An important feature of the multiaxial diagnostic system introduced in the DSM-IlI is the separate axis for the independent evaluation oi personality disorders. Axis Il encourages clinicians to consider the contributing effects of the patient's personality on the presentation, course, and treatment of Axis 1 clinical syndromes rather than forcing clinicians to choose arbitrarily between them.' There is now substantial clinical and research interest in the comorbidity of personality and Axis I disorders.
Personality disorders can also be the principal focus of treatment. Persons with a personality disorder are unable to respond flexibly or adaptiven to the changes and demands oï life, including medical disorders, developmental changes, career changes, and other stressors. Instead, they tend to create and exacerbate stress by provoking adverse reactions in others; by failing to make optimal social, occupational, and personal decisions; and by creating problematic and pathogenic situations that are at times, the fulfillment of their worst fantasics.
The diagnosis of personality disorders, then, is obviously of substantial clinical importance. This article reviews recent findings with respect to the prevalence and comorbidity of the personality disorders, including the comorbidity among the personality disorders themselves and with Axis I syndromes.
Personality disorders are very prevalent in clinical settings, both inpatient and outpatient, and therefore should be assessed in every patient. Most patients, in fact, have one or more maladaptive personality traits even if they do not meet the criteria for a specific personality disorder. It is then useful to note the presence of these maladaptive traits on Axis II, in addition to any Axis Il diagnoses that do apply. This will be particularly useful in nonpsychiatric medical settings where many patients will have maladaptive personality traits below the threshold for a psychiatric personality disorder diagnosis.
Prevalence of DSM-III Personality Disorders in Eight Studies
Prevalence estimates tend to be higher in studies that use structured interviews than in studies that obtain diagnoses from patients' charts.2,5 Practicing clinicians tend not to give a personality disorder diagnosis as often as a researcher who systematically assesses each of the features for every personality disorder. It might be that personality disorders still are not given as much consideration as they warrant, with attention being distracted by a more florid and immediate Axis I clinical condition.1 Also, structured interviews typically take about two hours to assess all of the personality disorders,5 which might not be practical in everyday clinical practice. Clinicians might wish to consider using one of the major self-report inventories as an initial screening device to alert them of the presence of particular personality disorders.5 The clinical interview then could focus on those personality disorders that were elevated on the self-report scales.
Table I provides prevalence estimates from eight studies.2,4-10 It is evident that these estimates vary substantially across studies, due in part to the use of different structured interviews, the threshold and criteria used by different interviewers for each item, and the variation in the prevalence of the disorders across settings. For example, the highest values were obtained by Widiger et al9 because they confined their sampling to state hospital inpatients who had a personality disorder and used nonprofessional interviewers who might have had a lower threshold for assessment. The lowest values were obtained by Kass et al6 because they included all outpatients and used a relatively less structured interview with clinicians who might have been reluctant to provide more than one or two personality diagnoses per patient.
There are. however, some consistencies across studies, particularly with respect to the rank order of the prevalence estimates. The most prevalent is borderline personality disorder, both in inpatient and outpatient settings. Prevalence rates for borderline have varied substantially across studies, but the best estimate based on all available studies is 1 % to 2% of the community. 8% of all outpatients. 15% of all inpatients, 27% of outpatients with a personality disorder, and 5f% of inpatients with a personality disorder." The next most common personality disorders tend to be schizotypal and histrionic, followed by dependent, avoidant, antisocial, passive-aggressive, paranoid, and narcissistic.
Two personality disorders that appear to be rarely diagnosed are the schizoid and the obsessive compulsive. Obsessive compulsive disorder is a relatively high functioning personality disorder that might be more common among private practice settings than the inpatient and outpatient settings that have been sampled in the research. The most prevalent personality disorders - borderline, schizotypal, and histrionic - are also the lowest functioning. These personality disorders are particularly common in inpatient settings, as they arc characterized in part by symptomatology resulting in hospitalization (eg. impulsive drug use, suicidal behaviors, and cognitive-perceptual aberrations). Outpatient and nonpsychiatric medical settings might find more patients with dependent and passiveaggressive, as well as obsessive compulsive disorder.
Prevalence, Multiple Diagnosis, and Co-occurrence Averaged Across Four Studies*
The very low prevalence found for the DSM-III schizoid personality disorder (many studies do not even find one patient) might have resulted from the very restrictive diagnostic criteria (eg. it is rare to find patients who one can confidently state have a complete absence o\ warm, tender feelings for others and are indifferent toward others). In addition, patients previously diagnosed as schizoid by DSM-II criteria arc now diagnosed by DSM-III criteria with avoidant or schizotypal personality disorder. Most introverted persons are anxious, insecure, and ambivalent (ie. avoidant) rather than apathetically indifferent, or they will have eccentricities of speech, thought, and behavior (ie. schizotypal). The low prevalence for schizoid personality disorder might not occur when the DSM-III-R criteria are used, as they are not as restrictive. They provide a set of seven optional criteria with a variety of subsets that can be used to make the diagnosis. The prevalence of DSM-III-R schizoid personality disorder is then likely to be higher.8
COMORBIDITY AMONG PERSONALITY DISORDERS
In any case one should not slop with one personality disorder diagnosis. Most patients who meet the criteria for one personality disorder will meet the criteria for another, particularly in inpatient settings. Table 2 provides the prevalence, percent of multiple diagnoses, and percent of co-occurrence among the personality disorders averaged across four studies.2,4,8,10 Again, it should be noted that these rates tend to vary across studies and that the values reported are best interpreted as relative (rank order) comparisons among the pairs of personality disorders rather than as absolute values for each pair.
It is evident from Table 2 that the highest comorbidity occurs with the borderline personality disorder, with almost half of the borderline and histrionic patients meeting the criteria for both personality disorders. There is also substantial co-occurrence of borderline with the antisocial, schizotypal, dependent, and passiveaggressive personality disorders. This substantial degree of comorbidity may suggest that the borderline diagnosis is best used as an indicator of dysfunction severity rather than as a distinct personality disorder.11
It is also of use to note that avoidant personality disorder co-occurs as much with the dependent as with the schizoid (and the schizotypal) personality disorder. It has been suggested that avoidant personality disorder cannot be distinguished from the schizoid, but it appears to occur more frequently than the schizoid and may have more clinical utility (see Table 1). The more problematic overlap for the avoidant personality disorder is with the dependent personality disorder. Avoidant persons will often become very dependent once they do become involved with someone. Features of dependent persons that might be helpful in differentiating these personality disorders are a lack of selectivity in finding new mates and the rapidity with which they select someone new once a relationship ends.12
The histrionic and the passiveaggressive personality disorders also have substantial comorbidity with other personality disorders. A substantial proportion of histrionic patients will also meet the DSM-III criteria for the passive-aggressive, narcissistic, antisocial, dependent, schizotypal, and/or the avoidant personality disorders, in addition to borderline. Many passive-aggressive patients will meet DSM-III criteria for the histrionic, borderline, antisocial, avoidant, dependent, and/or the paranoid personality disorders. These comorbid personality disorders should be considered during clinical assessment.
It is also important to note that the co-occurrence rates in Table 2 can be somewhat misleading if it is not recognized that they depend on the relative base rates (prevalence) of each disorder. The 8% co-occurrence of the histrionic and narcissistic personality disorders is low in part because there tends to be many more histrionic than narcissistic patients (see Table 1). For example, in the study by Dahl,4 the co-occurrence of histrionic and narcissistic also was only 8%, but 100% of the three narcissistic patients were histrionic whereas only 8% of the 36 histrionic patients were narcissistic. Most narcissistic patients tend to be histrionic, but few histrionic patients tend to be narcissistic. In other words, the co-occurrence rates presented in Table 2 will be somewhat higher for the disorder that occurs less frequently.
COMORBIDITY WITH AXIS I DISORDERS
A substantial comorbidity of the personality disorders with the Axis I disorders also exists.15 The comorbidity is not random, as personality disorders will predispose a person to the development of particular Axis I syndromes. Considerable research effort currently is focused on trying to outline the extent and pattern of this comorbidity. One hypothesis is that the comorbidity will be consistent with cluster organization of the DSM-III-R personality disorders. The personality disorders are organized in the DSM-III-R into three clusters: the odd or eccentric (eg, paranoid, schizoid, and schizotypal); the dramatic, emotional, or erratic (eg, antisocial, borderline, histrionic, and narcissistic); and the anxious or fearful (eg, avoidant, dependent, obsessive compulsive, and passive-aggressive). This might suggest comorbidity of these three clusters with psychotic, affective, and anxiety disorders, respectively. This does occur to some extent, particularly for the schizotypal, paranoid, borderline, avoidant, and dependent personality disorders, but there are also many notable exceptions. For example, passi ve - aggressive patients tend not to be as anxious as schizotypal patients. One should not take the clustering too seriously.
Most of the personality disorders in fact predispose a person to depression, although perhaps not for the same reasons.14 For example, dependent persons arc vulnerable to depression in response to loss or separation; narcissistic persons are not so vulnerable to separation but they are quite vulnerable to blows to a fragile and granthose self-image. Avoidant persons tend to be depressed as a result of their loneliness and low self-esteem, while selfdefeating persons tend to be depressed as a result of their pessimistic, self-blaming cognitive style. The presence of a personality disorder in depressed patients tends to be associated with an earlier onset, more prior episodes of depression, more suicidal potential and history, poorer social support, more life stressors, more separation and divorce, and poorer response to pharmacologic treatment. 13,15,16 Depression is particularly common in borderline patients, with rates ranging from 24% to 74% for major depression. 4% to 20% for bipolar, and 3% to 14% for dysthymia.15
A substantial comorbidity of anxiety disorders with personality disorders also tends to exist, although this area is not as well researched.15 The borderline, avoidant, dependent, obsessive compulsive, schizotypal. paranoid, and perhaps the antisocial personality disorders may be particularly prone to anxiety disorders.14 Social phobia often will be seen in avoidant patients, and posttraumatic stress disorder is often seen in borderline patients.
The impulsive and self-destructive tendencies of borderline patients can also be expressed in a variety of forms other than suicidal tendencies, including the Axis I syndromes of bulimia and substance use disorders. Borderline personality has been reported to occur in as many as 25% of bulimics.17 Clinicians, then, should be cognizant of the possible presence of additional syndromes in their borderline and other personality disorder patients that will warrant additional, special attention. However, one also must be careful not to overdiagnose borderline and other personality disorders in patients with impulse dyscontrol, anxiety, and affective disorders."18 Personality disorder diagnoses tend to be inflated in patients with Axis I symptomatology resembling and/or overlapping with the personality disorder symptomatology. This is particularly problematic in the case of the borderline personality disorder.1 ' Therefore, it is important to verify the behavior pattern as pervasive (ie. affecting all areas of the person's life, such as social, occupational, leisure, and family relations) and the behavior as chronic (eg, evident since adolescence).
Personality disorders are very prevalent in psychiatric settings, and many nonpsychiatric patients will have maladaptive personality traits below the threshold for a psychiatric diagnosis. Personality disorders, then, should be evaluated in every patient, including those seen in nonpsychiatric medical settings. These personality traits will often affect the presentation, course, and treatment of an Axis I psychiatric and nonpsychiatric condition. The psychiatrist also should not confine his other assessment to only one personality disorder, since many patients will present with a variety of maladaptive personality traits and may even meet the criteria for two. three, or even more personality disorders. However, the psychiatrist also must be careful not to overdiagnose personality disorders in patients with affective, anxiety, or impulsive dyscontrol disorders that are similar to and/or overlap with the personality disorders.
1. Frances A: The DSM-III personality disorders section: A commentary. Am I Psychiatry 1980: 157:1050-1054.
2. Pl'ohl B. Coryell W. Zimmerman M. et al: DSM-III personality disorders: Diagnostic overlap and internal consistency o\ individual DSM-III criteria. Conipr Psvchiatry 198b: 27:21-54.
3. Widiger T. Frances A: Interviews and inventories lor the measurement of personality disorders. Clinical Psychology Review 1987; 7:49-75.
4. Dahl A: Some aspects of the DSM-III personality disorders illustrated by a consecutive sample ol hospitalized patients. Acta Psych ia Ir Sea nd 198b: 73:61-66.
5. Frances A, Clarkin I. Gilmore M. et al: Reliability of' criteria for borderline personality disorder: A comparison of DSMIII and the Diagnostic Interview tor Borderline Patients. Am I Psychiatry 1984: 141:1080-1084.
6. Kass F. Skodol A. Charles L. et al: Scaled ratings o\' DSM-III personality disorders. Am I Psychiatry 1985: 142:627-650.
7. Loranger A. Susman V. Oldham |. et al: The personality disorder examination: A preliminary report, journal of Personality Disorders 1987: 1:1-15.
8. Morey L: Personality disorders in DSMIII and DSM-III-R: Convergence, coverage, and internal consistency. .4/?? / Psvchiatrv 1988: 145:575-577.
9. Widiger t. Trull T. Hurt S. et al: A multidimensional scaling of the DSM-III personality disorders. Arch Cen Psychiatry 1987: 44:557-565.
10. Zanarini M. Frankenburg F, Chauncey D. et al: The Diagnostic Interview for Personality Disorders: Interrater and testretest reliability. Compr Psvchiatrv 1987: 28:407-480.
11. Widiger T, Frances A: Epidemiology, diagnosis, and comorbidity of borderline personality disorder, in Gunderson | (ed): Review of Psychiatry. Washington. DC. American Psychiatric Press, in press.
12. Trull G, Widiger T, Frances A: Covariation of criteria sets lor avoidant, schizoid, and dependent personality disorders. Am J Psychiatry 1987; 144:767-771.
13. Docheity (. Rester S. Shea T: Syndrome diagnosis and personality disorder, in Hales R. Frances A (eds): Psychiatry Update. American Psychiatric Association's Annual Review. Washington, DC. American Psychiatric Press. 198b.
14. Widiger T. HvlerS: Axis I/Axis Il interactions, in Michels R. Cavenar ) teds): Psychiatry, Philadelphia. IB Lippincott. 1987. '
15. PiObJ B, Slang) D, Zimmerman M: The implications oi DSM-IIl personality disorders l'or patients with major depression. J Affective Disord 1984: 7:509-518.
16. Shea T. Glass D. Pilkonis P. et al: Frequency and implications of personality disorders in a sample of depressed outpatients. J Personality Disorders 1987: 1:27-42.
17 Levin A, Hyler S: DSM-III personality diagnosis in bulimia. Compr Psvchiatrv 198b: 27:47-55.
18. Pope H. Frankenburg F. Hudson l.et al: Is bulimia associated with borderline personality disorder? A controlled study. J Clin Psvchiatrv 1978: 48:448-455.
Prevalence of DSM-III Personality Disorders in Eight Studies
Prevalence, Multiple Diagnosis, and Co-occurrence Averaged Across Four Studies*