Exploring the relationship between eating disorder symptoms and substance use severity in women with comorbid PTSD and substance use disorders

Abstract

Background

Eating disorders (ED) and substance use disorders (SUD) commonly co-occur, especially in conjunction with posttraumatic stress disorder (PTSD), yet little is known about ED and ED symptoms in women presenting to addiction treatment programs.

Objective

We examined the association between ED symptoms and substance use frequency and severity in a sample of women with a DSM IV diagnosis of current SUD and PTSD enrolled in SUD treatment.

Method

Participants were 122 women from 4 substance abuse treatment sites who participated in a multi-site clinical trial through the National Institute of Drug Abuse Clinical Trials Network (NIDA CTN). The Eating Disorder Examination-Questionnaire (EDE-Q), the Clinician’s Administered PTSD Scale (CAPS) and the Addiction Severity Index (ASI) were administered at baseline and correlational analyses were performed. Variables that significantly correlated with EDE-Q total and subscale scores were entered into a linear regression analysis.

Results

Scores on the EDE-Q Global scale, as well as the Eating Concern, Weight Concern and Shape Concern subscales of the EDE-Q were significantly associated with Caucasian race/ethnicity, past 30 day opiate use, higher ASI Psychiatric Subscale score and lower ASI Employment Subscale score.

Conclusion

Although exploratory, these findings suggest that there may be a relationship between addiction severity, use of certain drugs of abuse and eating disorder symptoms, particularly those involving weight and shape concerns in women with comorbid PTSD and SUD.

Introduction

Eating disorders (ED) occur more frequently in substance use disorder (SUD) populations than in the general population, with the highest reported rates of co-occurrence in individuals with bulimia nervosa and alcohol use disorders (AUD) (1–5). The National Comorbidity Survey Replication Study found lifetime rates of any SUD in ED subgroups was between 23–37% (3). The National Center on Addiction and Substance Abuse estimated 50% of women with ED abuse substances (4). Binge eating and subthreshold eating disorders have also been more likely to occur in individuals with SUD than in those without SUD (5–8). The strongest associations between SUD and ED involve bulimic behaviors such as binge eating, purging, and laxative use (9).

Substances including alcohol, cocaine, amphetamines, methamphetamine, nicotine and opioids are often used to suppress appetite or increase metabolism, induce vomiting, and self- medicate negative affective mood and anxiety states (10–13). Several studies have linked certain shared risk factors that may increase vulnerability to the development of ED and SUD. Bulik and colleagues (2004) reported more depressive and anxiety disorders, cluster B personality disorders and impulsivity and perfectionism personality profiles in women with comorbid AUD and ED (14). Childhood adversity increases the odds of developing an ED, as well as a host of other comorbidities including AUD and SUD. Individuals with comorbid ED and SUD have higher rates of childhood trauma and PTSD than either disorder alone (15–17). Increased impulsivity may explain the association between childhood trauma, ED and SUD (18). Substance use and bingeing on food may be ways of coping with negative emotions in the absence of adaptive coping skills. This raises the possibility that in the absence of substance use, individuals may use food as an emotional coping response (19). Thus, these individuals may be at risk for either developing disordered eating behaviors or exacerbation of an already existing ED.

Recent research into food addiction and obesity has demonstrated neurobiological, physiologic and behavioral similarities with drug addiction (19–22). Similar to drug use, highly palatable foods have reinforcing properties that can alter dopamine function by increasing reward motivational drive and reducing inhibitory control in the limbic system leading to brain reward deficiencies (23–26). Behaviorally, substance misuse and pathologic food consumption (i.e. binge eating) also share similarities in modulating stress and emotional states (27). Gearhardt and co-investigators (2013) found that 41% of treatment seeking patients with binge eating disorder (BED) met the Yale Food Addiction Scale (YFAS) classification for food addiction. The YFAS was correlated with negative affect, low self-esteem, emotional dysregulation, binge eating and other ED psychopathology (28). Women in recovery from SUDs, particularly those with comorbid PTSD and depression, may use palatable foods as a means of regulating emotions, medicating negative affective states and correcting brain reward deficiencies. However, it should be noted, that in restricting anorexic patients, altered taste processing may interfere with the rewarding properties of food (29). Contrary to bulimia, food restriction and starvation are more hedonic and anxiolytic in patients with anorexia nervosa (30, 31).

Almost a third of women entering treatment for SUD report binge eating (32). Furthermore, women with PTSD and SUD who report binge eating behavior have more severe clinical courses and worse treatment outcomes than those with PTSD and SUD who do not report any binge eating (32). Less is known regarding the relationship between addiction severity, use of certain drugs and disordered eating behaviors and attitudes in women presenting for SUD treatment. The current study seeks to explore the relationship between addiction severity and use of certain substances on disordered eating behaviors and attitudes.

Methods

The present study is a secondary analysis of a National Institute of Drug Abuse Clinical Trials Network (NIDA CTN) study investigating the effectiveness of treatment for trauma and SUD in women presenting for treatment at community SUD treatment programs across the U.S. The study was reviewed and approved by an institutional review board, and participation involved informed consent. Participants were female outpatients with comorbid SUD and PTSD seeking treatment at psychosocial community treatment programs affiliated with the NIDA CTN (33). To be eligible, participants needed to have had at least one traumatic event in their lifetime and to have met DSM-IV-TR criteria for either full or sub-threshold PTSD (34). For sub-threshold PTSD, participants had to fulfill DSM-IV-TR criteria A (exposure to a traumatic stressor), B (re-experiencing symptoms), E (symptom duration of at least one month) and F (significant distress or impairment of functioning), and either C (avoidance and numbing symptoms) or D (symptoms of increased arousal), but not both as in full PTSD. Other inclusion criteria were: 1) between 18–65 years of age; 2) used alcohol or an illicit substance within the past six months and have a current diagnosis of drug or alcohol abuse or dependence; and 3) capable of giving informed consent.

Women were excluded if they had 1) advanced stage medical disease as indicated by global physical deterioration; 2) impaired cognition 3) significant risk of suicidal/homicidal intent or behavior; 4) history of schizophrenia-spectrum diagnosis; 5) a history of active (past two months) psychosis; 6) involvement in litigation related to PTSD; 7) non English-speaking; or 8) refused to be video- or audio-taped.

Participants, enrolled between 2004 and 2006, were randomized to receive 12 twice weekly group sessions of either Seeking Safety(35) or a Women’s Health Education intervention. A complete description of the parent study can be found elsewhere (33). A subset of 122 women, from four out of seven treatment programs completed the Eating Disorder Examination Questionnaire (EDE-Q) at baseline (32). The three other treatment programs cited resources and added assessment burden as reasons for nonparticipation.

Assessments

Substance diagnostic data were collected via the Composite International Diagnostic Interview for DSM-IV (CIDI) (36), a fully structured, interviewer-administered measure used to determine lifetime and current substance disorder diagnoses for alcohol, marijuana, stimulants, opioids, cocaine, and sedatives.

The Eating Disorder Examination Questionnaire (EDE-Q), a self-report version of the structured Eating Disorder Examination (EDE), assesses specific behavioral and attitudinal/cognitive features of ED psychopathology. It has consistently shown good correspondence with the EDE interview in a range of populations,(37) including females with SUD (38). The EDE-Q has been found to be an effective screening instrument for detecting the presence of ED behaviors. It has a global score and four subscale scores including concerns about restraint, eating, shape and weight. The EDE-Q consists of 36 items and it focuses on the timeframe of the past 28 days. A score of 4.0 or greater on global or subscale scores is in the clinical range (37, 39). Women in the current study completed the EDE-Q at baseline and at post intervention. Post intervention scores did not significantly differ from baseline scores. As such, baseline assessments were used for the current analyses.

The Clinician Administered PTSD Scale (CAPS) is an interviewer administered structured clinical assessment that measures frequency and intensity of signs and symptoms of PTSD, impairments in social and occupational functioning, and overall symptom severity over time (40, 41). Lifetime traumatic events were assessed using the Life Events Checklist, a 17-item self- report questionnaire (42).

The Addiction Severity Index (ASI) is a semi-structured interview that assesses alcohol and substance use severity in seven different life domains that are typically affected by addiction. Separate composite subscale scores are generated for medical, employment, legal, drug, alcohol, family/social and psychiatric severity. The composite subscales range from 0–1. Higher subscale scores indicate more severe problems. The ASI drug use subscale collects past 30 day use of substances of abuse (43).

Statistical Analysis

Descriptive statistics were used to describe the sociodemographic and clinical characteristics of the participants. Bivariate associations between variables of interest and EDE-Q global and subscale scores were computed using spearman’s rho correlation coefficient or point biserial correlation coefficient for categorical variables. Variables of interest included past 30 day ASI drug use frequency at baseline (cocaine, alcohol, opiates, marijuana and sedatives), ASI composite subscale scores, age, race/ethnicity, marital status, education and total CAPS scores. Modeled variables demonstrating significant associations with EDE-Q global and/or subscale scores in bivariate and point biserial correlations (p ≤ .05) were entered into a multivariate linear regression model using backward stepwise elimination to determine independent correlates of EDE-Q global and subscale scores. All tests were two-sided with significance levels set at p ≤.05 and assessed using SPSS Version 21. Diagnostics showed that normal distribution of the residuals for the regression analysis was assumed for EDE-Q global, weight and shape subscale scores (Kolmogorov-Smirnov test p ≥ .2 and Shapiro-Wilk test p ≥ .07). Multicollinearity was negligible as indicated by none of the variance inflation factors > 5 and tolerance scores > .25. Lastly, the frequencies of select individual EDE-Q items were explored to further characterize the relationship between weight and shape concerns.

Results

Sociodemographic and clinical characteristics are presented in Table 1. Women participating in the study were mostly Caucasian (42.6%) or African American (32.8%), divorced/separated (47.5%) or single (38.6%), with an average age of 38 years. The most frequently used substances were cocaine (72%), alcohol (66%) and opiates (26%). Ninety-three percent of the women reported lifetime history of physical abuse and 63% reported sexual abuse.

Table 1.

Sample Sociodemographic and Clinical Characteristics (N = 123)

	Mean (SD) or % (n)
Age (mean)	37.99 (9.96)
Race/ethnicity (%)
African American	32.8 (40)
Caucasian	42.6 (52)
Latina	9.8 (12)
Multiracial and other	14.8 (18)
Marital status (%)
Married	13.9 (17)
Single	38.6 (47)
Divorced/separated	47.5 (58)
Education (years)	12.43 (2.09)
Employed (%)	54.2 (194)
Most frequent substances used (%)
Alcohol	66.4 (81)
Cocaine	72.1 (88)
Opiates	25.8 (31)
Lifetime trauma exposure (%)
Physical abuse	93.1 (114)
Sexual abuse	62.9 (77)
CAPS scores (mean)	59.6 (18.4)

CAPS = Clinician Administered Scale for PTSD

The mean global score on the EDE-Q was 1.35 (range 0 – 4.95) which is in the normal community sample range for young adult women (44, 45). Approximately 21% of the women had a global EDE-Q scale score in the eating disturbance range (>2.3) (44, 46). Twenty percent of women reported one or more objective binge eating episodes (OBE) in the past 28 days. Using DSM 5 criteria, approximately 13.8% of the women had weight and shape concern subscale scores ≥ 3 and had at least one OBE in the past 28 days suggesting either full or subthreshold criteria for eating disorder pathology. Only 5.7% participants reported purging compensatory behaviors; however 18.7% reported excessive exercise as a means of controlling weight or shape.

The EDE-Q global, as well as the eating, weight and shape concern subscales were significantly correlated to Caucasian race/ethnicity, baseline number of days of opiate use in past 30 days, ASI medical, drug and psychiatric composite subscale scores, and were inversely related to the ASI employment subscale score. There were no significant correlations between these variables and the restraint subscale scores. Modeled variables that were p ≤ .05 on bivariate tests were entered into a multivariate linear regression model with EDE-Q global and subscale scores as dependent variables (Table 2). Caucasian race (β = −.202, p ≤ .01), past 30 day opiate use frequency (β =.204, p ≤. 01) and ASI employment (β = −.169, p ≤ .05) and psychiatric subscale scores (β = .305, p ≤ .001) all remained significant in the regression model for EDE-Q global scores. The overall model fit was R² = .28, p ≤ .001. These same variables were also significant for weight and shape subscale scores (Table 2). For the eating concern subscale the only significant variable in the regression analysis was the psychiatric subscale severity score (β =.34, p ≤.001).

Table 2.

Linear Regression Analyses EDEQ Global and Subscale Scores^*

Global Score	B	SE	beta	p
Constant	1.612	.430		.0001
Race	−.420	.165	−.202	.013
ASI Psychiatric subscale	2.013	.541	.305	.0001
ASI Employment subscale	−.698	.335	−.169	.039
Baseline days of opiate use	.045	.018	.204	.013
Weight Subscale Score	B	SE	beta	p
Race	−.593	.220	−.212	.008
ASI Psychiatric subscale	2.978	.719	.336	.0001
ASI Employment subscale	−.756	.445	−.136	.092
Baseline days of opiate use	.063	.024	.209	.010
Eating Subscale Score	B	SE	beta	p
Race	−.194	.146	−.117	.188
ASI Psychiatric subscale	1.394	.503	.264	.007
ASI Employment subscale	−.375	.294	−.113	.205
Baseline days of opiate use	.001	.018	.005	.957
Shape Subscale Score	B	SE	beta	p
Race	−.643	.236	−.214	.008
ASI Psychiatric subscale	3.002	.774	.316	.0001
ASI Employment subscale	−.998	.479	−.168	.039
Baseline days of opiate use	.063	.026	.195	.017

^*Restraint subscale not presented; B = unstandardized regression coefficient; SE = standard error; Beta = standardized coefficient

ASI = Addiction Severity Index

On individual EDE-Q items, 50% of the women had a definite fear of weight gain or becoming fat with 22% feeling this way on a daily basis. Fifty percent had a strong desire to lose weight with 28.5% feeling this way on a daily basis. Thirty percent of women were afraid of losing control over eating and 45.5% felt guilt about eating because of the effect on their weight and shape with 19% feeling this way on half or greater than half the days. Forty one percent of the women deliberately limited food to influence weight or shape with 11.4% doing so on a daily basis. Approximately 34% and 36% of the women were moderately to markedly dissatisfied with their shape and weight, respectively.

Discussion

Women in recovery from SUD and PTSD reported concerns about weight, shape and eating at a higher rate than the general US population of women (46). In one study that explored initial motivation for drug use, Brecht et al. (2004) found that 36% of women in treatment for methamphetamine use disorder initiated methamphetamine use to lose weight (47). In another study exploring weight concerns in women recovering from SUD, 45% of the women reported concerns that weight gain could trigger relapse and these concerns were associated with higher levels of body dissatisfaction, dieting behaviors, bulimic symptoms and weight-related drug use expectancies (48).

According to the DSM 5 criteria the number of objective binge eating episodes (OBE) required for the diagnosis of bulimia nervosa (BN) and binge eating disorder (BED) is ≥ 4 per month and the number of compensatory behaviors is ≥ 4 per month for BN. The number of OBE is < 4 and ≥ 1 per month for subthreshold BN and BED and the number of compensatory behaviors is < 4 and ≥ 1per month for subthreshold BN. In addition, EDE-Q weight and shape concerns subscale scores are ≥ 4 for full diagnostic criteria and ≥ 3 and < 4 for subthreshold BN and BED (49, 50). Thirty-eight percent of the women in the current study had weight or shape concerns in a range that puts them at risk for a DSM 5 ED diagnosis while 13.8% had a possible full or subthreshold ED. Using the new DSM 5 criteria for a lifetime BED diagnosis, the approximate prevalence rate in US adult women is estimated at 3.6% (51). Thus, these women may be at risk for eating disordered behaviors with the possibility of relapse to substances to cope with concerns related to weight, shape and eating. It has been reported that women in recovery from SUD not only have concerns about eating and weight but actually gain a substantial amount of weight after cessation of drug use (52). Women also report that addictive disorders affect their eating patterns (52).

The finding that past 30 day opiate use was associated with EDE-Q global scores as well as weight and shape subscale scores may indicate that opiates also play a substantial role in disordered eating symptoms. Use of opiates may act as an appetite suppressant for women with weight and shape concerns. In a meta-analysis of drug use in eating disorders, it was found that individuals with ED were just as likely to use opiates as other appetite suppressant drugs (6). Opiates and other central nervous system depressants may also be taken to counteract stimulant adverse effects such as restlessness and anxiety (6). In a recent large epidemiologic study of treatment seeking individuals with SUD, Mathews and colleagues (2013) found that opiate use was associated with lower BMI than stimulant use disorders (53). Recent evidence from animal and human studies demonstrates that the hedonic properties of palatable food stimulate endogenous opioid activity. Similar to exogenous opioids, chronic consumption of palatable food may downregulate endogenous opioid function. Thus, use of opioids and palatable food may both be used to satisfy aversive withdrawal states (54–56).

It is not surprising that the EDE-Q restraint subscale was the only subscale with no significant correlations with race/ethnicity, days of opiate use in past 30 days, and ASI composite subscale scores, Studies have consistently shown SUD to be most highly correlated with bulimic and binge eating disorders and less correlated with anorexia and restricting behaviors(6, 14).The finding that the ASI psychiatric subscale scores, but not PTSD severity, are associated with global, eating, weight and shape concerns may underscore the impact of comorbidity and emotional stressors experienced during recovery on eating cognitions and behaviors. All of the women enrolled in the current study had either PTSD or subthreshold PTSD, resulting in a restricted range of trauma symptom severity that may have influenced this finding. High rates of suicide attempts in ED are correlated with depression, SUD and childhood physical and sexual abuse (57). There are also several reports that show unhealthy diets and diets with more trans saturated fatty acids are associated with increased depression and anxiety (58, 59). Future studies should explore dietary intake of women in recovery. Further, early identification of patients with comorbid SUD and eating disorders would be important for suicide prevention.

As in other studies of ED, in the current study Caucasian women had higher EDE-Q scores than African American women (60). It has been reported that Caucasian women are more vulnerable to social pressure to be thin, have more body dissatisfaction and diet more than African American women (61). Although there are also studies indicating that this disparity has been overstated, in the current study Caucasian women had higher EDE-Q scores than African American women, there were no differences in self-reported OBEs. This result is consistent with reports that rates of Binge eating disorder, a new DSM 5 diagnostic disorder, are comparable in Caucasian and African American populations (62).

Compared to either disorder alone, rates of comorbid PTSD are highest in patients with bulimic ED and SUD (5). There are a number of integrated treatments that address both PTSD and SUD but few SUD programs address ED, and typically these disorders are treated separately and/or sequentially (i.e., not in an integrated way). None of the community programs in the current study had eating disorder treatment services other than offering health and nutrition education. Given the functional relationship between ED and SUD, patients may benefit from an integrated approach which may improve treatment outcomes for both disorders. Cognitive behavioral treatment strategies can target affect dysregulation and specific cognitions associated with both ED and SUD psychopathology (63). For example, addressing emotional eating and body image may be an important intervention for relapse prevention.

There are several limitations to the current study. This was a secondary analysis and not hypothesis driven. The exploratory nature of the findings requires further investigation to establish reliable relationships. Only a subsample of interested community treatment programs from the larger study participated in the EDE-Q survey. However, four geographically dispersed community treatment program sites did participate. EDE-Q scores were only representative of the past 28 days and lifetime history was not obtained. Disordered eating cognitions and behaviors may not be manifested until women are further along in recovery, particularly for women who may replace substances with food for negative affect and stress reduction. The current study did not obtain weights or body mass indices (BMI) which could provide additional information supporting eating, weight and shape concerns. Without knowing if patients in recovery from SUD are under or above BMI norms, it is difficult to ascertain if weight and shape concerns are consistent with disordered eating secondary to SUD or a pre-existing ED that is manifested or exacerbated in SUD recovery.

In conclusion, additional research is needed to further explore the relationship between SUD/ED, particularly for various substances of abuse, addiction severity, PTSD, and other psychiatric comorbidity. Assessing and addressing ED behaviors in SUD recovery may improve SUD outcomes. Interventions that address nutrition, body image and emotional eating in recovery may reduce the risk of relapse in patients who have eating and weight concerns. Particularly, women’s residential and intensive treatment programs may want to re-examine the types of food offered and consumed in their programs. Also, interventions that include focus on development of affect regulation skills would most likely be beneficial in helping patients interrupt the cycle of self-medicating with both food and drugs/alcohol. Comprehensive and integrated treatment approaches need to be developed to address this complex but common comorbidity.