Moderators and mediators of remission in family-based treatment and adolescent focused therapy for anorexia nervosa

Abstract

Few of the limited randomized controlled trails (RCTs) for adolescent anorexia nervosa (AN) have explored the effects of moderators and mediators on outcome. This study aimed to identify treatment moderators and mediators of remission at end of treatment (EOT) and 6- and 12-month follow-up (FU) for adolescents with AN (N = 121) who participated in a multi-center RCT of family-based treatment (FBT) and individual adolescent focused therapy (AFT). Mixed effects modeling were utilized and included all available outcome data at all time points. Remission was defined as ≥ 95% IBW plus within 1 SD of the Eating Disorder Examination (EDE) norms. Eating related obsessionality (Yale-Brown-Cornell Eating Disorder Total Scale) and eating disorder specific psychopathology (EDE-Global) emerged as moderators at EOT. Subjects with higher baseline scores on these measures benefited more from FBT than AFT. AN type emerged as a moderator at FU with binge-eating/purging type responding less well than restricting type. No mediators of treatment outcome were identified. Prior hospitalization, older age and duration of illness were identified as non-specific predictors of outcome. Taken together, these results indicate that patients with more severe eating related psychopathology have better outcomes in a behaviorally targeted family treatment (FBT) than an individually focused approach (AFT).

FIGURE 1. Remission rates over time for individuals with low and high baseline YBC Total Score

Note: The total sample was split based on the median YBC score into (a) 13 or less (N=56) and (b) 14 or more (N=49). In the time axis, EOT = end of treatment (12 months from the baseline), 6mF = follow-up at six months from EOT, and 12mF = follow-up at twelve months from EOT. We used a log transformation using ln(actual time + 1). In mixed effects modeling using three repeated measures of cure status (EOT, 6mF, 12mF), we assumed linear trends over this log-transformed time. Note that no one was remitted at baseline.

FIGURE 2. Baseline YBC as a moderator of treatment effect on change in remission status from baseline to EOT

Note: The total sample was split into low and high baseline YBC groups based on different cutpoints. The results reported in Figure 1 are based on the median split (bottom 50% as the low and top 50% as the high YBC groups). The top and bottom curves show the treatment effects in terms of success rate difference (SRD), where SRD is defined as remission rate under FBT minus remission rate under AFT. The middle dashed curve shows the interaction effect, which is the difference between the SRD of the top curve (high YBC) and the SRD of the bottom curve (low YBC). N=44/61 at 40/60%, N=56/49 at 50/50%, N=65/40 at 60/40%, N=75/30 at 70/30%.

FIGURE 3. Remission rates over time for individuals with low and high baseline EDE Global Score

Note: The total sample was split based on the median EDE global score of (a) 1.41 or less (N=53) and (b) greater than 1.41 (N=52). In the time axis, EOT = end of treatment (12 months from the baseline), 6mF = follow-up at six months from EOT, and 12mF = follow-up at twelve months from EOT. We used a log transformation using ln(actual time + 1). In mixed effects modeling using three repeated measures of remission status (EOT, 6mF, 12mF), we assumed linear trends over this log-transformed time. Note that no one was remitted at baseline.