254 - DXA Scans and Adolescent Restrictive Eating Disorders: Whose Bone Density is Worse

Publication Number: 3245.254

Eva M. Goren, University of Michigan, Wilmette, IL, United States; Lise DeShea, University of Oklahoma College of Medicine, Oklahoma City, OK, United States; Stephanie Ferrin, Oklahoma Childrens Hospital at OU Health, Oklahoma City, OK, United States; Amy B. Middleman, UH Rainbow Babies & Children's Hospital, Cleveland, OH, United States

Background: Dual-energy X-ray absorptiometry (DXA) has been recommended for adolescents with anorexia nervosa (AN) for >1 year or females who have had amenorrhea for >6 months. Less is known about bone mineral density (BMD) in patients with avoidant/restrictive food intake disorder (ARFID).

Objective: We would like to better understand the bone density risks associated with AN versus ARFID.

Design/Methods: Using a database approved by a university IRB, we gathered data on age, natal sex, eating disorder (ED) diagnosis, illness duration, menstrual status, %mBMI, and other health information from 70 adolescents who completed a DXA scan as part of their disordered eating assessment. Associations between predictors and DXA z scores were analyzed using generalized linear models with a Gaussian link for continuous outcomes and a binomial link for boolean outcomes.

Results: The 37 patients with AN and 33 patients with ARFID had a mean age of 16 years (SD 2.4), %mBMI 74.8% (SD 11.3); there were no significant differences between groups in age, %mBMI, race, ethnicity. Those with ARFID were more likely to be male (39%) than those with AN (14%), (OR=4.2, 95% CI 1.35, 14.63). Compared with the AN group, patients with ARFID had significantly lower lumbar DXA z scores (OR=0.58, 95% CI 0.36, 0.96), and total-body-less-head (TBLH) DXA z-scores (OR=0.35, 95% CI 0.21, 0.56) (Figure 1). Limiting the analyses to those who had achieved menarche, mean days from last menstrual period (LMP) for those with AN was 195 days (SD 171.9) vs 20 days (SD 18.2) for those with ARFID (p < 0.001). Number of days since LMP was unrelated to DXA z score. The final model included diagnosis, illness duration, and natal sex (Figure 2). Only illness duration emerged as a significant predictor of lumbar z scores; those with illness duration of >3 y had lower lumbar z-scores than patients with duration < 6 mo, OR=0.40 (95% CI 0.18, 0.91). Diagnosis (ARFID vs AN OR=0.37 (95% CI 0.21, 0.66)) and longer illness duration (>3 y vs < 6 mo, OR=0.39 (95% CI 0.17, 0.89)) were significant predictors of TBLH z scores.

Conclusion(s): These results reveal a lower BMD in patients with ARFID when compared to those with AN with a similar age and %mBMI; illness duration of >3 y was also associated with poorer bone density. Number of days since last menses was not associated with DXA results. Small sample size is a limitation to this study. A larger study is needed to further validate these results; however, providers treating patients with ARFID should have a low threshold to order DXA scans for their patients.

Figure 1
Figure 1 shows box plots comparing illness type to Lumbar DXA z score and TBLH DXA z score showing both ARFID mean z-scores were lower than AN mean z scores.

Figure 2
Duration of illness by sex on the x-axis, DXA z scores on the Y-axis with mean DXA scores shown as triangles. Mean DXA scores are lower with longer duration of illness; inferences pertaining to differences in outcomes by natal sex must be made with caution due to limited sample size for males.