هم ابتلایی در بی اشتهایی عصبی و بولیمیا(پرخوری عصبی): طبیعت، شیوع، و روابط علت و معلولی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|32487||2003||18 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Clinical Psychology Review, Volume 23, Issue 1, February 2003, Pages 57–74
Eating disorders are complex, multifactorially determined phenomena. When individuals with eating disorders present for treatment with comorbid conditions, case conceptualization is further complicated and, as a result, it may be difficult to determine optimal psychological or pharmacological treatment. This article reviews the evidence of the association between eating disorders (anorexia nervosa [AN] and bulimia nervosa [BN]) and Axis I depression, obsessive–compulsive disorder (OCD), substance abuse, and Axis II personality disorders, for the purposes of increasing awareness about the different options for case conceptualization. Although other diagnoses comorbid with eating disorders are of interest to clinicians (e.g., posttraumatic stress disorder [PTSD] and social phobia), their comprehensive review is currently premature due to a lack of empirical scrutiny. Finally, future directions for research, including suggestions for the use of particular assessment tools and more sophisticated research designs, are discussed.
The clinician working with individuals who have eating disorders is all too aware of the fact that comorbid psychiatric diagnoses often complicate case conceptualization and treatment planning. To assist in clarifying the nature of psychiatric comorbidity in anorexia nervosa (AN) and bulimia nervosa (BN), this review will describe their most commonly co-occurring Axis I and Axis II conditions, and will review the evidence as to the nature of these relationships. While other Axis I and Axis II conditions are of interest, we included only those disorders in which there is a sufficient quantity of data to allow for thorough discussion and generation of tentative statements regarding causality. A large body of recent research has attempted to define risk factors for development of general psychopathology and specific risk factors for eating disorders (e.g., Fairburn et al., 1999 and Fairburn et al., 1997). Some researchers have analyzed and condensed the results of this research to form etiological models for the development of eating disorders (e.g., Crowther & Mizes, 1992). Other researchers have utilized twin studies as a way of determining the relative contributions of genetic and environmental risk factors to the development of eating disorders. For example, Bulik, Wade, and Kendler (2001) evaluated monozygotic twins discordant for BN in order to clarify environmental risk factors for the disorder. However, twin studies have demonstrated divergent findings and are difficult to interpret when considered as a group (Fairburn, Cowen, & Harrison, 1999). In this review, we emphasize the discussion of Axis I and II comorbidity with eating disorders, rather than categorization of variables into general and specific risk factors for development of eating pathology. We provide a comprehensive descriptive analysis of the extant comorbidity literature, and discuss possible relationships between disorders while making tentative statements regarding causality. A considerable body of evidence, both empirically and clinically based, suggests a relationship between the eating disorders and specific Axis I and II psychopathology. While there is a wealth of empirical data in this area, heterogeneity in research results makes drawing clear conclusions about the nature of these associations challenging. Reasons for the diversity of findings in this area are that the diagnostic criteria for both AN and BN have changed over time (Halmi, Kleifield, Braun, & Sunday, 1999), discrepant methods of diagnosis are used across studies (e.g., structured clinical interview versus psychometric data only), small sample sizes, and a lack of comparison group data. Even with these caveats, it is possible to say that some disorders frequently coexist with particular subtypes of eating disorders (e.g., binge eating vs. restricting), and even make tentative statements about causality. Although eating disorders affect both males and females, the literature on males with eating disorders is sparse and will not be reviewed here. In the occasional instances in which males are included in a study, this will be noted. In this review, we have considered studies which use psychometric data to determine levels of psychopathology, as well as studies which use clinical interview-based methods. Although interview-based methods have advantages in terms of greater diagnostic specificity, they are not always practical in research settings. In contrast, psychometric evaluations provide quantitative information about the relative severity of a problem, but do not speak to diagnosis and often do not capture the real-life impairments associated with a problem (e.g., days missed at work). Consequently, neither method is viewed as flawless and both approaches will be considered here. Multimodal assessment (both interview and self-report) is suggested as preferable for research in this area. In the latter part of the paper, additional specific suggestions are made encouraging the development of assessment strategies and research designs, which will be of maximal benefit to the field at large. Researchers have long noted that comorbidity research focussing on individuals who present for treatment at clinics may create sampling bias, as it is likely that individuals with more than one psychological disorder will seek treatment more frequently than those with one. However, well-conducted research comparing community and clinic samples with BN (Fairburn, Welch, Norman, O'Connor, & Doll, 1996) failed to find differences in general psychopathology or substance abuse among the clinic and community samples, although high rates of comorbidity were noted among both groups. Based on higher levels of impaired social adjustment among the clinic group, the authors speculate that there may be higher levels of personality pathology among these individuals (Fairburn et al., 1996). As such, it appears that clinic samples are a reasonable starting point when conducting comorbidity research; however, it may be that greater caution should be extended when evaluating rates of Axis II psychopathology among this population.
نتیجه گیری انگلیسی
High rates of comorbid Axis I and II psychopathology exist among women with eating disorders. This review has discussed studies which reveal that major depression is very common in both AN and BN and that OCD is prevalent in those with AN, while substance abuse may not be as common in BN as was once thought. Instead, the prevalence of substance abuse in those with BN may depend on the co-existence of major depression, drug dependence, and Cluster B personality disorders. Furthermore, physiological changes due to starvation, and a premorbid, possibly inherited susceptibility to OCD and/or major depression, may combine to facilitate the expression of an eating disorder. Results from this review indicate that BPD is common in those with BN, and OCPD is frequently found in those with AN (restricting type). While a definitive statement is premature at this point, an underlying “multi-impulsive” syndrome (Lacey, 1993) may contribute to the comorbidity observed in a subgroup of individuals with BN, while obsessive and perfectionistic traits such as those observed in OCPD and OCD may create a vulnerability to the development of AN. Many of these exciting relationships warrant replication and future study. Unfortunately, a number of methodological issues hinder attempts to make causal connections regarding such relationships. These issues include differences in diagnostic criteria across studies, small sample sizes, difficulties in conducting prospective and longitudinal research, and heterogeneity in assessment methodology. Perhaps more importantly, given the complex and subtle nature of many of the relationships between AN, BN, and Axis I and II disorders, more consistency is needed in the assessment process. As the Structured Clinical Interview for DSM (SCID; First, Spitzer, Gibbon, & Williams, 1995) is the most commonly used clinical interview for Axis I and Axis II diagnoses in more recent research, its use would substantially enhance the reliability of results in the field, and allow for more accurate between-study comparisons. Although not always possible, the use of an interview is generally preferable to questionnaires in diagnosing personality disorders, as questionnaire data may artificially inflate estimates of personality psychopathology (e.g., Rosenvinge et al., 2000). The consistent use of self-report psychometric measures will further bolster the assessment of Axis I pathology. Based on a review of this literature, commonly used self-report measures of depression, obsessive–compulsive symptoms, alcoholism, and general distress due to Axis I conditions are the Beck Depression Inventory (BDI; Beck, Ward, Mendelsohn, Mock, & Erbaugh, 1961), the Yale-Brown Obsessive Compulsive Scale (YBOCS; Goodman et al., 1989), the CAGE questionnaire (Ewing, 1984), and the Symptom Checklist-90-Revised (SCL-90; Derogatis, 1983), respectively. As all of these measures have good reliability and validity, and are sensitive to change following psychological and pharmacological treatments, the research community would be well-served by incorporating these questionnaires into research projects to ease comparisons between studies, especially in cases where the use of a structured clinical interview is not feasible. Comorbidity researchers would also benefit from employing research and statistical designs which are parallel in sophistication to the types of constructs that they are trying to measure. For example, the field would benefit from the consistent use of a carefully selected control group. If one is using a clinical population in research, the comparison group should, at a minimum, be matched on level of distress, gender, and age. For example, inpatients with eating disorders could be contrasted to other types of inpatients, and outpatients with eating disorders could be compared to a similarly distressed, treatment-seeking population such as persons obtaining regular treatment for a medical problem or another Axis I disorder. As indicated in this review, some of the most interesting work in this area emerges from the examination of Axis I and Axis II features in individuals with AN and BN who have been successfully treated. This group warrants further study. Additionally, within the extant eating disorders comorbidity literature, passing reference is often made to mediating/moderating variables, but explicit tests of such models are lacking. These terms are often used interchangeably, but they are not the same. Although it is generally not possible to draw causal conclusions by distinguishing between mediator/moderator variables, this is not merely a semantic distinction; the correct determination of the nature and influence of these variables has implications for treatment, and may alter its focus should an important “third variable” or variable of influence come to light. It may also assist in determining whether there is a true overlap between disorders, or whether a third variable such as a common premorbid risk factor is creating a spurious relationship (Serpell et al., 2002). Moderators are variables that influence the predictive abilities of an independent variable (Gogineni, Alsup, & Gillespie, 1995). Fig. 1 provides an example of a possible moderating variable, based on this literature review. In this figure, a dieting episode could be conceptualized as a moderator if the interaction between dieting episode and OCD/OCPD symptoms was significant in predicting AN diagnosis. Similarly, a dieting episode could be viewed as moderating the ability of depression symptoms to predict AN symptoms. Hierarchical regression equations are often used to test for moderator variables. In one equation, the main effects are tested, while in the second, the interaction of the two predictor variables is also included (Gogineni et al., 1995). If one of the variables is a moderator, the difference in R2 between the two equations will be statistically significant in the presence of a moderator (Gogineni et al., 1995); that is, the existence of a moderator is supported if the interaction between the moderator (e.g., neurotransmitter imbalance) and the predictor variable (AN) is significant (Baron & Kenny, 1986). In contrast, a mediator is a “third variable,” which is responsible for all or some of the effect of the independent variable on the dependent variable (Gogineni et al., 1995); this third variable is involved in transmitting the effect of the predictor to the outcome variable (James & Brett, 1984). To test whether a variable is a mediator between two constructs, regression analysis is generally used (Gogineni et al., 1995), although computer-modeling approaches such as path analysis and structural equation modeling are increasingly employed (Gogineni et al., 1995). To test for mediation, it is necessary to utilize three regression equations. Initially, the outcome variable is regressed on the predictive variable; if this relationship is found to be significant, then the mediating variable is regressed on the predictor, and the third equation is calculated, whereby the outcome variable is regressed on both the mediating variable and on the predictor simultaneously (Lindley & Walker, 1993). If the relationship between the predictor and outcome variables becomes nonsignificant when you control for the significant relationships between the mediator and the predictor and outcome variables, then a mediating variable is at work (Baron & Kenny, 1986); if the relationship is significant, the variable may be a partial mediator (Baron & Kenny, 1986).