Whenever I read about a consistent association between two types of psychopathology, such as bulimia nervosa (BN) and substance abuse, for example, I always wonder “Why?” The ‘Why’ question is what really drives my interest in research on eating disorders and co-occurring conditions. Sexual trauma is one experience that is fairly consistently linked to eating disorder symptoms. Thus, I am delighted to have this opportunity to write about possible mechanisms of action linking trauma experiences to the development of eating disorder symptoms, and to share some new data testing these hypotheses. First, I would like to share some hypotheses developed with a colleague, Erin Hartzell, about how childhood abuse may lead to later symptoms of bulimia nervosa. Next, I would like to share a hypothesis and some new data about how later sexual trauma experiences may affect BN symptom development.
Child abuse, most often sexual, is considered a non-specific risk factor for eating disorder symptoms. It is considered a risk factor because there are elevated rates of child abuse in eating disordered samples, yet non-specific because those rates are similar to rates of child maltreatment in other samples of women with psychopathology (Thompson & Wonderlich, 2004; Jacobi, Hayward, & de Zwaan, 2004). A recent meta- analysis on this topic found that women who had been sexually abused had higher rates of eating disorder symptoms than those who had not, and that individuals with BN had higher rates of sexual abuse than those without BN (Smolak& Murnen, 2002). More recent data in a sample of women who had been sexually victimized as adults or children indicate that women with trauma histories had higher rates of eating disorder symptoms than did controls (Holzer, Uppala, Wonderlich, Crosby, & Simonich, 2008). Researchers have only recently begun to describe specific models linking sexual or physical abuse to later eating disordered symptoms, rather than examining rates of co-occurrence of the two conditions (Thompson & Wonderlich, 2004). For example, some data indicate that Post Traumatic Stress Disorder symptoms mediate the link between childhood sexual abuse (CSA) and later eating disorder symptoms (Holzer et al., 2008). In other words, in this sample, the relationship between CSA and later eating disorder symptoms was explained statistically by PTSD, such that the reason that relationship existed was because those who experienced CSA developed PTSD, which in turn influenced the development of eating disorder symptoms. Thus, the time appears ripe for specific mechanisms of action linking trauma to the development of eating disorder symptoms to be articulated and tested, as the field begins to move beyond descriptive data collection.
In contrast to CSA, emotional or psychological abuse is one form of child maltreatment that has received relatively little attention; however, it is also found in elevated rates in samples of women with obesity and eating disorder symptoms ( Grilo, Masheb, Brody, Toth, Burke-Martindale, & Rothschild, 2005; Grilo & Masheb, 2001). CSA does not likely occur in a vacuum; in families in which children are sexually or physically abused, the victims are also likely to be emotionally abused and neglected. Therefore, it seems likely that the overlooked factor of co-occurring emotional abuse may have a significant contribution to the development of psychopathology, including BN symptoms. In fact, most studies that examine CSA do not present data on emotional abuse.
The consequences of growing up in an emotionally abusive environment have several potential links to future eating disordered behavior, which can be incorporated into existing models of the development of BN symptoms. Bruch (1973) was the first to identify poor interoceptive awareness as a possible risk factor for disordered eating behavior, and to link this deficit to childhood environments. She described poor interoceptive awareness as being associated with difficulties in recognizing hunger and satiety cues, resulting from an environment in which a child's needs are not sufficiently met. In this type of environment, skills to recognize and manage physiological signals are not taught. Poor interoceptive awareness is also associated with difficulties in correct emotion recognition and identification (van Strien & Ouwens, 2007; Garner, 1991) and emotional eating (Larsen, van Strien, Eisinga, & Engels, 2006). For example, individuals who have difficulty distinguishing emotions and physical sensations may perceive themselves as hungry when they are feeling anxiety or arousal. A number of studies have linked alexithymia (poor emotion identification) and interoceptive awareness to eating disorders (Mitchell & Mazzeo, 2004). One prospective study demonstrated that poor interoceptive awareness predicts the onset of disordered eating in girls (Leon, Fulkerson, Perry, & Early-Zald, 1995).
Emotional abuse may lead to BN symptoms by a second mechanism, which can be incorporated in the cognitive behavioral model of the development of eating disorders. Individuals who are emotionally abused are often devalued, criticized, and belittled from a young age. This devaluation is likely to be integrated into the self concept of the victim of emotional abuse, resulting in the person believing that they are unlovable, or weak, or worthless. This low self esteem could lead to weight and shape concerns, and attempts to modify one's appearance through dieting. This cycle, of course, is the start of Fairburn’s Cognitive Behavioral model of the development of BN. While a discussion of dieting and the influence of dieting on BN symptoms is beyond the scope of this article, it does appear that severe caloric restriction, such as attempting to undereat as opposed to reduction of overeating (Herman, van Strien, & Polivy, 2008), cutting out food groups, and following food rules, leads to restraint that is almost impossible to sustain. This leaves one susceptible to breaking food rules and 'black and white thinking' about those rules, which causes individuals to overeat after a prolonged period of restraint. The low self esteem that is hypothesized to lead to attempts to modify weight and shape through diet is one mechanism by which emotional abuse can influence later BN symptoms.
Finally, emotional abuse can lead to the development of BN symptoms by deficits in adaptive emotion regulation skills. Heatherton and Baumeister (1991) suggest that the function of binge eating is to distract from an abstract stimulus (negative thoughts about self; painful emotions) with a concrete stimulus (food). Several studies have been conducted with the use of EMA technology, and all consistently find that increases in negative affect precede binge eating behavior (Stein, Kenardy, Wiseman, Dounchis, Arnow, & Wilfley, 2007; Engelberg, Steiger, Gauvin, & Wonderlich, 2007; Smyth, et al., 2007). Moreover, binge eating behavior may be negatively reinforced via this pathway because it acts to decrease or block negative affect, at least in the short term, by relieving the experience of painful emotions (Arnow, Kennedy, & Agras, 1992; Heatherton & Baumeister, 1991; Polivy & Herman, 1993; Whiteside, Chen, Neighbors, Hunter, Lo, & Larimer, 2007). Linehan’s (1993) conceptualization of Borderline Personality Disorder and the role of the invalidating environment provides theoretical support for the relationship between early abuse and disorders of emotion dysregulation. While the role of the invalidating environment has not explicitly been described or studied in relation to later BN symptoms, we believe that this model provides a framework for describing one mechanism by which emotional abuse may lead to eating disorders. Only one study to date has attempted to examine the relationship of the invalidating environment and eating disordered behavior. In this study, women with eating disordered symptoms endorsed self report items indicating that their family environments had several features of the invalidating environment described by Linehan (Mountford, Corphostine, Tomlinson, & Waller, 2007).
Deficits in emotion identification and labeling, a consequence of the invalidating environment and emotional abuse, often then lead to deficits in emotion regulation skills. If one is not provided with modeling of healthy emotion regulation strategies, such as in an abusive home, then one may turn to immediately reinforcing, maladaptive strategies to alleviate distress or distract oneself from negative emotion. As described previously, several studies have already linked BN symptoms to negative emotional antecedents and poor distress tolerance skills. We hypothesize that poor distress tolerance skills mediate the link between the experience of emotional abuse and later binge eating. In short, emotional abuse leads to the development of poor emotion regulation skills, and these maladaptive skills lead to binge eating.
So, where does sexual trauma fit in here? As previously stated, it is unlikely that CSA occurs without concurrent emotional abuse. Therefore, my colleagues and I hypothesize that it is actually the emotional abuse that co-occurs with CSA that leads to deficits in self esteem, interoceptive awareness, and especially emotion regulation skills, which mediate the link between child abuse and later ED symptoms. We had the opportunity to test one of these hypotheses in a large sample of undergraduate women (N = 1302). Using a statistical technique known as structural equation modeling (SEM), we tested a mediational hypothesis. Specifically, we first hypothesized that emotional abuse would contribute the most variance to BN symptoms when three types of childhood abuse were simultaneously modeled: emotional, sexual, and physical (Hartzell & Fischer, manuscript submitted for publication). Second, we hypothesized that deficits in emotion regulation skills would mediate the relationship between childhood abuse and BN symptoms. Overall, results indicated that the model fit the data well (CFI = .99, RMSEA = .03) and supported our hypotheses. However, we were surprised about one finding. When all three types of childhood abuse were simultaneously modeled, only emotional abuse had a significant direct effect on multiple eating disorder symptoms (restriction, binge eating, and weight and shape concerns). Sexual abuse, in fact, did not have a significant direct relationship to any of the ED symptoms in our sample. With regards to the mediation, all three types of abuse had significant, positive associations with deficits in emotion regulations skills, which in turn had significant positive associations with all ED symptoms modeled (Hartzell & Fischer, manuscript submitted). (Sobel tests of the significance of the mediation indicated that the indirect effects of emotional abuse through emotion dysregulation were significant, but not those of sexual abuse).
I would like to now turn to a discussion of some hypotheses of how personality can interact with trauma experiences to influence BN symptoms. Of course, genetic influences on temperament and personality are in place from birth, and are also constantly influenced by environmental events. Therefore, childhood trauma experiences may also influence personality development. For example, Borderline Personality Disorder symptoms in individuals with BN may be uniquely associated with childhood sexual abuse experiences (Dohm et al., 2002).
This aside, individuals often experience sexual trauma in adulthood as well as childhood. The first few years of college may be a particularly high risk time for sexual trauma in young adult women. Therefore, I am interested in examining the interacting effects of personality and new trauma experiences on the development of BN. Cross sectional research has indicated that negative urgency, the tendency to act rashly in response to negative affect or distress, is positively associated with binge eating symptoms (Fischer & Smith, 2008; Fischer, et al., 2004; Anestis, et al., 2007). However, this trait is also positively associated with alcohol abuse (Whiteside & Lynam, 2003; Fischer & Smith, 2008; Miller, Flory, Lynam, & Leukefeld, 2003) risky sexual behavior (Miller, et al., 2003 ) and pathological gambling (Fischer & Smith, 2008). Theoretical models of personality risk for pathology hypothesize that personality traits are genetically heritable factors that increase vulnerability to several types of psychopathology. However, environmental events may actually cause an individual to cross the 'liability threshold' into pathology (Bulik, Sullivan, & Kendler, 2003). Thus, genetic trait by environment interactions may be linked to bulimia nervosa and other eating disorders (Bulik, et al., 2003).
My current research is testing hypotheses based on the idea that one such gene -environment transaction may exist between negative urgency and sexual trauma. Our hypothesis is that individuals with high levels of negative urgency, who act rashly or impulsively when distressed, may not provide themselves with the opportunity to learn longer term, more adaptive strategies to cope with negative affect. They thus tend to turn to maladaptive emotion regulation strategies, such as binge eating, purging, and substance abuse. I hypothesize that individual differences in negative urgency may interact with the experience of sexual assault to influence the development of or increase in eating disorder symptoms. I had the opportunity to collect data to test this hypothesis in a longitudinal study of women across their first year of college (N = 550), a high risk time for the development of BN symptoms (Fischer, Stojek, & Collins, manuscript in preparation). My colleagues and I collected data on eating disorder symptoms, impulsivity related personality constructs, and sexual assault at three different time points across the year. Thus, we were able to capture pre-existing sexual assault, new occurrences of sexual assault, pre-existing BN symptoms, and the emergence or increase of BN symptoms over the year.
At time 1, we found that pre-existing sexual assault and negative urgency had unique direct positive effects on BN symptoms. We then controlled for the presence of BN symptoms at time 1, and tested the interaction effect of time 1 levels of negative urgency and new occurrences of sexual assault at both time 2 and time 3. At both time points, the interaction effect was significant (Fischer, et al., manuscript in preparation). Thus, women who had been sexually assaulted in the interim between time 1 and time 2, and time 2 and time 3, and had high levels of negative urgency, experienced increases in BN symptoms at time 2 and time 3. My colleagues and I are very excited about this data, as it is one of the only prospective studies demonstrating both the influence of sexual assault on the development of symptoms and testing a hypothesized mechanism of action.
Of course, both childhood and adult experiences can interact to influence personality, interact with personality, and influence BN symptoms. The effect of an emotionally labile temperament may be mitigated when an individual grows up in an environment that puts forth effort to teach adaptive emotion regulation skills. However, when such an individual is placed in an emotionally abusive environment in which they are not taught adaptive emotion regulation tools, the result may be psychopathology such as bulimia nervosa. Of course, this interaction could lead to any of the pathways previously outlined. For example, high negative urgency individuals in an abusive environment may be more likely to develop weight and shape concerns as a result of low self esteem, or may be more likely to experience deficits in interoceptive awareness because of their tendency to quickly block negative emotions. Poor distress tolerance may be both a mediator and a moderator of emotional abuse to later binge eating. Sexual abuse experiences can also influence neurotransmitter and personality development itself. For example, the neurotransmitter that is hypothesized to play a key role in the development of eating disordered behavior is serotonin, or 5HT (Steiger, 2004). This neurotransmitter has been associated with satiety and disordered eating, but also with personality traits associated with BN, such as affective instability and impulsivity (Lesch, Bengel, Heils, et al., 1996; Steiger, Richardson, & Joober, 2007). Steiger and colleagues found increased levels of psychopathology in bulimic women who both had been abused and were a carrier of an allele that regulates serotonin reuptake (Steiger, et al., 2007). This indicated that a gene by environment interaction was present; that individuals who had a genetic vulnerability to personality instability had greater psychopathology when they had experienced abuse. This is supportive of the notion that personality may interact with environmental events to influence the development of psychopathology. Overall, there are many avenues to be explored in this area of research, and it is an exciting time to study these transactions.
If you are interested in learning more about these topics, we recommend that you browse our newly redesigned online store. Products are listed by category (e.g., eating disorders, ADHD) and are geared for both professionals and non-professionals.