Correlations between the offensive subtype of social anxiety disorder and personality disorders


Toshihiko Nagata, MD, PhD, Department of Neuropsychiatry, Osaka City University, Graduate School of Medicine, 1-4-3 Asahimachi, Abenoku, Osaka 545-8585, Japan. Email:


Aims:  Recent studies have revealed the possibility that the offensive subtype of social anxiety disorder (SAD) may no longer be a culture-bound syndrome; however, detailed clinical pictures have never been reported. This study investigated the differences between the offensive and non-offensive subtypes of SAD in terms of the background and axis I and II comorbidity.

Methods:  A total of 139 patients with SAD based on DSM-IV criteria were studied by conducting a semi-structured interview including the Structured Clinical Interview for DSM-IV axis I and II disorders, and the Liebowitz Social Anxiety Scale.

Results:  Fifty-two (37%) patients were classified with the offensive subtype. There were no significant differences in most demographic variables and axis I lifetime comorbidity between offensive and non-offensive subtype patients. On logistic regression analysis, offensive subtype patients showed a more frequent history of parental physical abuse, higher Liebowitz Social Anxiety Scale scores, and more frequently exhibited obsessive–compulsive personality disorders than non-offensive subtype patients.

Conclusion:  Yamashita (1977) reported that the majority of offensive subtype patients were doted on by their parents, although current offensive subtype patients are more likely to have had a troubled childhood, show severer forms of SAD, and more frequently exhibit an inflexible personality. This study suggested that the offensive subtype might not be essentially different from the non-offensive subtype (quantitative rather than qualitative).

SOCIAL ANXIETY DISORDER (SAD), characterized by an excessive fear of exposure to situations that involve potential scrutiny by others, is one of the most frequently occurring anxiety disorders. According to DSM-IV criteria,1 it can be subdivided into a specific subtype, such as fear of public speaking or eating in public, and a generalized subtype, in which most social situations lead to fear or avoidance. Recent studies have suggested that SAD is the most prevalent anxiety disorder in not only Western but also Asian countries.2

Conversely, long before the introduction and expanding (from just a specific into a generalized subtype) of the concept of SAD in Western countries, taijin-kyofusho (TKS) was considered a prototype of ‘Neurose’ in Japan by Morita Shoma (since the 1930s),3 and Kasahara4 and Yamashita5 focused on cases bordering between neurotic and psychotic. Morita considered TKS as the manifestation of a vicious cycle based on a ‘hypochondriacal’ temperament (a combination of neuroticism and obsessionality).3 Based on this hypochondriacal temperament, small (internal or external) events evoke oversensitivity and promote a more deviated, narrow attention toward small events (especially interoceptive awareness), and result in perpetuating a vicious cycle.3 Morita established Morita therapy to break such a vicious cycle.3 However, focuses of later researchers, such as Kasahara4 and Yamashita,5 moved to how severe (almost delusional) and paroxysmally favorable the outcome of the offensive subtype of TKS could be. Two subtypes of TKS were described: tension and offensive (or convinced) subtypes (for the diagnostic criteria of TKS, see Table 1).5–7 Although the tension subtype with ‘fear of being noticed’ resembles SAD, these Japanese psychiatrists concentrated on describing severe clinical pictures and contrasting them with the favorable outcome of the offensive subtype of TKS in the literature.4,5 They insisted that most Japanese psychiatrists believed that the offensive subtype of TKS differed from psychotic disorders, despite a superficial similarity. Along these lines, Takahashi8 introduced TKS as ‘an individual's intense fear that his or her body, its parts or functions, displease, embarrass, or are offensive to other people in appearance, odor, facial expressions, or movements’ to the West. At present, only the offensive subtype of TKS is known as TKS in the West (as seen in appendix of DSM-IV), as a culture-bound syndrome, and its nosological positioning remains unresolved in the DSM diagnostic system. SAD involves a fear of humiliating oneself, whereas in TKS (e.g. the offensive subtype of TKS), the focus of fear is offending others by embarrassing them or by making them uncomfortable because of a personal flaw or shortcoming (e.g. emitting an unpleasant body odor). This variation was thought to be explained by cultural differences: Western culture engenders an independent, self-construal governed by personal abilities, desires, and autonomy, whereas East-Asian cultures, with interdependent self-construals, emphasize the importance of relationships, agreeableness, conformity, cultural indentity, and modesty.9

Table 1.  A. Diagnostic criteria for taijin-kyofusho
  • In this study, a diagnosis of the offensive subtype required the patient to meet criterion B(2).

A. At least one of the following features:
(1) Fear of blushing in the presence of others
(2) Fear of stiffening of facial expression, of trembling of the head, hands, feet, or voice, of sweating while facing others
(3) Fear of physical deformities being noticed
(4) Fear of emitting body odors
(5) Fear of line-of-sight becoming uncontrollable
(6) Fear of uncontrollable flatus in the presence of others
B. Either of the following two, because of the above fear(s))
(1) Tension subtype: Fear of being looked at (noticed) by others
(2) Offensive subtype: Fear of offending or embarrassing others
C. At most points during the course of the disorder, the person recognizes that the fear is excessive or unreasonable.
D. The fear(s) interferes significantly with the person's normal routine, occupational (academic) functioning, or social activities or relationships, or there is marked distress about having the fear(s)
E. The symptoms must have been present for at least 1 year. In individuals under age 18 years, the duration should have been at least 6 months.
Table 1.  B. Modified diagnostic criteria for social anxiety disorder (SAD) based on DSM-IV (American Psychiatric Association, 1994)
  • Modified.

A. A marked and persistent fear of one or more social and performance situations in which the person is exposed to unfamiliar people or to possible scrutiny by others. The individual fears that he or she will act in a way (or show anxiety symptoms) that will be humiliating or embarrassing.
B. Exposure to the feared social situation almost invariably provokes anxiety, which may take the form of a situationally bound or predisposed panic attack.
C. At most points during the course of the disorder, the person recognizes that the fear is excessive or unreasonable. The person recognizes that the fear is excessive or unreasonable.
D. The feared social or performance situation are avoided or else are endured with intense anxiety or distress.
E. The avoidance, anxious anticipation, or distress in the feared social or performance situation(s) interferes significantly with the person's normal routine, occupational (academic) functioning, or social activities or relationships, or there is marked distress about having the phobia.
F. In individuals under age 18 years, the duration is at least 6 months.
G. The fear or avoidance is not due to the direct physiological effects of a substance (e.g., a drug of abuse, a medication) or a general medical condition and is not better accounted for by another mental disorder (e.g., panic disorder with or without agoraphobia, separation anxiety disorder, body dysmorphic disorder, a pervasive developmental disorder, or schizoid personality disorder).
H. If a general medical condition or another mental disorder is present, the fear in Criterion A is unrelated to it, e.g., the fear is not of stuttering, trembling in Parkinson's disease, or exhibiting abnormal eating behavior in anorexia nervosa or bulimia nervosa.

Although TKS has drawn little attention from Western audiences, recently, Choy et al.10 found that symptoms of TKS (offensive subtype of SAD) are not rare even in Western countries, in contrast to former views. They speculated that SAD patients in Western countries were not assessed regarding offensive subtype symptoms. We are of the opinion that the clinical presentation of neurotic disorder has altered in Japan due to the marked socioeconomic change caused by the economic depression and the subsequent change in industry and the family structure. Yamashita5 classified the TKS family into four groups: doted on (20%), moderately doted on (29%), average (46%), and abusive (5%), and reported that patients were very sensitive to criticism because they had never been rebuked by their parents. However, currently, we believe that many neurotic disorder patients in Japan had a troubled childhood. The Japanese Ministry of Health, Labour and Welfare reported that cases of parental child abuse reported to consultation centers markedly increased eightfold in a single decade (from 5352 to 40 639 between 1997 and 2007). In addition, in contrast to the previous view of former Japanese psychiatrists that SAD is more common in Japan than in Western countries because Japan has a high rate of TKS, recent studies have shown that the 12-month prevalence of SAD is 6.8% in the USA,11 whereas it is just 0.8% in Japan.2 However, the cohort effect of anxiety disorders in Japan is higher than in other countries.12

To the best of our knowledge, no previous study has investigated the differences between offensive and non-offensive subtypes among SAD patients in terms of the clinical background and axis I and II comorbidities. It is important to study childhood maltreatment and the presence of personality disorders in order to understand how the clinical features of the offensive subtype of SAD (and TKS) changed, although few previous studies focused on this issue. The aim of the current study was to investigate whether the offensive subtype of SAD essentially differs from the non-offensive subtype in terms of the clinical background and axis I comorbidities. In addition, we hypothesized that the clinical presentation of offensive subtype SAD patients was not essentially different from the non-offensive type (quantitative rather than qualitative); however there were some differences compared with 40 years ago.



Subjects consisted of a consecutive series of 168 patients with SAD according to DSM-IV1 who sought treatment for SAD (not for TKS). The diagnosis of SAD was based on DSM-IV criteria, and item C was expanded to ‘At most points during the course of disorder, the person recognizes that the fear is excessive or unreasonable’, to include patients with the offensive subtype in the current study. Item G was not modified because SAD and body dysmorphic disorder can be diagnosed in one patient.13 In addition, the diagnosis of TKS (SAD) offensive subtype was based on TKS diagnostic criteria (Table 1). This employed Takahashi's summary of the Japanese literature,8 as used in our previous study.6,7 Patients had to show a marked fear of offending others as a result of their imagined deficits. All patients provided written informed consent before entering the study.


All subjects underwent a direct (face-to-face) 2- to 3-h assessment conducted by T. N. or H. Y. on the third or fourth consultation. This assessment included a modified version of the anxiety section of the Structured Clinical Interview for DSM-IV (SCID-p),14 mood disorder and substance use disorder sections of the SCID-p DSM-III-R,15,16 and the Structured Clinical Interview for DSM-IIIR Personality Disorders17 because the official Japanese version of DSM-IV SCID-p and SCID-II had not yet been published at the start of this study. However, the avoidant and paranoid sections of the Structured Clinical Interview for DSM-IV Personality Disorders (SCID-II)18,19 were added. The Japanese version of the Liebowitz Social Anxiety Scale (LSAS)6,20 was employed later on in the study; thus, not all subjects completed the LSAS. Parental physical abuse was defined as a history of frequent physical punishment by caregivers, which at least caused bruising.21


Subjects without LSAS scores were excluded from analysis to control the severity of SAD. Fisher's exact test, the χ2-test, and independent t-test with Levene's test for the equality of variances were performed to compare demographic features. To reduce the risk of type I error, we used a more stringent significance level of <0.01 in comparisons of demographic variables. Logistic regression analysis was performed to compare axis-I and -II comorbidities, and the severity of SAD (LSAS score) was entered as a confounding factor. Finally, hierarchical logistic regression analysis was performed to predict the offensive subtype (spss for Windows version 11.0, Chicago, IL, USA).


A total of 168 patients sought treatment for SAD. Twenty-nine subjects had not undergone LSAS scoring. A total of 139 subjects were finally included in this study; 52 (37%) of the subjects met the criteria for the offensive subtype. No subjects required the expanding of SAD criterion C, contrary to our expectations. All subjects recognized that their fear was excessive or unreasonable, even regarding offensive symptoms.

As presented in Table 2, there were no significant differences in demographics, such as age, sex, years of education, Global Assessment of Function scores, marital status, employment status, or rate of suicide attempts, between offensive and non-offensive subtype cases. As expected, subjects with the offensive subtype showed significantly higher scores on the LSAS than the non-offensive subtype subjects. Contrasting with Yamashita's report, subjects with the offensive subtype were significantly more likely to have been physically abused by their parents.

Table 2.  Demographic background
nOffensive subtypeNon-offensive subtypeX2/t/ Fisher (P)
  • *

    P < 0.01.

  • Mean and SD.

  • Axis V of DSM-IV.

  • LSAS, Liebowitz Social Anxiety Scale; SAD, social anxiety disorder; TKS, taijin-kyofusho.

Age (years)27.4 (8.2)29.9 (11.4)1.4 (0.17)
Men (%)22 (42%)38 (44%)(1.0)
Length of education (years)13.4 (2.3)13.5 (2.6)0.2 (0.86)
Marital status   
 Single44 (85%)75 (86%)1.4 (0.49)
 Married4 (8%)9 (10%) 
 Divorced4 (8%)3 (4%) 
Employment status   
 Full-time20 (39%)47 (54%)3.2 (0.21)
 Part-time7 (14%)9 (10%) 
 Unemployed25 (48%)31 (36%) 
Global Assessment of Function53 (15)57 (15)1.4 (0.16)
Childhood parental loss5 (9%)8 (9%)(1.0)
Parental physical abuse11 (21%)*5 (6%)(0.01)
Onset of SAD (years)11.7 (4.1)13.4 (5.3)2.0 (0.05)
Generalized type51 (96%)76 (87%)(0.13)
Suicide attempt13 (25%)11 (13%)(0.07)
Onset of TKS (years) 17.8 (4.3) 
A1) Blushing20 (38%)*14 (16%)(0.004)
A2) Facial stiffening24 (45%)*21 (24%)(0.009)
A3) Physical deformity24 (45%)*16 (18%)(0.001)
A4) Odor11 (21%)*1 (1%)(<0.001)
A5) Eye contact22 (42%)*3 (3%)(<0.001)
A6) Flatus9 (17%)4 (5%)(0.02)
LSAS95.7 (25.4)*79.7 (24.0)3.7 (<0.001)

The period between the onset of SAD and the offensive subtype was 4.3 years (SD = 4.6, range: −4 to 17). The offensive symptoms preceded the SAD symptoms in only two patients, by 2 and 4 years, respectively. The SAD symptoms preceded in 37 subjects (71%), and the offensive and SAD symptoms coincided in 13 subjects (25%). As expected, subjects with the offensive subtype were significantly more likely to show TKS-related symptoms (all of A items in Table 1A), except for flatus.

As presented in Table 3, there were no significant differences in lifetime axis-I comorbidity between subjects with and without the offensive subtype, if LSAS scores were entered into logistic regression analysis. Conversely, only obsessive–compulsive personality disorder (OCPD) was significantly more prevalent in subjects with the offensive than non-offensive subtype. Paranoid personality disorder narrowly failed to reach significance (Table 4).

Table 3.  Lifetime axis I and II comorbidity
nOffensive subtypeNon-offensive subtypeOddsP
  • *

    P < 0.05.

  • n(%).

  • Comorbidity according to DSM-IIIR criteria.

Major depressive disorder27 (52%) 42 (48%)0.8 (0.4–1.8)0.66
Dysthymia11 (21%)11 (13%)1.9 (0.7–5.1)0.19
Bipolar I1 (2%)1 (1%)1.5 (0.1–24.7)0.77
Alcohol abuse2 (4%)3 (3%)2.8 (0.2–33.0)0.41
Alcohol dependence3 (6%)1 (1%)6.2 (0.6–66.1)0.13
Panic disorder6 (11%)1 (1%)8.2 (0.9–75.6)0.06
Specific phobia9 (17%)15 (17%)0.8 (0.3–2.1)0.65
Obsessive–compulsive disorder4 (8%)5 (6%)1.3 (0.3–5.5)0.69
Generalized anxiety disorder7 (14%)6 (7%)1.6 (0.5–5.6)0.47
Axis II personality disorders    
Paranoid14 (27%)9 (10%)2.6 (0.9–6.8)0.05
Schizotypal1 (2%)1 (1%)1.9 (0.1–34.9)0.66
Schizoid02 (2%) 
Borderline04 (8%) 
Histrionic1 (2%)2 (2%)0.8 (0.1–11.0)0.89
Narcissistic1 (2%)2 (2%)1.1 (0.1–14.3)0.95
Avoidant50 (96%)70 (81%)3.5 (0.7–16.9)0.12
Dependent14 (27%)11 (13%)2.3 (0.9–5.7)0.09
Obsessive–compulsive18 (35%)14 (16%)2.4* (1.0–5.8)0.04
Table 4.  Results of logistic regression analysis
 Offensive subtypeP-value
Odds (95% confidence interval)
  1. X2 of model = 30.2 (d.f. = 4), P ≤ 0.001, *P < 0.05.

  2. LSAS, Liebowitz Social Anxiety Scale.

Parental physical abuse6.0*(1.7–21.0)0.005
Avoidant personality4.9(0.7–28.9)0.06
Obsessive–compulsive personality3.5*(1.4–9.0)0.007

Employing hierarchical logistic regression analysis, the dependent variable was the presence of the offensive subtype, and independent variables were divided into three blocks. The first block included parental physical abuse in addition to the LSAS score (both were forced entry variables), because we identified a significant correlation with parental physical abuse, in contrast with Yamashita's reports. The second block included all axis I disorders. The third block included all personality disorders. The second and third blocks were entered with variable forward stepwise selection employing Wald statistics. The second block was not significant (χ2 was 9.5, d.f. = 9, P = 0.40). In addition to parental physical abuse and the LSAS score, only OCPD reached the level of significance.


To the best of our knowledge, this is the first study reporting the comorbidity of axis I and II disorders among Asian patients with SAD. Around half of the subjects showed a lifetime comorbidity of major depressive disorder, and 10–20% showed some type of anxiety disorder. Compared with previous reports from Western countries,22–24 the rate of major depressive disorder-associated comorbidity ranks among the highest, and that of anxiety-disorder-related comorbidity among the lowest. The differences in comorbidity of anxiety disorders can be understood by a lower base prevalence in the general Japanese population.2 The discrepancy regarding the 12-month prevalence of anxiety disorders (5.3 vs 18.2%, respectively) was greater than that for mood disorders (3.1 vs 9.6%, respectively) between Japan and the USA.25 Considering the lower prevalence of anxiety disorders in Japan, the comorbidity rate cannot increase any further (reaching a ceiling), even with a severer subtype of SAD indicated by high LSAS scores and early onset.

In addition, almost all of the subjects met the DSM-IV criteria of avoidant personality disorder. Recent studies showed that differences in clinical features, including the treatment response, between groups of patients with and without avoidant personality disorder have tended to be more quantitative than qualitative.26 Avoidant personality disorder is now conceptualized at the severe end of the spectrum.26

The prevalence of the offensive subtype of SAD among the general population, and even the clinical population, remains unknown. However, Choy et al.10 reported a similar rate of the offensive subtype among Koreans with SAD (25/64, 39%). In addition, Korean offensive subtype patients complained of similar symptoms (blushing: 24%; stiff facial expression: 44%; physical appearance: 16%; body odor: 28%; inability to make eye contact: 28%; and intestinal gas: 24%). They also reported that even US subjects showing a fear of offending others due to the above symptoms ranged from 15.5 to 39.2%. These results suggest that offensive symptoms and the prevalence of the offensive subtype among clinical subjects are very similar in Japanese, Korean, and US SAD subjects.

We found that OCPD significantly predicted the offensive subtype on logistic analysis (after controlling for SAD severity using LSAS scores). Although, in DSM-IV, personality disorders are classified into three clusters from A to C, a recent study showed that only OCPD among cluster C personality disorders is unique in terms of the cause (heritability and influences of the environment).27 An axis II/II diagnostic co-occurrence study showed an unexpectedly marked overlap between OCPD and paranoid personality disorders (up to 68%), which are theoretically at the opposite ends of the spectrum.28 More recently, Hummelen et al.29 reported that only the association between OCPD and paranoid personality disorder was significantly closer than expected among many other overlapping personality disorders. Also, in this study, paranoid personality disorder narrowly failed to reach significance, although paranoid personality disorder was not selected on variable selection. Paranoid personality disorder and OCPD share the feature of ‘unreasonable insistence that others submit to their wishes, indecisiveness, and lack of generosity (p.217)’.28 Hummelen et al.29 reported that OCPD had two components: one is perfectionism and the second is rigidity. This latter component, rigidity, offers a bridge of clinical pictures that share features of paranoid personality disorders. This rigid personality coincides with our clinical impressions of the offensive subtype, although the majority of those with the offensive subtype did not meet the diagnostic criterion of OCPD (just 35%).

The presence of parental physical abuse significantly predicted offensive subtypes. Previous studies revealed that individuals with social anxiety disorder do not show evidence of high rates of childhood maltreatment or other specific forms of early-onset psychosocial adversity; very few previous studies were interested in the correlations between parental physical abuse and generalized SAD.30 Generalized SAD runs in families;31 parents also exhibit a tendency to show generalized SAD, lacking self-assertiveness, and are warm in temperament. Thus, the offensive subtype differs from usual SAD in terms of the childhood environment.

We identified a significant correlation between LSAS scores and the offensive subtype on univariate analysis, and logistic regression analysis identified these scores as the strongest predictor. Both Choy et al.10 and Kim et al.32 noted a significant correlation between their offensive subtype scales (TK offensive symptoms score and TKS offensive subscale) and SAD scale scores, including the Social Phobia Scale (SPS), Social Interaction and Anxiety Scale (SIAS), and LSAS scores. These findings consistently suggest that the offensive subtype is a severe form of SAD. There were no significant differences in the comorbidity rates of axis I disorders other than SAD. The offensive subtype belongs to SAD as a severer form rather than any other anxiety disorders (quantitative rather than qualitative).

In addition, the present study suggested that offensive subtype patients of today and those of 40 years ago probably grew up in very different family environments. Yamashita5 reported that most TKS patients were doted on by their parents, and that patients were very sensitive to criticism because they had never been rebuked by their parents. The vicious cycle of wishing to please everybody and being oversensitive to an imagined personal flaw or shortcoming will perpetuate. Yamashita stated that even persons without a troubled childhood still excessively desire to please everybody (excessive sensitivity to criticism or disapproval), and have excessive concerns over physical shortcomings (excessive concern over physical appearance). These statements spotlight the changes in the family environment over these 4 decades.

There are some limitations in the present study. There is no gold standard regarding TKS diagnostic criteria or a semi-structured interview for TKS, although our criteria were used in our previous studies.7 Second, the subjects were limited to one facility that specializes in the treatment of anxiety, eating, and personality disorders. This has led to a sampling bias in the results. In spite of these limitations, this study showed the possibility that the offensive subtype of SAD is no longer a culture-bound syndrome, but a severe subtype of SAD. Further international studies are needed to clarify whether offensive subtype subjects require specific treatment or if that applied for SAD is equally effective.


We appreciate the very helpful comments of Professor M.R. Liebowitz.