Summary
One hundred and seventy-five offspring of parents in two psychiatrically ill groups and of normal controls in the New York High-Risk Project (NYHRP) were assessed for Axis II personality traits and disorders as defined by the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R). These offspring include: subjects at high risk for schizophrenia (HRSz, n = 48), all of whom have a parent with schizophrenic disorder; subjects at high risk for affective disorder (HRAff, n = 40), all of whom have a parent with affective disorder; and subjects at no increased risk for psychiatric illness (NC, n = 87), whose parents are psychiatrically normal. The trained interviewers, who administered a standardized direct interview, were blind to parental clinical status and to previous clinical status of the offspring.
The rates for any personality disorder (PD) ranged from 7% to 20%. Comorbidity between Axis I and Axis II disorders was high for all groups.
Introduction
Personality disorders were defined in the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R) (APA, 1987) as persistent, maladaptive personality styles beginning in adolescence and leading to distress or vocational and social impairment. While the eleven personality disorder categories on Axis II were not presumed to be mutually exclusive, three clusters were identified, which may be more discrete. These are the odd or eccentric cluster (schizotypal, schizoid and paranoid disorders), the dramatic, emotional or erratic cluster (borderline, histrionic, antisocial and narcissistic personality disorders) and the anxious, fearful cluster (passive–aggressive, dependent, obsessive–compulsive and avoidant personality disorders).
It has been hypothesized that Axis II features and disorders aggregate in families where Axis I disorders are also present (Baron, Gruen, Asnis, & Kane, 1983; Baron, Gruen, Kane, & Asnis, 1985a; Docherty, Fiester, & Shea, 1986). For example, it has been suggested that there is a familial co-occurrence of Axis I schizophrenic disorders with traits and disorders from the odd, eccentric cluster (Gottesman & Shields, 1982; Gunderson, Siever, & Spaulding, 1983; Kendler, 1988a,b; Kety, 1983, 1985, 1987, 1988; Shields, Heston, & Gottesman, 1975; Siever & Gunderson, 1983; Stephens, Atkinson, Kay, Roth, & Garside, 1975). Others have reported a familial association between Axis I affective disorders, and traits and disorders from the erratic, dramatic cluster (Baron, Gruen, Asnis, & Lord 1985b; Loranger, Oldham, & Tulis 1982; McGlashan, 1983; Pope, Jonas, Hudson, Cohen, & Gunderson, 1983; Soloff & Millward, 1983)
The New York High-Risk Project (NYHRP) allows a systematic evaluation of rates of personality features and disorder in offspring of informative psychiatric and control groups. To explore the specific relationships of Axis I and Axis II disorders, our interviewers—blind to the clinical status of parents—made standardized, direct Axis II assessments of offspring. Informative multiple risk groups (psychiatric and normal controls) permit estimates of relative risks and direct evaluation of the specificity of familial associations.
If familial associations between Axis II and Axis I disorders are confirmed by independent studies using similar standardized methods, clues may emerge about diagnostic boundaries of the so-called spectrum disorders and about shared etiological components.
Methods
Subjects and Assessment Approach
Subjects are members of Sample A of the New York High-Risk Project (NYHRP) (Erlenmeyer-Kimling & Cornblatt, 1987) who were assessed for Axis II disorders in young adulthood (mean age of 25 years). This group represents 86% of the initial Sample A study subjects.
Axis II assessments were made by direct interviews using the Personality Disorder Examination (PDE; Loranger, Susman, Oldham, & Russakoff, 1987). The PDE is a standardized, semi-structured clinical interview assessing DSM-III-R personality disorders. The PDE was given by seven clinical psychologists and social workers trained in its use, blind to parental diagnostic status, and blind to results of previous clinical assessments of the offspring. Eighty-six percent of the direct interviews were face-to-face interviews, with the remainder being conducted by telephone.
The PDE developers (Loranger et al., 1987) reported reliabilities of the instrument for schizotypal, histrionic, borderline, antisocial and compulsive disorders of .80, .77, .96, .70, and .88, respectively. No formal site specific reliability assessment was undertaken in our study. Our interviewers were trained and supervised as follows: Dr. Loranger and/or a colleague conducted a workshop in the use of the instrument for our clinically trained interviewers (psychologists and psychiatric social workers). Interviewers then trained with pilot subjects. Finally, all interviews with sample subjects were audiotaped and reviewed by clinical supervisors.
The interviewers assessed the subjects’ personality traits and features in terms of how traits were expressed, their impact on the individual’s functioning, and their persistence. Interviewers did not form a diagnostic conclusion as to the presence of a given disorder. The diagnostic algorithm provided by the PDE summarizes the number and pattern of clinical features and yields a threshold or subthreshold clinical trait summary for a given subject. Subjects were evaluated for Axis I disorders if they responded positively to certain screening items for these conditions. If an individual expressed schizotypal features, these could in fact prove to be prodromal or residual signs of schizophrenia.
Table 1 shows the distribution of sex and age for those subjects from Sample A who received the PDE. Offspring sex and age distributions do not differ by parental diagnostic group.
Table 1.
Demographic Characteristics of Sample Receiving PDE Assessment: Sex and Age by Group
| Characteristic | HRSz n = 48 n (%) | HRAff n = 40 n (%) | NC n = 87 n (%) |
|---|---|---|---|
| Sex* | |||
| Males | 24 (50.0) | 15 (37.5) | 49 (56.3) |
| Females | 24 (50.0) | 25 (62.5) | 38 (43.7) |
| Age* | |||
| Range: 18–28 years | 24(1.28) | 24 (2.0) | 23.7 (2.06) |
Note. HRSz = High risk for schizophrenia group. HRAff = High risk for affective disorders group. NC = Normal comparison group.
p = .145.
p = .166.
Results
Rates of Probable and Definite Disorder
Table 2 shows both the proportion of subjects from each group expressing probable disorder (3 or more personality disorder features), and the proportion of subjects with definite personality disorder (defined by the DSM-IIIR as the expression of 4 to 5 features, depending on the specific disorder). The rate for definite disorder is low for all groups for most disorders. Paranoid, borderline, avoidant, and schizotypal are the only disorders with rates higher than 4% for any group. The subclinical threshold of 3 or more features for probable disorder ranges from 1 % to 25% over all disorders. With conservative significance levels (i.e., p < .0056, corresponding to an alpha level appropriate for 9 independent tests at .05), only paranoid and borderline disorders, probable or definite, show significant group differences. In general, the rates for the HRSz and HRAff groups are higher than those for the NC group (except for obsessive–compulsive disorder).
Table 2.
Rates of Probable and Definite DSM-III-R Personality Disordersa
| DSM-III-R Personality disorders | HRSz n = 48 (A) | HRAff n = 40 (B) | NC n = 87 (C) | Significance levelsb | |||
|---|---|---|---|---|---|---|---|
| A vs B | A vs C | B vs C | |||||
| Paranoid | (Probable) | .06 | .15 | .05 | — | — | .04 |
| (Definite) | (.02) | (.10) | (.00) | — | — | (.003) | |
| Schizoid | (Probable) | .02 | .03 | .01 | — | — | — |
| (Definite) | (.00) | (.00) | (.00) | — | — | — | |
| Schizotypal | (Probable) | .13 | .18 | .08 | — | — | — |
| (Definite) | (.04) | (.08) | (.01) | — | — | (.06) | |
| Antisocial Borderline | (Definite) | (.04) | (.03) | (.03) | — | — | — |
| (Probable) | .25 | .25 | .06 | — | .002 | .002 | |
| (Definite) | (.00) | (.18) | (.03) | (.002) | — | (.006) | |
| Histrionic | (Probable) | 0.04 | 0.03 | 0.01 | — | — | — |
| (Definite) | (.00) | (.03) | (.01) | — | — | — | |
| Narcissistic | (Probable) | .06 | .15 | .03 | — | — | 0.2 |
| (Definite) | (.00) | (.00) | (.00) | — | — | — | |
| Avoidant | (Probable) | .08 | .18 | .05 | — | — | .02 |
| (Definite) | (.02) | (.08) | (.01) | — | — | (.06) | |
| Dependent | (Probable) | .13 | .13 | .05 | — | — | — |
| (Definite) | (.00) | (.00) | (.00) | — | — | — | |
| Obsessive–Compulsive | (Probable) | .06 | .08 | .07 | — | — | — |
| (Definite) | (.00) | (.00) | (.01) | — | — | — | |
| Passive–Aggressive | (Probable) | .13 | .15 | .07 | — | — | — |
| (Definite) | (.03) | (.00) | — | (.04) | — | ||
| Any personality disorder | (Probable) | .46 | .40 | .22 | — | .004 | .04 |
| (Definite) | (.10) | (.20) | (.07) | — | — | (.04) | |
Note. Probability level for “multiple independent tests” = 1-(1-α)1/number of tests for α = .05 and 9 tests, p = .0057. HRSz = High risk for schizophrenia group. HRAff = High risk for affective disorders group. NC = Normal comparison group.
Probable personality disorder is present if a subject exhibited 3 or more features. Definite personality disorder is present if a subject exhibited diagnostic threshold number of features which range between 4–5 for all disorders.
Fleiss’s (1981) z statistic for proportion difference (one-tail significance levels).
Table 3 shows rates of disorder (including clusters of disorders) and relative risk for disorders broken down by risk group. The table includes the proportion of subjects expressing (1) definite personality disorder (i.e., having the 4 to 5 features necessary to meet full criteria for any of the eleven personality disorders); (2) probable personality disorder (i.e., having 3 or more features of any given disorder); (3) probable disorder in the odd, eccentric cluster (i.e., 3 or more features of any one of the following disorders: schizoid, schizotypal or paranoid); (4) probable disorder in the dramatic, erratic cluster (i.e., 3 or more features of any one of the following disorders: borderline, histrionic, narcissistic or antisocial); and (5) probable disorder in the anxious, fearful cluster (i.e., 3 or more features of any one of the following disorders: avoidant, dependent, obsessive–compulsive or passive–aggressive).
Table 3.
Rates of Personality Disorder and Clusters for NYHRP Subjects
| DSM-III-R Personality disorder and clusters | HRSz n = 48 (A) | HRAff n = 40 (B) | NC n = 87 (C) | Relative risk (Confidence Limits for RR) Significance levelsa− | ||
|---|---|---|---|---|---|---|
| A vs B | A vs C | B vs C | ||||
| Any definite personality disorder | .10 | .20 | .07 | — | — | 2.9 (1.1, 7.8) .04 |
| Any probable personality disorder | .46 | .40 | .22 | — | 2.1 (1.3, 3.5) .004 |
1.8 (1.1, 3.2) .04 |
| Probable odd, eccentric cluster | .17 | .20 | .09 | — | 1.8 (.7, 4.5) .21 |
2.2 (.9, 5.4) .10 |
| Probable dramatic, erratic cluster | .29 | .30 | .10 | — | 2.8 (1.3, 6.0) .01 |
2.9 (1.3, 6.3) .01 |
| Probable anxious, fearful cluster | .29 | .30 | .15 | — | 1.9 (1.0, 3.8) .05 |
2.0 (1.0, 4.0) .05 |
Note. HRSz = High risk for schizophrenia group. HRAff = High risk for affective disorders group. NC = Normal comparison group.
Likelihood ratio 2×2 table significance level.
As Table 3 shows, there are no significant differences between the HRSz and the HRAff groups in rates of personality disorders. Subjects from the HRSz group have significantly higher rates than the NC subjects for probable disorder, probable disorder in the fearful cluster, and probable disorder in the dramatic cluster. Subjects from the HRAff group were significantly higher than the NC subjects on all measures except the probable odd, eccentric cluster.
Further analyses were undertaken to assess the role of affective disorder subtype (unipolar vs bipolar disorder) in the parent on the rate of probable and definite DSM-III personality disorders in the offspring. Of the 40 offspring from the HRAff group, 17 have a parent with bipolar affective disorder subtype and 23 have a parent with unipolar affective disorder subtype. It might be expected that the elevated rates of personality features and disorders in subjects from the HRAff group arose through those families with parental bipolar affective disorder. Surprisingly, this was not the case. All disorders except definite antisocial disorder were more prevalent in the unipolar parental group. The effect of offspring gender on the pattern of rates of personality features and disorder was examined by (1) noting the significance of the difference in rates of probable and definite disorder between males and females within parental diagnostic groups, and by (2) noting the significance of pair-wise parental diagnostic effects while controlling for gender effects. The only significant difference in rates of expression of probable or definite personality disorder by gender occurred in the rates of probable avoidant disorder in the offspring from the HRSz group. In this group males exhibited probable avoidant disorder at a rate of 17%, while there were no females exhibiting this probable disorder. We have compared the significance levels for pair-wise group differences in expression of probable and definite personality disorder seen in Table 2 with the significance levels obtained using logistic regression controlling for gender effects. Of the fourteen significant or near significant (uncorrected) pair-wise group differences seen in Table 2, four changed from significant to nonsignificant when controlling for gender effects. These group differences include the pair-wise difference between the HRAff versus the NC groups for definite paranoid disorder and definite schizotypal disorder. (The rate for males from the HRAff group for definite paranoid disorder was 13%, while the rate for females was 8%. The rate for males from the HRAff group for definite schizotypal disorder was 7%, while the rate for females was 8%. The rate for males from the NC group for definite schizotypal disorder was 4%, while there were no females from the NC group expressing definite schizotypal disorder.) The two other pair-wise group differences that become nonsignificant when controlling for gender effects include the pair-wise difference between the HRSz versus the HRAff groups for definite borderline disorder and the pair-wise difference between the HRSz versus the NC group for definite passive aggressive disorder. (The rate for males from the HRAff group for definite borderline disorder was 13%, while the rate for females was 20%. The rate for males and females from the HRSz group for definite passive–aggressive disorder was 4%, while no males or females from the NC group expressed definite passive–aggressive disorder.)
Tables 4 to 6 address this question: Within each risk group, is the expression of a probable or definite personality disorder associated with any major Axis I categories? Major categories in Axis I are defined as according to Research Diagnostic Criteria (RDC, Spitzer, Endicott, & Robins, 1978). Lifetime RDC diagnoses were assigned to subjects based on SADS-L interviews (Schedule for Affective Disorders and Schizophrenia—Life-time Version, Third Edition, Spitzer, & Endicott, 1979) conducted at average subject age of 20 or 23 years or on both testing occasions. The resulting RDC categories are then grouped in the following hierarchical order: psychosis includes schizophrenia, schizoaffective disorder and unspecified functional psychosis; affective disorders include major depression and bipolar II disorder; substance abuse includes drug abuse and alcohol abuse; affective disorders include hypomania and intermittent depression; anxiety disorders include generalized anxiety disorder, panic disorder and phobic disorder.
Table 4.
Rates of Personality Disorder and Clusters for HRSz Subjects With Major Categories of Axis I Disorders (n = 48)
| DSM-III-R Personality disorder and clusters | Subjects without Axis I disorders n = 15 | Subjects with psychotic disorders n = 9 | Subjects with affective disorder n = 17 | Subjects with substance abuse n = 12 | Subjects with anxiety disorders n = 10 |
|---|---|---|---|---|---|
| Any definite personality disorder (.10)a | .07 | .11 | .12 | .25 | .00 |
| Any probable personality disorder (.46)a | .33 | .89 | .53 | .67 | .20 |
| Probable odd, eccentric cluster (.17)a | .07 | .67* | .18 | .25 | .00 |
| Probable dramatic, erratic cluster (.29)a | .20 | .44 | .35 | .67 | .10 |
| Probable anxious, fearful cluster (.29)a | .27 | .56 | .24 | .42 | .10 |
Note. HRSz = High risk for schizophrenia group.
Significance level of less than .05 for contrast between proportion of subjects with PD cluster in Axis I major category versus proportion of subjects with PD cluster in “complement” of Axis I major category, i.e., all subjects without the given Axis I major category.
Rate of personality disorder for all HRSz subjects.
Table 6.
Rates of Personality Disorder and Clusters for NC Subjects With Major Categories of Axis I Disorders (n = 87)
| DSM-III-R Personality disorder and clusters | Subjects without Axis I disorders n = 35 | Subjects with Psychotic disorders n = 2 | Subjects with affective disorder n = 26 | Subjects with substance abuse n = 24 | Subjects with anxiety disorders n = 18 |
|---|---|---|---|---|---|
| Any definite personality disorder (.07)a | .00 | .50 | .08 | .21* | .06 |
| Any probable personality disorder (.22)a | .14 | .50 | .31 | .42* | .17 |
| Probable odd, eccentric cluster (.09)a | .03 | .50 | .12 | .21 | .00 |
| Probable dramatic, erratic cluster (.10)a | .03 | .50 | .15 | .29* | .11 |
| Probable anxious, fearful cluster (.15)a | .14 | .50 | .15 | .17 | .06 |
Note. NC = Normal comparison group.
Significance level of less than .05 for contrast between proportion of subjects with PD cluster in Axis I major category versus proportion of subjects with PD cluster in “complement” of Axis I major category, i.e., all subjects without the given Axis I major category.
Rate of personality disorder for all NC subjects.
Rates of Personality Disorders in HRSz Subjects
Table 4 shows the rates of personality disorders including the disorder clusters, in HRSz subjects with and without major categories of Axis I disorders. Rates of personality disorders and clusters consistently (if not significantly) increase in the presence of psychotic disorders, affective disorder (except for the anxious, fearful cluster) and substance abuse. Subjects with anxiety disorder do not appear to be at increased risk for personality disorders or clusters. In fact, the risks for personality disorders and clusters were slightly lower in these subjects, although the reduction was not statistically significant.
Rates of Personality Disorders in HRAff Subjects
Table 5 shows the rates of personality disorder including clusters for HRAff subjects with and without major categories of Axis I disorders. Rates of disorder clusters increase in the presence of psychotic disorder, affective disorder, substance abuse, and anxiety disorder. This increase is significant only for definite personality disorder and disorders from the odd, eccentric cluster in psychotic subjects.
Table 5.
Rates of Personality Disorder and Clusters for HRAff Subjects With Major Categories of Axis I Disorder (n = 39)
| DSM-III-R Personality disorder and clusters | Subjects without Axis I disorders n = 9 | Subjects with psychotic disorders n = 2 | Subject with affective disorder n = 20 | Subject with substance abuse n = 10 | Subjects with anxiety disorders n = 14 |
|---|---|---|---|---|---|
| Any definite personality disorder (.18)a | .00 | 1.00* | .20 | .30 | .21 |
| Any probable personality disorder (.39)a | .22 | 1.00 | .45 | .60 | .36 |
| Probable odd, eccentric cluster (.18)a | .00 | 1.00* | .20 | .30 | .21 |
| Probable dramatic, erratic cluster (.28)a | .11 | 1.00 | .30 | .50 | .29 |
| Probable anxious, fearful cluster (.28)a | .11 | 1.00 | .35 | .30 | .29 |
Note. HRAff = High risk for affective disorders group.
Significance level of less than .05 for contrast between proportion of subjects with PD cluster in Axis I major category versus proportion of subjects with PD cluster in “complement” of Axis I major category, i.e., all subjects without the given Axis I major category.
Rate of personality disorder for all HRAff subjects.
Rates of Personality Disorders in NC Subjects
Table 6 shows the rates of personality disorders, including clusters, for NC subjects with and without major categories of Axis I disorder. Rates of personality disorders and clusters increase in subjects with psychotic, affective, or substance abuse disorder. For subjects with anxiety disorders, only the rates of definite personality disorder and of the dramatic, erratic cluster are elevated. These increases are significant only for definite disorder, probable disorder, and disorders from the dramatic, erratic cluster among subjects with substance abuse.
Rates of Personality Disorders in the Entire Study Sample
Table 7 shows the rates of personality disorders, and clusters for all subjects with and without major categories of Axis I disorder. The presence of any major Axis I category (except anxiety disorders) increases the likelihood of expression of all definite and probable personality disorders. Statistically significant increases are seen for any definite personality disorder, any probable personality disorder, the probable odd, eccentric cluster, and the probable dramatic, erratic cluster among subjects with psychotic and substance abuse disorders. Occurrence of disorders in the probable anxious, fearful cluster is significantly increased among subjects with psychosis.
Table 7.
Rates of Personality Disorder and Clusters for the Total Sample of Subjects With Major Categories of Axis I Disorders
| DSM-III-R Personality disorder and clusters | Subjects without axis I disorders n = 59 | Subjects with psychotic disorders n = 13 | Subjects with affective disorder n = 63 | Subjects with substance abuse n = 46 | Subjects with anxiety disorders n = 42 |
|---|---|---|---|---|---|
| Any definite personality disorder (.10)a | .02* | .31* | .13 | .24* | .10 |
| Any probable personality disorder (.32)a | .20* | .85* | .41 | .52* | .24 |
| Probable odd, eccentric cluster (.13)a | .03* | .69* | .16 | .24* | .07 |
| Probable dramatic, erratic cluster (.20)a | .09* | .54* | .25 | .44* | .17 |
| Probable anxious, fearful cluster (.22)a | .17 | .62* | .24 | .26 | .14 |
Significance level of less than .05 for contrast between proportion of subjects with PD cluster in Axis I major category versus proportion of subjects with PD cluster in “complement” of Axis I major category, i.e., all subjects without the given Axis I major category.
Rate of personality disorder for all subjects.
(It should be noted that the statistical significance was evaluated with a significance level of less than .05 for the contrast between the proportion of subjects with personality disorder clusters in a given Axis I major category vs the proportion of subjects with personality disorder clusters in the “complement” of the given Axis I major category, i.e., all subjects without the given Axis I major category. The significance level for rates of personality disorder clusters among subjects in the category “Subjects without Axis I disorders” was evaluated by comparing these rates with rates for subjects with Axis I disorders, i.e., with at least one Axis I disorder. The tests used were the likelihood ratio chi-square tests.)
Discussion
Defining Study Populations
Comparing our present results with those of previous studies is problematic. Different definitions of personality disorders, different assessment approaches, and different criteria for subject selection are the limiting factors.
Before DSM-III was in use, personality disorders were conceptualized and described chiefly in terms of presumed etiological processes, in contrast to description of clinical signs and symptoms. In the absence of clear phenomenological description, identification of cases and comparison of rates among investigations was difficult.
In DSM-III, the personality disorders are phenomenologically defined. Tools for assessment include self report, interview checklists, interviews with informants, and direct interviews with subjects. These assessment approaches may vary in sensitivity as a function of the specific personality disorder in question. For example, antisocial personality disorder may be most accurately characterized by informant report. In contrast, schizotypal personality disorder may best be observed and described by a trained interviewer.
Obviously, subject selection affects observed rates of personality traits and disorders. Subjects assessed by standardized measurements and DSM-III criteria have included community samples, “ever-treated” subjects, outpatients, inpatients, and relatives of psychiatric index cases.
Rates of Personality Traits and Disorder in “Non-Patients” as Reported in Recent Studies
Given the use of different diagnostic criteria, assessment approaches, and subject selection, it is remarkable that past studies actually do report similar rates of “any personality disorder”. Studies prior to DSM-III-R (reviewed by Merikangas & Weissman, 1986) reported rates of about 9%. Since DSM-III-R, Reich, Yates and Nduaguba (1989) using self-report assessments have reported rates of 11%. Zimmerman and Coryell (1990) reported that rates of any personality disorder were 13.5% for “non-patients” whether they were relatives of schizophrenic and affective disorder index cases or of normal controls. Zimmerman and Coryell’s rates were not stratified by clinical status of the index case. It is possible that the rates for this combined sample may be biased upward because of likely higher rates in relatives of index cases. Our rate of 7% for the NC group is not significantly different from the 11% that Reich and colleagues found (1989).
Table 8 shows the rates for definite and probable personality disorder among “normals” in three studies including this one. Schizotypal personality disorder was reported in 3 to 5.17% of Reich’s sample, 3% of Zimmerman’s sample, and was seen in 1% of our NC group. Kendler, reviewing studies using DSM-III criteria for schizoid, schizotypal, and paranoid disorder observed rates of 0 to 11.3% in normal controls, with most studies reporting rates between 1.7 and 2.7% (Kendler, 1988a).
Table 8.
Rates of Probable and Definite DSM-III-R Personality Disorders in Community Samples
| DSM-III-R Personality disorder | Reich n = 235 | Zimmeman n = 697 | NYHRP n = 87 | |||
|---|---|---|---|---|---|---|
| Self report | Interview | |||||
| Probable | Definite | Self report | Interview | Probable | Definite | |
| Paranoid | .009 | .008 | .004 | .004 | .050 | .000 |
| Schizoid | .013 | .009 | .009 | .007 | .010 | .000 |
| Schizotypal | .132 | .051 | .056 | .030 | .080 | .010 |
| Antisocial | .004 | .004 | .009 | .030 | .030 | |
| Borderline | .013 | .004 | .046 | .017 | .060 | .030 |
| Histrionic | .038 | .020 | .027 | .030 | .010 | .010 |
| Narcissistic | .004 | .004 | .004 | .000 | .030 | .000 |
| Avoidant | .004 | .000 | .013 | .014 | .050 | .010 |
| Dependent | .145 | .051 | .067 | .017 | .050 | .000 |
| Obsessive–compulsive | .149 | .064 | .040 | .017 | .070 | .010 |
| Passive–aggressive | .000 | .000 | .004 | .030 | .070 | .000 |
| Any personality disorder | .290 | .110 | .103 | .135 | .220 | .070 |
Rates in HRSz and HRAff Groups of NYHRP
Most studies of “schizophrenia-related” personality traits and disorders (e.g., schizotypal, schizoid, and paranoid) in relatives of schizophrenic probands have observed a statistically significant elevation in rates (compared to rates in relatives of normal controls). For example, Kendler (1988a), in a review of such studies, noted rates of between 5% and 34%. Compared with the associated control rates, the relative risks ranged from 2.5 to 9.1.
In those few reports, examining rates of “schizophrenia-related” traits and disorders in relatives of probands with nonschizophrenic disorder (e.g., affective disorders) rates of such conditions have been shown to be elevated compared to normal controls (Ingraham & Kety, 1987, 1988; Squires-Wheeler et al., 1988, 1989, 1992).
In the present study, the rates of probable and definite paranoid and schizotypal personality disorder are elevated in the HRSz and the HRAff groups compared to the NC group. This elevation is only significant, however, for definite paranoid personality disorder in the HRAff group. As for the rates of the odd, eccentric cluster (i.e., schizotypal, schizoid, and paranoid traits and disorder), both the HRSz and HRAff groups have higher rates than the NC group. Although the respective relative risks exceed 1.0, the confidence intervals include 1.0, which indicates a nonsignificant relative risk.
Previous studies have reported elevated rates of borderline personality traits and disorder in relatives of probands with affective disorder (Pope et al., 1983). However, the lack of normal control groups complicates the interpretation of these studies. (Most studies of the association between affective disorder and borderline traits in families have evaluated the risk of affective disorder in the relatives of probands with borderline traits and disorder. As reviewed by Baron (Baron et al., 1985a,b), these studies have shown significantly increased rates of affective disorder in the relatives of probands with borderline personality disorder compared to controls.)
In the NYHRP, definite borderline personality disorder is significantly higher in the HRAff group than in either of the other groups. However, for probable borderline personality disorder, both the HRAff group and the HRSz group had significantly higher rates than the NC Group.
With respect to the dramatic, erratic cluster (e.g., borderline, histrionic, narcissistic, and antisocial personality traits and disorder), both the HRAff group and the HRSz groups showed relative risks greater than 2.0, with associated probability levels of .007 and .006, respectively.
These data support the expectation that affective disorders in parents are associated with personality features from the erratic, dramatic cluster in offspring. A link between schizophrenic disorder in parents and personality features from the odd, eccentric clusters in offspring is also observed at a trend level. However, these associations are not specific to parental diagnostic groups.
Other investigators have noted complex, nonspecific familial aggregation of Axis I and Axis II disorders. Gunderson and colleagues (Gunderson & Elliott, 1985; Gunderson & Phillips, 1991) found a “surprisingly weak and nonspecific relationship” between Axis I affective disorder and Axis II borderline personality disorder in individuals and within families. Boyd et al. (1984) reported that within individuals, “there was a general tendency toward co-occurrence” (of Axis I disorders and antisocial personality disorder) “so that the presence of any disorder increased the odds of having almost any other disorder, even if DSM-III does not list it as a related disorder.” Fyer, Frances, Sullivan, Hurt and Clarkin (1988) have reported a nonspecific increase in affective disorders in individuals with any of several other personality disorders, as well as the anticipated increase in subjects with borderline personality disorder.
In the NYHRP, both risk groups showed somewhat increased (albeit nonsignificant) odds of expressing a personality disorder in the presence of affective disorder or substance abuse. Pfohl, Stangl and Zimmerman (1984) reported an analogous finding using a different design. They showed that if a proband with depressive illness also had a personality disorder, the probands’s first-degree relatives are at higher risk for depression, alcoholism, and antisocial personality disorder. The risk both for depression and for alcoholism increased still further when the proband had personality disorders.
Summary
The hypothesis of specific aggregation of Axis II features and disorder in families identified by parental Axis I disorders was examined. This expectation was not robustly confirmed. The offspring exhibited comorbidity of both Axis I and Axis II disorders yet this comorbidity was not confined to the expected pattern. For example, the expectation that personality features and disorder in the erratic, dramatic cluster would be seen to aggregate selectively in the offspring of parents with affective disorder was not observed. Further, the expectation that personality features in the odd, eccentric cluster would be seen to aggregate selectively in the offspring of parents with schizophrenic disorder was not observed.
This pattern may be consistent with a view that the major (presumably discrete) nosological categories do in fact share etiological components. Before this alternative view can be entertained seriously, it is important to review limitations of the study and to outline required next steps. The confidence in the generality and robustness of these results must be tempered by considerations of ascertainment, assessment, sample size and the relative youth of the offspring subjects.
It should be recalled that the rates of Axis II probable and definite disorders should not be thought of as representative of either population rates (from the normal control sample) or relative recurrence risks (from the offspring of psychiatric disorder parent probands). The normal controls are not selected from epidemiologic catchment areas and the parent probands are not necessarily representative of patient probands. The sample selection criteria consisted of identifying married parents treated at state psychiatric hospitals whose children (at age 7–12 years) expressed no psychopathology. Further, the families had agreed to participate in a study of adolescent development and have maintained contact and commitment to the study for more than two decades. These criteria may ensure that the parent probands have higher functioning than more representative patient probands.
The assessment instrument used was the PDE which evaluates all eleven of the DSM-III-R Axis TI disorders. We did not exclusively focus on schizophrenic or affective disorder spectra candidates such as schizotypal or borderline personality disorders. It may be argued that a more directed and expansive inquiry on these disorders would have yielded higher rates. However, the relative risks shown in Table 3 would remain the same in so far as increased sensitivity of assessment would be expected to increase rates for all groups.
A further limitation, in addition to the ascertainment and assessment approaches described above, is the small sample size. With larger ns the trends observed here would be statistically significant, i.e., the rates for most personality features and disorders would be significantly higher for the offspring of parent probands in contrast to the rates for offspring of normal control parents.
Next Steps
The present study extended our previous research on how Axis II disorders are distributed among the relatives of probands with Axis I disorders by evaluating comorbid Axis I and Axis II features and disorders in the offspring of all three parental groups. One of our main goals is to learn whether certain disorders share etiological components, i.e., represent a pathological spectrum rather than distinct entities. To achieve this, a longitudinal study of Axis I and Axis II must document both the onset and the course of those disorders in both probands and relatives. In the ongoing clinical assessments, we are noting the natural history of both Axis I and Axis II traits and disorders. Longitudinal assessment of the course of Axis I and Axis II traits and disorders, as advocated by Docherty et al. (1986), will shed light on whether comorbidity and family clustering of psychiatric disorder is causal, artifactual, or a product of chance (Hirschfeld et al., 1983).
Acknowledgments
The New York High-Risk Project is supported in part by a grant from the National Institute of Mental Health (MH19560), and by the Office of Mental Hygiene of the State of New York. Past support has come also from the Scottish Rite Schizophrenia Research Program and the William T. Grant Foundation. Data analyses are supported in part by grant MH 30906 to the Computer Center at the New York State Psychiatric Institute. We thank Dr. Donald Rock, whose help with statistical advice throughout the project has been invaluable. We are also grateful: (1) to our colleagues, Drs. John D. Rainer, Michael Stone, Irving Gottesman, Leonard Heston, and Jean Endicott, for their work in establishing the diagnoses of the parents in the NYHRP, as well as Dr. Clarice Kestenbaum and her colleagues, and to Dr. Endicott and her excellent staff of diagnosticians, for continuing assistance in clarifying the diagnoses of the offspring; (2) to Barbara Maminski Shapiro for her invaluable assistance; (3) to our social-work staff; (4) to our interviewers; and (5) to Mimi Simon and Sky Pape for their enormous aid in preparation of the manuscript. We thank Anne West for her valuable editorial help.
References
- American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 3. Washington, DC: Author; 1987. rev. [Google Scholar]
- Baron M, Gruen R, Asnis L, Kane J. Familial relatedness of schizophrenia and schizotypal states. American Journal of Psychiatry. 1983;140:1437–1442. doi: 10.1176/ajp.140.11.1437. [DOI] [PubMed] [Google Scholar]
- Baron M, Gruen R, Kane J, Asnis L. Modern research criteria and the genetics of schizophrenia. American Journal of Psychiatry. 1985a;142(6):697–701. doi: 10.1176/ajp.142.6.697. [DOI] [PubMed] [Google Scholar]
- Baron M, Gruen R, Asnis L, Lord S. Familial transmission of schizotypal and borderline personality disorders. American Journal of Psychiatry. 1985b;142(8):927–933. doi: 10.1176/ajp.142.8.927. [DOI] [PubMed] [Google Scholar]
- Boyd JH, Burke JD, Gruenberg E, Holzer CE, III, Rae DS, George LK, Karno M, Stoltzman R, McEvoy L, Nestadt G. Exclusion criteria of DSM-III: A study of co-occurrence of hierarchy-free syndromes. Archives of General Psychiatry. 1984;41:983–989. doi: 10.1001/archpsyc.1984.01790210065008. [DOI] [PubMed] [Google Scholar]
- Docherty JP, Fiester SJ, Shea T. Syndrome diagnosis and personality disorder. In: Frances AJ, Hales RE, editors. Annual review of the American Psychiatric Association. Vol. 5. Washington, DC: Author; 1986. pp. 315–355. [Google Scholar]
- Erlenmeyer-Kimling L, Cornblatt B. The New York High Risk Project: A follow-up report. Schizophrenia Bulletin. 1987;13:451–461. doi: 10.1093/schbul/13.3.451. [DOI] [PubMed] [Google Scholar]
- Fyer MR, Frances AJ, Sullivan T, Hurt SW, Clarkin J. Comorbidity of borderline personality disorder. Archives of General Psychiatry. 1988;45:348–352. doi: 10.1001/archpsyc.1988.01800280060008. [DOI] [PubMed] [Google Scholar]
- Gottesman II, Shields J. Schizophrenia: The epigenetic puzzle. Cambridge: Cambridge University Press; 1982. [Google Scholar]
- Gunderson JG, Siever LJ, Spaulding E. The search for a schizotype. Archives of General Psychiatry. 1983;40:15–22. doi: 10.1001/archpsyc.1983.01790010017002. [DOI] [PubMed] [Google Scholar]
- Gunderson JG, Elliott GR. The interface between borderline personality disorder and affective disorder. American Journal of Psychiatry. 1985;142(3):277–288. doi: 10.1176/ajp.142.3.277. [DOI] [PubMed] [Google Scholar]
- Gunderson JG, Phillips KA. A current view of the interface between borderline personality disorder and depression. American Journal of Psychiatry. 1991;148(8):967–975. doi: 10.1176/ajp.148.8.967. [DOI] [PubMed] [Google Scholar]
- Hirschfeld RMA, Klerman GL, Clayton PJ, Keller MB, McDonald-Scott P, Larkin BH. Assessing personality: Effects of the depressive state on trait measurement. American Journal of Psychiatry. 1983;140:695–699. doi: 10.1176/ajp.140.6.695. [DOI] [PubMed] [Google Scholar]
- Ingraham LJ, Kety SS. Schizophrenia Spectrum Disorders: Exclusively Related to Schizophrenia?. Poster presented at the International Congress on Schizophrenia Research; 28 March-1 April 1987.1987. [Google Scholar]
- Ingraham LJ, Kety SS. Schizophrenia spectrum disorders. In: Nasrallah AH, editor. Handbook of schizophrenia, Volume 3: Nosology, epidemiology and genetics of schizophrenia. Amsterdam: Elsevier; 1988. pp. 117–137. [Google Scholar]
- Kendler KS. The genetics of schizophrenia and related disorders: A review. In: Dunner DL, Gershon ES, Barrett JE, editors. Relatives at risk for mental disorder. New York: Raven Press; 1988a. pp. 247–266. [Google Scholar]
- Kendler KS. Familial aggregation of schizophrenia and schizophrenia spectrum disorders. Archives of General Psychiatry. 1988b;45:377–383. doi: 10.1001/archpsyc.1988.01800280095013. [DOI] [PubMed] [Google Scholar]
- Kety SS. Mental illness in the biological and adoptive relatives of schizophrenic adoptees; Findings relevant to genetic and environmental factors in etiology. American Journal of Psychiatry. 1983;140:720–727. doi: 10.1176/ajp.140.6.720. [DOI] [PubMed] [Google Scholar]
- Kety SS. Schizotypal personality disorder: An operational definition of Bleuler’s latent schizophrenia? Schizophrenia Bulletin. 1985;2:590–594. doi: 10.1093/schbul/11.4.590. [DOI] [PubMed] [Google Scholar]
- Kety SS. The significance of genetic factors in the etiology of schizophrenia: Results from the National Study of Adoptees in Denmark. Journal of Psychiatric Research. 1987;21:423–429. doi: 10.1016/0022-3956(87)90089-6. [DOI] [PubMed] [Google Scholar]
- Kety SS. Schizophrenic illness in the families of schizophrenic adoptees: Findings from the Danish National Sample. Schizophrenia Bulletin. 1988;14:217–222. doi: 10.1093/schbul/14.2.217. [DOI] [PubMed] [Google Scholar]
- Loranger AH, Oldham JM, Tulis EH. Familial transmission of DSM-III borderline personality disorder. Archives of General Psychiatry. 1982;39:795–799. doi: 10.1001/archpsyc.1982.04290070031007. [DOI] [PubMed] [Google Scholar]
- Loranger AH, Susman VL, Oldham JM, Russakoff LM. The Personality Disorder Examination: a preliminary report. Journal of Personality Disorders. 1987;1:1–13. [Google Scholar]
- McGlashan TH. The borderline syndrome: II. Is it a variant of schizophrenia or affective disorder? Archives of General Psychiatry. 1983;40:1319–1323. doi: 10.1001/archpsyc.1983.01790110061011. [DOI] [PubMed] [Google Scholar]
- Merikangas KR, Weissman MM. Epidemiology of DSM-III Axis II Personality Disorders. In: Frances AJ, Hales RE, editors. Annual review of the American Psychiatric Association. Vol. 5. Washington, DC: American Psychiatric Press; 1986. p. 258278. [Google Scholar]
- Pfohl B, Stangl D, Zimmerman M. The implications of DSM-III personality disorders for patients with major depression. Journal of Affective Disorders. 1984;7:309–318. doi: 10.1016/0165-0327(84)90052-1. [DOI] [PubMed] [Google Scholar]
- Pope HG, Jr, Jonas JM, Hudson JI, Cohen BM, Gunderson JG. The validity of DSM-III borderline personality disorder. Archives of General Psychiatry. 1983;40:23–30. doi: 10.1001/archpsyc.1983.01790010025003. [DOI] [PubMed] [Google Scholar]
- Reich J, Yates W, Nduaguba M. Prevalence of DSM-III personality disorders in the community. Social Psychiatry and Psychiatric Epidemiology. 1989;24:12–16. doi: 10.1007/BF01788194. [DOI] [PubMed] [Google Scholar]
- Shields J, Heston LL, Gottesman II. Schizophrenia and the schizoid: the problems for genetic analysis. In: Fieve RR, Rosenthal D, Brill H, editors. Genetic research in psychiatry. Baltimore: The Johns Hopkins University Press; 1975. pp. 167–197. [Google Scholar]
- Siever LJ, Gunderson JG. The search for a schizotypal personality: Historical origins and current status. Comprehensive Psychiatry. 1983;24:199–212. doi: 10.1016/0010-440x(83)90070-6. [DOI] [PubMed] [Google Scholar]
- Soloff PH, Millward JW. Psychiatric disorders in the families of borderline patients. Archives of General Psychiatry. 1983;40:37–44. doi: 10.1001/archpsyc.1983.01790010039004. [DOI] [PubMed] [Google Scholar]
- Spitzer RL, Endicott J. Schedule for affective disorders and schizophrenia, lifetime version (SADS-L) 3. New York: New York State Psychiatric Institute, Department of Biometrics; 1977. [Google Scholar]
- Spitzer RL, Endicott J, Robins E. Research Diagnostic Criteria (RDC) New York: Department of Biometrics, New York State Psychiatric Institute; 1978. [Google Scholar]
- Squires-Wheeler E, Skodol A, Erlenmeyer-Kimling L. The assessment of schizotypal features over two points in time. Schizophrenia Research. 1992;6:75–85. doi: 10.1016/0920-9964(91)90023-k. [DOI] [PubMed] [Google Scholar]
- Squires-Wheeler E, Skodol AE, Bassett A, Erlenmeyer-Kimling L. DSM-III-R Schizotypal personality traits in offspring of schizophrenic disorder, affective disorder, and normal control parents. Journal of Psychiatric Research. 1989;23(3/4):229–239. doi: 10.1016/0022-3956(89)90028-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Squires-Wheeler E, Skodol AE, Friedman D, Erlenmeyer-Kimling L. The specificity of DSM-III schizotypal personality traits. Psychological Medicine. 1988;18:757–765. doi: 10.1017/s0033291700008461. [DOI] [PubMed] [Google Scholar]
- Stephens DA, Atkinson MW, Kay DWK, Roth M, Garside RF. Psychiatric morbidity in parents and sibs of schizophrenics and non-schizophrenics. British Journal Psychiatry. 1975;127:97–108. doi: 10.1192/bjp.127.2.97. [DOI] [PubMed] [Google Scholar]
- Zimmerman M, Coryell WH. Diagnosing personality disorders in the community. Archives of General Psychiatry. 1990;47:527–531. doi: 10.1001/archpsyc.1990.01810180027005. [DOI] [PubMed] [Google Scholar]