Issues regarding the proposed DSM-5 personality disorders in geriatric psychology and psychiatry

The official introduction of the psychiatric diagnosis of personality disorders (PDs) in the Diagnostic and Statistical Manual of Mental Disorders (DSM) began in 1952 with the publication of the first edition (American Psychiatric Association, 1952). DSM-I contained 12 main types of PDs with a total description for all types in only two paragraphs. In the following DSM-II (American Psychiatric Association, 1968), just 10 specific types of PDs were described, including a very brief general definition of PDs. The DSM-III (American Psychiatric Association, 1980) included a significant paradigm shift from the medical model by incorporating the design of a multi-axial approach, in which the combinations of symptoms of more than five primary axes were used to describe the pathological state and formulate the diagnosis. Notably, the PDs were placed on a separate axis (Axis II) to distinguish their long-standing nature from the more episodic clinical disorders placed on Axis I. PDs were recognized as important formal diagnoses and included a more comprehensive listing of polythetic diagnostic criteria for each specific PD.

This introduction of Axis II led to a sharp increase in research studies devoted to PDs. Severity was debated heavily in the DSM-III Task Force and scheduled for use in several diagnostic categories (Millon, 2011). Severity is known to be the best predictor of therapeutic outcome (e.g., Crawford et al., 2011; Morey et al., 2011) and, therefore, especially important for clinical practice. However, rather than working out severity distinctions (i.e., different degrees of personality disturbance or functioning) between PDs, DSM-III (American Psychiatric Association, 1980) organized PDs into three superordinate clusters based on presumed common underlying themes. Cluster A groups the paranoid, schizoid, and schizotypal PDs, in which individuals often appear odd or eccentric. Cluster B includes antisocial, borderline, histrionic, and narcissistic PDs, in which individuals appear to be dramatic or erratic. Cluster C contains avoidant, dependent, obsessive–compulsive PDs (including passive aggressive PD in DSM-III and DSM-III-R), in which individuals frequently appear fearful or anxious. The three-cluster system was retained in DSM-III-R, DSM-IV, and DSM-IV-TR (American Psychiatric Association, 1987; 1994; 2000), although empirical studies did not support this structure (e.g., Bastiaansen et al., 2011). Finally, the diagnosis of “Personality Disorder Not Otherwise Specified (PD-NOS)” can be assigned to cases in which the patient has clear signs of a PD, but fails to meet the threshold for any specific PD. Problematic is the high prevalence of this PD-NOS, indicating that personality pathology is seldom confined to a single category (e.g., Verheul and Widiger, 2004; Clark, 2007). Also problematic is the frequent comorbidity between Axis I and II categories (e.g., Widiger and Schea, 1991). This again calls for an important paradigm shift. For example, Krueger (2005) suggested a unified model of personality, personality disorders, and clinical disorders. Widiger et al. (2005) identified the dysfunction-dimensional conceptualization for PDs as a first step in the near future.

We can conclude that ever since DSM-III, the criteria of most specific PDs have gone through several changes, and that ideas on how to conceptualize PDs continue to evolve. However, in the evolution of this classification, little attention has been given to the specific context of older adults. Also, in the current DSM-IV edition, age-specific criteria are lacking and empirical research has shown that the actual Axis-II criteria are problematic and could result in either under- or over-diagnosis in older adults (Balsis et al., 2007a; 2007b). In contrast to DSM-IV criteria, research has shown that dimensional personality traits are well suited to the measurement of personality over the life span, as dimensional traits evidence less measurement bias across age groups than the categorical approach of DSM (Oltmanns and Balsis, 2011; Van den Broeck et al., 2012). In addition, although geriatric variants of all of the PDs have been proposed (Solomon, 1981; Agronin and Maletta, 2000; Segal et al. 2006; Van Alphen et al., 2012a) and clinically endorsed, these variants to date lack empirical validation.

The publication of the fifth edition of DSM (DSM-5) is expected in May 2013 and is poised to mark one of the most anticipated events in the field of mental health (www.dsm5.org). Notably, the proposed DSM-5 revisions regarding PDs reflect major changes, consequently resulting in a heavy debate among theoreticians, researchers, and clinicians. Livesley and Verheul recently decided to leave the DSM-5 task force. Together with Tyrer, chair of the working group for the revision of PDs in International Classification of Diseases 11th Revision (ICD-11), they are critical of the DSM-5 proposal as it currently stands. Dominant criticisms of the proposal include that it does not provide a coherent framework for reliable diagnosis, lacks empirical support, and is far too complex for the average clinician to reasonably use (Emmelkamp and Power, 2012).

The DSM-5 hybrid model, with new general criteria for PD, proposes indicators of both severity of personality dysfunction in terms of levels of personality functioning, and trait dimensions in terms of five domains comprising 25 pathological traits. New criteria are proposed for only six types of PDs (schizotypal, antisocial, borderline, narcissistic, avoidant, and obsessive–compulsive). This is very different from the way the current DSM-IV manual presents PDs in which 10 discrete diagnostic categories are ordered in three overarching clusters. The new proposal aims to address three important shortcomings of the DSM-IV (Verheul, 2012). First, in DSM-5 PDs the specific commonalities of various PDs are now better conceptualized by formulating general diagnostic criteria and levels of personality functioning. Second, these levels also address variation in severity of PD. Third, since the DSM-IV, 10 categories did not capture the variability and complexity of PD, dimensions are introduced.

Using obsessive–compulsive PD as an example, the DSM-5 proposal describes significant impairments in personality functioning manifested under the construct of impairments in self-functioning by identity (sense of self derived predominantly from work or productivity; constricted experience and expression of strong emotions) or self-direction (difficulty completing tasks and realizing goals associated with rigid and unreasonably high and inflexible internal standards of behavior; overly conscientious and moralistic attitudes). Moreover, impairments in interpersonal functioning are addressed under the construct of empathy (difficulty understanding and appreciating the ideas, feelings, or behaviors of others) or intimacy (relationships seen as secondary to work and productivity; rigidity and stubbornness negatively affect relationships with others). The pathological personality traits are identified as compulsivity (characterized by rigid perfectionism) and negative affectivity (characterized by perseveration, followed by temporal stability, non-normative trait expressions). None of these impairments are directly attributable to and not solely due to the direct physiological effects of a substance or a general medical condition.

This new proposal is a major reconceptualization and certainly will have consequences for geriatric psychology and psychiatry. First, the complexity of the new construct of PD is challenging, especially diagnosing impairments in self (identity or self-direction) in older adults. To appropriately utilize the new diagnostic structure, an extensive amount of diagnostic expertise will be required of psychiatrists, geriatricians, and psychologists working in psychogeriatrics. It is debatable whether most practitioners actually have, or can realistically achieve, this level of expertise. Second, in the DSM-5 proposal, severity dimensions in PD diagnosis, although posed to be helpful, are also poised to be problematic. Do we have enough information about the kinds of severity and expressions of PD in old age (see Van Alphen et al., 2012b)? In what kind of context, for example, compared with younger adults or even compared with peer groups of healthy older adults, do we apply the severity appraisal? Do we need specific gero-cut-off points on the DSM-5 severity dimensions?

Certainly, an appropriate, valid, and reliable assessment instrument needs to be developed. Researchers are just beginning to develop specific personality assessments for older adults to measure age-specific features and contexts. To date, three measurement instruments have been specifically developed and validated for older people, namely the Gerontological Personality Disorders Scale (Van Alphen et al., 2006), the Hetero-Anamnestic Personality questionnaire (Barendse, et al., submitted), and a hybrid PD scale (Balsis and Cooper, submitted). Although these measures are helpful for older adults, at the moment we do not know how they will integrate with the DSM-5 proposal, and how they might actually identify issues relevant for older patients not captured by the DSM-5.

As far as we know, the Personality Inventory DSM-5 (PID-5; Krueger et al., 2011) is the first and only instrument based on the five maladaptive trait domains and 25 specific trait facets of DSM-5 PDs. This questionnaire contains 220 items, both in self-report and informant versions using a four-point Likert scale. Wright et al. (in press) replicated the initially reported five-factor structure in a sample of students, and as previously observed, a number of scales have significant cross-loadings (i.e., depressivity, perseveration, restricted affectivity, and risk taking). Moreover, it remains to be seen whether clinicians can rate these traits (and the disorders based on them) reliably and regard these as being useful. At the moment, there are also no scoring rules for determining types on the basis of traits. In addition, we do not know whether the factor structure will replicate in older adults. Clearly, the later life context was not explicitly investigated during the development of the PID-5. Nevertheless, Van den Broeck et al. (submitted) concluded that most PID-5 traits are measured equally well across ages. The PID-5 facets were examined for differential item functioning comparing a group of students and an adult community sample aged 60 years and older. Differential Test Functioning (DTF) analyses revealed a lack of measurement invariance for only four (withdrawal, attention seeking, rigid perfectionism, and unusual beliefs) of the 25 PID-5 facets. However, the relationship between types and trait domains was not empirically investigated and will need to be examined more carefully (e.g., Clarkin and Huprich, 2011). Concerning clinical utility, this self-report list of 220 items is likely to be fatiguing for some older adults, especially the oldest old and those most frail. Moreover, several items are also likely to be overly complex for the very old, such as items of the Psychoticism domain, like “It's weird, but sometimes ordinary objects seem to be a different shape than usual,” “I often see vivid dream-like images when I'm falling asleep or waking up,” or “Sometimes I feel ‘controlled’ by thoughts that belong to someone else.” This could influence the reliability and validity of the PID-5 for older adults. Both a semi-structured interview of the PID-5 and informant version of the PID-5 in psychogeriatric would seem to be preferable.

Another challenge is how to compare this new definition of PDs in DSM-5 with epidemiological studies with older adults related to the now outdated DSM-IV (TR) PDs? With the change in the DSM iterations, there is a question of what happens with the paranoid, schizoid, histrionic, and dependent PDs? We can be certain that the dysfunctional traits and behaviors associated with these deleted PDs will still exist, but we will lack appropriate diagnostic labels by which to identify them. Moreover, reducing the number of PDs does not eliminate comorbidity (e.g., Zimmerman et al., 2012). This is of course a problem that does not only apply to older adults. For patients of all age groups, it is important that useful clinical information from DSM-IV studies is not lost. Clinicians need to have a clear view on how the “old” DSM-IV categories and new DSM-5 traits relate.

Another potential stumbling block is the lack of clarity regarding what is meant by “personality functioning” and the criterion requiring that the individual's personality trait expression is “relatively stable” across time. How is “relatively stable” being operationalized? Does this mean stable across adulthood and into old age or for a finite period of years? With regard to older adults, it is known that the expression of PDs varies considerably according to the unique contexts and frequently occurring challenges of later life (Segal et al., 2006). It can be expected that forms of disturbed interpersonal relationships may be aggravated in old age in the various contexts of care (Sadavoy, 1987; Segal et al., 2006; Van Alphen et al., 2012a). Finally, some criteria of specific PDs are likely still age biased, such as the above-mentioned obsessive-compulsive PD with a focus on work and productivity, among others.

The DSM-5 PD proposal, as it now stands, includes a highly complex dimensional system to describe individuals’ personalities in terms of impairments in personality functioning and trait dimensions that are maladaptive in their extremity. Nearly all aspects of the proposed criteria lack sufficient empirical grounding, and may be especially problematic when applied to older adults. In addition, its sheer complexity portends enormous challenges to its utility. Clearly, the utility to clinicians and benefit to older patients must be investigated.

Future research also needs to attend to the temporal stability of the DSM-5 PDs into old age, the age-neutrality of the specific criteria for DSM-5 PDs, gero-cut-off points in the severity dimensions in PD diagnosis, and the validity of both the patient and informant versions of the PID-5 in community dwelling, inpatients, outpatients, and other contexts of living and care for older adults. A more comprehensive severity measure needs to be developed.

Conflict of interest

None.