.
A number of questions remain at both theoretical and clinical level. For example, to date
there is no research investigating how coping strategies inﬂuence levels of belief in catastrophic
outcomes. At a clinical level, are there still occasions when it is appropriate to encourage
clients to perform behaviours that have the potential to function as safety behaviours (e.g.
breath control, use of anxiolytic medication)? Does this purely depend upon the idiosyncratic
formulation of the individual and/or the stage of therapy?
Despite the emphasis on safety-seeking behaviours in current treatments of anxiety, we
are still in an early stage in our understanding of their role in maintaining anxiety and the
precise mechanisms involved. Current theory would suggest that clinicians should devote time
and attention to understanding the idiosyncratic function of patient behaviours in order to
discriminate between helpful (or harmless) coping strategies and safety behaviours that are
believed to hamper new learning by the patient and thus limit the beneﬁt received from CBT.
Safety seeking behaviours 187
Thanks to Shirley Platz for helpful suggestions, including the title. Grateful thanks to both
anonymous referees who provided helpful comments.
Casey, L. M., Oei, T. P. S. and Newcombe, P. A. (2004). An integrated cognitive model of panic
disorder: the role of positive and negative cognitions. Clinical Psychology Review, 24, 529–555.
Clark, D. M. (1997). Panic disorder and social phobia. In D. M. Clark and C. G. Fairburn (Eds.), Science
and Practice of Cognitive Behaviour Therapy (pp. 119–154). Oxford: Oxford University Press.
Clark, D. M. (1999). Anxiety disorders: why they persist and how to treat them. Behaviour Research
and Therapy, 37, S5–S27.
Clark, D. M. (2001). A cognitive perspective on social phobia. In W. R. Crozier and L. E. Alden (Eds.),
International Handbook of Social Anxiety: concepts, research and interventions relating to the self
and shyness (pp. 405–430). London: Wiley & Sons.
Clark, D. M. and Ehlers, A. (1993). An overview of the cognitive theory and treatment of panic disorder.
Applied and Preventive Psychology, 2, 131–139.
Clark, D. M. and McManus, F. (2002). Information processing in social phobia. Biological Psychiatry,
Clum, G. A. and Knowles, S. L. (1991). Why do some people with panic disorders become avoidant?
Areview.Clinical Psychology Review, 11, 295–313.
Craske, M. G. and Barlow, D. H. (1988). A review of the relationship between panic and avoidance.
Clinical Psychology Review, 8, 667–685.
Craske, M. G. and Barlow, D. H. (2001). Panic disorder and agoraphobia. In. D. H. Barlow (Ed.),
Clinical Handbook of Psychological Disorders (pp. 1–59). New York: Guilford Press.
Davey, G. C., Burgess, I. and Rashes, R. (1995). Coping strategies and phobias: the relationship
between fears, phobias and methods of coping with stressors. British Journal of Clinical Psychology,
Gelder, M. (1997). The scientiﬁc foundations of cognitive behaviour therapy. In David M. Clark and
Christopher G. Fairburn (Eds.), Science and Practice of Cognitive Behaviour Therapy (pp. 27–46).
Oxford: Oxford University Press.
Heimberg, R. G. (2002). Cognitive-behavioural therapy for social anxiety disorders: current status and
future directions. Biological Psychiatry, 51, 101–108.
Hughes, I. (2002). A cognitive therapy model of social anxiety problems: potential limits on its
effectiveness? Psychology and Psychotherapy: Theory, Research and Practice, 75, 411–435.
Hughes, I., Budd, R. and Greenaway, S. (1999). Coping with anxiety and panic: a factor analytic study.
British Journal of Clinical Psychology, 38, 295–304.
Kamphuis, J. H. and Telch, M. J. (1998). Assessment of strategies to manage or avoid perceived threats
among panic disorder patients: the Texas Safety Manoeuvre Scale (TSMS). Clinical Psychology and
Psychotherapy, 5, 177–186.
Marks, I. (1987). Fears, Phobia and Rituals. New York: Oxford University Press.
Morgan, H. and Rafﬂe, C. (1999). Does reducing safety behaviours improve treatment response in
patients with social phobia? Australian and New Zealand Journal of Psychiatry, 33, 503–510.
Powers, M. B., Smits, J. A. and Telch, M. J. (2004). Disentangling the effects of safety-behaviour
utilization and safety-behaviour availability during exposure-based treatment: a placebo-controlled
trial. Journal of Consulting and Clinical Psychology, 72, 448–454.
Rachman, S. (1984a). Agoraphobia: a safety signal perspective. Behaviour Research and Therapy, 22,
188 R. Thwaites and M. H. Freeston
Rachman, S. (1984b). The experimental analysis of agoraphobia. Behaviour Research and Therapy, 22,
Richards, J. C. and Richardson, V. (2002). The relative contributions of negative cognitions and
self-efﬁcacy to severity of panic attack in panic disorder. Behaviour Change, 19, 102–112.
Salkovskis, P. M. (1991). The importance of behaviour in the maintenance of anxiety and panic:
a cognitive account. Behavioural Psychotherapy, 19, 6–19.
Salkovskis, P. M. (1996). Avoidance behaviour is motivated by threat beliefs: a possible resolution of
the cognitive-behaviour debate. In P. M. Salkovskis (Ed.), Trends in Cognitive and Behavioural
Therapies (pp. 25–42). Chichester: Wiley.
Salkovskis, P. M., Clark, D. M. and Gelder, M. G. (1996). Cognitive-behaviour links in the
persistence of panic. Behaviour Research and Therapy, 34, 453–458.
Salkovskis, P. M., Clark, D. M., Hackmann, A., Wells, A. and Gelder, M. G. (1999). An exp-
erimental investigation of the role of safety-seeking behaviours in the maintenance of panic disorder
with agoraphobia. Behaviour Research and Therapy, 37, 559–574.
Wells, A. (1997). Cognitive Therapy of Anxiety Disorders: a practice manual and conceptual guide.
Wells, A., Clark, D. M., Salkovskis, P., Ludgate, J., Hackman, A. and Gelder, M. G. (1995).
Social phobia: the role of in-situation safety behaviours in maintaining anxiety and negative beliefs.
Behaviour Therapy, 26, 153–161.
Wells, A. and Papageorgiou, C. (1998). Social phobia: effects of external attention in anxiety, negative
beliefs, and perspective taking. Behavior Therapy, 29, 357–370.