Our results provided some clarity regarding the nature of the association between impulsivity, obsessions, and compulsions and binge eating and purging in AN. Purging, but not binge eating, was associated with all measures of impulsivity (as measured by the BIS-11 and novelty seeking and harm avoidance scales of the TCI), general obsessions and compulsions (as measured by the Y-BOCS), and obsessions and compulsions related to eating (as measured by the preoccupations and rituals scales of the YBC-EDS). These findings confirm previous demonstrations of associations between purging behavior in AN and impulsivity (Favaro, et al., 2005
; Root, Pinheiro, et al., 2010
) as well as obsessionality and compulsivity (Favaro & Santonastaso, 1996
). We did not find support for an association between binge eating and impulsivity, obsessions, or compulsions. Moreover, the absence of a binge eating × purging interaction also supports purging as the stronger correlate of impulsivity in AN. Purging together with associated impulsivity, obsessions, and compulsions can adversely affect the course of illness in AN (Bulik, et al., 2008
; Milos, Spindler, Ruggiero, Klaghofer, & Schnyder, 2002
Although a different measure of impulsivity was used, results from this larger sample mirror those ofFavaro et al. (2005)
who found that purging behavior, but not binge eating, was significantly associated with the presence of impulsive behaviors in a sample of individuals with AN and BN. Behaviors indicative of impulsivity such as substance abuse/dependence, self-injury, and suicide attempts have also been associated with the presence of purging behavior specifically in AN rather than the larger AN-BP subtype (Favaro & Santonastaso, 1996
; Root, Pinheiro, et al., 2010
). In the present study, we also found purging, but not binge eating behavior, to be significantly associated with novelty seeking, a personality trait that has been associated previously with impulsive behavior in individuals with AN and BN (Favaro, et al., 2005
). Several past studies reporting an association of binge eating with measures of impulsivity in AN did not explore the independent effects of purging in their analyses, although many individuals reporting binge eating in their samples also engaged in purging (Casper, et al., 1980
; Fernandez-Aranda, et al., 2008
; Garfinkel, et al., 1980
). Thus, given our findings, purging may be the stronger predictor of impulsivity in AN.
In terms of obsessions and compulsions, we found that purging, but not binge eating, was significantly associated with general obsessions and compulsions measured with the Y-BOCS as well as preoccupations and rituals related to eating measured with the YBC-EDS. Purging, but not binge eating, was also significantly associated with higher scores of harm avoidance on the TCI. In line with this finding, individuals with OCD (without eating disorders) score higher on harm avoidance than healthy participants (Alonso et al., 2008
). If purging is more strongly associated with obsessions and compulsions in AN, then it may be that discrepant findings in past studies of obsessions and compulsions across AN subtypes are the result of AN-BP symptom heterogeneity (Halmi, et al., 2003
; Speranza et al., 2001
). Based on DSM-IV criteria, these studies included individuals with binge eating, purging, or both behaviors in subtypes designated as AN-BP. It may be that studies that found no significant difference in obsessions or compulsions across subtypes (Halmi, et al., 2003
) had a greater percentage of individuals engaging in binge eating only in their AN-BP sample.
Strengths of this study include the large sample size, interview-based assessment of objective binge eating and purging behaviors, and interview-based assessments of obsessions and compulsions. Although a major challenge in the past has been consistent classification of binge eating and purging due to differing cutoffs for frequency and duration of the behaviors, we employed a more inclusive definition for purging behavior relative to DSM-IV criteria for BN (twice a week for three months), considering even those individuals who responded that they “rarely” engaged in the behavior as positive for purging. This represents a more conservative approach, as inclusion of those with lower frequency of the behavior would be assumed to bias results toward the null. Using a similar classification scheme,Eddy et al. (2002)
found that individuals with AN-BP were more likely to engage in several impulsive behaviors when compared with an AN-R group that only included individuals who had “never” engaged in binge eating or purging.
There are several limitations that should be considered when interpreting our results. First, participants were interviewed at one point in time and asked to retrospectively describe their eating behaviors (SIAB), worst ever obsessions and compulsions (Y-BOCS), worst ever eating preoccupations and rituals (YBC-EDS), and current impulsivity (BIS-11 and TCI). Within this sample, some individuals were currently symptomatic whereas others were considered recovered (symptom free for the year prior to interview) at the time of assessment. In order to account for the influence that status of illness may have had on responses to assessments, we included status of illness as a covariate in all analyses. Second, some affected relatives of probands in the GAN study were not diagnosed with AN at least three years prior to study entry. As research has shown that most binge eating/purging develops within the first three years of illness in AN (Bulik, et al., 1997
; Eckert, et al., 1995
; Eddy, et al., 2002
; Strober, et al., 1997
; Tozzi, et al., 2005
), those individuals who were diagnosed less than three years prior to study entry and did not endorse binge eating or purging behaviors at the time of interview may later develop these behaviors. Third, this was a cross-sectional study, so no inferences can be made regarding the presence or direction of causality with respect to the association between binge eating or purging and impulsivity, obsessions, and compulsions. Lastly, participants from the GAN study represented individuals with familial AN who might have different symptom profiles than individuals with AN in the general population, limiting the generalizability of our findings.
In conclusion, purging but not binge eating is associated with impulsivity, general obsessions and compulsions, and preoccupations and rituals related to food and eating in women with AN. A recent review discussing the utility of subtyping individuals with AN concluded that one of the major issues with the current subtyping system is determination of what constitutes binge eating in individuals with AN (Peat, et al., 2009
). They also contend that subtyping individuals with AN has the potential to be useful for treatment planning by clinicians, although there are currently no empirical data to support this practice. Although evaluation of binge eating (whether subjective or objective) represents an integral component of assessment and treatment, our results suggest that assessment of purging behavior can provide independently valuable information about associated symptom profiles (impulsivity, obsessions, and compulsions) that may be useful to guide further assessment and treatment planning in AN.