Trajectories of change in body mass index during inpatient treatment for severe anorexia nervosa during adolescence: predictive factors and hospitalization outcomes.

Background: Weight restoration is one of the main goals of treatment for AN. Weight trajectories derive from various elements including baseline personal characteristics and factors linked to the course of treatment. The aim of our research was to identify different BMI trajectories during inpatient treatment, to examine whether patient characteristics were predictive of the nature of these trajectories, and to examine how they affect hospitalization outcomes.

Methods: The study population consisted of 310 female AN inpatients. To analyse the data, we examined trajectories of change in BMI, using a clustering algorithm: k-means for longitudinal data.

Results: We chose a four-trajectory model. The most common was the A trajectory, which we labelled "severe and compliant" (38.71%, N = 120). The second most frequent (28.71%, N = 89) was B trajectory, labelled "the least severe with weight fluctuations before discharge", it is situated above the others over the whole period. Trajectory C, which we labelled "severe, with high dissatisfaction scores and long lengths of stay" included 25.16% (N = 78) of patients. The D trajectory, which we labelled "resistant and non-compliant", was the smallest with only 23 subjects (7.42%), situated below the others. Significant differences were found across trajectories concerning: lifetime, admission and target BMI, satisfaction with target BMI, menarcheal status and the duration of amenorrhea, previous inpatient treatments and parental psychiatric disorders. Factors that differed in the course of treatment were: length of stay, dropout, discharge BMI, changes in target weight, tube feeding and transfers to intensive care unit.

Conclusion: This study is one of the few to examine BMI trajectories during treatment for AN. It shows that different trajectories lead to different outcomes. A better understanding of the underlying clinical profiles associated with trajectories could enable more personalized care and an improved outcome.