- Top of page
Rates of substance use and substance use disorders (abuse/dependence) are high among women with eating disorders (ED). This association is greatest in women with bulimia nervosa (BN) and anorexia nervosa (AN) binge purge subtype and is exhibited for both alcohol and illicit drug disorders.1–5 Findings from the few studies investigating chronology of EDs and substance use disorders suggest a bidirectional association. Cross-sectional studies reveal somewhat similar rates of women reporting onset of a substance use disorder (SUD) to precede an ED and vice versa.6, 7 Moreover, in a 9-year longitudinal examination, 18% of women with AN and 30% of women with BN were diagnosed with a drug use disorder for the first time over the course of the study, suggesting that the risk for a drug use disorder continues over time in women with EDs.8 Similar ongoing risk has been found for alcohol disorders.9
In addition to illicit drug and alcohol use disorders, regular nicotine use is also frequent in women with an ED. A higher proportion of women with BN are regular smokers compared to controls or to women with AN.5, 10-12 This association may arise because of the commonly held belief that nicotine use can aid in weight loss. Therefore, women's concerns about body weight and shape, which are particularly salient for women with an ED, may increase their risk for cigarette use.5
It remains unclear why EDs and SUDs frequently coexist; these associations are complex and likely have several biological and psychosocial influences. For example, when women begin to remit from an ED, they may substitute substances (i.e., dramatic increase in alcohol consumption) for the ED symptoms (i.e., binge eating) or vice versa. In addition, a common familial diathesis has been proposed. Several family studies show increased rates of SUDs in relatives of women with BN.1, 2, 13 However, several reports also suggest that these two disorders are transmitted independently.3, 14–16 In contrast to these latter family studies, twin studies often show a familial relationship between EDs and SUDs.
Research has shown evidence for shared genetic influences between BN and illicit drug abuse/dependence.6 Authors reported that 83% of the phenotypic correlation between BN and illicit drug abuse/dependence was accounted for by genetic factors. However, a report examining the overlap between BN and alcohol use disorders demonstrated that the two disorders load on distinct genetic factors.14 Examining the genetic covariance between specific ED symptoms and substance use also reveals common genetic factors. For example, shared genetic factors were shown to account for a portion of the covariance between weight preoccupation and binge eating and alcohol use in both males and females.17 Thus, it may be important to examine specific ED symptoms and their relation to SUDs rather than focusing on specific diagnostic categories (AN vs. BN). This is an important consideration as recent work has emphasized the importance of examining specific symptoms of an ED noting that, “A DSM-IV diagnostic category … might actually represent an occasionally co-occurring yet etiologically diverse mixture of genetically and environmentally influenced symptoms …” (p. 191).18
Previous work that has examined associations between ED symptoms and SUDs has been phenotypic in nature and shows that, in general, the more severe the ED symptoms, the greater the number of substance classes used.12, 19, 20 Independent of diagnostic category, specific phenotypic associations have been shown between caloric restriction and amphetamine use and binge eating and tranquilizer use.20 Severe bingeing is also consistently associated with alcohol use,19, 21, 22 whereas purging behaviors have been shown to predict the use of a multitude of substances including alcohol, cocaine, cigarettes, stimulants, and amphetamines.9, 19-21, 23
This study aimed to extend previous research in this area in several ways. First, we examined the phenotypic associations between ED diagnoses (AN and BN) as well as ED symptoms. It was specifically hypothesized that significant associations would be found between binge eating and an alcohol use disorder, purging behaviors and an alcohol use disorder, illicit drug use disorder, and regular smoking and between body image and an alcohol use disorder due to previous associations found between these variables.9, 21 Second, we also examined the chronology of comorbid EDs and SUDs. Finally, follow-up twin analyses were conducted between ED symptomatology and SUDs to examine for genetic covariance. We hypothesized that ED symptomatology and SUDs would show a moderate amount of genetic covariance. This report is one of the first to integrate all the above-mentioned aspects into one report using a population-based sample.
This report also adds to previous research examining similar types of associations using the Virginia Adult Twin Study of Psychiatric and substance use disorders in several ways. First, while the association between BN and illicit drug and alcohol disorders has been examined within this sample, our previous reports have only examined associations at the diagnostic level and have not examined relations between BN (at the diagnostic or symptom-level) and smoking or caffeine disorders. Second, our previous reports have neglected to include AN. Third, we examine ED diagnoses in two distinct ways (absence or presence of diagnosis and symptom count). Last, we include a detailed examination of the prevalence of comorbid ED symptomatology and SUDs including an examination of self-reported chronology of symptoms.
- Top of page
The purpose of this study was to examine the association between ED diagnoses, their symptomatology, and SUDs. Several important findings emerged. First, consistent with the previous literature,2 ED diagnoses were significantly related to SUDs. Specifically, women with AN had a twofold increased risk of having an alcohol use disorder and being regular smokers. Women with BN were two to three times more likely to have an alcohol or illicit drug use disorder and be regular smokers. However, in contrast to previous reports, which generally find SUDs to be significantly more common in women with BN compared to AN,2 our results showed no significant differences between the prevalence rates of SUDs in women with AN and BN. Additionally, while results were not statistically significant, a caffeine disorder and regular smoking were more prevalent in women with AN compared to BN. These differences could be due to the fact that our sample is community-based as opposed to clinical. The nonsignificant increased rates of a caffeine disorder and regular smoking in women with AN also could be due to the common belief that both these substances increase metabolism. The association between EDs and a caffeine disorder is also not well studied.
Second, results revealed that, in general, BN manifests before a SUD. Of women with age of onset information for both disorders, most reported binge eating preceded symptoms of a SUD. This may be occurring because of several possibilities. First, these women may be beginning to remit from BN and be substituting substances for the ED symptoms. Second, the women may be using substances to dampen bulimic urges or negative affect. Third, both disorders may be caused by a common factor, whereas BN symptoms simply have an earlier onset. In contrast, more women with AN reported SUD symptoms preceded their age of lowest weight. The only exception to this was AN and an alcohol use disorder. It could be hypothesized that women with AN begin experimenting with substances before the onset of AN (or before a diagnosis can be made) in an effort to lose weight. Additional possibilities for this pattern could be the age of onset information used for AN in this report was the age at lowest weight. Second, AN diagnosis requires a specific amount of weight to be lost before a diagnosis can be made; therefore, this finding may be a reflection of our data or of AN diagnostic criteria.
Third, as hypothesized, several of the ED symptoms examined were significantly associated with SUDs. For AN symptomatology, only the variables reflecting body image (fear of gaining weight and belief overweight when underweight) were significantly associated with a SUD. The presence of negative body image in women who reported having been under weight previously increased the risk for a caffeine use disorder by ∼2. A similar pattern was revealed for BN-related symptoms. Concern about weight and shape in women with a history of binge eating was significantly associated with all of the SUDs examined, again increasing risk by ∼2. It is important to note, however, that these body image symptoms are likely more common in our sample and show greater variability compared to other ED symptomatology, which may account for the large number of significant associations revealed. Finally, binge eating and purging behaviors were significantly related to a multitude of SUDs. Binge eating was associated with all SUDs, whereas purging behaviors in women with a history of binge eating showed significant associations with all expect a caffeine use disorder. These results confirm our hypotheses, based on previous reports, that binge eating, purging behaviors, and body image would be associated with an alcohol use disorder and that purging behaviors would be significantly associated with illicit drug use disorders and regular smoking. However, we did not find a significant relation between AN body image and an alcohol use disorder identified by previous research.9
Forth are the findings from the twin analyses. To date, few studies have examined the genetic and environmental covariance between ED symptomatology and SUDs. Moreover, a limited number of investigations have examined BN using a symptom count. The univariate BNSC genetic and environmental estimates are similar to our previous examination of broadly defined BN diagnosis in VATSPSUD.6
Bivariate analyses indicate a shared etiology between ED symptomatology and SUDs. Although our data may not have been sufficient to differentiate between genetic and shared environmental effects on comorbidity, our results do indicate that familial factors are important for the overlap between these two disorders, which is consistent with previous research.6, 17 Because no re confidence intervals included 1.00, we know that individual-specific environmental factors are not sufficient to account for the overlap. Estimated genetic correlations from the ACE models ranged from +.25 to +.53, with the highest correlation between BNSC and an alcohol use disorder. This contrasts a previous VATSPSUD report investigating genetic covariance between alcoholism and BN diagnosis (along with four other disorders).14 Kendler and colleagues14 noted that most of the genetic factors responsible for alcoholism were disorder specific and unrelated to the factors that influence BN (as well as mood and anxiety disorders). Differences could have arisen, because the previous report used a broad definition of alcoholism, a differing definition of BN, a best-fit model, and a multivariate common factor model.
Results also indicate that the relationship between BNSC and SUDs may not be substance specific as a majority of the SUDs examined showed common familial influences with BNSC. This result makes sense in the light of substance use literature that shows genetic influences on both substance use and SUDs is not substance specific, but produces a general vulnerability to use or misuse.36–38 This common familial susceptibility to BN-related behaviors and substance misuse may simply place an individual at risk for both, whereas additional specific factors determine the exact type of substance used (e.g., availability). This may also provide limited evidence for BN as an addictive disorder. It could be that there is a general vulnerability to BN and substance misuse and that additional specific factors (e.g., personality) determine which behaviors arise.
Finally, this is one of the first studies to examine the comorbidity of ED symptomatology and a caffeine disorder. Although neither AN nor BN diagnoses were significantly associated with a caffeine disorder in GEE analyses, several ED symptoms were. Body image related to both AN and BN as well as binge eating were significantly related to a caffeine disorder, corroborating research showing a correlation between binge frequency and caffeine use.10
There are limitations to this study that warrant discussion. First, and perhaps most noteworthy, is the way our ED symptom variables were created, specifically the use of entry questions. Although the use of entry questions is common practice in large scale, interview-based studies to ease the burden of participants; this can be problematic when examining symptomatology. Although the entry questions are part of the diagnostic criteria for AN or BN, we still are not able to obtain “true” symptom counts as it is possible to have participants whom purge but not binge eat, for example.
Second is our insufficient power to differentiate between genetic and shared environmental effects. Although our sample size is large in terms of both general population and twin samples, we still did not have the power to reject the AE model if the CE model was the true, best-fitting model. However, for a small effect size, an exceptionally and possibly unreasonably large sample of twin pairs would be needed. Despite the inability to differentiate between common genetic and shared environmental effects, results still provide clear evidence for the importance of familial factors impacting the comorbity between BNSC and SUDs. Third is the retrospective nature of the data. This method can have an impact on the reliability and validity of the twins' reports. However, BN diagnosis within our Wave 3 sample has been shown to be reliable.28
A fourth potential limitation is merging substance abuse and dependence into a combined variable. Importantly, however, examinations within our own data have never produced results, suggesting distinct genetic risk factors for alcohol abuse versus dependence25, 39 and other substances.40 Therefore, collapsing categories would not have a significant impact on results. Other limitations include the fact that women who were born between 1934 and 1968 were ascertained for the VATSPSUD, so there could be age or cohort effects on results and the use of broader definitions of AN and BN.
Despite the limitations of this report, it also has several strengths including the use of ED diagnoses, symptoms, and a BN symptom count variable. As discussed, research is emphasizing the importance of examining ED symptoms as opposed to diagnoses, which might only represent occasionally co-occurring etiological diverse symptoms.18 We attempted to address this issue by including symptoms rather than relying solely on diagnoses in our investigations. Second, we thoroughly assessed the temporal relationship of AN, BN, and SUDs. A limited amount of studies have examined this in both AN and BN. Third, we used a population-based sample, increasing the generaliability of our results.
Current findings have important implications for future research as well as clinical work. Further investigations regarding genetic and environmental influences on associations among specific ED symptoms and SUDs are important for treatment and prevention efforts. Moreover, women presenting with either an ED or a SUD should be assessed for symptoms of the other disorder. Importantly, women with subthreshold levels of EDs may be vulnerable to developing a SUD, as specific symptoms as well as overall diagnosis, were related to substance use. EDs continue to be among the most difficult psychiatric disorders to treat. Continued elucidation of predisposing and maintaining factors, including comorbid relationships, is essential in addressing these pernicious behaviors.
Earn CE credit for this article!
Visit: http://www.ce-credit.com for additional information. There may be a delay in the posting of the article, so continue to check back and look for the section on Eating Disorders. Additional information about the program is available at www.aedweb.org