Dropout from anorexia nervosa inpatient treatment programs is frequent and is linked to a poorer outcome. This study aimed to identify predictive factors for dropout among anorexia nervosa inpatients. Between 1988 and 2004, 601 consecutive female inpatients with anorexia, restrictive (AN-R) or binge/purging (AN-B/P) subtype (Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV)), were assessed at admission (clinical, socio-demographic, and psychological data). A stepwise logistic model was developed. Dropout rates were respectively 50.0% and 56.2% for AN-R and AN-B/P. Seven predictive factors were identified in multivariate analysis: having one or more children, low desired body mass index (BMI), a low minimum BMI, high scores on the SCL-90 paranoid ideation and the Morgan and Russell eating behavior subscales, and low educational status. Early dropouts had a particular profile: lower desired BMI, higher score on SCL90 paranoid subscale, and more impulsive behaviors (alcohol use, suicide attempts). Dropout appeared as a multifactorial event. In clinical practice, certain factors could serve as warning messages reflecting the severity of the illness (high EDI score and low minimum BMI); while others could be targeted before hospitalization (having at least one child and low desired BMI).
Anorexia nervosa (AN) patients are difficult to treat (Halmi et al., 2005 and Vandereycken & Van Humbeeck, 2008) and frequently drop out of treatment (Halmi et al., 2005). Inpatient treatment for AN is indeed known for its high rates of dropout (20% to 51%) (Vandereycken & Pierloot, 1983, Kahn & Pike, 2001, Surgenor et al., 2004, Woodside et al., 2004, Zeeck et al., 2005 and Masson et al., 2007). Patients who drop out from inpatient care have an increased risk of relapse within the first year, exhibit more numerous eating disorder symptoms at follow-up (Baran et al., 1995), and have a more chronic and severe course of illness (Strober et al., 1997). We recently published a review of the literature on dropout among AN inpatients (Wallier et al., 2008). The seven studies selected were difficult to compare because of four methodological issues. There were differences and variability in sample composition, treatment protocols, definition of dropout, and factors explored as predictors of dropout (for details see Wallier et al. (2008)). Because of the comparatively small samples (from 77 to 268 inpatients) and the diversity of the factors tested, there is still considerable uncertainty regarding factors predicting dropout from AN inpatient treatment programs. Despite these limitations, weight on admission, AN subtype, and the absence of depression appear to be related to dropout from inpatient care. In addition, further research is needed including a larger sample, as it is likely that a number of factors explain dropout from inpatient care.
In response to these elements, the aim of the present study was to more fully explore predictors of premature discharge from inpatient treatment for anorexia nervosa, in order to identify clinical signs that could alert the clinician to the risk for dropout, and/or possible targets for treatment.
