This article is from a psychologist’s table paper titled “Personality Disorder, Its Nature, Stigmatization, Relationship to Mental Illness and Its Treatment Possibilities”, while we can appreciate that it is too wordy and complicated for most people to follow along with those who can will appreciate its publication here. Anyway, let’s get one with it…
The study of co-morbidity between major mental disorders and personality disorders is increasingly important in psychiatric research. Epidemiological surveys suggest that co-morbidity is highly prevalent in persons with mental disorders and is associated with severity and chronicity. The literature on Axis I/Axis II co-morbidity is reviewed in this chapter and the evidence for statistical associations between different diagnostic categories is presented. Axis I/Axis II co-morbidity is traditionally associated with poor response to treatment. But the research evidence for this may now need reconsideration. In the future, clinicians should develop programmes designed specifically for patients with Axis I/Axis II co-morbidity.
Co-morbidity is generally taken to imply the occurrence of independent psychiatric disorders in an individual patient. This is of considerable clinical and theoretical importance as the presence of one disorder can markedly affect the treatment, course, and phenomenology of another. Furthermore, the extent and pattern of co-morbidity will also indicate areas of overlap and redundancy among diagnoses. The study of co-morbidity is therefore of major importance in the refinement of psychiatric diagnosis. However, the process of separating personality disorders from major mental disorders into different axes of a multi-axial system is not without problems. The introduction of the DSM-III-R Manual pointed out that “there is no assumption that each mental disorder is a discrete entity with sharp boundaries (discontinuity) between it and other mental disorders”. Vize and Tyrer (1994) were critical of this system, concluding that considerable confusion remains and that many studies show a continuum between mental state and personality disorders for certain categories. Nevertheless, both the DSM and ICD glossaries persist with the categorical rather than the dimensional approach, as does most published research on personality disorders.
Epidemiology of Axis I co-morbidity
Considerably more is known about the co-morbidity of Axis I than Axis II conditions, with the exception of antisocial personality disorder (ASPD). The epidemiologic catchment area (ECA) study was the first to document that co-morbidity is widespread, not only amongst patients but also in the general population (Robins et al 1991). Over 54% of ECA respondents with a lifetime history of at least one DSM III psychiatric disorder were found to have a second diagnosis. Similar results were found in the more recent National Co-morbidity survey in the USA (Kessler 1995) where 56% also had one or more disorders. Associations were found between almost all disorders, especially between major depression and dysthymia and mania. Disorders of a single type, e.g. affective disorders, tended to be more strongly related to each other than another type e.g. substance use disorders. Only 21% of all lifetime disorders occurred in a sub-sample of respondents who had no additional lifetime co-morbidity. Thus 79% of disorders occurred with lifetime co-morbidity. Furthermore, over 50% of lifetime disorders occurred in 14% of the population who had a history of three or more disorders. Overall, co-morbidity was associated with both severity and chronicity
Axis I – Axis II co-morbidity
Excessive co-morbidity within Axis II can result from a series of diagnostic artefacts. Caron and Rutter (1991) pointed out that the DSM and ICD follow quite different approaches. Problems can occur with the DSM system where diagnoses are made on algorithms based on specified constellations but without regard to the presence or absence of accompanying symptomatology of a different kind, apart from a few exceptions. Consequently, when a patient has one diagnosis there is usually at least one other diagnosis. The ICD-10 system allows multiple diagnoses, but tends to discourage them by adoption of a pattern approach. Caron and Rutter argue that the underlying concept of the WHO approach is probably correct because it is more likely that a patient would have a single disease rather than several.
Loranger (1992) reviewed the literature on research diagnostic instruments and concluded that self-reporting inventories are not suitable for making personality disorder diagnoses, although they may have potential use as screening devices. Self-report instruments produce considerably more Axis II categories than research diagnostic interviews, indicating a high level of false positives and inevitably leading to excessive co-morbidity. Earlier versions of interviews also experienced problems. Artificial co-morbidity could arise when the same criterion appeared in more than one diagnostic category. This problem is largely eliminated from recent interview schedules. But subtle differences between superficially similar criteria in two or more different categories does require considerable sophistication in interviewing skills.
Additional problems can arise when disorders defined in terms of one main symptom complex are subdivided into various sub-categories, particularly in the DSM system. In some cases, one disorder may represent an early manifestation of another, or could be part of, or a secondary manifestation of, another. However, studying artefactual co-morbidity may be helpful in throwing further light on the nature of the disorders involved. Caron and Rutter (1991) proposed four potential processes which can lead to true co-morbidity
Caron and Rutter (1991) proposed four potential processes which can lead to true co-morbidity:
1) Shared risk factors: Overlap may occur when two disorders share the same risk factor or factors. Many psychiatric disorders are multi-factorial in origin and many aetiological factors are not diagnosis-specific.
2) Overlap between risk factors: Even when the risk factors for two disorders are distinct and different, there may still be co-morbidity because the risk factors themselves are associated. In this case, the individual may be at risk for two separate conditions.
3) The co-morbid pattern constitutes a meaningful syndrome: A constellation with a co-morbid set of symptoms which shows relatively poor response to treatment compared to a “pure” constellation of symptoms would tend to suggest that there are indeed two separate conditions present.
4) One disorder creates an increased risk for the other Life stressors can be risk factors for psychiatric disorder. Evidence suggests that individuals shape and select their own environments and that various psychiatric disorders play a part in generating stress and adversity. For example, stress induced by one disorder, such as ASPD, could lead to a higher risk of other disorders, e.g. depression.
Clinical difficulties in differential diagnosis
In the UK, undergraduate and post-graduate training schemes do not routinely teach how to diagnose personality disorders. In the USA, residents training in psychiatry are taught to consider both Axis I and Axis ||| disorders when carrying out a diagnostic formulation. But even with training in the multi-axial approach, it can still be difficult to make a satisfactory differential diagnosis. For example, it can be exceptionally difficult to make an Axis Il diagnosis in the presence of pervasive Axis I disorder, especially severe psychotic illness. Furthermore, personality psychopathology may complicate both the diagnosis and the treatment of Axis | disorder. Even in the absence of Axis | disorder, personality disorder psychopathology may not be initially apparent and some features of Axis Il disorder may not be diagnosed until the clinician knows the patient well. The theoretical orientation of the clinician may also influence whether an Axis I or Axis Il diagnosis is made. Nevertheless, it is essential that an adequate differential diagnosis is made because Axis Il disorders can be as severe and disabling as Axis I disorders.
Clinical difficulties are influenced by several theoretical difficulties. It remains unclear whether personality disorders should always be separated from clinical syndromes in a multi-axial system. Is there a continuum between mental state and personality disorders? For example, it has been argued that borderline personality disorder (BPD) may be a variant of affective disorder (Coid 1993). Similarly, there is increasing consensus that there is no clear dividing line between avoidant personality disorder and social phobia. A recent study of aetiology of personality disorders further supported the argument that not all should be contained in a separate Axis (Coid 1999). Preliminary findings suggest that avoidant and BPD fail to maintain diagnostic stability over time in a significant percentage of patients, undermining the criterion of temporal stability within the definition of personality disorder.
Studies of inter-Axis co-morbidity
A literature search was carried out of studies of inter-axis co-occurrence using Medline and Psychlit. This was supplemented by a hand search of references from available publications. Only studies which employed a standardised assessment procedure to establish both Axis I and Axis II diagnoses were included. Studies which used self-report data to establish Axis Il diagnoses were excluded. Only two previous studies were found which examined the full range of disorders from both Axes and then carried out a statistical examination of co-morbidity between the individual categories (Alnaes and Torgerson 1988, Zimmerman and Coryell 1989). Four other studies carried out statistical analyses after combining personality disorders into clusters or
nical syndromes into broad groupings (Oldham et al 1995. Maier et al 1991. Nestadt et al 1992), or else limited the number of conditions studied (Jackson et al 1991).
An alternative and less satisfactory method of examining Axis I/Axis Il co-morbidity is to include studies which selected a single Axis | category but examined the full range of Axis Il disorders. This method is limited in that few Axis Il disorders occur in isolation over the lifespan in a single subject, as has been discussed above. Differences between studies may have emerged due to the presence of additional but unreported Axis | conditions which confound the findings. The differences between samples and research diagnostic instruments will also contribute to the variability between studies. However, bearing these limitations in mind, the main findings from this review are summarised in Table I.
The suggestion that mood disorders correlate with Cluster B disorders (Siever and Davis 1991), particularly BPD, received moderate support from the review of literature. There was greater co-morbidity with avoidant and dependant personality disorders from Cluster C. There was a strong association between dysthymia and BPD, but this was not always the case for depressive disorder. The relationship between BPD and mood disorders is not entirely resolved (Kroll and Ogata 1987, Gunderson and Phillips 1991). The defining features of BPD include affect dysregulation, physically self-damaging acts, chronic feelings of emptiness, and intense relationships. BPD may represent a variant of mood (and impulse) pathology in the same way that schizotypal personality disorder is a characterologic variant of schizophrenic pathology. It has been argued that BPD represents a diagnosis that is literally on the boundary of mood disorders and may thereby represent both a personality and a mood disorder (Widiger 1989).
Several studies demonstrate high levels of co-morbidity between schizophrenia and related psychotic disorders and schizotypal personality disorder and, in certain studies, with BPD and paranoid personality disorders. The latter demonstrate moderate consistency across studies (Zimmerman and Coryell 1989, Oldham et al 1995, Jackson et al 1991). Cluster A (schizotypal, paranoid, schizoid) disorders are more prevalent in subjects with lifetime schizophrenia and schizoaffective disorders compared to other categories. Schizotypal personality disorder was a new addition to the DSM-III and intended to describe certain psychopathological characteristics which are usually stable and assumed to be genetically related to a spectrum of disorders, including chronic schizophrenia (Spitzer et al 1979). However, the ICD-10 now includes this condition in the section with schizophrenia and not as a personality disorder.
Substance use disorder
Co-morbidity between substance use disorders and Axis II is complex. There is strong evidence that personality disorder may act as a risk factor for the development of substance abuse. Axis Il disorders also modify the course of dependence on alcohol and drugs. Subjects with co-morbidity tend to have earlier onset of substance abuse, more additional psychiatric symptoms, and higher rates of polysubstance abuse. Similarly, features of Axis II psychopathology may emerge as a consequence of substance abuse, although there is less evidence of this phenomenon.
Drug dependency has been found to have a high level of co-morbidity with ASPD followed by BPD. However, the internal consistency of this research is compromised by the questionable independence of ASPD and substance use disorders (Gerstling et al 1990). Several criteria are similar in each condition. When early onse of drug abuse occurs, it can be difficult to decide whether the criteria for ASPD are the result of drug-seeking behaviour, including a criminal lifestyle.
Studies are consistent in demonstrating co-morbidity between avoidant and phobic disorders, especially social phobia, followed by dependent personality disorder. BPD also appeared moderately prevalent and was somewhat more so in the context of lifetime studies of Axis | diagnostic data. However, as has already been noted, there no longer appears to be strong evidence for retaining avoidant personality disorder within a separate Axis from social phobia (Herbert et al 1992, Turner et al 1993).
Anxiety and panic disorder
Anxiety and panic disorders demonstrate a similar pattern of Axis Il co-morbidity with Axis Il studies to the phobic disorders, although the overall prevalence of personality disorders tends to be somewhat lower and with less consistency observed between studies. Overall, these tend to show moderately elevated associations with borderline, avoidant, and dependent personality disorders.
Obsessive compulsive disorder The review demonstrated a moderate degree of consistency across five studies, demonstrating as expected, that the highest level of co-morbidity was between obsessive-compulsive disorder, compulsive (now obsessive-compulsive) personality disorder and dependent personality disorder. The relationship between obsessive-compulsive disorder and compulsive personality disorder continues to be a matter of debate between those who traditionally view the two conditions along a continuum and those who strongly support separation between the two elements.
There was a relatively high degree of consistency for association between eating disorders and BPD, followed by avoidant and dependant personality disorders. Binge-eating can be a sign of generalised impulsivity and is included among the criteria for impulsivity in the diagnosis of BPD, indicating a strong potential for overlap. Lacey and Evans (1986) suggest that a “multi-impulsive” subgroup of bulimics exists with many features of BPD and a poor response to treatment.
Post-traumatic stress disorder
Post-traumatic stress disorder is associated with high levels of co-morbidity with several other Axis | conditions. There have been few studies of association with Axis Il disorders. The evidence suggests that BPD is most prevalent, followed by compulsive, avoidant and paranoid personality disorders. This could reflect an overlap between diagnostic criteria within two Axes. Several criteria for avoidant personality disorder are similar to symptoms of withdrawal and emotional numbing within PTSD. Similarly, hypervigilance and diminished trust overlap with features of paranoid personality disorder.
There are two views of character pathology within PTSD. Firstly, that personality disorder traits were present in these patients before the trauma occurred and have thereby contributed to the actual development of PTSD. Secondly, severe trauma may be sufficient to cause secondary change in character in individuals who have normal pre-morbid functioning. Thus, the high prevalence of BPD could be partly explained by trauma.
Many of these findings are preliminary, but demonstrate the importance of further research and that clinicians should not always assume poor outcome for mental disorders with co-morbid personality disorder. However, there remain a series of problems to be overcome. These include artefactual co-morbidity resulting from poor diagnostic validity and reliability, and diagnostic practices that lead to the misclassification of both Axis I and Axis Il disorders. Future studies should include epidemiologically representative samples, particularly samples drawn from the community. Many previous studies have examined a single mental disorder then measured the prevalence of personality disorders in highly selected clinic samples, increasing the risk of Berkson’s bias. Statistical error as a result of confounding is rarely addressed within these studies and analysis may require logistic regression for categorical data or multiple regression when using dimensional scores. It will also be necessary to incorporate aetiological measures in an attempt to examine whether apparently separate diagnostic categories have similar risk factors.
The majority of studies suggest that co-morbid personality disorder leads to a poorer outcome for major mental disorder. But it is time to re-evaluate this evidence as this may have been overstated for certain conditions, and plan for co-morbid Axis II psychopathology in future treatment programmes. In the past, patients have sometimes been excluded from inpatient services in the UK on the basis of a diagnosis of personality disorder, even when suffering from severe mental illness, but where difficult and unco-operative behaviour has been anticipated. The development of new treatments for personality disorder remains in the early stages. But in future, it will be increasingly important and necessary to make accurate clinical diagnoses of personality disorder when co-morbid with major mental disorder, devise suitable treatment programmes, and evaluate their effectiveness.