Objective
It is often stated that a higher body mass index (BMI) at the beginning of treatment predicts a better weight outcome at the end of treatment in persons with anorexia nervosa (AN). However, this interpretation is based on the between-persons relationship of BMI at the two measurements, which primarily reflects the fact that the rank-ordering of persons according to their BMI is quite stable over time. In contrast, a lower BMI at baseline is related to a larger BMI change, which primarily reflects the fact that the variance of BMI at the end of treatment is larger than that at baseline. This study aimed to demonstrate these relationships empirically and caution against interpreting BMI at baseline as a predictor of BMI at discharge or BMI change.
Method
Changes of BMI from admission to discharge were analyzed based on 4863 persons with AN (97% female) who received inpatient treatment between 2015 and 2024.
Results
BMI at admission was positively related to BMI at discharge (r = 0.55) but negatively related to BMI change from admission to discharge (r = −0.39).
Discussion
While it is true that a higher BMI at baseline is associated with a higher BMI at the end of treatment, a lower BMI at baseline is actually related to a larger weight gain during treatment. Yet, concluding that the treatment is more effective for patients with low or high BMI at baseline would be incorrect in either case, as the independent and dependent variables are the same variables measured at different time points.