Occurrence of bipolar spectrum disorder and comorbidities in women with eating disorders

There is a growing body of evidence concerning bipolar disorder (BD) comorbidity in patients with eating disorder (ED) (Blinder et al. 2006;McElroy et al. 2005). Higher rates of BD compared to the general population have been reported among ED patients (Hudson et al. 2007;McElroy et al. 2005) and have also been associated with early onset, suicidal ideation, residual symptoms, treatment resistance, and worsening course (Wildes et al. 2007;McElroy et al. 2001;Fedorowicz et al. 2007;Kaye et al. 2004).

The wide range in rates of BD comorbidity in patients with ED is likely due to the use of different definitions of bipolarity and the boundaries between ED and BD (McElroy et al. 2005). Halmi et al. (1991) found higher rates of atypical BD in female anorexia nervosa (AN) patients than in the control group (13.3% vs. 0%). Evaluating a community sample, Hudson et al. (2007) found the occurrence of bipolar I (BP I) and II (BP II) disorders ranging from 3% in AN to 17.7% in bulimia nervosa (BN), including 12.5% in binge eating disorder (BED) and 10.5% in subthreshold BED.

Moreover, the prevalence of EDs not meeting the formal criteria for AN or BN, such as ‘eating disorders not otherwise specified’ (EDNOS) or subthreshold ED spectrum disorders is considerably higher in BD patients (up to 3.6% to 10%) (McElroy et al. 2005). Evaluating 50 AN patients regarding manic and depressive symptoms, Wildes et al. (2007) found that 86% presented threshold symptoms for depression and 60% for mania, and it was associated with a history of self-induced vomiting and suicidality.

Despite the recognized association between mood disorders and ED, few studies described the relationship with BD and its broader definition of bipolar spectrum (BS). Most studies have been conducted using the DSM-III-R criteria, and the estimates of prevalence of comorbid disorders in psychiatric settings may be skewed by the imperfect recognition of BD (Mantere et al. 2004).

To our knowledge, the prevalence of soft forms of BD (i.e., bipolar spectrum) in ED patients is not yet reported. There is a need to analyze the prevalence and correlates of BS in ED, first because BS may be underdiagnosed due to phenomenology overlap and the severity of ED. Second, the axis I comorbidity pattern could be clinically useful in helping recognize BS in this population. Third, understanding this overlap raises the question of shared vulnerability and, possible, physiopathology. The aim of this study was to determine the prevalence of BS in a well-defined sample of ED patients, evaluating the prevalence of other axis I disorders.

The methods of ESPECTRA project are described in greater details elsewhere (Campos et al. 2011). Women aged 18 to 45 years were recruited from the Eating Disorder Unit (AMBULIM) of the Institute of Psychiatry of the Clinical Hospital of the University of São Paulo, School of Medicine. Considering that ED is much more frequent in women than in men (Woodside et al. 2001), we just enrolled females in order to exclude a possible gender effect. Patients were included if they met the DSM-IV-TR criteria for AN, BN, or EDNOS. BED patients were excluded, and the absence of other comorbidities is not an exclusion criterion (Campos et al. 2011).

The evaluation included completion of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) (First et al. 1996) to establish the diagnosis of eating disorders and comorbidities (psychotic disorders, mood disorders, anxiety disorders, alcohol and substance use disorders). Substances evaluated were amphetamine, cannabis, cocaine, hallucinogen, opioids, and sedative/hypnotic/anxiolytic. A clinical interview based on the Zurich criteria identified the BS, and structured questionnaires determined demographic factors, illness information, and family history.

The protocol was approved by the local ethics committee, and it conformed to the 2008 Helsinki Declaration (CAPPesq protocol number: 1154/09). Written informed consent was obtained from all participants prior to any study-related activities.

Data were analyzed using SPSS for Windows (version 14.0, SPSS, Chicago, IL, USA). Descriptive statistics included means and standard deviations (SD) for continuous variables and percents for categorical variables. Chi-square test and student's t test or Mann–Whitney were used for group comparisons. All statistical tests were two-tailed, and a significant level of 5% was considered for all tests.

A total of 69 individuals with ED were evaluated. The mean current age was 27.9 years (SD 6.4), most of the participants were not married (72.5%), and the mean offspring number was 0.46 (SD 0.98). Patients reported 13.3 years (SD 3.5) of education, 63.8% were not working, and 36.2% had no religion.

The present study shows a high prevalence (68.1%) of BS in a well-defined population of 69 ED patients. This high rate of BS is compatible with the findings of Alciati et al. (2007) of 89% in 83 severely obese patients seeking surgical treatment. The definition of BS has been supported by many studies showing that it is a highly prevalent condition with clinical implications, and there is a need to redefine comorbidities according to this knowledge (Akiskal et al. 2000). The Zurich definition of BS (Angst et al. 2003), epidemiology based, enriched the understanding of soft bipolarity challenging the current DSM concepts. The broader definitions result directly in strong increase of the prevalence of BS. A reanalysis of the database of the Epidemiological Catchment Area (ECA) study indicated that 6.4% of the general population met the criteria for bipolar spectrum (Judd and Akiskal 2003), while the original study found 0.8% for BD. Data from the Sao Paulo Epidemiological Catchment Area study showed that the lifetime prevalence of BS was 8.3% (the softer forms of BD representing 6.6%) and it was of great clinical importance (Moreno and Andrade 2005;Moreno and Andrade 2010). Merikangas et al. (2011) in a large study across 11 countries found that the overall prevalence of BS was 2.4%. In addition to the findings mentioned above, we extended previous investigations by demonstrating that ED is associated with BS in its broader definition.

The BS is underdiagnosed in clinical practice (Angst et al. 2003;Bschor et al. 2012), and comorbidity often is an important factor that contributes to the delay in its identification. Patterns of psychiatric comorbidity suggest one possible strategy for improving recognition of bipolar disorder among patients presenting with depressive symptoms, and these patients may benefit from additional screening for bipolar disorder (Matza et al. 2005). We compared the overall comorbidity profile of ED patients with or without bipolar spectrum to obtain a comprehensive view of the differences of current comorbidity with ED. We found patterns of psychiatric comorbidity in ED with and without BS to differ somewhat qualitatively. This finding is in line with the results of previous studies that indicate an association of ED with greater substance use (Bulik et al. 2004).

Because of the severity of ED, comorbidities with other axis I disorders may be neglected. It is important to note that anxiety disorders and substance use are also highly prevalent, representing conditions that should be systematically assessed. The understanding of how comorbidities can modify the course and outcome in this population can be of clinical utility both in the diagnostic and treatment field. Once these comorbidities become more frequent than would be expected, it raises the question about pathophysiological implications.

These findings must be considered in view of some methodological limitations. First, historical illness variables were obtained retrospectively. Second, our group of ED patients was selected in a tertiary and very specialized research center and may not be representative of the community, and results may not be generalized. Our study has several strengths such as the use of specific and validated criteria for a broader definition of BD, and structured interview and validated instruments for the evaluation of axis I comorbidities.

This study supports previous data (McElroy et al. 2005;Brietzke et al. 2011;Hudson et al. 2007) concerning the relationship of ED and bipolar disorders, and shows the importance of carefully evaluating ED patients. BS is highly prevalent in women with ED and strongly associated with substance use disorder comorbidity. A common liability may underlie the expression of these disorders in a single externalizing spectrum. One possibility is that this liability is continuous and the risk of externalizing disorders is graded. Another possibility is that the liability underlying the externalization of spectrum disorders is categorical such that individuals fall into groups of liability (Markon and Krueger 2005). A broader evaluation of psychiatric disorders beyond the standardized categorization criteria can contribute to a more realistic understanding of comorbidities.

A much more comprehensive assessment and measurement of psychopathology can enhance operational diagnoses and lead clinicians to broader examinations (Angst 2007), and lifetime substance use is a correlate that can be of great importance in differentiating disorders that have common characteristics. This kind of information can build a body of evidence supporting a dimensional approach in diagnosing these conditions, and as long as new studies incorporate this wider approach, the results will be directly seen in the number of patients diagnosed.