The role of negative affect regulation in the maintenance of restriction in anorexia nervosa spectrum disorders.

Restrictive eating disorders, including typical and atypical anorexia (i.e., AN and AAN, or termed here AN spectrum disorders [ANSD]), are notoriously difficult to treat (Abbate-Daga et al., 2013; Bulik, 2014). Only about one-third of patients with AN achieve remission following treatment (Le Grange et al., 2016; Lock & Le Grange, 2019), and even among those who do recover, relapse rates range from 9-52% (Khalsa et al., 2017). Although few studies have directly compared course and outcome between AN and AAN, emerging evidence indicates that AN and AAN patients are re-hospitalized at similar rates following inpatient treatment (Shachar-Lavie et al., 2022), underscoring the recalcitrant nature of both disorders. Restrictive eating, or reduced caloric consumption related to weight/shape concerns (Haynos & Fruzzetti, 2015), is the core behavioral symptom of ANSD (Birgegård et al., 2023; Sysko et al., 2005). Restrictive eating is also among the most persistent symptoms; even when treatment leads improvements in body weight and other psychological symptoms, restrictive behaviors often remain (Mayer et al., 2012). To develop more effective treatments for ANSD, it will be critical to clarify (and ultimately, disrupt) the functional processes involved in the maintenance of restrictive eating. Affect regulation models of psychopathology posit that a key function of disordered behaviors is to modulate emotional experiences (Ford & Gross, 2024). In the absence of more adaptive emotion regulation strategies, individuals with eating disorders may rely on disordered behaviors to either increase positive affect (e.g., provide feelings of self-assurance) or decrease negative affect (e.g., reduce feelings of shame or guilt). Consequently, disordered behaviors are maintained through feedback loops of positive and/or negative reinforcement. Whereas some disordered eating behaviors (e.g., binge eating, purging) have been relatively well-established as maladaptive affect regulation strategies (Haedt-Matt & Keel, 2015; Schaefer et al., 2020), the role of restrictive eating in regulating affect is less clear. Clarifying the relationship between restrictive eating and affect necessitates the use of methods which can capture the effects of dietary restriction on emotion in real-time, and vice versa. Ecological momentary assessment (EMA) is a particularly useful tool for examining how behaviors and emotional states occur in an individual’s natural environment. Prior EMA studies have detected a clear link between restriction and affect (Engel et al., 2013; Fitzsimmons‐Craft et al., 2015), but elucidating the functional role of restriction requires establishing whether restriction reliably produces affective shifts. To this end, Haynos and colleagues (2017) tested the extent to which affective changes preceded and followed engagement in restrictive eating in a sample of individuals with AN and sub-threshold AN. Their results suggested that restriction decreased guilt and increased self-assurance but did not exert effects on higher-order domains of positive and negative affect. Although this study shed additional light on the complex relation between affect and restriction, several gaps nevertheless remain in the existing literature. First, although prior studies have leveraged multilevel modeling techniques capable of disaggregating within- and between-person effects, these approaches are typically used to examine static relationships between variables, rather than capturing truly dynamic processes. To advance our understanding of the complex interplay between affect and restriction, it is necessary to consider the dynamics of the system; that is, to take into account lagged relations between variables over time. This facilitates an enhanced understanding of how processes unfold over time at the within-person level (Hamaker et al., 2018). Second, traditional multilevel modeling approaches do not allow for the estimation of bidirectional causal effects; however, it is likely that changes in affect lead to restriction, and vice versa. A multivariate framework is necessary to estimate more than one potential causal direction at a time (Falkenström et al., 2022). Third, the preponderance of evidence in this domain is drawn from studies conducted in “underweight” (i.e., BMI < 18.5) individuals, therefore, the extent to which results would generalize to individuals with significant restriction who are not currently underweight, including those across the AN spectrum, remains unclear. This knowledge gap is critical to address, as individuals with AAN may demonstrate even higher levels of dietary restraint than those with AN, for example (Garber et al., 2019; Sawyer et al., 2016). Given these considerations, the primary aim of the present study is to evaluate the dynamic temporal interplay between negative affect and restrictive eating in a sample of individuals with ANSD. To further extend upon prior findings, we will also explore whether the function of restriction, with regards to affect regulation, differs between individuals with diagnoses of AN and AAN. To achieve these aims, we will leverage a dynamic structural equation modeling (DSEM) approach, which is an extension of conventional multilevel modeling that facilitates the identification of temporal dependencies and feedback loops between variables of interest using intensive longitudinal data.