Evaluating individual- and sample-level response to treatment for inpatients with eating disorders: Is change clinically significant?

Abstract

Evaluating treatment efficacy solely on whether sample-level change is statistically significant does not indicate whether clinically significant change (CSC) has occurred at the individual-level. We assessed whether change in measures of eating disorder psychopathology was statistically significant at the sample-level and clinically significant at the individual-level for inpatients treated in a hospital-based eating disorder program. Participants (N=143) were consecutive underweight distinct admissions diagnosed with anorexia nervosa or other specified feeding and eating disorder. The Eating Disorder Examination Questionnaire (EDEQ) and Eating Disorder Recovery Self-efficacy Questionnaire (EDRSQ) were assessed at admission and program discharge. CSC was defined as individual score change that was both statistically reliable and shifted from dysfunctional to normative. Mean EDRSQ and EDEQ scores significantly improved with treatment across the sample; effect sizes were moderate to large. Individual-level analyses demonstrated that 85%, 50%, and 20-35% of participants had CSC or statistically reliable change in BMI, eating symptomatology, and body image respectively. One-third of participants showed CSC on BMI and on at least one self-report measure. Individual-level analyses offer more nuanced outcome data that could identify patients at higher risk of relapse who may benefit from adjunctive interventions during or immediately post-discharge. This study examined change in eating pathology for inpatients with eating disorders using sample- and individual-level analyses, including whether change was statistically reliable and clinically significant (scores statistically improved and moved into the healthy range). Only half of patients responded robustly to treatment, which may be related to high relapse rates following discharge. Individual-level analyses provided a detailed view of treatment response and may identify patients at higher relapse risk.