Research paperPinpointing core and pathway symptoms among sleep disturbance, anxiety, worry, and eating disorder symptoms in anorexia nervosa and atypical anorexia nervosa
Introduction
Every 52 minutes an individual dies from an eating disorder (ED) (Deloitte Access Economics, 2020). Anorexia nervosa (AN) is one of the most fatal psychiatric illnesses and compounding this issue, there are no evidence-based treatments for adults with AN (Arcelus et al., 2011; Berends et al., 2018; Chesney et al. 2014). AN and atypical-AN (A-AN), share most diagnostic features including restriction of food and fear of weight gain. The primary difference between the conditions is body mass index (BMI); individuals with AN have a BMI below 18.5 and those with A-AN have a BMI above 18.5 criteria (American Psychiatric Association [APA], 2013). Regardless of weight status, both AN and A-AN cause significant and comparable levels of impairment (Sawyer et al., 2016; Moskowitz and Weiselberg, 2017). Sleep disturbances, which include trouble falling or staying asleep, changes in sleep pattern, loss of energy, and feeling tired or fatigued, are strongly related to psychiatric illnesses such as anxiety and depression (Freeman et al., 2020), and a common complaint for individuals with an ED diagnosis (Cooper Loeb and McGlinchey, 2020), including AN and A-AN. In fact, sleep disturbance has been proposed to be a clinical marker in EDs (Abdou et al., 2018), and more specifically, that subjective sleep disturbance may be a clinical marker for illness severity of AN (Bat-Pitault et al., 2020). In light of the strong association between sleep disturbances and EDs, there is surprisingly little research examining the relationship between sleep disturbance and AN symptoms, and in particular the mechanisms of how sleep disturbance and AN symptoms interrelate with each other.
The recent, but limited, research available regarding sleep disturbances suggests that abnormalities in the sleep-wake and rest-activity circadian rhythms are relevant to AN symptomology. Specifically, these disruptions are associated with restriction of eating in the morning, increased daily mood fluctuations, and increased rates of sleep-related irregularities, such as hypersomnia, a neurological disorder expressed by excessive exhaustion and sleeping during the day (Menculini et al., 2019). Further, anxiety and worry are often co-morbid with EDs (e.g., Levinson et al., 2018a; Lloyd et al., 2020; Startup et al., 2013), with up to 75% of individuals with AN meeting criteria for a co-occurring anxiety disorder. Moreover, worry and anxiety are also related to sleep problems, with worry predicting sleep problems (e.g., McGowan et al., 2016) and sleep problems predicting anxiety the next day (e.g., Narmandakh et al., 2020), leading to the recommendation that addressing anxiety and worry may improve sleep.
One way that sleep disturbance might relate to AN symptoms is through anxiety and worry. Heightened worry increases the likelihood of sleep disturbance and sleep problems predict anxiety symptoms (e.g., McGowan et al., 2016; Narmandakh et al., 2020). It is likely that these factors are bidirectionally associated. That is, worry and anxiety increases sleep problems, and then sleep problems in turn increase worry and anxiety (e.g., Cox and Olatunji, 2016; McGowan et al., 2016). In addition to the significant association between anxiety (e.g., Steinglass et al., 2013), worry (e.g., Lloyd et al., 2020), and sleep problems (e.g., Bat-Pitault et al., 2020) with AN symptoms, subjective sleep disturbances indicate greater severity of illness, more negative emotionality, and worse quality of life in individuals with AN (Bat-Pitault et al., 2020). Still, it is unclear how specific types of anxiety, worry, and sleep disturbance symptoms are interrelated with specific ED symptoms in AN.
Based on this literature, we developed a theoretical conceptual model of how ED symptoms, anxiety, worry, and sleep disturbances may inter-relate (see Fig. 1). Specifically, we propose that restriction and fasting are used by individuals diagnosed with AN and A-AN to manage judgement of shape and weight cognitions and to cope with increased worry and anxiety (Frank et al., 2019). Restriction and fasting also connect to feeling tired and fatigued and loss of energy from low caloric intake (Casper et al., 2020). We expect a reciprocal relationship between feeling tired and loss of energy with worry and anxiety. Worry and anxiety connect to sleeping disturbances (tiredness and loss of energy), which connect back to more worry and anxiety (Cox and Olatunji, 2016; McGowan et al., 2016). Worry and anxiety also connect to restriction, due to anxiety inducing food restriction in AN (Frank et al., 2019). Therefore, the literature supports a model with multiple interrelations among AN symptoms, sleep disturbances, anxiety, and worry.
Network analysis is the ideal method to test our theoretically-based conceptual model because of the ability to test specific interrelations among these symptoms, while accounting for all other symptom relations, and because of the ability to identify potential bi-directional relationships. Further, network analysis can test if these pathways among symptoms are similar between AN and A-AN. Despite similar diagnostic criteria, few studies have distinguished between AN and A-AN or compared clinical co-morbidity or clinical targets between diagnoses. Network analysis is a data-driven method which pinpoints the central maintaining symptoms (i.e., high centrality symptoms), as well as specific bridge comorbidity symptoms or illness pathways among diagnostic clusters (i.e., anxiety, worry, sleep problems, and EDs symptoms in AN). Pinpointing pathways that bridge between specific symptom clusters (i.e., psychiatric conditions) may provide specific treatment targets to intervene and disrupt illness pathways between AN and co-morbid clusters, such as sleep and anxiety. Importantly, while network analysis has often been utilized as a method to explore central maintaining symptoms and interconnectedness between symptoms, recent research has suggested that network analysis should also be applied to a-priori theoretical models (see e.g., Robinaugh et al., 2020), which select specific variables based on psychological theory (see e.g., De Beurs et al., 2019). In other words, there is a strong need to use network analysis beyond only as an exploratory method.
Thus, the current study used network analysis to test our theoretical conceptual model to pinpoint central features and bridge pathways among specific types of sleep disturbance, anxiety, worry, and ED symptoms in AN and A-AN, and to test if our conceptual model matches the data. We hypothesized that there would be several sleep disturbance, anxiety and worry, and ED symptoms that would be specifically interconnected in AN. Specifically, see our theoretical conceptual model of AN and A-AN at Fig. 1. Based on this model, we hypothesized that a central symptom would be judgement based on shape and weight, which is a core AN and A-AN diagnostic feature and drives these pathologies and leads to restriction. We additionally hypothesized that restriction would connect to worry and anxiety, which would connect to sleep disturbances, such as loss of energy and feeling tired or fatigued. We predicted fatigue and loss of energy would link both restriction and more worry, which would produce a feedback loop to hypervigilance of fear of gaining weight or becoming fat, leading to restriction, and connect to feeling more tired or fatigued and loss of energy. In sum, we hypothesized that the bridge symptoms would be worry/anxiety, feeling tired or fatigued, and loss of energy. We also hypothesized there would be no differences between AN and A-AN networks.
Section snippets
Participants
Two-hundred-and-sixty-seven participants (n = 253; 94.8% female; n = 4; 1.5% male; n = 10; 3.7% did not list their sex), 14 to 61 years-of-age (M = 26.72; SD = 8.40) with a current diagnosis of AN (n = 162; 60.7%) or atypical AN (A-AN; n = 105; 39.3%) were combined from two non-overlapping studies. Most of the participants were White (n = 231; 86.5%), followed by Hispanic (n = 11, 4.1%); the remainder of the sample identified themselves as Asian (n = 3;1.2%), Black (n = 2, 0.7%), multi-racial (n
Stability and centrality
The network was stable (strength stability = .52). The symptoms with the greatest strength centrality were judgement based on shape (strength [S] = 1.86; significantly stronger than 71.43% of the other symptom estimates [ps < .05]), restriction (S = 1.33; significantly stronger than 42.86% of the other symptom estimates [ps < .05]), and feeling tired or fatigued (S = .91; significantly stronger than 21.43% of the other symptoms estimates [ps < .05]; see Fig. 2).
Bridge symptoms
BEI was stable (BEI
Discussion
Although sleep disturbance, anxiety, and worry are associated with and key complaints of individuals with EDs, it has been unclear how these specific symptoms are interrelated in AN and A-AN. The current study tested a theoretically-based conceptual model by utilizing the data-driven approach of network analysis to pinpoint the maintaining central features and pathways among specific types of sleep disturbance, anxiety, worry, and ED symptoms in a large sample of individuals with AN or A-AN
Limitations and future direction
The limitations in the present study may inform future studies. First, the measure of sleep disturbance may be extended from specific items drawn from multiple measures to an entire sleep disturbance measure. Second, anxiety sensitivity was measured with the ASI, which had a high amount of missing data in our study. Third, anxiety sensitivity is a very specific and important aspect of anxiety, but as interoceptive deficits (difficulty interpreting internal sensations) are common in AN, perhaps
Conclusion
In conclusion, the results of this study suggest that targeting identified specific sleep, anxiety, worry, and ED symptoms could improve and inform novel and effective treatments for AN and A-AN. Importantly, there were no significant differences between AN and A-AN. Future research investigating sleep disturbances in AN and A-AN are needed to further understand how sleep may function among the EDs.
Funding
This research received funding from NIMH R15MH121445 (CAL & CRN), T32 5T32DA007261 (CAL), and NIH 5P20GM103436-18 (CAL & CRN) .
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Declaration of Competing Interest
Authors have no conflicts of interest to declare.
Acknowledgement
Thank you to all of our participants that made this research possible, and for our EAT Lab team.
References (59)
- et al.
Sleep and eating disorders: current research and future directions
Current Opinion in Psychology
(2020) - et al.
A systematic review of sleep disturbance in anxiety and related disorders
J. Anxiety Disord.
(2016) - et al.
Exploring the psychology of suicidal ideation: A theory driven network analysis
Behav. Res. Ther.
(2019) - et al.
Motivation to eat and not to eat–The psycho-biological conflict in anorexia nervosa
Physiol. Behav.
(2019) - et al.
Sleep disturbance and psychiatric disorders
The Lancet Psychiatry
(2020) Eating disorder symptoms and core eating disorder fears decrease during online imaginal exposure therapy for eating disorders.
Journal of Affective Disorders
(2020)- et al.
Obsessions are strongly related to eating disorder symptoms in anorexia nervosa and atypical anorexia nervosa
Eat. Behav.
(2019) - et al.
Examining the relationship between worry and sleep: a daily process approach
Behav. Ther.
(2016) - et al.
Development and validation of the penn state worry questionnaire
Behav. Res. Ther.
(1990) - et al.
Validity of the eating disorder examination questionnaire (EDE-Q) in screening for eating disorders in community samples
Behav. Res. Ther.
(2004)