Symposia
Improved Use of Research Evidence
Bethany Harris, PhD
Graduate Student
State University of New York (SUNY) at Albany
Albany, New York
James F. Boswell, Ph.D.
Associate Professor
University at Albany, SUNY
Albany, New York
Hallie M. Espel-Huynh, Ph.D.
Postdoctoral Research Fellow
Alpert Medical School of Brown University
Providence, Rhode Island
Heather Thompson-Brenner, Ph.D., FAED
Research consultant
The Renfrew Center
Cambridge, Massachusetts
Background: Eating disorders (EDs) have one of the highest mortality rates of all mental disorders (Arcelus et al., 2011; Galmiche et al., 2019). Anorexia Nervosa restricting subtype (AN-R), Anorexia Nervosa binge-purge subtype (AN-BP), and Bulimia Nervosa (BN) are especially common. Hospitalization and residential treatment rates have continued to rise over the last decade (Guarda et al., 2018), however existing outcome measures initially developed for outpatient settings do not fit these inpatient settings properly. Routine outcome monitoring tools can help identify non-responders and increase patient engagement and satisfaction, but most data have been gathered in general outpatient settings. The recently developed Progress Monitoring for Eating Disorders (PMED, Espel-Huynh et al., 2020) assessment bridges this gap by evaluating domains deemed more pertinent to intensive inpatient ED treatment, and is the first transdiagnostic ED measure for such settings. Of concern now is whether the PMED measures the same constructs across ED diagnoses, a question of measurement invariance (MI), which is not often tested when creating new measures (Putnick & Bornstein, 2016). The aim of this study was to determine if the PMED is measuring consistent constructs across different principal EDs.
Methods: Participants (N = 1300) were routinely presenting women treated for EDs in residential settings, aged 13-67 (M=24.4, SD=10.2 years); 81.9% White. Diagnoses were assigned at intake, and patients completed the PMED at admission and multiple follow-up timepoints. The established PMED factor model includes: weight and shape concern; emotion avoidance; ED behaviors/urges; relational connection; and adaptive coping. Data were analyzed with nested structural equation models for MI at baseline.
Results: Chi-square was significant in all models. Change in CFI and RMSEA (Δ >.01 units) from metric to scalar models indicated rejection of the latter. AIC was more than 3 units lower from configural to metric, and more than 3 units higher from metric to scaler, indicating metric again as the best fitting model.
Conclusions: Inpatient ED treatment lacks setting-appropriate measures to properly monitor and improve care. While transdiagnostic assessments can reduce clinician and clinic burden, it’s important to ensure these measures are assessing different subgroups in consistent ways. The PMED demonstrates configural and metric MI across the three Eds, supporting the generalizability of the factor model among key subgroups.