ReviewComorbidity of personality disorders in mood disorders: A meta-analytic review of 122 studies from 1988 to 2010
Introduction
Unipolar depressive disorders (DD) and bipolar disorders (BD) may be severe and disabling. DDs are considered the 3rd leading cause of disease burden world-wide (WHO, 2008), but ranked as number one in terms of DALY (number of life years lost due to disability) (Wittchen et al., 2011). They have the second highest life-time prevalence (21%) of all mental disorders, only beaten by anxiety disorders (29%) (Kessler et al., 2005). The life-time prevalence of BDs is much lower, about 2.4% (Merikangas et al., 2011), and the burden of disease (WHO, 2008) and DALY (Wittchen et al., 2011) are also lower compared with DDs. However, occupational functioning is usually more compromised in BDs (Goldberg and Harrow, 2004, Judd et al., 2008) as the number of work days lost is about 2.5 times higher than in DD (Kessler et al., 2006).
Simultaneous co-occurrence of a personality disorder (PD) contributes strongly to the increased disease burden and DALY. Having a comorbid PD impairs the outcome or prognosis for several reasons: it affects treatment adherence negatively (Pompili et al., 2009) and thereby treatment response indirectly, it implies a higher level of general psychopathology (Sanderson et al., 1992), it more than doubles the risk for a poor outcome following antidepressant treatment (Newton-Howes et al., 2006), it hampers psychosocial and occupational functioning considerably (Cummings et al., 2011, Markowitz et al., 2007, Skodol et al., 2005), and the risk of developing additional axis I disorders, such as anxiety disorders (Stein et al., 1993), is markedly higher. Patients with additional PDs are also more costly to the health care system than patients with depression or anxiety alone (Soeteman et al., 2008). Having a comorbid PD is therefore a top risk factor for a poor prognosis or outcome (Holzel et al., 2011, Skodol et al., 2011). Compared with comorbid axis I disorders, a comorbid PD will have a more negative impact on the course and treatment of depression (Reich, 2003) through an elevated risk of drop-out (McFarland and Klein, 2005), reduced motivation and less positive treatment expectations (Martino et al., 2012), and a more fragile therapeutic alliance (Bienenfeld, 2007, Cummings et al., 2011, Martino et al., 2012). The degree of impairment may however co-vary with the type of PD, and is for example more pronounced in schizotypal or borderline PDs than in obsessive-compulsive (OC-PD) or avoidant PDs (Skodol et al., 2002).
We did not locate any previous meta-analyses examining the proportions of PD comorbidity in mood disorders, except a qualitative review by Corruble et al. (1996) on depressed patients only. In their study, cluster A PDs were the least and cluster C the most prevalent (Corruble et al., 1996). Within cluster A, the schizoid and schizotypal PDs varied the most, while paranoid PD had consistently lower estimates. Within cluster B, borderline and histrionic PDs varied the most. The cluster C PDs had the highest prevalence and also the largest variance between studies (ranging from 5% to 65%), except for the OC-PD subtype, which had low estimates. The review (Corruble et al., 1996) was however limited by a low number of studies (25 papers), it only included studies of depression and did not use statistical methods providing weighted average comorbidity estimates. In a qualitative review by Torgersen (2009) on the prevalence of PDs in the general population, a similar internal ranking was evident (highest prevalence of cluster C and lowest of cluster A) although the prevalences were much lower.
A meta-analysis may more precisely determine the true average proportion of comorbid PDs in mood disorder patients, and thus indicate the level of attention that should be paid to constellations of mood and personality disorders in clinical assessment and treatment. In particular, the large observed between-study variances calls for an extended examination of heterogeneity in PD comorbidity, and in case of significant variance in comorbidity after partialling out sampling error variance, a meta-analysis may examine to what extent methodological or clinical features co-vary with comorbidity. Potential moderating factors may be diagnostic system (DSM-III-R versus DSM-IV), diagnostic methods (interview versus questionnaire), patient characteristics (inpatients versus outpatients), mean age of onset and duration of a mood disorder.
A consistent finding has been that PD assessments based on questionnaire methods yield higher comorbidity estimates than assessments based on diagnostic interviews in eating and anxiety disorders (Friborg et al., 2013, Ramklint et al., 2010, Rosenvinge et al., 2000). A similar pattern is expected for mood disorders. Intuitively, one may expect a higher proportion of PDs among inpatients than outpatients, as inpatients generally are more severely ill (Holma et al., 2008). In a meta-analysis on anxiety disorders (Friborg et al., 2013), the PD comorbidity was higher among inpatients than outpatients with panic disorder, while the opposite trend was seen in OC-PD. A meta-analysis on eating disorders found higher rates of comorbid PDs among inpatients than outpatients (Rosenvinge et al., 2000), as expected. However, in the review by Corruble et al. (1996) on PD comorbidity in depression, outpatients were worse off than inpatients. Hence, a clear-cut prediction for the present meta-analysis could not be made. With respect to age of onset, the findings were more clear-cut, as we found six studies which all indicated more prevalent PDs among early compared with late onset of a DD (Abrams et al., 1994, Camus et al., 1997, Fava et al., 1993, Klein et al., 1999, Rothschild and Zimmerman, 2002, Sato et al., 1999). An early age of onset was therefore expected to yield a higher comorbidity incidence. As the duration of a DD is partly related to age of onset, one may expect that a longer duration of DD also would imply a higher degree of comorbid PDs, albeit not as clearly. The between-study variation in comorbidity was estimated as random-effects in order to quantify whether the degree of heterogeneity was high enough warranting further moderator analyses.
Mood disorders represent a heterogeneous group of psychiatric disorders, and sub-groups had to be defined to reduce the number of analyses. First, we preferred to keep the established division between bipolar and unipolar disorders, due to a long clinical tradition as well as evidence supporting it. Recurrent depressions are for example more disruptive for the long-term outcome in BD than in DD (Goldberg and Harrow, 2011) and crossover from DD to BD occurs in 1 of 5 mood disorder patients (Scott, 2011), while the reverse is rare. We did not separate between bipolar I and II, as studies on bipolar II were scarce.
Second, we chose to further divide the unipolar group into dysthymic disorders (DYS) and major depressive disorder (MDD). These two conditions share many risk factors and comorbidity patterns (Blanco et al., 2010), and they sometimes occur together, a condition called double depression. The degree of functional loss across these two groups seem comparable, as some studies indicating lower level of function in MDD (Judd et al., 2000, Rapaport et al., 2005), others in DYS (Buist-Bouwman et al., 2006, Rhebergen et al., 2010). However, clinically they are quite different, as the diagnostic duration criterion is two years for DYS versus two weeks for MDD (APA, 1994). Another noticeably difference is that patients with DYS have markedly lower remission and recovery rates than MDD patients (Klein et al., 2006, Rhebergen et al., 2010), which is a bit contra-intuitive, given the normally higher symptom load in MDD. The mean age of onset in DYS is lower than in MDD (Yang et al., 2011). In combination with the two-year diagnostic duration criterion, we expected patients with DYS to have more personality problems than MDD patients (Klein et al., 1999, Ryder et al., 2001) Weighted average proportions for MDD and DYS were therefore estimated separately.
The purpose of this paper was to examine (1) the proportion of comorbid PDs across the main diagnostic groups of mood disorders, i.e., bipolar disorder (BD), major depressive disorder (MDD) and dysthymic disorder (DYS), (2) how the proportion of PDs vary across studies, and (3) to what extent the rate of comorbid PDs co-vary with clinical and methodological features, such as gender, age of onset, duration of illness, diagnostic system, diagnostic methods and sample characteristics?
Section snippets
Literature search
The current study was part of a larger meta-analytic project on the comorbidity between axis-I symptom disorders (depression, anxiety- and eating disorders) and axis-II PDs, of which two papers (on anxiety and eating disorders) have been published (Friborg et al., 2013, Friborg et al., 2013). Studies published in English or German between 1988 and 2010 on subjects having a current or a lifetime unipolar DD and a comorbid PD were located using the databases PsychINFO, Embase and Medline. We used
Sample characteristics
The total number of patients in the 122 studies was N=24 867. The study pool constituted k=59 studies on outpatient samples (n=10,367, 41.7%,), 28 studies on inpatient samples (n=4866, 19.6%), 16 studies on patients receiving both out- and inpatient services (n=7113, 28.6%), five studies on recruited samples (n=1069, 4.3%), five studies on recruited outpatient samples (n=425, 1.7%), and nine unspecified samples (n=1027, 4.1%). Most studies were conducted in America (k=76: 71 in USA, five in
Summary and consideration of the results
The present meta-analysis was based on 122 studies. Most of the data on MDD and DYS came from outpatient samples, while studies on BD more often included inpatient samples. The mean age of onset was comparable for MDD and BD (in the late twenties). It was considerably lower in DYS (beginning in the early twenties), and these patients also had a longer history of illness.
Conclusion
The comorbidity between mood disorders and personality disorders is strong, and the risk of having at least one comorbid PD was high across all mood disorders and highest in dysthymic disorder. Cluster C PDs dominated in unipolar depression, while cluster B and C PDs were comparably frequent in bipolar disorders. The reported comorbidity was lower when diagnoses were based on structured clinical interviews. Compared to self-report measures, clinical interviews are time consuming, nevertheless
Role of funding source
The work with the paper has not been externally funded. Funding has been provided internally by the Department of Psychology at the Faculty of Health Sciences, University of Tromsø, Norway.
Conflict of interest
None of the authors have any conflict of interests to declare.
Acknowledgments
None.
References1 (80)
- et al.
Impact of anxiety disorder comorbidity on quality of life in euthymic bipolar disorder patients: differences between bipolar I and II subtypes
Journal of Affective Disorders
(2008) - et al.
Are personality disorders more frequent in early onset geriatric depression?
Journal of Affective Disorders
(1997) - et al.
Comorbidity of personality disorders and unipolar major depression: a review
Journal of Affective Disorders
(1996) - et al.
Personality disorders in elderly patients with dysthymic disorder
American Journal of Geriatric Psychiatry
(2000) - et al.
Recollection deficiencies in patients with major depressive disorder
Psychiatry Research
(2010) - et al.
Comorbidity of personality disorders in anxiety disorders: a meta-analysis of 30 years of research
Journal of Affective Disorders
(2013) - et al.
Consistency of remission and outcome in bipolar and unipolar mood disorders: a 10-year prospective follow-up
Journal of Affective Disorders
(2004) - et al.
Risk factors for chronic depression-A systematic review
Journal of Affective Disorders
(2011) - et al.
Psychosocial disability and work role function compared across the long-term course of bipolar I, bipolar II and unipolar major depressive disorders
Journal of Affective Disorders
(2008) - et al.
Age of onset in chronic major depression: relation to demographic and clinical variables, family history, and treatment response
Journal of Affective Disorders
(1999)
Early- versus late-onset dysthymic disorder: comparison in outpatients with superimposed major depressive episodes
Journal of Affective Disorders
Mental health service use by patients with dysthymic disorder: treatment use and dropout in a 7 1/2-year naturalistic follow-up study
Comprehensive Psychiatry
Assessing personality disorders in eating disordered patients using the SCID-II: influence of measures and timing on prevalence rate
Personality and Individual Differences
Trajectories of recovery of social and physical functioning in major depression, dysthymic disorder and double depression: a 3-year follow-up
Journal of Affective Disorders
Axis II comorbidity in euthymic bipolar disorder patients: no differences between bipolar I and II subtypes
Journal of Affective Disorders
Prevalence of personality disorders in patients with major depression and dysthymia
Psychiatry Research
The size and burden of mental disorders and other disorders of the brain in Europe 2010
European Neuropsychopharmacology
Age at onset of major depressive disorder in Han Chinese women: relationship with clinical features and family history
Journal of Affective Disorders
Personality disorder correlates of late and early onset depression
Journal of the American Geriatrics Society
Comorbid personality disorders in subjects with bipolar I disorder
International Journal of Psychiatry in Clinical Practice
Diagnostic and Statistical Manual of Mental Disorders
Diagnostic and Statistical Manual of Mental Disorders
Diagnostic and Statistical Manual of Mental Disorders
Cognitive therapy of patients with personality disorders
Psychiatric Annals
The epidemiology of chronic major depressive disorder and dysthymic disorder: results from the national epidemiologic survey on alcohol and related conditions
Journal of Clinical Psychiatry
Comprehensive Meta-analysis (Version 2.) [Computer Software]
Introduction to Meta-analysis
Functional disability of mental disorders and comparison with physical disorders: a study among the general population of six European countries
Acta Psychiatrica Scandinavica
The long-term course of rapid-cycling bipolar disorder
Archives of General Psychiatry
Navigating therapeutic alliance ruptures in cognitive therapy for avoidant and obsessive–compulsive personality disorders and comorbid axis I disorders
International Journal of Cognitive Therapy
Personality traits and personality disorders
British Journal of Psychology
Mania, depression, and mood dependent memory
Cognition and Emotion
Patterns of personality disorder comorbidity in early-onset versus late-onset major depression
The American Journal of Psychiatry
Anger attacks in unipolar depression, part 1: clinical correlates and response to fluoxetine treatment
American Journal of Psychiatry,
The problems in using fixed-effects models of meta-analysis on real-world data
Understanding Statistics
Personality Disorders in Eating Disorder NOS (EDNOS) and Binge Eating Disorder (BED): a meta-analysis of comorbidity studies
Journal of Mental and Nervous Disease
Patterns of relation and premorbid personality in late paraphrenia and depression
Psychopathology
A 15-year prospective follow-up of bipolar affective disorders: comparisons with unipolar nonpsychotic depression
Bipolar Disorders
Cited by (157)
Validity of PDQ4+ self-screening to assess the impact of personality disorder traits on the course of bipolar disorder
2024, Journal of Affective Disorders ReportsCourse and clinical correlates of obsessive-compulsive disorder with or without comorbid personality disorder
2024, Journal of Affective DisordersIndicators of Clinical Severity in the Comorbidity of Bipolar Disorder and Borderline Personality Disorder: A Systematic Review
2023, Revista Colombiana de PsiquiatriaHow can the DSM-5 alternative model of personality disorders advance understanding of depression?
2023, Journal of Affective Disorders
- 1
Note. Papers marked with an asterix (⁎) were included in the meta-analysis.