2.1.1. What is social anxiety disorder?

Social anxiety disorder (previously termed ‘social phobia’) was formally recognised as a separate phobic disorder in the mid-1960s (Marks & Gelder, 1965). The term ‘social anxiety disorder’ reflects current understanding, including in diagnostic manuals, and is used throughout the guideline. As set out in the International Classification of Diseases, 10th Revision (ICD-10) (World Health Organization, 1992) and in the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Text Revision (DSM-IV-TR) (American Psychiatric Association, 2000) social anxiety disorder is a persistent fear of one or more social situations where embarrassment may occur and the fear or anxiety is out of proportion to the actual threat posed by the social situation as determined by the person's cultural norms. Typical social situations can be grouped into those that involve interaction, observation and performance. These include meeting people including strangers, talking in meetings or in groups, starting conversations, talking to authority figures, working, eating or drinking while being observed, going to school, shopping, being seen in public, using public toilets and public performance including speaking. While anxiety about some of the above is common in the general population, people with social anxiety disorder can worry excessively about them and can do so for weeks in advance of an anticipated social situation. People with social anxiety disorder fear that they will say or do (involuntarily or otherwise) something that they think will be humiliating or embarrassing (such as blushing, sweating, shaking, looking anxious, or appearing boring, stupid or incompetent). Whenever possible, people with social anxiety disorder will attempt to avoid their most feared situations. However, this is not always feasible, and they will then endure the situation, often with feelings of intense distress. Usually the condition will cause significant impairment in social, occupational or other areas of functioning.

Children may manifest their anxiety somewhat differently from adults. As well as shrinking from interactions, they may be more likely to cry or ‘freeze’ or have behavioural outbursts such as tantrums. They may also be less likely to acknowledge that their fears are irrational when they are away from a social situation. Particular situations that can cause difficulty for socially anxious children and young people include participating in classroom activities, asking for help in class, activities with peers (such as team sports or attending parties and clubs), participating in school performances and negotiating social challenges.

2.1.2. How common is social anxiety disorder?

There are no UK epidemiological surveys that specifically report data on social anxiety disorder in adults; however, the prevalence of social anxiety disorder has been included in large general population surveys in other western European countries, the US and Australia. Prevalence estimates vary, with much of the variability probably being due to differences in the instruments used to ascertain diagnosis. However, it is clear that social anxiety disorder is one of the most common of all the anxiety disorders. Lifetime prevalence rates of up to 12% have been reported (Kessler et al., 2005a) compared with lifetime prevalence estimates for other anxiety disorders of 6% for generalised anxiety disorder, 5% for panic disorder, 7% for post-traumatic stress disorder (PTSD) and 2% for obsessive-compulsive disorder (OCD). Twelve-month prevalence rates as high as 7% have been reported for social anxiety disorder (Kessler et al., 2005b). Using strict criteria and face-to-face interviews in the US, the lifetime and yearly prevalence figures are halved to 5% and 3%, respectively (Grant et al., 2005b), but it is still more common than the major autoimmune conditions (rheumatoid arthritis, ulcerative colitis, Crohn's disease, systemic lupus erythematosus, diabetes mellitus type I, multiple sclerosis, uveitis, hypothyroidism and hyperthyroidism) put together (American Autoimmune Related Diseases Association, 2011). Data from the National Comorbidity Survey reveals that social anxiety disorder is the third most common psychiatric condition after major depression and alcohol dependence (Kessler et al., 2005a).

Women and men are equally likely to seek treatment for social anxiety disorder, but community surveys indicate that women are somewhat more likely to have the condition (Kessler et al., 2005a). Turk and colleagues (Turk et al., 1998) reported that in a clinical sample women feared more social situations and scored higher on a range of social anxiety measures. It therefore seems that although women are more likely to experience social anxiety, men may be more likely to seek treatment and to do so with less severe symptoms.

Population rates of social anxiety disorder in children and young people have been investigated in several countries. As in adult studies, a range of methods have been used for diagnosis, which probably explains the wide variability in prevalence estimates. A large New Zealand study reported that 11.1% of 18-year-olds met criteria for social anxiety disorder (Feehan et al., 1994). However, a large, British epidemiological survey (Ford et al., 2003) reported that just 0.32% of 5- to 15-year-olds had the disorder, a rate that was higher than that for PTSD, OCD and panic disorder, but lower than separation anxiety disorder, specific phobia and generalised anxiety disorder. Rates of diagnosis in this British study were higher in males than females, and increased slightly with age. A large US-based study reported very similar rates in 9- to 11-year-olds (Costello et al., 2003), while a German study estimated rates of 4% for 14- to 17-year-olds (Wittchen et al., 1999b).

2.1.3. When does social anxiety disorder start and how long does it last?

Social anxiety disorder typically starts in childhood or adolescence. Among individuals who seek treatment as adults the median age of onset is in the early to mid-teens with most people having developed the condition before they reach their 20s. However, there is a small subgroup of people who develop the condition in later life. Some people can identify a particular time when their social anxiety disorder started and may associate it with a particular event (for example, moving to a new school or being bullied or teased). Others may describe themselves as having always been shy and seeing their social anxiety disorder as a gradual, but marked, exacerbation of their apprehension when approaching or being approached by other people. Others may never be able to recall a time when they were free from social anxiety.

Several studies (Bruce et al., 2005; Reich et al., 1994a; Reich et al., 1994b) have followed-up adults with social anxiety disorder for extended periods of time. These studies have generally found that it is a naturally unremitting condition in the absence of treatment. For example, Bruce and colleagues (2005) reported a community study in which adults with various anxiety disorders were followed up for 12 years. At the start of the study, individuals had had social anxiety disorder for an average of 19 years. During the next 12 years 37% recovered, compared with 58% for GAD and 82% for panic disorder without agoraphobia.

Prospective longitudinal studies with children, although more sparse than those with adults, have confirmed that anxiety disorders are very likely to start by adolescence, and that this is particularly the case for social anxiety disorder. However, there is also evidence that some socially anxious young people will outgrow the condition (albeit still maintaining a high risk for other anxiety disorders) (Pine et al., 1998). Putting the adult and child prospective studies together, it appears that a significant number of people who develop social anxiety disorder in adolescence may recover before reaching adulthood. However, if social anxiety disorder has persisted into adulthood, the chance of recovery in the absence of treatment is modest when compared with other common mental disorders.

2.1.4. What other mental disorders tend to be associated with social anxiety disorder?

Four-fifths of adults with a primary diagnosis of social anxiety disorder will experience at least one other psychiatric disorder at sometime during their life (Magee et al., 1996). Among adults, social anxiety disorder is particularly likely to occur alongside other anxiety disorders (up to 70%), followed by any affective disorder (up to 65%), nicotine dependence (27%) and substance-use disorder (about 20%) (Fehm et al., 2008; Grant et al., 2005b). As social anxiety disorder has a particularly early age of onset, many of these comorbid conditions develop subsequently. It is of interest that comorbid anxiety predicts poorer treatment outcomes for people with bipolar affective disorder and major depressive disorder (Fava et al., 2008; Simon et al., 2004) and also that 25% of people presenting with first episode psychosis have social anxiety disorder (Michail & Birchwood, 2009), yet the relevance of this to clinical practice has been somewhat neglected. When people meet criteria for social anxiety disorder and another anxiety disorder, social anxiety comes first in 32% of people; in people with social anxiety and affective disorders or substance misuse, social anxiety precedes these comorbid conditions in 71% and 80%, respectively (Chartier et al., 2003); and in individuals who present with major depressive and social anxiety disorder, the depressive episode may be secondary. This may reflect a common aetiology or a despondency about the way in which social anxiety disorder prevents the person from realising their full potential, or it may be an indication of different peak incidence. One study of adult outpatients presenting for treatment for social anxiety disorder found that 53% had had a previous episode of a depressive disorder, with the average number of episodes being 2.2 in a cohort that had a mean age of 33 years. Similarly, substance misuse problems can develop out of individuals' initial attempts to manage their social anxiety with alcohol and drugs. Of course, the relationship between social anxiety disorder and other clinical conditions can also work the other way. For example, some individuals with scars and/or other physical problems in the context of PTSD may subsequently develop social anxiety disorder when they become concerned about how they will appear to other people. Some individuals who are usually socially confident may develop social anxiety during a depressive episode and recover once the depression lifts. The picture is similar in adolescence: comorbidity is 40% for anxiety disorders, 40% for affective disorders and 16% for substance misuse (Ranta et al., 2009); in one large German study of young people (aged up to 24 years) social anxiety preceded the additional anxiety diagnosis in 64.4% of people, the mood disorder diagnosis in 81.6% and the substance misuse diagnosis in 85.2% (Wittchen et al., 1999b).

There is also a significant degree of comorbidity between social anxiety disorder and some personality disorders. The most common is avoidant personality disorder (APD), with as much as 61% of adults who seek treatment for social anxiety also meeting criteria for a personality disorder (Sanderson et al., 1994). However, there is some controversy about the significance of this finding. There is a marked overlap between the criteria for social anxiety disorder and APD, and some experts consider APD a severe variant of social anxiety disorder. As many people develop their social anxiety disorder in childhood, some researchers have argued that much of the association with APD is simply due to the chronicity of the anxiety disorder. However, research studies have succeeded in identifying a few characteristics that tend to distinguish people with social anxiety disorder alone from those with social anxiety disorder plus APD. These include interpersonal problems, in particular problems with intimacy, increased functional impairment and lower levels of social support (Marques et al., 2012), although the differences have not always been replicated. Whatever the relationship between social anxiety disorder and APD, there is some evidence that successful psychological treatment of social anxiety also reduces the incidence of APD (Clark et al., 2006; McManus et al., 2009a). Similarly, Fahlen (1995) reported that abnormal personality traits wane with successful pharmacological treatment. Besides APD, comorbidity rates with other personality disorders are low and not higher than with other anxiety disorders or depression.

Among children and young people, comorbidity of anxiety disorders is also very high, as is comorbidity between anxiety and mood and behavioural disorders (Ford et al., 2003). The specific comorbidities of social anxiety in this age group are less well explored, but in a large sample of young people (aged 14 to 24 years) Wittchen and colleagues (1999b) found that 41.3% of those with a diagnosis of social anxiety disorder also had a diagnosis of substance misuse (including nicotine), 31.1% a mood disorder and 49.9% another anxiety disorder (compared with 27.9%, 12.1% and 20.8% of participants without a diagnosis of social anxiety disorder, respectively). Social anxiety is a substantial predictor of nicotine use in adolescence (Sonntag et al., 2000). In some people, social anxiety may be expressed as selective mutism (Viana et al., 2009).

2.1.5. How does social anxiety disorder interfere with people's lives?

Social anxiety disorder should not be confused with normal shyness, which is not associated with disability and interference with most areas of life. Educational achievement can be undermined, with individuals having a heightened risk of leaving school early and obtaining poorer qualifications (Van Ameringen et al., 2003). One study (Katzelnick et al., 2001) found that people with generalised social anxiety disorder had wages that were 10% lower than the non-clinical population. Naturally, social life is impaired. On average, individuals with social anxiety disorder have fewer friends and have more difficulty getting on with friends (Whisman et al., 2000). They are less likely to marry, are more likely to divorce and are less likely to have children (Wittchen et al., 1999a). Social fears can also interfere with a broad range of everyday activities, such as visiting shops, buying clothes, having a haircut and using the telephone. The majority of people with social anxiety disorder are employed; however, they report taking more days off work and being less productive because of their symptoms (Stein et al., 1999b). People may avoid or leave jobs that involve giving presentations or performances. The proportion of people who are in receipt of state benefits is 2.5 times higher than the rate for the general adult population. Katzelnick and colleagues (2001) also report that social anxiety disorder is associated with outpatient medical visits.

2.1.6. Are there different types of social anxiety disorder?

Individuals with social anxiety disorder vary considerably in the number and type of social situations that they fear and in the number and range of their feared outcomes. These two features (feared situations and feared outcomes) can vary independently. For example, some people fear just one or two situations but have multiple feared outcomes (such as, ‘I'll sound boring’, ‘I'll sweat’, ‘I'll appear incompetent’, ‘I'll blush’, ‘I'll sound stupid’ or ‘I'll look anxious’). Others can fear many situations but have only one feared outcome (such as ‘I'll blush’). Because of this variability, researchers have considered whether it might be useful to divide social anxiety disorder into subtypes. Several subtypes have been suggested, some of which are defined by specific feared outcomes (fear of blushing, fear of sweating and so on). The most common distinction is between generalised social anxiety disorder, where individuals fear most social situations, and non-generalised social anxiety disorder, where individuals fear a more limited range of situations (which often, but not always, involve performance tasks such a public speaking); however, some authors have suggested that the difference between these subtypes is a difference in degree. The generalised subtype is associated with greater impairment and higher rates of comorbidity with other mental disorders (Kessler et al., 1998). The generalised subtype also has a stronger familial aggregation, an earlier age of onset and a more chronic course. While most psychological therapies are applied to both subtypes, evaluations of drug treatments have mainly focused on generalised social anxiety disorder.

2.2.1. What do we know about the causes of social anxiety disorder?

As with many disorders of mental health, the development of social anxiety disorder is probably best understood as an interaction between several different biopsychosocial factors (Tillfors, 2004).

Genetic factors seem to play a part, but genes may influence the probability of developing any anxiety or depressive disorder rather than developing social anxiety in particular. Higher rates of social anxiety disorder are reported in relatives of people with the condition than in relatives of people without the condition, and this effect is stronger for the generalised subtype (Stein et al., 1998a). Further evidence for a genetic component comes from twin studies. Kendler and colleagues (1992; 1999) found that if one twin is affected, the chance of the other twin being affected is higher if the twins are genetically identical (monozygotic) than if they only share 50% of their genes (dizygotic). However, heritability estimates are only 25 to 50%, indicating that environmental factors also have an important role in the development of the condition for many people.

Stressful social events in early life (for example, being bullied, familial abuse, public embarrassment or one's mind going blank during a public performance) are commonly reported by people with social anxiety disorder (Erwin et al., 2006). Parental modelling of fear and avoidance in social situations plus an overprotective parenting style have both been linked to the development of the condition in some studies (Lieb et al., 2000).

The success of selective serotonin reuptake inhibitors (SSRIs), serotonin and noradrenaline reuptake inhibitors (SNRI) and monoamine oxidase inhibitors (MAOIs) in treating social anxiety disorder suggests that dysregulation of the serotonin and dopamine neurotransmitter system may also play a role, but studies that establish a causal relationship for such dysregulation in the development of the condition have not yet been reported.

Neuroimaging studies so far suggest different activation of specific parts of the brain (the amygdalae, the insulae and the dorsal anterior cingulate – all structures that are involved in the regulation of anxiety) when threatening stimuli are presented compared with healthy volunteers.

2.3.1. How well is social anxiety disorder recognised?

Recognition of social anxiety disorder in adults, children and young people by general practitioners (GPs) is often poor. The problem of under-recognition for anxiety disorders in general has recently been highlighted by evidence that the prevalence of PTSD is significantly under-recognised in primary care (Ehlers et al., 2009). In part this may stem from GPs not identifying the disorder, a general lack of understanding about its severity and complexity, and a lack of clearly defined care pathways. But it may also stem from service users' lack of knowledge of its existence, their avoidance of talking about the problem and stigma.

The early age of onset and effects on educational achievement mean that recognition of social anxiety disorders in educational settings is also an issue. As well as underachieving, children with social anxiety disorder may be particularly likely to be the targets of bullying and teasing. Teachers and other educational professionals may have limited knowledge of how to recognise and oversee the management of the condition.

In primary care many service users report being misdiagnosed as having ‘pure’ major depression. Missing the diagnosis may also occur in secondary care if an adequate history has not been taken. This is a serious omission because having a comorbidity has treatment and outcome implications.

2.3.2. How many people seek treatment?

Despite the extent of suffering and impairment, only about half of adults with the disorder ever seek treatment, and those who do generally only seek treatment after 15 to 20 years of symptoms (Grant et al., 2005a). Likely explanations for low rates and delays include individuals thinking that social anxiety is part of their personality and cannot be changed (or in the case of children, that they will grow out of it), lack of recognition of the condition by healthcare professionals, stigmatisation of mental health services, fear of being negatively evaluated by a healthcare professional, general lack of information about the availability of effective treatments and limited availability of services in many areas.

2.3.3. How can we know whether a treatment is effective?

Randomised controlled trials (RCTs) are the main way of determining whether a treatment is effective. Individuals who are diagnosed with social anxiety disorder are randomly allocated to the treatments under investigation or a control condition. Assessments are conducted at pre-treatment/control and post-treatment/control. The treatment is considered to be effective if significantly greater improvement is observed in the treatment condition than the control condition. In order to determine whether the improvements obtained by treatment are sustained, ideally participants should be systematically followed up for an extended period after the end of treatment.

RCTs are the best way of dealing with threats to internal validity (for example, ‘are the improvements that are observed due to the treatment or would they have happened in any case?’). However, they do not necessarily deal well with threats to external validity (for example, ‘would the results that are obtained with the rather selective group of participants that were studied in the RCT generalise to most people with social anxiety disorder?’). For this reason, it is helpful if data from RCTs are supplemented by data from large cohorts of relatively unselected people who receive the same treatment.

Researchers have traditionally distinguished between specific and non-specific treatment effects. The specific treatment effect refers to the amount of improvement that is attributable to the unique features of a particular treatment. The non-specific treatment effect refers to the amount of improvement that is attributable to features that are common to all (or most) well-conducted therapies.

In RCTs of pharmacological interventions the main contrast is always between the active drug and a placebo. The placebo controls for the non-specific effects of seeing a competent clinician, having one's symptoms consistently monitored, receiving a plausible treatment rationale and taking a tablet. The comparison between active drug and placebo is therefore only an index of the specific treatment effect attributable to a particular chemical. As most chemicals have side effects, some of which are severe, it is generally accepted that a drug must show a specific effect in order to warrant its use. However, it is important to note that service users are likely to show substantially greater improvements than implied by the active drug versus placebo effect size because giving a placebo also produces a further non-specific benefit.

In RCTs of psychological interventions the focus is less exclusively on establishing specific treatment effects. Commonly the control condition is a waitlist. In this case, the observed difference between the treatment and the control condition will be the sum of the relevant non-specific and specific effects. As psychological interventions are generally thought to have few side effects, it seems reasonable for researchers to have a primary interest in determining whether the treatment has any beneficial effects compared with no treatment. However, it is also important that evaluations of psychological interventions attempt to determine whether the treatment has specific effects as this gives us greater confidence in knowing exactly which procedures therapists should be taught in order to replicate the results that the treatment has obtained in RCTs. If a psychological intervention is known to have a specific effect, it is clear that therapists need to be trained to deliver the procedures that characterise that treatment. If a treatment has only been shown to have a non-specific effect people should be informed and it should not usually be offered in a publicly funded system.

In social anxiety disorder it seems highly plausible that part of the improvement that is observed in treatment is simply due to the non-specific effect of meeting someone who is (initially) a stranger while talking about one's emotions and numerous embarrassing topics. In other words, almost all interventions for social anxiety disorder involve a substantial amount of potentially beneficial exposure to feared social situations.

How does one determine whether a psychological intervention has a specific effect? Essentially one needs to demonstrate that the treatment is superior to an alternative treatment that includes most of the features that are common to various psychological interventions (such as seeing a warm and empathic therapist on a regular basis, having an opportunity to talk about one's problems, receiving encouragement to overcome the problems, receiving a treatment that seems to be based on a sensible rationale and having one's symptoms measured regularly). RCTs approach this requirement in one of three ways, each of which has strengths and weaknesses. In the first approach the alternative/control condition is a treatment that was specifically designed for the study and is intended to include non-specific features only, a good example of which is the education-support condition used by Heimberg and colleagues (1990; 1998). In the second approach, the alternative treatment might be something that is used routinely in clinical practice and is considered by some to be an active intervention but it turns out to be less effective than the psychological intervention under investigation, despite involving a similar amount of therapist contact. In the third approach, the psychological intervention is compared with pill placebo, which controls the many non-specific factors but often fails to fully control for therapist contact time because this is usually less in a medication-based treatment.

The fact that RCTs of medications almost always only focus on assessing specific treatment effects, whereas RCTs of psychological intervention may focus on assessing specific, non-specific or both types of effect, means that caution needs to be exercised when comparing the findings of such evaluations. In an ideal world, it should be possible to obtain an estimate of the effectiveness of each type of treatment against controls for specific effects as well as the overall benefit of treatment (compared with no treatment). The network meta-analysis (NMA) that underpins this guideline attempts to provide such information by inferring how medications would fair against no treatment even though most RCTs of medication use placebo controls and do not include a waitlist (no treatment) control (see Chapter 3 for further information about the NMA).

The next section outlines the different psychological and pharmacological interventions that have been tested for efficacy in social anxiety disorder.

2.3.4. Psychological interventions

In the mid 1960s, when social anxiety disorder was formally recognised as a separate phobic disorder (Marks & Gelder, 1965), the dominant evidence-based psychological interventions for anxiety disorders involved repeated exposure to the phobic stimulus in imagination. The first RCTs of psychological interventions for social anxiety disorder used two variants of this approach (systematic desensitisation and flooding) and obtained modest improvements. However, in anxiety disorders in general imaginal exposure treatment soon became superseded by treatments that involved confronting the feared stimulus in real life. Marks (1975) published a seminal review arguing that real life (‘in vivo’) exposure was more efficacious than imaginal exposure. This review had a substantial effect on treatment development work in all anxiety disorders. Subsequent behavioural and cognitive behavioural interventions for social anxiety disorder have therefore focused on techniques that involve real life confrontation with social situations, to a greater or lesser extent.

Exposure in vivo is based on the assumption that avoidance of feared situations is one of the primary maintaining factors for social anxiety. The treatment involves constructing a hierarchy of feared situations (from least to most feared) and encouraging the person to repeatedly expose themselves to the situations, starting with less fear-provoking situations and moving up to more difficult situations as confidence develops. Exposure exercises involve confrontation with real-life social situations through role plays and out of office exercises within therapy sessions and through systematic homework assignments. Many people with social anxiety disorder find that they cannot completely avoid feared social situations and they tend to try to cope by holding back (for example, by not talking about themselves, staying quiet or being on the edge of a group) or otherwise avoiding within the situation. For this reason, exposure therapists devote a considerable amount of time to identifying subtle, within-situation patterns of avoidance (safety-seeking behaviours) and encouraging the person to do the opposite during therapy.

Applied relaxation is a specialised form of relaxation training that aims to teach people how to be able to relax in common social situations. Starting with training in traditional progressive muscle relaxation, the treatment takes individuals through a series of steps that enables them to relax on cue in everyday situations. The final stage of the treatment involves intensive practice in using the relaxation techniques in real life social situations.

Social skills training is based on the assumption that people are anxious in social situations partly because they are deficient in their social behavioural repertoires and need to enhance these repertoires in order to behave successfully and realise positive outcomes in their interactions with others. The treatment involves systematic training in non-verbal social skills (for example, increased eye contact, friendly attentive posture, and so on) and verbal social skills (for example, how to start a conversation, how to give others positive feedback, how to ask questions that promote conversation, and so on). The skills that are identified with the therapist are usually repeatedly practiced through role plays in therapy sessions as well as in homework assignments. Research has generally failed to support the assumption that people with social anxiety disorder do not know how to behave in social situations. In particular, there is very little evidence that they show social skills deficits when they are not anxious. Any deficits in performance seem to be largely restricted to situations in which they are anxious, which suggests that they are an anxiety response rather than an indication of a lack of knowledge. Nevertheless, social skills therapists argue that practising relevant skills when anxious is a useful technique for promoting confidence in social situations.

Cognitive restructuring is a technique that is included in a variety of multicomponent therapies and has also occasionally been used on its own, although this has usually been as part of a research evaluation assessing the value of different components of a more complex intervention. The therapist works with the person to identify the key fearful thoughts that they experience in anxiety-provoking social situations, as well as some of the general beliefs about social interactions that might trigger those thoughts. The person is then taught largely verbal techniques for generating alternative, less anxiety-provoking thoughts (‘rational responses’), which they are encouraged to rehearse in anticipation of, and during, social interactions. To facilitate this process, they regularly complete thought records, which are discussed with therapists in the treatment sessions. Some practitioners argue that it is not essential that they fully believe a rational response before they start rehearsing it in fear-provoking situations (Marks, 1981).

Cognitive behavioural interventions encompass various well-recognised and manualised approaches including cognitive behavioural therapy (CBT). However, most cognitive behavioural interventions involve exposure in vivo and cognitive restructuring. Some programmes also include some training in relaxation techniques and/or social and conversational skills training. In recent years, research studies have identified several processes that appear to maintain social anxiety in addition to avoidance behaviour. These include self-focused attention, distorted self-imagery and the adverse effects of safety-seeking behaviours, including the way they change other people's behaviour. Some cognitive behavioural interventions have included techniques that aim to address these additional maintaining factors, for example, training in externally focused and/or task-focused attention, the use of video feedback to correct distorted self-imagery and demonstrations of the unhelpful consequences of safety-seeking behaviours. CBT can be delivered in either an individual or group format. When it is delivered in a group format, other members of the group are often recruited for role plays and exposure exercises. Sessions tend to last 2 to 2.5 hours with six to eight people in a group and two therapists. When CBT is delivered in an individual format, therapists may need to identify other individuals who can sometimes join therapy sessions for role plays.

Cognitive therapy (CT) developed by Clark and Wells (1995) is based on a model of the maintenance of social anxiety disorder that places particular emphasis on: (a) the negative beliefs that individuals with social anxiety hold about themselves and social interactions; (b) negative self-imagery; and (c) the problematic cognitive and behavioural processes that occur in social situations (self-focused attention, safety-seeking behaviours). A distinctive form of CT that specifically targets the maintenance factors specified in the model has developed. The procedures used in the treatment overlap with some of the procedures used in more recent CBT programmes, therefore CT can validly be considered to be a variant of CBT. However, it is distinguished from many CBT programmes for social anxiety disorder by the fact that it takes a somewhat different approach to exposure (with less emphasis on repetition and more on maximising disconfirmatory evidence) and it does not use thought records. Instead, the key components of treatment are: developing an individual version of Clark and Wells' (1995) model using the service user's own thoughts, images and behaviours; an experiential exercise in which self-focused attention and safety behaviours are manipulated in order to demonstrate their adverse effects; video and still photography feedback to correct distorted negative self-images; training in externally focused non-evaluative attention; behavioural experiments in which the person tests specific predictions about what will happen in social situations when they drop their safety behaviours; discrimination training and memory rescripting for dealing with memories of past social trauma.

The treatment is usually delivered on an individual basis. However, there is a need for the therapist to be able to call on other people to participate in within-session role plays. It is common for the therapist and the person with social anxiety disorder to also leave the office to conduct behavioural experiments in the real world during therapy sessions. This is easier to do if sessions are for 90 minutes, rather than the usual 50 minutes.

Interpersonal psychotherapy (IPT) was originally developed as a treatment for depression but was modified by Lipsitz and colleagues (1997) for use in social anxiety disorder. Treatment is framed within a broad biopsychosocial perspective in which temperamental predisposition interacts with early and later life experiences to initiate and maintain social anxiety disorder. There are three phases to the treatment. In the first phase, the person is encouraged to see social anxiety disorder as an illness that has to be coped with, rather than as a sign of weakness or deficiency. In the second phase, the therapist works with the person to address specific interpersonal problems particularly in the areas of role transition and role disputes, but sometimes also grief. Role plays encouraging the expression of feelings and accurate communication are emphasised. People are also encouraged to build a social network comprising close and trusting relationships. In the last phase, the therapist and the person review progress, address ending of the therapeutic relationship, and prepare for challenging situations and experiences in the future. Sessions are typically 50 to 60 minutes of individual treatment.

Psychodynamic psychotherapy sees the symptoms of social anxiety disorder as the result of core relationship conflicts predominately based on early experience. Therapy aims to help the person become aware of the link between conflicts and symptoms. The therapeutic relationship is a central vehicle for insight and change. Expressive interventions relate the symptoms of social anxiety disorder to the person's underlying core conflictual relationship theme. Leichsenring and colleagues (2009a) consider that in social anxiety disorder the core conflictual relationship theme consists of three components: (1) a wish (for example, ‘I wish to be affirmed by others’); (2) an anticipated response from others (for example, ‘others will humiliate me’); and (3) a response from the self (for example, ‘I am afraid of exposing myself’). Supportive interventions include suggestion, reassurance and encouragement. Clients are encouraged to expose themselves to feared social situations outside therapy sessions. Self-affirming inner dialogues are also encouraged.

Mindfulness training is a psychological intervention that has developed out of the Buddhist tradition and encourages individuals to gain psychological distance from their worries and negative emotions, seeing them as an observer, rather than being engrossed with them. Treatment starts with general education about stress and social anxiety. Participants then attend weekly groups in which they are taught meditation techniques. Formal meditation practice for at least 30 minutes per day using audiotapes for guidance is also encouraged.

2.3.5. Pharmacological interventions

Several different pharmacological interventions have been used in the treatment of social anxiety, many of which were originally developed as antidepressants. Antidepressants used in the treatment of social anxiety disorder come from four different classes: SSRIs, SNRIs, noradrenaline and selective serotonin antagonists and MAOIs. A fifth class, tricyclic antidepressants (TCAs), have also been used in the past but this is no longer the case.

SSRIs were initially marketed in the 1980s, having been developed as more selective agents following work on the TCAs and MAOIs. They are thought to act by increasing serotonin concentration in the brain and, after obtaining licences for major depression, many pharmaceutical companies carried out additional studies that indicated their efficacy in social anxiety disorder as well as in other anxiety disorders. The only SNRI that has been studied extensively is venlafaxine and it is possible that its effects in social anxiety disorder are mediated solely through changes in serotonin at usually prescribed doses.

MAOIs inhibit the breakdown of noradrenaline, dopamine, serotonin, melatonin, tyramine and phenylethylamine. This effect is not limited to the brain and affects other parts of the body rich in monoamine oxidase (MAO), for example, the gut. Therapeutic effects in social anxiety disorder are again thought to be related to increased levels of serotonin and dopamine in the brain. However, inhibition of MAO may result in a potentially dangerous interaction with foods containing tyramine which may lead to episodes of dangerously high blood pressure. This risk is much reduced with moclobemide as it is ‘reversible’ – this means that in the presence of other relevant substances, moclobemide ‘comes off the enzyme’. Because of this, moclobemide prescription comes with far fewer dietary restrictions than the older MAOIs, such as phenelzine. MAOIs are now rarely prescribed because of their perceived risks.

Benzodiazepines are restricted by the fact that it is preferable not to administer them for prolonged periods of time because of potential tolerance and dependence. In addition they may complicate some of the more prevalent comorbidities such as PTSD and depression.

Finally, alpha2delta calcium gated channel blockers, such as pregabalin, reduce neuronal excitability but it is not at all clear why these should work when other anticonvulsants have no known therapeutic effects in social anxiety disorder.