Antisocial Personality Disorder and Psychopathy in Cocaine-Dependent Women

The antisocial personality disorder diagnosis has undergone several changes since its original description in the first DSM. Before DSM-III, the diagnostic conceptions of antisocial personality disorder were strongly influenced by the works of Cleckley (1), McCord and McCord (2), and Karpman (3). These individuals described a psychopathic personality type that lacked dynamic features of personality such as guilt, anxiety, and loyalty. Early diagnostic criteria for antisocial personality disorder in DSM-I and DSM-II focused on the absence of these dynamic features of personality. In response to concerns over the reliability and standardization of the antisocial personality disorder diagnosis, the conceptualization of antisocial personality disorder in DSM-III and subsequent DSMs emphasized assessment of antisocial acts more than assessment of features of personality. This shift in focus was also evident in the behaviorally based research criteria for antisocial personality disorder such as the Feighner criteria (4) and the Research Diagnostic Criteria (RDC) (5). Although the DSM-III, DSM-III-R, and DSM-IV criteria, Feighner criteria, and the RDC for antisocial personality disorder are all behaviorally based, they differ in the required number and type of childhood and adult behaviors (Table 1) and whether behaviors related to substance use are included. These differences affect the rates of antisocial personality disorder and levels of diagnostic agreement among the five systems.

The formulations of antisocial personality disorder that appeared in DSM-I (1952) and DSM-II (1968) did not specifically require the presence of childhood symptoms. The germinal work of Robins (6), using exclusively male subjects, revealed sufficient consistency between childhood and adult antisocial behaviors to consider antisocial personality disorder a unitary syndrome beginning in childhood. These findings strongly influenced subsequent formulations of antisocial personality disorder. The Feighner criteria, published in 1972 (4), was the first widely disseminated and used classification system for antisocial personality disorder that explicitly required the early onset of antisocial behavior (i.e., at least one type of antisocial behavior before age 15). The RDC (5) for antisocial personality disorder, published in 1978, required the early onset (before age 18) of at least three types of antisocial behaviors, with at least one beginning before age 15. DSM-III (1981) and DSM-III-R (1987) required the presence of three or more antisocial behaviors before age 15. More recently, DSM-IV (1994) requires “evidence of conduct disorder with onset before age 15.” No specific number of required types of behavior is stated, although at least in the research community, the convention of three childhood symptoms still appears to be the rule. For example, the structured diagnostic interviews such as the Structured Clinical Interview for DSM-IV Axis II Disorders (7) and the Structured Interview for DSM-IV Personality (8) require three childhood symptoms in order to diagnose antisocial personality disorder.

As shown in Table 1, there is substantial variation in antisocial personality disorder criteria among the five systems. Only three early-onset criteria (truancy, ran away, and often lied) and five adult criteria (inconsistent work behavior, repeated antisocial acts, irritable or aggressive behavior, impulsivity, and no regard for the truth) are assessed in all five diagnostic systems. Compared with the other systems, DSM-III-R and DSM-IV include more violent and aggressive childhood behaviors such as forcing someone to engage in sexual activity and physical cruelty to animals or people. An arrest or referral to juvenile court is an early-onset item assessed only in the Feighner criteria, RDC, and DSM-III diagnostic systems. Engaging in adult behaviors such as substance use or sexual activity before age 15 are included as criteria only in the RDC and DSM-III diagnostic systems. While there is more similarity in adult diagnostic criteria than childhood criteria among the systems, only the RDC assesses an individual’s ability to sustain adult, responsible, and reciprocal relationships. The ability to feel remorse is assessed only in DSM-III-R and DSM-IV.

Many studies have demonstrated that antisocial personality disorder is more common among substance abuse patients than in the general population. The degree to which behaviors that are a result of substance use are considered in making an antisocial personality disorder diagnosis is another difference among the classification systems. Early diagnostic systems for antisocial personality disorder such as the DSM-I, DSM-II, and Feighner criteria made no explicit mention of excluding behaviors that are a consequence of substance use. In comparison, the RDC indicate that behaviors that occur solely as a result of substance abuse should not be considered when diagnosing antisocial personality disorder. DSM-III, DSM-III-R, and DSM-IV stipulate that if an individual demonstrates antisocial behavior related to substance use, a diagnosis of antisocial personality disorder may be appropriate if there is evidence that antisocial behavior was present in childhood and continued into adulthood or if substance abuse and antisocial behavior began in childhood and continued into adulthood. In the differential diagnosis sections of these DSMs, however, it is also stated that the diagnosis of antisocial personality disorder should not be based solely on behaviors that are consequences of substance use, withdrawal, or associated activities. Hence, it is not absolutely clear how to consider substance-related behaviors when diagnosing antisocial personality disorder according to the DSM systems.

Hare’s Revised Psychopathy Checklist (9) offers an alternative conceptualization of antisociality to the diagnosis of antisocial personality disorder. Hare’s measure of psychopathy is based largely on Cleckley’s conceptualization of psychopathy. In addition to a global rating of psychopathy, two factors have been identified in the Revised Psychopathy Checklist. Factor 1 assesses psychopathic personality traits with the following eight items: glibness/superficial charm, grandiose sense of self-worth, pathological lying, conning/manipulative, lack of remorse or guilt, shallow affect, callous/lack of empathy, and failure to accept responsibility for one’s actions. Factor 2, antisocial lifestyle, includes the following nine items: need for stimulation, parasitic lifestyle, poor behavioral controls, early behavioral problems, lack of realistic long-term goals, impulsivity, irresponsibility, juvenile delinquency, and revocation of conditional release. Three items, promiscuous sexual behavior, many short-term marital relationships, and criminal versatility, are included only in the Revised Psychopathy Checklist total score. Behaviors that are considered related to drug use are included in the assessment of Revised Psychopathy Checklist items.

Although the Revised Psychopathy Checklist concept of psychopathy and the antisocial personality disorder diagnoses are theoretically related to the same underlying construct of psychopathy first described by Cleckley (1), antisocial personality disorder and Revised Psychopathy Checklist psychopathy differ in significant ways. A diagnosis of antisocial personality disorder in the five most current diagnostic systems requires the presence of a specific number and type of early-onset behaviors; the two Revised Psychopathy Checklist early-onset items, early behavior problems and juvenile delinquency, are not required for a diagnosis of psychopathy.

Another important difference between Revised Psychopathy Checklist psychopathy and the more recent antisocial personality disorder diagnoses pertains to the emphasis placed on personality traits. Although the earlier measures of antisocial personality disorder such as DSM-I and DSM-II focused on assessment of personality traits related to the construct of psychopathy, later formulations of antisocial personality disorder assess more behavioral traits and fail to capture the personality traits central to psychopathy. There is little overlap in criteria assessing personality traits between the antisocial personality disorder diagnostic systems and the Revised Psychopathy Checklist. It is impossible to get a high psychopathy score without the presence of a constellation of psychopathic personality traits such as grandiosity, egocentricity, and lack of empathy. It is, however, possible to receive a diagnosis of antisocial personality disorder without the presence of such personality traits.

The differences between antisocial personality disorder and psychopathy outlined earlier result in those with a diagnosis of psychopathy constituting a more homogeneous and severely impaired group than those qualifying for only a diagnosis of antisocial personality disorder. Psychopathic individuals are cold-hearted, grandiose, and manipulative, display shallow and labile emotions, are unable to form long-lasting bonds, and lack genuine empathy, remorse, and guilt. Whereas individuals diagnosed with antisocial personality disorder typically demonstrate behaviors that go against social norms, they may or may not demonstrate the personality traits common to psychopathic individuals listed earlier.

This article examines the rates of antisocial personality disorder and relationships among antisocial personality disorder diagnoses made with the Feighner criteria, RDC, DSM-III, DSM-III-R, and DSM-IV classification systems in a study group of cocaine-dependent women. The prevalence of psychopathy, as defined by the Revised Psychopathy Checklist, is also examined. Next, the relationship between antisocial personality disorder diagnoses in these five classification systems and psychopathy is investigated. Finally, differences on important treatment admission variables are examined on the basis of the presence of antisocial personality disorder or psychopathy diagnoses or both.

METHOD

Participants

Participants were 137 treatment-seeking, cocaine-dependent women recruited from an urban community hospital. The average woman was African American (85.5%), was 32 years old (SD=6), had one 10-year-old child (mean=1, SD=2) living with her, had a high school diploma or its equivalent (mean=11.8 years of education, SD=1.6), was currently unemployed (84.8%), and was currently single (89.9%). The majority of women had never been treated for psychiatric problems during their lifetime as inpatients (73.2%) or as outpatients (79.7%). Participants reported engaging in an average of 5 days (SD=9) of illegal activity during the past month. However, only 39.9% of the study group had ever been arrested. Of those arrested, an average of two arrests (SD=12) and 1.2 months of incarceration (SD=1.9) were reported. Participants reported a mean of two (SD=2) prior drug treatments and one (SD=2) prior alcohol treatment. The typical participant reported a history of 7 years (SD=5) of cocaine use, 10 years (SD=8) of regular alcohol use, 8 years (SD=8) of marijuana use, and 1 year (SD=4) of sedative use. In the month before entering treatment, women reported using cocaine for 18 days (SD=10), alcohol for an average of 12 days (SD=11), marijuana for 6 days (SD=9), and sedatives for 2 days (SD=6).

Procedure

All participants were screened on the basis of their clinical intake battery. All women judged by the unit psychiatrist as not currently psychotic or mentally retarded and without a history of schizophrenia were considered eligible. A research technician described the study in detail to all potential participants within a week of their admission to treatment for cocaine dependence. All participants who agreed to partake in the study signed informed consent forms before the first assessment. Participants were compensated $25 for their time.

The Addiction Severity Index (10) was used to assess psychosocial functioning. The Revised Psychopathy Checklist interview (9) was expanded to included questions that allowed interviewers to make diagnoses of antisocial personality disorder through use of the Feighner, RDC, DSM-III, DSM-III-R, and DSM-IV criteria.

Five individuals (one Ph.D.-, two M.A.-, and two B.A.-level) administered all interviews in this study. All interviewers had extensive experience in administration of the Revised Psychopathy Checklist and in assessing antisocial personality disorder. Training for this study consisted of several didactic sessions in which antisocial personality disorder criteria of the five diagnostic systems and Revised Psychopathy Checklist items were reviewed. The interviewers then videotaped cocaine-dependent woman being interviewed with the expanded Revised Psychopathy Checklist interview. Ten videotaped interviews were then observed and scored by all research staff. Interrater reliability was established for the Revised Psychopathy Checklist by using consensus scores provided by Hare’s group and for the antisocial personality disorder diagnoses by using consensus ratings established by the first two authors.

The five interviewers had an overall intraclass correlation (ICC model 3) (11)of 0.97 for the total Revised Psychopathy Checklist score, 0.89 for factor 1, and 0.97 for factor 2 for the 10 videotaped interviews. The individual ICCs for the total score ranged from 0.83 to 0.98 (mean ICC=0.91); for factor 1 the range was 0.55 to 0.94 (mean ICC=0.78), and for factor 2 the range was 0.86 to 0.97 (mean ICC=0.91).

Coefficients alpha for the entire Revised Psychopathy Checklist scale in this study group was 0.80; the mean interitem correlation was 0.16. For factor 1, alpha was 0.72 and the mean interitem correlation was 0.24. Factor 2 homogeneity was less than that found for the total scale or factor 1, as indicated by a coefficient alpha of 0.61 and mean interitem correlation of 0.15.

Interrater reliability for the antisocial personality disorder classification systems was evaluated by comparing the number of symptoms endorsed as well as by diagnostic agreement. Overall intraclass correlations (11) for the antisocial personality disorder classification systems all exceeded 0.93, and the ICCs of individual raters all exceeded 0.90. Kappa coefficients for the five individual interviewers using the 10 tapes were as follows: for DSM-III, 0.78 to 1.00; for DSM-III-R, 0.74 to 1.00; for DSM-IV, all interviewers had perfect agreement on diagnoses (kappa=1.00); for RDC, 0.11 to 0.29; and for Feighner criteria, 0.78 to 1.00. Coefficients alpha, a measure of internal consistency, for the antisocial personality disorder measures ranged from 0.56 for the Feighner criteria (mean interitem correlation=0.13) to 0.78 for DSM-III criteria (mean interitem correlation=0.14).

RESULTS

As shown in Table 2, there was a wide range in the percentage of women who received an antisocial personality disorder diagnosis under the five systems. Rates of women who met adult and childhood antisocial personality disorder criteria are also presented in Table 2. It is somewhat difficult to compare separately the Feighner criteria for adult and child antisocial personality disorder because several of the criteria could be considered either early-onset or adult items (e.g., trouble with police), depending on the age at first occurrence. If a behavior first occurs before age 15, it is an early-onset item. If it first occurs after age 15, it is an adult item. In addition, although five manifestations of antisocial behavior are required for a diagnosis of antisocial personality disorder according to the Feighner criteria, only one item needs to have an early onset (76% met this criterion). Therefore, we have used the presence of four adult behaviors and one early-onset behavior when comparing rates of full, adult, and childhood antisocial personality disorder diagnoses based on the Feighner criteria to the other diagnostic systems.

With the exception of the Feighner criteria and RDC, more than 75% of women endorsed adult antisocial personality disorder criteria according to the three most recent DSM systems. The percent of participants meeting the childhood antisocial personality disorder criteria varied greatly among the diagnostic systems. It should be noted that if three early-onset items were required for the Feighner criteria, as in the other systems, only 35% of the women would have received a childhood antisocial personality disorder diagnosis. This rate is comparable to that found for DSM-III-R.

The joint distribution of the antisocial personality disorder diagnoses in the five classification systems is presented in Table 3. Generally, there was relatively little agreement across the five systems. No women received only one diagnosis of antisocial personality disorder according to RDC, DSM-III-R, or DSM-IV. In this study group, 22.6% of the women were diagnosed with antisocial personality disorder under two systems, 9.5% were diagnosed in three systems, and 19.0% were diagnosed in four systems. Only 5.8% (N=8), however, received the diagnosis in all five systems.

Next, phi correlations between the antisocial personality disorder diagnoses in the different systems were determined. There was a strong correlation of 0.72 between the DSM-III-R and DSM-IV systems. The other correlations were generally moderate, ranging from 0.51 between DSM-III and DSM-III-R to 0.20 between the Feighner criteria and the RDC. The correlations between the adult and childhood antisocial personality disorder diagnoses were very low, ranging from 0.06 for the RDC adult and childhood diagnoses to 0.24 for the DSM-III adult and childhood diagnoses.

The possible range of Revised Psychopathy Checklist scores is as follows: total score=0–40, factor 1 score=0–16, and factor 2 score=0–18. In this group of women, the mean Revised Psychopathy Checklist total score was 14.2 (SD=6.0), factor 1 was 5.7 (SD=2.9), and factor 2 was 6.2 (SD=2.9). These scores are significantly lower than the mean scores reported by Hare (9) for male prisoners (total=23.6, SD=7.9, t=16.9, df=1426, p<0.001; factor 1=8.9, SD=3.9, t=11.8, df=1426, p<0.001; factor 2=11.7, SD=3.9, t=20.3, df=1426, p<0.001) and those reported for male methadone patients (12) (total=17.9, SD=3.3, t=4.5, df=223, p<0.001; factor 1=7.1, SD=3.7, t=3.0, df=223, p<0.01; factor 2=8.1, SD=3.5, t=4.2, df=223, p<0.001) but are quite similar to those reported for female methadone patients (13) (total=13.8, SD=7.0, t=0.4, df=193, p>0.05; factor 1=5.6, SD=3.6, t=0.2, df=193, p>0.05; factor 2=6.4, SD=2.9, t=0.4, df=193, p>0.05).

The Pearson product moment correlations between the total number of antisocial personality disorder symptoms, adult symptoms, and childhood symptoms and the Revised Psychopathy Checklist scores were also examined. The Feighner criteria were not included in the childhood and adult correlational matrices because of the difficulty in separately evaluating adult and child antisocial personality disorder criteria in this system, as noted previously. Correlations between antisocial personality disorder symptoms and Revised Psychopathy Checklist scores were generally similar across the classification systems for each Revised Psychopathy Checklist score. For all of the classification systems, the correlations between the total number of antisocial personality disorder symptoms and factor 1 (psychopathic personality traits) scores were significantly lower (t values for nonindependent r values [14] all >4.44, df=136, p<0.001) than those between the number of antisocial personality disorder symptoms and factor 2 (antisocial lifestyle). Correlations between the number of adult antisocial personality disorder symptoms and Revised Psychopathy Checklist scores were also remarkably similar for the four diagnostic systems. When correlations between childhood antisocial personality disorder symptoms and Revised Psychopathy Checklist scores were examined, most striking were the low correlations between the number of childhood symptoms in DSM-III-R and DSM-IV and the factor 2 scores (0.10 and 0.11, respectively). These correlations are significantly lower than those found for the RDC (0.60) and DSM-III (0.62, t values for nonindependent r values all >6.47, df=136, p<0.001).

Typically, a cutoff score of 30 is used to diagnose psychopathy (9). Scores of 25, however, have been used to diagnose psychopathy in forensic patients (15) and in male and female methadone patients (12, 13). Scores of 20 and above indicate at least a moderate level of psychopathy (9). In this study group, only two women (1.5%) had Revised Psychopathy Checklist total scores of 30 or more, and just five (3.6%) scored above 25, but 26 (19.1%) had scores of 20 or more. Therefore, Revised Psychopathy Checklist total scores of 20 and above, which are considered at least a moderate level of psychopathy, will be used for comparative purposes in the following analyses.

The percentage of women with Revised Psychopathy Checklist scores of 20 and above who received an antisocial personality disorder diagnosis in the various classification systems was examined next. Of the 26 women with scores over 20, only 23.1% received an RDC diagnosis of antisocial personality disorder, and those six women represented 40% of all who received an RDC diagnosis of antisocial personality disorder. Approximately half of the women with scores of 20 and above received DSM-III-R and DSM-IV diagnoses of antisocial personality disorder, representing about a third of the women with antisocial personality disorder in these systems. Women with scores of 20 and above all received antisocial personality disorder diagnoses under DSM-III and the Feighner criteria; yet this represented only 31% and 25%, respectively, of all women diagnosed with antisocial personality disorder in those systems.

Finally, we examined differences in treatment admission variables based on the number of antisocial personality disorder diagnoses given and on the combination of antisocial personality disorder and Revised Psychopathy Checklist diagnoses. As the number of antisocial personality disorder diagnoses that a woman received increased, so did the severity of her drug use history, number of lifetime arrests, months of incarceration, and employment problems seen when she entered treatment. Years of regular alcohol use and years of cocaine use, however, were the only variables in which a statistically significant difference was found (table available from Dr. Rutherford on request). Women who had no antisocial personality disorder diagnoses (N=24) and a Revised Psychopathy Checklist score less than 20, any antisocial personality disorder diagnosis and a checklist score less than 20 (N=87), or a checklist score more than 20 (N=26) and an antisocial personality disorder diagnosis were then compared on several treatment admission variables by using stepwise linear discriminate analyses with the jackknife procedure. The only significant differences between the groups revealed in this analysis were that women with moderate psychopathy (Revised Psychopathy Checklist score more than 20 and an antisocial personality disorder diagnosis) were involved in illegal activities more than antisocial women (checklist score less than 20 and any antisocial personality disorder diagnosis) and nonpsychopathic nonantisocial women (checklist score less than 20 and no antisocial personality disorder diagnoses: F=4.3, df=2, 132, p=0.02). Evaluation of involvement in illegal activities, however, appears to be of little help in classifying women into these diagnostic groups (overall classification rate=25.2%). Whereas 87.5% of nonantisocial nonpsychopathic women were correctly classified, just 38.5% of women considered to be moderately psychopathic and only 3.5% of nonpsychopathic antisocial women were correctly classified.

DISCUSSION

Results from this study indicate that although there is clearly some similarity in antisocial personality disorder diagnoses among these five classification systems, they are by no means interchangeable. Moreover, findings demonstrate that psychopathy, as measured by the Revised Psychopathy Checklist, and antisocial personality disorder are not synonymous diagnoses. Results also indicate that for women, there is little relationship between the behaviors seen as children and antisocial behaviors exhibited as adults in any of the antisocial personality disorder diagnostic systems.

As discussed in the introduction, three sources of variation may explain the differences in rates of antisocial personality disorder among the five classification systems. First is the manner in which behaviors that are a consequence of substance use are considered in making the diagnosis. This appears to be a primary reason that rates of antisocial personality disorder in the RDC, which excludes substance-related behaviors, were low relative to those in the other diagnostic systems, which do not explicitly exclude such behaviors. Second, the number of childhood criteria required for an antisocial personality disorder diagnosis by the various classification systems also affected the number of women diagnosed with antisocial personality disorder. Unlike the RDC and DSM classification systems, which require the presence of at least three early-onset items, the Feighner criteria require only one early-onset behavior. This resulted in considerably higher rates of antisocial personality disorder based on the Feighner criteria compared with the other classification systems. Third, the specific criteria and differences in wording of the criteria influenced rates of antisocial personality disorder diagnoses. For example, the RDC criterion regarding the inability to establish adult, responsible, and reciprocal relationships is unique to the RDC system and must be present for a diagnosis of antisocial personality disorder. When this criterion was deleted, the rate of RDC diagnoses of antisocial personality disorder increased to 37.9% (N=52), which was more comparable to the other systems. Even when these differences are taken into account, it is surprising that there is such a wide variation in diagnostic rates and diagnostic agreement across the five classification systems.

Antisocial personality disorder and psychopathy are often viewed as interchangeable diagnoses. Findings from the present study do not support this view. In the present study more than a quarter of the women received a DSM diagnosis of antisocial personality disorder, but only 1.5% of women were diagnosed with a moderate level of psychopathy. Only the Feighner criteria and DSM-III classification systems diagnosed the 26 women with a moderate level of psychopathy as also having antisocial personality disorder. This suggests that antisocial personality disorder may be a heterogeneous diagnosis. Some of the women diagnosed with antisocial personality disorder appear to engage in antisocial behaviors but do not demonstrate the personality traits associated with psychopathy. Other women diagnosed with antisocial personality disorder appear to engage in antisocial behaviors and demonstrate a lack of empathy and a callous disregard for others. These women would likely be considered psychopathic. The prognosis and treatment requirements for women diagnosed with antisocial personality disorder will differ greatly depending on whether the women possess the personality traits associated with psychopathy.

In all of the diagnostic systems, there was a weak relationship between the childhood and adult antisocial behaviors seen in women. Our previous work (16) demonstrated that DSM-III-R’s inclusion of violent and aggressive childhood criteria, while excluding more rule-breaking types of behaviors seen in DSM-III, resulted in a strengthening of the relationship between childhood and adult antisocial behaviors for men but not for women. Research has shown that precursors of antisocial behavior in women may be more related to the early taking on of adult roles and norm-breaking behaviors as a way of obtaining adult possessions (e.g., money, clothes, drugs [17]).

The Feighner criteria, RDC, and DSM-III childhood criteria, which do not include aggressive behaviors but do include rule- and norm-breaking behaviors, had the strongest correlations to adult antisocial behavior as measured by the Revised Psychopathy Checklist. This indicates that although norm- and rule-breaking behaviors in childhood for girls do not predict specific adult antisocial behaviors as measured by the DSM systems, they do predict the more general antisocial adult traits assessed by the Revised Psychopathy Checklist. DSM-IV added several childhood criteria that were nonviolent, as well as requiring earlier ages at onset. Despite these changes, the early-onset behaviors assessed by DSM-IV still appear to have little relationship to adult antisocial behaviors in women.

A longitudinal study by Zoccolillo and colleagues of 381 girls, recruited in kindergarten and followed for the next 6 years, found that DSM-III-R conduct disorder criteria identified only 3% preadolescent girls who were rated by themselves, parents, and/or teachers as having early-onset, pervasive, and persistent antisocial behavior, whereas DSM-III conduct disorder criteria identified 22% of these girls (18). In light of their results, Zoccolillo et al. suggested two modifications to conduct disorder, lowering the required number of criteria for women and including a specific criterion related to violation of rules.

The first suggestion of lowering the required number of conduct disorder symptoms for girls from three to two resulted in a fivefold increase in conduct disorder rates for those identified as persistently antisocial by parents, teachers, or both; there was a minimal increase (1%) in conduct disorder diagnoses for girls who were not rated as persistently antisocial (18). The second suggestion is to require inclusion of the criterion of frequent violation of rules at home or at school in the conduct disorder criteria set and require at least one other additional symptom. This resulted in the highest rate of conduct disorder diagnoses in the persistently antisocial group (35%) and a minimal increase (1%) in conduct disorder diagnoses in girls not rated by parents and/or teachers as antisocial. Moreover, 80% of the girls who met one criterion and the violation of rules criterion were rated as persistently antisocial.

Our findings support the notion that the DSM system would be improved by broadening its criteria for both adult and childhood antisocial personality disorder. More broadly defined criteria, as found in the Revised Psychopathy Checklist and Feighner diagnostic systems, allow for a variety of behaviors to be considered as indicators of traits rather than viewing only a single specific behavior as an indication of a more general trait.

Making the conduct disorder criteria more inclusive would allow for a wider range of behaviors to be considered when making a conduct disorder diagnosis. As an illustration of this, in DSM-IV “serious violation of rules” does not include suspension at school or trouble with the police, which can also indicate a serious violation of rules. If the definition of this criterion were broadened, many other behaviors could also be included. For example, behaviors such as early use of drugs or alcohol and early sexual behaviors, as well as repeated suspensions from school and truancy after age 13, could be included. Broadening the definition of specific conduct disorder criteria would allow for more of the precursors of antisociality in girls to be captured, in turn allowing for the early identification of young girls with persistent antisocial behavior and the development and implementation of appropriate treatment interventions.

DSM-IV broadened the adult antisocial personality disorder criteria to include more behaviors as exemplars of a trait; specific behaviors are no longer listed as indicators of a trait. For example, in DSM-III-R the trait of recklessness for one’s own or others’ personal safety was indicated only by driving while intoxicated or recurrent speeding. This criterion excluded many women, and men, who did not drive but were nevertheless reckless in their regard for the safety of others, as indicated by behaviors such as having unprotected sex and using drugs while pregnant. In DSM-IV the criterion is the same, but clinicians can use their judgment to determine if the criterion is met on the basis of the behaviors exhibited by the individual, and behaviors such as unprotected sexual activity could be considered as evidence of recklessness.

Another possibility is to eliminate the necessity of childhood behaviors entirely from antisocial personality disorder for women. Results from this study demonstrate that for women, there is very little relationship between the childhood behaviors and adult antisocial personality disorder behaviors assessed by DSM-IV. In this study, the Revised Psychopathy Checklist, which does not require the presence of childhood antisocial behaviors, was as successful as antisocial personality disorder in identifying women with the most problematic drug use and legal histories. Moreover, antisocial personality disorder is the only personality disorder that currently requires the presence of childhood behaviors, although all personality disorders are believed to have an onset in adolescence or early adulthood.

At the very minimum, modification of the conduct disorder criteria seems essential if we are to identify girls at an early age who are at risk for life-persistent antisocial behavior (19). There is a need for further research similar to that of Zoccolillo and colleagues (18) but employing a much longer follow-up to determine the most salient precursors of persistent antisociality in women. Furthermore, additional retrospective and prospective studies of adult women who are and are not antisocial, psychopathic, or both are also necessary. This study is limited by the small number of psychopathic women included. Studies that include a larger group of psychopathic women would provide further evidence regarding the importance of personality traits, antisocial behaviors, or both with respect to general functioning.

Presented in part at the 150th annual meeting of the American Psychiatric Association, San Diego, May 17–22, 1997. Received Sept. 3, 1997; revisions received May 8, Oct. 2, and Dec. 18, 1998; accepted Jan. 20, 1999. From the Alcohol and Drug Abuse Institute, University of Washington; Department of Psychiatry, University of Pennsylvania School of Medicine, Philadelphia; and the VA of Philadelphia. Address reprint requests to Dr. Rutherford, Alcohol and Drug Abuse Institute, University of Washington, 3937 15th Ave., N.E., Seattle, WA, 98105-6696; [email protected] (e-mail). Supported by the VA Medical Research service, National Institute on Drug Abuse research grant DA-05858, and Center grant DA-05186.

TABLE 1

Enlarge table

TABLE 2

Enlarge table

TABLE 3

Enlarge table