Mental health articles

The categorical versus dimensional status of the Diagnostic and Statistical Manual of Mental Disorders’s (DSM-IV; APA, 1994) axis II is a particularly important question, which taxometric research should play a major role in resolving. Although the DSM-IV represents PDs as distinct categories and diagnoses them as simply present or absent, dimensional models of PDs are popular among theorists and researchers, much as they are popular in the study of normal personality. Thus, there have been a number of proposals to overhaul the classifi cation of PDs in a dimensional fashion (Livesley, Schroeder, Jackson, & Jang, 1994; McCrae, 1994; Widiger & Clark, 2000).

Consequently, the latent structure of PDs is not simply an abstract issue for theorists, but a very practical matter with implications for the shape of future psychiatric classifications. Several considerations argue in favor of the dimensional view of PDs (Haslam, 2003). First, many disorders are strongly associated with dimensions of normal personality, such as the Five-Factor Model (Costa & Widiger, 2002), and most appear to shade imperceptibly into normal personality variation. Second, dimensional measures of PD symptomatology often predict clinical phenomena better than categorical diagnoses (Nakao et al., 1992). Third, the troublingly high levels of “comorbidity” among some PDs—the fact that several PDs are often diagnosed together—might be understood better if the disorders are not discrete categories. Apparent comorbidity might not  refl ect the co-occurrence of two or more discrete disorders, but only the ways in which features associated with different diagnoses share associations with underlying dimensions of abnormal personality. These arguments against the prevailing system of categorical diagnosis provide a strong rationale for taxometric analyses of PDs.

Given the profound theoretical and practical implications of the categorical versus dimensional issue in the study of PDs, it is surprising how little relevant taxometric research has been conducted. One reason for this neglect may be the taken-for-grantedness of the dimensional view among most research psychologists. Whatever the reasons, taxometric attention has only been paid to three of the ten DSM-IV PDs, and some of this work has only an indirect relevance to PDs. We will review the work conducted to date that bears on the latent structure of schizotypal, antisocial, and borderline PDs, and conclude this section of the chapter with some integrating remarks and a plea for further research in this area. Taxometric research on schizotypy has obvious relevance for the latent structure of personality pathology, and as we note above no latent variable has received more taxometric scrutiny.

However, these studies have not usually directly addressed schizotypal PD (SPD), and the taxon that they point to does not correspond to the disorder straightforwardly. The Tyrka et al. (1995b) study comes closest to being an analysis of SPD, using indicators that corresponded to DSM-III (APA, 1980) symptoms of the disorder, but most studies have addressed particular schizotypal traits rather than the entire syndrome. Moreover, the taxon that has been identifi ed does not correspond  precisely with SPD as it is defi ned within the DSM system. Most important, the population prevalence of SPD (about 3%) is substantially below the 10% taxon base rate that repeatedly emerges in taxometric studies of normal undergraduates. Similarly, the base rates derived in research on psychiatric inpatient and high-risk (offspring of one parent with schizophrenia) samples—41% (Golden & Meehl, 1979) and 47% or 46%  (ErlenmeyerKimling et al., 1989; Tyrka et al.,1995a,b), respectively—easily exceed plausible rates of SPD. One interpretation of these fi ndings is that schizotypy is taxonic, and SPD is a relatively severe and impairing variant of it. The schizotypal traits that have been investigated taxometrically are elevated in SPD, and both the taxon and the disorder have been shown to confer increased risk of schizophrenia spectrum disorders. It is therefore plausible to view SPD as nested within the broader taxon. Current research does not  allow us to determine whether SPD is a discrete variant of schizotypy (i.e., a taxon within a taxon) or simply a quantitative variant. To answer this question a taxometric analysis restricted to a large sample of taxon members would have to be conducted.

However, this issue is perhaps not of great importance for larger question of the categorical versus dimensional status of PDs. If SPD is nested within a taxon, whether or not it is itself taxonic, the categorical view of PDs becomes more plausible and any conceptualization of SPD must take the existence of a category boundary into account. Another issue germane to SPD must also be mentioned. If the disorder is rooted in a signifi cantly more prevalent taxon, then it could be argued that DSM-IV draws the diagnostic boundary incorrectly, excluding a large proportion of schizotypes from the diagnosis. Perhaps, if diagnostic systems should “carve nature at the joints” and the joint in this case defi nes 10% of humanity as taxon members, then the DSM-IV’s diagnostic rule should be relaxed to become more inclusive (e.g., by requiring that only four of the nine symptoms be present). Although we are generally sympathetic to the idea that taxometric fi ndings should be used to revise diagnostic systems, we believe that this proposal would be a mistake. The existence of a psychopathology-related taxon does not require that a clinical diagnosis should map its boundary. Diagnosis should always be sensitive to practical issues of clinical severity and impairment, rather than simply refl ecting structural realities, and if it is only the most severe fraction of schizotypal taxon members who suffer signifi cant impairment then there is little reason to recalibrate the SPD diagnosis. In this situation there is also a very real concern that relaxing the requirements for SPD diagnosis would amount to a needless and perhaps destructive pathologizing of ordinary eccentricity, a charge that the DSM system already faces from its critics. Antisocial PD (APD) has also been the subject of a limited amount of taxometric research, although as with SPD most of it has had only a glancing relevance to the disorder as it is defi ned in DSM-IV.

Most studies have examined psychopathy rather than APD, or have investigated antisociality in children, in whom PDs cannot be diagnosed, rather than adults. Nevertheless, these studies have yielded quite consistent results, which again tend to favor taxonic models. In the fi rst published study, Harris, Rice, and Quinsey (1994) found convergent evidence for taxonicity using a variety of psychopathy indicators. However, only some aspects of the psychopathy construct—those representing chronic antisocial behavior beginning in childhood—were taxonic, and a nontaxonic model seemed more appropriate for the interpersonal and affective aspects of psychopathy and for adult criminality per se. Ayers, Haslam, Bernstein, Tryon, and Handelsman (1999) replicated this taxonic fi nding in a study of antisocial PD features among polysubstance-abusing adults. Similarly, Skilling, Harris, Rice, and Quinsey (2001) found evidence of a taxon using measures of psychopathic tendencies and DSM-IV antisocial PD indicators in a study of persistently antisocial adult offenders. Finally, Skilling, Quinsey, and Craig (2001) detected a taxon of antisocial boys of middle-school age in a community sample, consistent with Harris et al.’s fi nding regarding the discreteness of chronic antisociality. These four studies consistently support a taxonic model of antisociality, which is characterized by a variety of features associated with psychopathy and APD. The  evidence for this taxon is less extensive than for schizotypy, but it has at least addressed diagnostic criteria to a greater degree. The antisocial taxon does not appear to be reducible to criminality and is detectable relatively early in life, implying that it does not originate in adult criminal careers or simply refl ect a social role or niche. It is interesting that evidence for the taxon is strongest precisely in those aspects of the psychopathy construct (i.e., chronic antisocial conduct) that are captured by the DSM-IV APD diagnosis. The third and fi nal PD that has submitted to taxometric inquiry is borderline PD (BPD), the most well-researched member of the Axis II family. Trull, Widiger, and Guthrie (1990) conducted the fi rst taxometric analysis of any PD in an investigation of BPD features judged from a chart review of a large sample of psychiatric outpatients. Their findings were somewhat ambiguous, but Trull et al. (1990) inferred support for a nontaxonic model. Several writers (e.g., Korfi ne & Lenzenweger, 1995) later disputed this interpretation, and suggested that the analysis probably revealed a low base rate taxon. Although Trull et al.’s (1990) findings proved controversial, their interpretation has been borne out by three more recent studies. Two unpublished investigations by Simpson (1994) and Ayers et al. (1999) also yielded nontaxonic findings for borderline PD in clinical samples, although they were based on a less reliable self-report assessment of BPD symptoms. More recently, Rothschild, Cleland, Haslam, and Zimmerman (2003) conducted a more complete taxometric analysis of BPD features assessed by structured interview, in a large outpatient sample. Obtaining similar results to Trull et al. (1990), they showed that these were more consistent with a nontaxonic model than with a borderline taxon, and that the source of the original ambiguity was most likely the positive skew of BPD indicators.

We return to the issue of skewness later, but now it seems reasonable to conclude that BPD is probably not taxonic. What, then, can we learn about the latent structure of PDs from taxometric research? Most simply, it can be argued that SPD may or may not be taxonic but it is probably rooted in a broader schizotypal taxon, that APD is probably taxonic, and that BPD is probably nontaxonic. By implication, PDs are likely to be a mixed bag when it comes to fundamental questions of latent structure. For some disorders a categorical approach may be most warranted, and for others a purely dimensional approach may be preferable.

Consequently, taxometric research argues against any dogmatic preference for categorical or dimensional understandings of Axis II. Equally, it does not support a universal, one-size-fi ts-all approach to the classifi cation of PDs, whether it be the categorical status quo or the dimensional alternative. Perhaps the most important contribution of taxometric research on PDs, however, is raising the plausibility of taxonic models in this domain. Although it has been claimed that the evidence for the dimensional view of PDs is “overwhelming” (Livesley, 1996, p. 224), taxometric research now makes it unreasonable to dismiss categorical models as questionable in principle. Entertaining the possibility that a PD may have taxonic components does not mean that it can no longer be measured quantitatively, given complex causal explanations, or modeled in terms of established dimensional systems of personality description (Haslam, 2003). It does mean that nonarbitrary distinctions between normal and abnormal personalitymust be admitted as possibilities and given proper theoretical attention. Determining how many PDs are best understood as categories or as dimensions remains an urgent scientifi c question, given that only three of the ten disorders have been studied taxometrically to date.

Tags: disorders, personality, personality disorders, taxometric