Comorbid Eating, Depressive, and Anxiety Psychopathology is Associated with Elevated Shame in Women with Food Insecurity

Abstract

Objective:

To compare participants with current food insecurity and different psychopathology profiles on shame, guilt, anxiety, and depression using a cross-sectional design.

Method:

Women with current food insecurity (n=99; 54% White) were placed into four groups based on their endorsement of symptoms of psychopathology: eating disorder with depression/anxiety comorbidity (ED-C group; n=17), depression/anxiety only (Depression/Anxiety group; n=34), eating disorder only (ED group; n=12), and no-diagnosis (n=36). Groups were compared on self-report measures of shame, guilt, depression, and anxiety using analysis of covariance.

Results:

The presence of an eating disorder was associated with quadruple the risk of screening positive for comorbid depression and anxiety. The ED-C group reported elevated shame relative to the ED and no-diagnosis groups. The ED-C group reported the highest levels of anxiety, followed by the Depression/Anxiety group, and the ED and no-diagnosis groups.

Discussion:

The presence of an eating disorder with comorbidity among women with food insecurity is associated with heightened shame. Given shame’s status as a transdiagnostic predictor of psychopathology, it may serve as a putative mechanism underlying the relationship between food insecurity and eating disorder comorbidity.

Introduction

Food insecurity, defined as the lack of consistent access to sufficient quantity or variety of affordable, nutritious food (USDA, 2018), is a pressing concern in the U.S. Recently, 10.2% of U.S. households reported experiencing food insecurity (Coleman-Jensen et al., 2022). Food insecurity is associated with nutritional deficiencies, chronic health problems, poor overall health, and psychopathology, especially among women (Leung et al., 2020; Maynard et al., 2018; Weaver & Fasel, 2018).

Historically, food insecurity has demonstrated strong, consistent associations with both depression and anxiety (Arenas et al., 2019; Maynard et al., 2018). However, emerging research suggests food insecurity also is linked to eating disorder (ED) psychopathology, including ED diagnoses (Hazzard et al., 2020). Among college students, the correlation between food insecurity and ED diagnosis remained significant even when controlling for the presence of depression and anxiety (Zickgraf et al., 2022), indicating a unique relationship. Because research on food insecurity and EDs is relatively nascent, more study is needed to understand the correlates of this relationship as an important first step in identifying targets for prevention and intervention.

Although research is clear that independent associations exist between food insecurity and ED, depressive, and anxiety psychopathology, less is understood about the co-occurrence of these problems, and the implications of co-occurrence, among women with food insecurity. Given the unique challenges of living with limited or uncertain access to food, the experience of simultaneous disordered eating could elevate risk for anxiety and depression among women with food insecurity.

Among samples not selected for food insecurity, EDs are highly comorbid with depression and anxiety (Garcia et al., 2020), and the presence of comorbid anxiety and depression predicts deleterious outcomes among those with EDs, including poor health-related quality of life (Singleton et al., 2019), earlier treatment drop out (Vall & Wade, 2015), and suicide attempts (Ahn et al., 2019). Previous research has not investigated the extent to which disordered eating overlaps with anxiety and depression among women experiencing food insecurity and the potential correlates of such overlap.

One potential correlate of comorbid psychopathology among women with food insecurity is the experience of shame, a complex emotion characterized by an evaluation of the self as irreparably flawed, incapable, and unacceptable to others (Tangney & Dearing, 2002). Shame is part of a family of negatively-valenced, self-conscious emotions; others include guilt and embarrassment. Shame is thought to be the most harmful of these emotions due to its emphasis on the global self as overwhelmingly bad or wrong (Tangney & Dearing, 2002), while guilt and embarrassment represent negative evaluations of a specific behavior. Importantly, shame and guilt can, and do often, co-occur (Tangney & Dearing, 2002). However, it is the shame part of the emotional response that is considered more painful (Tangney et al., 2007).

Importantly, in both cross-sectional and longitudinal studies of participants not selected for food insecurity, shame consistently emerges as the self-conscious emotion most strongly associated with, and predictive of, psychopathology including depression (Kim et al., 2011), anxiety (Cândea & Szentagotai-Tătar, 2018), and disordered eating (Blythin et al., 2020; O’Loghlen et al., 2022). Guilt, on the other hand, has inconsistent associations with psychopathology (Tangney et al., 2007) and is viewed as a potentially adaptive emotion that may motivate one toward positive, reparative behavior that is protective from psychological harm (Treeby et al., 2018; VanDerhei et al., 2014). Shame is therefore considered a transdiagnostic indicator of, and risk factor for, psychopathology, and may serve as a mechanism underlying psychiatric comorbidity (Davis et al., 2019; Davis et al., in press). Yet, associations among shame, EDs, depression, and anxiety have not been tested in food insecure populations.

There is theoretical and preliminary empirical reason to believe individuals with food insecurity may be particularly vulnerable to shame. Feelings of shame often result from moral transgressions, perceptions of one’s behavior as deviant, and/or failing to live up to societal standards (Brown, 2006). Being without enough food to support oneself or one’s family, and feeling forced to acquire food in unconventional or unexpected ways, may be viewed as deviant or failing and elicit shame. Indeed, qualitative research suggests food insecurity is highly stigmatized; individuals report feeling blamed for their food insecurity (Thompson et al., 2018), and receiving food assistance is associated with shame (Middleton et al., 2018).

Despite accumulating qualitative evidence, quantitative data regarding level of shame in food insecure populations are limited. One study used a single-item measure – “Did you feel shame if somebody, including friends, knew that you are without food?” – to assess feelings of shame related to food insecurity among children. Results supported a positive association between food insecurity and feelings of shame among children (Bernal et al., 2016).

Although there is strong scientific evidence differentiating shame and guilt in the prediction of symptoms of psychopathology and behavior (Kim et al., 2011; Sheehy et al., 2019; Tangney & Dearing, 2002), because they often co-occur in response to the same event, it can be difficult for participants to differentiate the two emotions. As a result, qualitative findings regarding shame may conflate feelings of guilt with feelings of shame, or vice versa. Because quantitative measures are validated to differentiate shame and guilt, quantitative data on shame and guilt in food insecure populations may provide valuable information that (1) clarifies distinctions between shame and guilt among individuals with food insecurity, (2) complements findings of qualitative research, or (3) uncovers new information that can inform research moving forward.

Given the strong, consistent links between shame and psychopathology (Blythin et al., 2020; Cândea & Szentagotai-Tătar, 2018; Kim et al., 2011; O’Loghlen et al., 2022) and food insecurity and psychopathology (Hazzard et al., 2020; Maynard et al., 2018), it is possible that individuals with food insecurity and co-occurring ED, depression, and anxiety psychopathology experience heightened levels of shame. If true, shame may be understood as a putative mechanism underlying psychiatric comorbidity in those with food insecurity, and a potential treatment target for patients presenting with co-occurring food insecurity and eating psychopathology.

The Current Study

The current study sought to evaluate cross-sectional associations among shame, EDs, depression, and anxiety in a sample of women with food insecurity. We first evaluated the presence of depression/anxiety comorbidity in participants with EDs, and then formed groups with different profiles of psychopathology. We anticipated that the presence of an ED would predict the presence of depression/anxiety comorbidity. We compared the resulting groups on shame, guilt, depression, and anxiety, and we predicted that:

1. Participants who screened positive for an ED, depression, or anxiety would report greater shame than participants who did not meet criteria for an ED or screen positive for depression and/or anxiety.

2. Participants with ED comorbidity would report the highest level of shame in the sample.

To confirm the validity of group membership, we predicted that groups that screened positive for depression and/or anxiety would report higher depression and anxiety scores, respectively, compared with groups that did not screen positive for depression and/or anxiety.

Method

Participants

Participants (N=99, mean (SD) age=40.26 (14.33) years) were self-identified women currently experiencing food insecurity in the U.S. Self-identified women were the focus of the current study given previous research that women with food insecurity are at particularly high risk for psychopathology (Maynard et al., 2018). Participants were recruited using flyers on bulletin boards, social media advertisements, and ResearchMatch, a national health volunteer registry. Telephone screens were conducted to verify eligibility. Inclusion criteria were as follows: identifying as female, being 18 years of age or older, meeting United States Department of Agriculture (USDA) criteria for household food insecurity, and having internet access to complete study questionnaires online. Exclusion criteria were as follows: unable to read English, pregnant at the time of the study, or unable to access the internet. Participant self-reported demographic information is reported in Table 1.

Table 1.

Self-reported demographic information across participants (N = 99)

Race	N	(%)
White	54	54.5%
Black	30	30.3%
Asian	4	4.0%
Pacific Islander	1	1.0%
American Indian	1	1.0%
Biracial	6	6.1%
Other (not specified)	3	3.0%
Ethnicity
Hispanic/Latinx	10	10.1%
Non-Hispanic/Latinx	79	79.8%
Not reported	10	10.1%
Level of education
High school diploma	4	4.0%
Some college	23	23.2%
Associate’s degree	17	17.2%
Bachelor’s degree	34	34.3%
Master’s degree	17	17.2%
Doctoral or professional degree	2	2.0%
Unreported	2	2.0%
Employment status
Full-time job	30	30.3%
Part-time job	28	28.3%
Stay-at-home parent/homemaker	7	7.1%
Student	4	4.0%
Unable to work	20	20.2%
Unemployed	5	5.1%
Retired	4	4.0%
Other (unspecified)	1	1.0%
Family composition
Child(ren) in household	38	38.4%

Measures

USDA Household Food Security Survey Module (HHFSM; Bickel et al., 2000).

The USDA’s HHFSM is a structured interview used to assess food security over the past 12 months, including 18 questions for individuals who endorse having children in the household, and 10 questions for individuals who do not endorse having children in the household. Total scores range from 0 to 10 for households with no children, and 0 to 18 for households with children, and exist on a continuum of four classifications: full, marginal, low, and very low food security. “Full food security” (score of 0), indicates a household with very little to no trouble obtaining food. “Marginal food security” (score of 1–2), describes a household that experiences anxiety acquiring food, with no reduction in the quantity, quality, or variety of foods consumed. “Low food security” (score of 3–5 for households with no children, score of 3–7 for households with children), refers to households that experience trouble or anxiety obtaining high quality and/or a variety of foods, with no reduction in the amount of food consumed. “Very low food security” (score of 6–10 for households with no children, score of 8–18 for households with children) is defined by an inadequate amount and variety of food consumed by the household. Research supports the construct validity of the HHFSM (Marques et al., 2015). In previous studies, the HHFSM demonstrated good to excellent internal consistency and high test-retest reliability (Derrickson et al., 2000; Derrickson et al., 2001). In the current study, internal consistency was excellent (α =.92).

ED Diagnostic Interview (EDDI; (Stice et al., 2017).

The EDDI is a semi-structured interview that generates diagnoses of EDs based on DSM-5 criteria. The EDDI demonstrates excellent test-retest reliability (Stice et al., 2012).

Body Mass Index (BMI).

Participants self-reported their height and weight, which were used to calculate BMI. Good agreement has been found between self-reported BMI and direct anthropometric measurements (Davies et al., 2020).

Patient Health Questionnaire-9 (PHQ-9; (Kroenke, Spitzer, et al., 2001).

The PHQ-9 is a brief, 9-item self-report questionnaire used to screen for and classify the severity of depression. Participants indicate how often they have experienced symptoms of depression over the past two weeks, using a Likert scale anchored at 0 (“not at all”) and 3 (“nearly every day”). Total scores range from 0–27 with higher scores indicating more severe depression. The PHQ-9 demonstrates high reliability and convergent validity with both longer self-report and interview measures of depression (Kroenke, West, et al., 2001). In the current study, internal consistency was good: α =.89.

General Anxiety Disorder-7 (GAD-7; (Spitzer et al., 2006).

The GAD-7 is a brief, 7-item self-report questionnaire used to screen for and classify the severity of generalized anxiety disorder (GAD). Participants indicate how often they have experienced symptoms of anxiety over the past two weeks, using a Likert scale anchored at 0 (“not at all”) and 3 (“nearly every day”). Scores range from 0 to 21. Higher scores indicate more severe anxiety symptoms. The GAD-7 demonstrates high internal consistency, construct validity, and convergent validity with other anxiety disorder measures among clinical (Johnson et al., 2019) and non-clinical samples (Löwe et al., 2008). In the current study, internal consistency was excellent: α=.93.

Personal Feelings Questionnaire-2 (PFQ-2; (Harder & Zalma, 1990).

The PFQ-2 is a self-report questionnaire that uses an adjective checklist to measure the degree of generalized shame and guilt experienced by a participant. Items are rated on a 4-point Likert scale anchored at 0 (never experience the feeling) to 4 (experience the feeling continuously or almost continuously). There are 10 items on the shame subscale, (e.g., “embarrassed”, “humiliated”) and six items on the guilt subscale (e.g., “remorse”, “regret”). Though shame and guilt tend to correlate, it is shame independent of guilt that is associated with harms (Tangney et al., 2007), so it is important to control for guilt when evaluating shame. The PFQ-2 is characterized by good internal consistency and adequate test-retest reliability (Di Sarno et al., 2019; Sekowski et al., 2020). In the current study, internal consistency for each of the two subscales was good: guilt, α=.85 and shame, α=.88.

Procedure

This study received Institutional Review Board approval from the Medical Center’s human subjects committee. Participants provided written informed consent prior to participation. We used a cross-sectional study design. Data were collected between January 2021 – January 2022. Procedures included the completion of two 45-minute virtual study visits: an electronic questionnaire visit and an interview visit conducted via telephone or Zoom with a member of the research staff. Participants received a $60 electronic gift card for participation in both study visits.

Creation of study groups.

We created a transdiagnostic ED diagnostic group based on DSM-5 (APA, 2013) criteria using participants’ responses to the EDDI. Criteria for each diagnosis are presented in Table 2.

Table 2.

Criteria used for eating disorder diagnoses among participants

Diagnosis	Criteria
Anorexia Nervosa (AN)	(1) Body mass index (BMI) less than 18.5 kg/m2 (2) Definite fear of weight gain more than 75% of the days for at least 3 months (3) Self-evaluation unduly influenced by body shape or weight, or an intense fear of becoming fat.
Bulimia Nervosa (BN)	(1) At least 4 binge eating episodes per month for at least 3 months (2) At least 4 compensatory behavior episodes per month for at least 3 months (3) Weight and shape were one of the main aspects of self-evaluation.
Binge-Eating Disorder (BED)	(1) At least 4 binge eating episodes per month for at least 3 months (2) < 1 compensatory behavior on average per month for 3 months (3) Binge eating characterized by 3 or more of the following: rapid eating; eating until uncomfortably full; eating large amounts when not physically hungry; eating alone because of embarrassment; feeling disgusted, depressed, or guilty after overeating.
Purging Disorder (BED)	(1) Purging episodes (e.g. self- induced vomiting, laxative use, or diuretic use) occurring, on average, at least once per week for the past 3 months (Keel & Striegel-Moore, 2009). (2) The absence of objectively large binge eating episodes. (3) Self-evaluation unduly influenced by body shape or weight, or an intense fear of becoming fat.
Atypical AN (AAN)	(1) Body mass index (BMI) greater than 18.5 kg/m2. (2) Significant weight loss, defined by greater than 5% weight loss (Forney et al., 2017), and determined by subtracting current weight from lifetime highest weight, and dividing amount of weight loss by lifetime highest weight to determine percentage of weight loss (Shaumberg et al., 2016). (3) Self-evaluation unduly influenced by body shape or weight, or an intense fear of becoming fat. A diagnosis of DSM-5 BED, BN, OSFED BED, OSFED BN, or PD trumped a diagnosis of AAN.
OSFED, BN of low frequency or limited duration	(1) At least 1 binge eating episode per month for at least 3 months, or at least 3 episodes over a shorter period. (2) At least 1 compensatory behavior episode per month for at least 3 months or at least 3 episodes over a shorter period. (3) Self-evaluation unduly influenced by body shape or weight.
OSFED, BED of low frequency or limited duration	(1) At least 1 binge eating episode per month for at least 3 months, or at least 3 episodes over a shorter period. (2) < 1 compensatory behavior on average per month for 3 months. (3) Binge eating characterized by 3 or more of the following: rapid eating; eating until uncomfortably full; eating large amounts when not physically hungry; eating alone because of embarrassment; feeling disgusted, depressed, or guilty after overeating.

Note: OSFED = Other Specified Feeding or Eating Disorder.

Participants’ responses to the PHQ-9 and GAD-7 were used to indicate their depression and anxiety screening status, respectively. Cut-off scores ≥ 10 were used to identify individuals who were likely to meet criteria for depression (Levis et al., 2019) and anxiety (Löwe et al., 2008).

Participants who did not meet criteria for an ED diagnosis and screened negative for depression and anxiety were assigned to the no-diagnosis group. Participants who met criteria for an ED diagnosis but screened negative for depression and anxiety were assigned to the ED (ED) group. Participants who met criteria for an ED diagnosis and screened positive for depression, anxiety, or both, were assigned to the ED comorbidity (ED-C) group. Participants who screened positive for depression, anxiety, or both depression and anxiety, and did not meet criteria for an ED diagnosis, were assigned to the Depression/Anxiety group.

Data Analytic Plan

Statistical analyses were performed using SPSS version 24 (IBM, 2016). In advance of statistical analyses, descriptive statistics were calculated. Means were computed for participant characteristics (age, BMI), and scores on study variables (food insecurity, shame, guilt, depression, and anxiety) and presented by study group (ED, ED-C, depression/anxiety, and no-diagnosis groups). Differences in participant characteristics and study variables by group were examined using chi-square tests for categorical variables (food insecurity status) and analysis of variance (ANOVA) for continuous variables (age, BMI).

An odds ratio and 95% confidence interval was calculated using binary logistic regression comparing the rate of comorbid depression and/or anxiety screening for participants diagnosed with an ED compared to participants without an ED. ED diagnosis (dichotomous; present or absent) was entered as the outcome variable. Comorbidity status was entered as the predictor variable, and was scored dichotomously as follows: “1” represented present for two or more of the following: depression, anxiety, or ED diagnosis; “0” represented present for only one of the following: depression, anxiety, and ED diagnosis, or absent for all three.

Analysis of covariance (ANCOVA) with Bonferroni correction was used to evaluate group differences in shame, controlling for guilt, and group differences in guilt, controlling for shame. Shame and guilt were entered as covariates in ANCOVA to control for overlap between the two constructs (Tangney & Dearing, 2002) as is standard in shame assessment (Tangney et al., 2007). ANOVA was used to test hypotheses regarding group differences in depression and anxiety. A Bonferroni-adjusted p-value was used to evaluate statistical significance of post-hoc comparisons. For variables that violated homogeneity of variance (age, depression, anxiety), results from the Welch’s test (adjusted F statistic and significance level) are reported.

Effect sizes (Cohen’s f) were calculated for each significant result. f=0.1 represents a small effect size, f=0.25 represents a medium effect size, and f=0.4 represents a large effect size (Cohen, 1988).

Results

Composition of participant groups.

Thirty-six participants were placed in the no-diagnosis group. The ED group (n=12) consisted of the following diagnoses: Atypical anorexia nervosa (AAN; n=5; 41.6%), binge-eating disorder (BED; n=4; 33.3%), other specified feeding or eating disorder (OSFED) BED (n=2; 16.7%), and bulimia nervosa (BN; n=1; 8.3%). The ED-C group (n=17) was comprised of: AAN (n=5; 29.4%), OSFED BED (n=5; 29.4%), BN (n=3; 17.6%), BED (n=2; 11.8%), purging disorder (PD; n=1; 5.9%), and OSFED BN (n=1; 5.9%). Of participants in the ED-C group, the majority (n=13; 76.4%) screened positive for both depression and anxiety; 2 participants (11.8%) screened positive for only depression, and 2 participants (11.8%) screened positive for only anxiety. The Depression/Anxiety group consisted of n=34 participants who screened positive for both depression and anxiety (n=19; 55.9%), only depression (n=12; 35.3%), and only anxiety (n=3; 8.8%).

Demographics.

As shown in Table 3, the four groups did not differ significantly by race, ethnicity, or food security status. Over half of participants in each group were classified as having “very low food security.” More than 24% of participants in each group were classified as having “low food security”, and 8 – 25% of participants in each group were classified as having “marginal food security”. Age differed significantly across the four participant groups. Specifically, participants in the ED and ED-C groups were significantly older on average than participants in the Depression/Anxiety group. Although the omnibus test indicated that the groups differed significantly by BMI, post-hoc contrasts revealed no significant pairwise differences.

Table 3.

Characteristics of study groups on demographic variables

	ED (n = 12)	ED-C (n = 17)	Depression/Anxiety (n = 34)	No-diagnosis (n = 36)
Variable	n (%)				X 2	p
Food Security Statusa					4.89	.56
Marginal food security	3 (25%)	1 (14.3%)	3 (8.8%)	6 (16.7%)
Low food security	3 (25%)	7 (41.2%)	8 (23.5%)	10 (27.8%)
Very low food security	6 (50%)	9 (52.9%)	23 (67.6%)	20 (55.6%)
Race					19.62	.35
White	6 (6.1%)	12 (12.1%)	16 (16.2%)	20 (20.2%)
Black	5 (5.1%)	2 (2%)	14 (14.1%)	9 (9.1%)
Asian	0	0	1 (1%)	3 (3%)
Pacific Islander	0	0	1 (1%)	0
American Indian	0	0	0	1 (1%)
Biracial	1 (1%)	3 (3%)	0	2 (2%)
Other (unspecified)	0	0	2 (2%)	1 (1%)
Ethnicity					6.54	.69
Hispanic/Latinx	0	1 (1%)	6 (6.1%)	3 (3%)
Non-Hispanic/Latinx	11 (11.1%)	14 (14.1%)	25 (25.3%)	29 (29.3%)
Not reported	1 (1%)	2 (2%)	3 (3%)	4 (4%)

	Mean (SD)				F	p	Post-hoc Bonferroni tests
Age	48.00 (8.25)	45.00 (13.68)	33.91 (11.88)	41.44 (16.22)	4.54	.005	ED > Depression/Anxiety ED-C > Depression/Anxiety
BMI	34.73 (7.34)	32.06 (10.30)	26.87 (7.54)	29.55 (10.20)	2.72	.049

Note: ED = Eating Disorder only group; ED-C = Eating Disorder Comorbidity group; BMI = Body Mass Index. BMI was calculated using self-reported height and weight.

^aFood security status was measured using the USDA Household Food Security Survey Module.

Hypothesis Tests

Overlap between depression/anxiety and ED diagnostic status.

Consistent with our hypothesis, binary logistic regression analyses indicated participants who met criteria for an ED were 4.39 [95% CI (1.76, 10.98)] times more likely to screen positive for comorbidity than participants who did not meet criteria for an ED (p < .01).

Shame.

Significant differences in the mean levels of shame, controlling for guilt, were reported by the various study groups (f=0.40; see Table 4). In partial support of our hypotheses, the ED-C group reported significantly higher levels of shame than the ED (f=0.58) and no-diagnosis groups (f=0.46). However, contrary to our hypotheses, there were no significant differences in shame between the ED, no-diagnosis, and Depression/Anxiety groups. There also were no significant differences in shame between the ED-C and Depression/Anxiety groups. This pattern of results did not differ when guilt was removed from the model. There were no significant group differences in guilt scores when controlling for shame. When shame was removed from the model, only the ED-C group reported significantly higher guilt than the no-diagnosis group. The ED-C, ED, and depression/anxiety groups did not differ on guilt.

Table 4.

Results of Analysis of Covariance and Analysis of Variance.

	ED (n = 12)	ED-C (n = 17)	D/A (n = 34)	ND (n = 36)
Variable	Mean (SD)				F
PFQ-2 Shame1	11.67 (6.75)	21.81 (6.65)	16.85 (7.41)	10.94 (7.00)	4.876**
PFQ-2 Guilt2	8.67 (5.33)	12.71 (5.80)	9.65 (4.47)	6.81 (4.49)	.73
PHQ-9	4.73 (3.00)	17.15 (5.67)	14.21 (4.17)	5.06 (2.82)	55.58**
GAD-7	3.25 (3.08)	14.76 (4.92)	11.44 (4.18)	3.47 (2.27)	52.63**

Note:

^**p < .01,

^***p < .001.

OR = Odds ratio. ED = Eating Disorder only group; ED-C = Eating Disorder Comorbidity group. D/A = Depression/Anxiety group. ND = No Diagnosis group.

^1.Adjusted for guilt

^2.Adjusted for shame

Depression and anxiety.

Mean level of depression (f=2.63) and anxiety (f=1.25) differed significantly across groups (Table 4). Confirming the validity of group membership, post-hoc tests revealed the ED-C group had significantly higher depression scores than the no-diagnosis (f=2.09) and ED (f=1.67) groups. The ED-C group also had significantly higher anxiety scores than the no-diagnosis (f=1.03) and ED (f=.82) groups. As expected, the Depression/Anxiety group had significantly higher depression scores compared with the no-diagnosis (f=1.94) and ED (f=1.43) groups. The Depression/Anxiety group also had significantly higher anxiety scores compared with the no-diagnosis (f=.89) and ED (f=.65) groups. The ED-C and Depression/Anxiety groups did not differ significantly on level of depression. The ED-C group reported significantly higher anxiety than the Depression/Anxiety group (f=.30), suggesting uniquely high levels of anxiety among the ED-C group.

Discussion

This study found that, in women with food insecurity, the combination of an ED diagnosis and screening positive for depression/anxiety was associated with elevations in the harmful, maladaptive emotion of shame when compared to women with only an ED, or no diagnosis. Findings build upon a broader literature linking food insecurity and psychopathology. Although novel in a sample with known food insecurity, these findings are unsurprising when considering the strong associations between shame and EDs, depression, and anxiety across clinical and nonclinical samples not selected for food insecurity (Blythin et al., 2020; Cândea & Szentagotai-Tătar, 2018; Kim et al., 2011; O’Loghlen et al., 2022).

Notably, study groups did not differ on guilt when adjusting for shame, but study groups differed on shame whether or not guilt was included in the model, indicating a stronger association between shame and psychopathology than guilt and psychopathology. This finding is consistent with previous research indicating that when both shame and guilt are included in a predictive model, shame generally emerges as significant (Tangney et al., 2007; Davis et al., in press), suggesting shame is the more potent correlate and predictor of psychopathology. Guilt demonstrates inconsistent links with, and is theorized to be protective against, psychopathology (Tangney et al., 2007; Tilghman-Osborne et al., 2010). Such associations may explain why the ED-C group demonstrated particularly elevated levels of shame, relative to the ED and no-diagnosis groups. Given independent associations between shame and each of the forms of psychopathology examined in this study (Blythin et al., 2020; Cândea & Szentagotai-Tătar, 2018; Kim et al., 2011), it is possible that there is a cumulative effect of psychopathology on shame. The combination of food insecurity and multiple forms of psychopathology may be associated with more self-blame, self-disparagement, and perceived stigma (all of which are consistent with shame) than with blaming of external factors, such as the inequitable food system in the U.S. The current findings, and the emphasis on shame rather than guilt, align with previous qualitative research supporting a link between shame and food insecurity (Swales et al., 2020).

The associations among ED psychopathology, depression, anxiety, and elevated shame in our food insecure sample may be explained, in part, by the unique interplay among these constructs. ED psychopathology and food insecurity are each associated with a preoccupation with food (Hazzard et al., 2020). Depression and anxiety often involve anxious rumination, and, in the case of food insecurity, could include worries about both past decisions that led to food insecurity and the uncertain future of one’s (or one’s family’s) access to food (Stack & Meredith, 2018). Potentially, disordered eating cognitions, pressure to obtain foods that one deems “healthy” despite having limited access to nutritious foods, and worry about the sufficient or equitable distribution of food among one’s family members, could lead to feelings of inadequacy or failure, and thus may promote the experience of shame.

Future research investigating the temporal relationship between variables may allow for a better understanding of the role of shame in the development of symptoms of psychopathology within food insecurity. Qualitative findings suggest individuals with food insecurity feel stigmatized and ostracized by society, which may bring about distress in the form of shame (Pineau et al., 2021; Swales et al., 2020). As a transdiagnostic risk factor for disordered eating, depression, and anxiety (Blythin et al., 2020; Cândea & Szentagotai-Tătar, 2018; Kim et al., 2011; O’Loghlen et al., 2022), elevated shame following continued food insecurity may predict the onset of such symptoms. Indeed, longitudinal research suggests, for some individuals, food insecurity precedes the development of disordered eating (Hazzard et al., 2022) and depression (Heflin et al., 2005). However, there also is evidence for prospective bi-directionality between depression and food insecurity (Huddleston-Casas, 2009). Given associations between shame and food insecurity (Swales et al., 2020) and shame and psychopathology (Blythin et al., 2020; Cândea & Szentagotai-Tătar, 2018; Kim et al., 2011; O’Loghlen et al., 2022), shame may serve as a correlate of, or putative mechanism underlying, the relationship between food insecurity and psychopathology regardless of the direction. Future studies should address this possibility using prospective, longitudinal designs.

Shame is associated with substantial psychosocial costs, including social isolation and loneliness (Rostami & Jowkar, 2016). These consequences may be even more costly among women with food insecurity. Given qualitative evidence suggesting that women who use charitable food programs feel shame about doing so (Pineau et al., 2021), shame could serve as a barrier to obtaining the resources needed to break out of the food insecurity cycle and, if food insecurity is shown to be a prospective risk factor for psychopathology in this population, the ED-depression-anxiety cycle. Previous research also suggests individuals with food insecurity feel shame in others knowing they are without food (Bernal et al., 2016), which is likely a contributing factor to known hesitations in asking for help (Swales et al., 2020). In this way, shame may also be a consequence of experiencing food insecurity, thus serving as a maintenance factor for ongoing food insecurity. Addressing shame in the psychological treatment of those with food insecurity may help patients to feel more comfortable reaching out to others in their families or networks for help, increase the use of community food pantries and governmental assistance, and in turn disrupt the cycle of food insecurity, which may additionally address symptoms of psychopathology. Importantly, it also is imperative that national policies are enacted to improve food access and address the mental health correlates of food insecurity.

The current findings support the possibility that shame may maintain both food insecurity and psychopathology. As such, clinicians treating patients with food insecurity, an ED, and depression/anxiety should consider the incorporation of interventions that target shame. Treatments focused on self-compassion (Gilbert, 2009) have shown effectiveness in treating shame in individuals with disordered eating (Kelly et al., 2014), and even brief, experimental interventions, such as mindfulness techniques, writing compassionate letters to oneself, and sensory-focused self-soothing exercises, have demonstrated significant reductions in shame among individuals experiencing EDs, depression, and anxiety (see Goffnett et al., 2020 for a review). Studies testing self-compassion interventions in samples with food insecurity and ED comorbidity are needed, and research to modify such interventions to be suitable and effective for individuals with food insecurity may be necessary.

Although not a core objective of the current study, we did find uniquely high levels of anxiety in the ED-C group compared to the other study groups, including the Depression/Anxiety group. This finding is consistent with the common overlap between eating and anxiety disorders (Blinder et al., 2006; Elran-Barak & Goldschmidt, 2021; Swinbourne & Touyz, 2007) and anxiety and food insecurity (Arenas et al., 2019). Importantly, a recent study found that ED symptoms are most severe when comorbid anxiety is present (Elran-Barak & Goldschmidt, 2021). Future studies may further tease apart this finding by examining specific components of anxiety. For example, given the tendency for worry about where one’s next meal may come from, and empirical evidence for the cross-sectional overlap among food insecurity, disordered eating, and worry (Becker et al., 2017), it is possible that the risk process for disordered eating, anxiety, and food insecurity emphasizes worry in particular.

Notably, there were high rates of EDs in our sample – almost a third of participants met criteria for BN, AAN, PD, BED, OSFED BN, or OSFED BED. Across the ED and ED-C groups, the most common ED was AAN, followed by BED and OSFED BED. The prevalence of BED and OSFED BED is not surprising, given findings that food insecurity is uniquely associated with, and serves as a risk factor for, binge eating (Hazzard et al., 2022). The relatively high prevalence of AAN is somewhat surprising; previous studies have reported a link between food insecurity and dietary restraint (Middlemass et al., 2021), a core feature of AAN (Forney et al., 2017). However, to our knowledge, this is the first study to report rates of AAN in a sample of women with food insecurity. Future studies of EDs in larger samples with food insecurity should assess criteria for AAN.

The presence of an ED diagnosis was associated with a four-fold increase in the likelihood of screening positive for two forms of psychopathology. The likelihood of comorbidity in this sample was similar to that observed in samples not selected for food insecurity. In a recent population sample study, individuals who met criteria for any ED (excluding anorexia nervosa) were 3.16 times more likely to meet criteria for a comorbid psychiatric disorder (Momen et al., 2022), which is slightly lower than the odds ratio of 4.39 observed in the current study. However, similarities should be interpreted with caution, as our sample was much smaller than that of Momen et al. (2022), and may not have been representative of the population of all individuals living with food insecurity in the U.S. Future studies may use prospective designs to investigate temporal relationships among ED, depression, and anxiety symptoms in women with food insecurity. It is possible that ED symptoms precede the development of depression and anxiety, as has been shown in samples not selected for food insecurity (Davis et al., 2019; Puccio et al., 2017; Tanofsky-Kraff et al., 2011). If true, ED symptoms may serve as a risk factor for the development of depression and anxiety symptoms in this population, and thus a risk factor for heightened shame that maintains food insecurity. Alternatively, evidence for a bidirectional relationship between depression and ED psychopathology among samples not selected for food insecurity (Puccio et al., 2017) suggests it is possible that the presence of depression prior to the experience of food insecurity could heighten risk for disordered eating.

Strengths of this research include the use of standardized, semi-structured interviews and self-report measures with strong psychometric properties. Additionally, racial and ethnic minority group representation was similar to population studies of food insecurity (Gomez & Perez, 2022). Our inclusion of only women is both a strength and potential limitation; women with food insecurity may be at higher risk for psychopathology (Maynard et al., 2018). Additionally, early conceptualizations of ED symptom development implicated shame as a factor that placed women at higher risk for the development of eating pathology (Rodin et al., 1984; Silberstein et al., 1987), given evidence that women are more prone to shame, compared to men (Else-Quest et al., 2012; Gross and Hansen, 2000; Lewis, 1971). Thus, an understanding of relations among shame, food insecurity, and psychopathology in women is warranted. However, we cannot know if results would differ in a more gender-diverse sample. To examine whether the role of shame is specific to women or relevant to understanding food insecurity and psychopathology across genders, future research of shame in food insecure samples should include men and other genders.

There were limitations to this research. First, we used positive screenings on the PHQ-9 and GAD-7 to determine depression and anxiety symptomatology, respectively, rather than a structured diagnostic interview. However, we used established, conservative cut scores demonstrating high sensitivity and specificity (Kocalevent et al., 2013; Löwe et al., 2008). Indeed, in general population studies, only 1% of respondents have GAD-7 scores > 10 (Löwe et al., 2008) and 5.4% of women have PHQ-9 scores > 10 (Kocalevent et al., 2013). Consistent with previous research (Arenas et al., 2019), participants in our study screened positive for depression (n=31, 31.3%) and anxiety (n=22, 22.2%) at much higher rates than previous population studies (Kocalevent et al., 2013; Löwe et al., 2008) indicating the high prevalence of such symptoms among women with food insecurity. Because depression and anxiety encompass a wide array of symptoms, future studies should use diagnostic assessments to determine if certain mood or anxiety disorders correlate most with EDs and shame within food insecurity. Relatedly, participants in our sample who screened positive for depression and/or anxiety were combined into one group, due to small numbers of participants screening positive for only anxiety or only depression. As such, our study was underpowered to detect differences between those who screened positive for only depression or only anxiety. Future studies may seek to replicate and extend our findings in larger samples to investigate potential differences in combinations of symptoms (e.g., ED plus anxiety only, ED plus depression only, and ED plus anxiety and depression).

Second, although our small sample size was appropriate for a preliminary test of these associations, this study should be replicated in larger samples that include other high-risk groups (e.g., college students, single mothers, welfare recipients). Third, the inclusion of only English-speaking women and women with Internet access may limit our understanding of shame, food insecurity, and ED comorbidity among non-English speaking women and women without Internet access. Fourth, it is possible that participants who self-selected into the study differed on variables of interest compared to individuals who would not volunteer under similar circumstances. We cannot know if such differences contributed to the pattern of results observed in the study. Participants were drawn from a combination of advertisements at local establishments and ResearchMatch, and therefore represent a convenience sample. We cannot know if findings would differ among a randomly-selected sample of women with food insecurity. Fifth, it is possible that shame is associated with other forms of psychopathology common in food insecurity but not assessed in our study, such as substance misuse (Pryor et al., 2016). Finally, the cross-sectional design precludes inferences of causality; this research can be viewed as a first step toward the development of a causal model of food insecurity, shame, and psychopathology.

Researchers should continue to investigate associations among food insecurity, disordered eating comorbidity, and shame to disentangle the temporal relationship of these variables in diverse populations. This, in turn, may inform prevention and intervention efforts. Clinicians should be sensitive to elevations in shame among individuals with food insecurity and disordered eating comorbidity, and compassion-focused interventions may prove helpful in treatment.

Public Significance Statement:

Women with food insecurity and an ED were more likely to also screen positive for depression and/or anxiety than women with food insecurity and no ED. Overlap between ED, depression, and anxiety was associated with elevated shame, a harmful, maladaptive emotion with negative psychosocial consequences.

Availability of data and material:

The data that support the findings of this study are available upon reasonable request from the University of Chicago Medicine EDs Program.