Objective:
We sought to clarify the correlates of severity in women with bulimia nervosa. Method: We studied 114 bulimic women at entry to a randomized clinical trial. We used eight definitions of severity that encompassed the DSM-Ill-R criteria set for bulimia nervosa (i.e., binging, purging, and body dissatisfaction in the current and worst lifetime time frames) and global functioning and symptomatology. We operationalized 28 possible correlates of severity that covered demographic, Axis I psychiatric, personality, family background, symptoms, and neuropsychological domains. We used regression techniques to investigate the associations between the correlates and severity indices. Results: The regression models were all statistically significant and tended to account for a sizeable proportion of variance. In particular, earlier age of onset of bulimia and lower character scores (measured by the Temperament and Character Inventory) correlated with greater severity. Worse mean parental care (from the Parental Bonding Instrument), the absence of any lifetime anxiety disorder, and the presence of any lifetime mood disorder emerged as independent correlates in several models. Conclusions: As these were hypothesis-generating analyses, confirmation or refutation of these results awaits further study. In particular, age of onset of bulimia and personality traits related to character may be important modifiers of the severity of illness in bulimia nervosa. 0 1 996 by John Wiley & Sons, Inc.
The factors related to severity of illness in women with bulimia nervosa are unclear. For example, stratifying bulimic women by a number of potentially relevant factors did not identify subgroups with increased severity of "core" bulimic symptoms (i.e., the fre-