Eating disorders (ED) tend to have a prolonged trajectory (Scmidt et al., 2016; Steinhausen, 2002). Whereas the majority of patients eventually reach either partial or full recovery, 20- 30% develop a chronic course with an illness duration of more than 10 years (Keel & Brown, 2010; Steinhausen & Weber, 2009). Although recovery rates increase with longer duration of follow-up, outcome studies indicate that the reduction of symptoms peaks in the first few years after onset, with a gradual decline in recovery rates (Robinson, 2014). Hence, it is argued that for patients with severe and enduring ED, a shift from cure to care might be a better investment for all parties involved (Hay & Touyz, 2015; Robinson, 2014; Reay et al., 2022; Wonderlich et al., 2012). An increased understanding of both longstanding ED and prognostic factors is needed to guide treatment.
The present naturalistic study follows the long-term course of patients admitted to the eating disorder unit at Modum Bad psychiatric hospital in Norway. The participants were evaluated at admission, discharge, 1-, 2-, 5- and 17-years, using both structured diagnostic interviews and self-report measures to assess ED, personality disorders (PD), and childhood trauma.
Paper I found a significant improvement from the 5- to the 17-year assessment, where the number of patients no longer fulfilling criteria for any ED increased from 35.5% to 50%. This illustrates that change is achievable even for patients with an illness duration of nearly 13 years upon admission. However, when cognitive ED symptoms were included in the recovery criteria, only 29% of the patients were fully recovered, 21% achieved partial recovery and half the sample was classified with no recovery. Most of the change happened early on in the follow-up phase, highlighting the need for early interventions as symptoms seem to consolidate over time. The proportion of patients with comorbid disorders, remained high at both the 5- and 17- year assessments, with a prevalence of more than 70%, however lower for the patients who attained full recovery. Quality of life, self-esteem and interpersonal problems remained unchanged from the 5- to the 17-year assessment.
While paper I examined the prevalence of axis I disorders in the sample of ED patients, paper II focused on the relationship between ED and axis II disorders. PD are defined as long-lasting and reluctant to change, however, our study found a marked decrease in both number of PD diagnoses in the sample, from 80 to 22, and the proportion of patients with these illnesses, from 74% to 24%. The most common PD were avoidant, obsessive-compulsive, borderline, paranoid and dependent PD. The anxious-avoidant PD (Cluster C) remained high throughout the follow-up period. Using multilevel modeling, we found a reciprocal association between ED and PD, where the initial level of either illness corresponded with a similar level of the other across the 17 years. We also found that borderline PD predicted a poor ED outcome. This between-person effect signals that a comprehensive assessment and attention to both PD and ED is advisable in therapy, however, it is unclear whether this holds true on an individual level, as there were no significant within-person effects.
Childhood maltreatment (CM) is a precursor to a variety of psychiatric disorders, ED included. In paper III, we investigated the prevalence of physical abuse, sexual abuse, emotional abuse, emotional neglect, and physical neglect. Our findings revealed that 71% scored above the clinical cut-off for one or more forms of CM. We explored the accumulative effect of additive non-sexual trauma, as well as childhood sexual abuse (CSA), and discovered that CSA was far more frequent among the patients with the poorest ED outcomes.
Furthermore, using growth mixture modeling, four distinct trajectories emerged based on the patients ED symptoms over time; “continuous improvement”, “high and declining”, “consistently low” and “high and increasing”. The level of symptoms at discharge remained stable throughout decades, indicating that group membership could present a prognostic value for ED outcome. The heterogeneity in course and outcome also supports tailored treatment, as patients with high levels of comorbidity as well as exposure to CSA experience long-lasting struggles calling for enhanced treatment efforts. Further research is needed to examine the trajectories in a larger sample.