Psychometric properties of the Compulsive Exercise Test in a large sample of female adolescent and adult inpatients with anorexia nervosa and bulimia nervosa.

OBJECTIVE The Compulsive Exercise Test (CET) was developed to assess compulsive exercise in patients with eating disorders (EDs), but originally validated in a nonclinical sample, and psychometric properties were only investigated in small clinical samples. Therefore, the aim of this study was to examine its psychometric properties in a large clinical sample of adolescent and adult inpatients with anorexia nervosa and bulimia nervosa. METHOD A sample of 2,535 German female inpatients with EDs completed the CET and other instruments at admission and discharge. Factor structure (confirmatory [CFA] and exploratory factor analyses [EFA]), internal consistency and construct validity, measurement invariance across age and diagnostic groups, group comparisons of means, as well as sensitivity to change during treatment were assessed. RESULTS The CET showed high internal consistency, very good construct validity, and sensitivity to change. CFA indicated a better fit of four-factor and three-factor solutions compared to the original five-factor model. However, subsequent EFA identified an optimum for a five-factor model. Only three subscales were satisfactorily invariant to measurement, but not the CET total score. Only small differences in scores between patient groups were observed. DISCUSSION Results support internal consistency, construct validity, and sensitivity to change, whereas factor structure remains inconclusive, questioning the theoretical basis of the CET. There is limited support for using the lack of enjoyment subscale, and only moderate support for using the rigidity subscale in patients with EDs. It is recommended to further explore and/or revise the original CET, including investigation in other samples, for example, male samples.