Abstract
BACKGROUND: Few measures have been validated to screen for eating disorders (ED) in youth with chronic pain. We conducted confirmatory (CFA) of two established factor structures of the Eating Attitudes Test-26 (EAT-26) in a sample of youth with chronic pain attending an intensive interdisciplinary pain treatment (IIPT) program and examined the validity of the best-fitting model in predicting ED diagnoses in this sample. METHODS: Participants were 880 adolescents (M age = 16.1, SD = 2.1) consecutively admitted into an IIPT program who completed the EAT-26 upon admission. CFA was conducted and in the case of inadequate fit, EFA was planned to identify alternative models. Factors of the best-fitting model were included in a logistic regression analysis to predict ED diagnoses. RESULTS: The TLIs (0.70; 0.90), RMSEAs (0.09; 0.07) and CFIs (0.73; 0.92) suggested poor fit of one model and adequate of the second model. Goodness of fit indices from EFA (TLI:0.85, RMSEA:0.06) did not outperform the fit of the second CFA. As such, the second model was retained with the exception of one factor. The items loaded onto a 16-item, five factor model: Fear of Getting Fat, Social Pressure to Gain Weight, Eating-Related Control, Eating-Related Guilt and Food Preoccupation. Based on chart review, 19.1% of the participants were diagnosed with an eating disorder. Logistic regression analyses indicated the new 16-item measure and Fear of Getting Fat, significantly predicted an ED diagnosis that did not include avoidant restrictive food intake disorder (ARFID) and Social Pressure to Gain Weight significantly predicted a diagnosis of ARFID. CONCLUSIONS: An alternative 16-item, 5-factor structure of the EAT-26 should be considered in screening for EDs with youth with chronic pain.