Participant Characteristics
The study participants consisted of 166 patients (163 female and three male patients) who met the DSM-IV criteria for anorexia nervosa. At admission, they had a mean age of 27.1 years (SD=9.0) and a mean body mass index of 14.9 (SD=1.8). The mean duration of illness was 6.7 years (SD=7.6). The average length of stay in treatment was 10.6 weeks (SD=6.3), and the mean weight gain in treatment was 10.1 kilograms (SD=6.9). Seventy-four percent (N=123) were single, 19% (N=31) were married or in common-law relationships, and 7% (N=12) were separated or divorced. Sixteen percent (N=26) lived alone before their hospital admission, and the remaining 84% (N=140) lived with parents/relatives, a partner/spouse, or friends. The majority of participants were unemployed (41%, N=68), 27% (N=45) were employed, and 32% (N=53) were students. Of the participants who provided information about ethnic background, 90.4% were Caucasian, 4.2% were Asian, 2.4% were East Indian, and the remaining 3.0% were Middle Eastern, African Canadian, or Hispanic. Ninety-one participants (55%) met the DSM-IV criteria for the binge eating/purging type of anorexia nervosa according to the Eating Disorder Examination interview; the remaining 75 patients (45%) met criteria for the restricting type of anorexia nervosa.
Predictors of Premature Termination
The first significant predictor of premature termination of treatment was anorexia nervosa type (Mantel-Cox χ
2=24.32, df=1, p≤0.0001).
Figure 3 shows the survival curves for patients with the binge eating/purging type and those with the restricting type of anorexia nervosa. (Patients completing the program are censored at the time at which they complete the program.) The overall survival rate was significantly lower for the patients with the binge eating/purging type (28%), compared with the patients with the restricting type (65%).
Figure 4 shows the hazard profiles of the patients with the binge eating/purging type and those with the restricting type. Almost one-half (45%) of the patients with the binge eating/purging type experienced premature termination of treatment by week 7, compared to only 15% of the patients with the restricting type. In addition, the mean time to premature termination was significantly earlier for the patients with the binge eating/purging type (11.5 weeks, 95% CI=9.5–13.3), compared with the patients with the restricting type (20.6 weeks, 95% CI=18.4–22.8).
The second significant predictor of premature termination was body mass index at admission (hazard ratio=1.15, 95% CI=1.02–1.30, p<0.03). Patients with higher body mass indexes at admission were more likely to terminate treatment prematurely. Third, Beck Depression Inventory scores at admission were significantly associated with the probability of premature termination (hazard ratio=1.04, 95% CI=1.01–1.08, p<0.02), with higher levels of depression at baseline predicting lower survival rates.
As for the severity of eating disorder psychopathology, admission Eating Disorder Examination subscale scores for weight concern (hazard ratio=2.15, 95% CI=1.25–3.70, p=0.006) and restraint (hazard ratio=0.67, 95% CI=0.51–0.88, p=0.004) were significant predictors of premature termination of treatment. Higher weight concern scores and lower restraint scores were associated with a higher likelihood of premature termination. Finally, higher Eating Disorder Inventory maturity fears subscale scores (hazard ratio=1.05, 95% CI=1.00–1.11, p<0.04) significantly predicted premature termination.
On the basis of the hypothesis that some of these results may be accounted for by baseline differences between patients with the restricting type and patients with the binge eating/purging type, the predictor analyses were repeated with adjustment for anorexia nervosa type by forcing this variable into the Cox regression equation in the first step. After adjustment for type, only higher weight concern scores (hazard ratio=2.15, 95% CI=1.25–3.70, p=0.006), lower restraint scores (hazard ratio=0.67, 95% CI=0.51–0.88, p=0.004), and higher maturity fears scores (hazard ratio=1.07, 95% CI=1.02–1.12, p<0.02) were significantly associated with the probability of premature termination of treatment. Body mass index (p=0.37) and depression (p=0.09) no longer had a significant effect on the cumulative probability of premature termination.
Several other variables showed no significant effect on the cumulative probability of premature termination of treatment, including marital status (p=0.25), employment status (p=0.40), living situation (p=0.71), age at onset (p=0.12), duration of illness (p=0.60), age at admission (p=0.23), maximum (p=0.55) and minimum (p=0.63) previous weight, previous use of specialized treatment for an eating disorder (p=0.43), and the frequency of binge eating (p=0.10) and purging (p=0.61) behavior in the 3 months before admission.
A post hoc decision was made to use a series of t tests to compare patients with the binge eating/purging type who had early premature termination of treatment (i.e., at 7 weeks of treatment or before) with those who had later premature termination (i.e., after 7 weeks). One result of this analysis approached significance: the patients with early premature termination had a monthly binge frequency of 42.8 episodes (SD=49.6), compared to 18.3 episodes (SD=21.7) for the patients with later premature termination (t=1.92, df=32, p=0.07).