REVIEW OF: Bindseil I et al, J Child Adolesc Psychopharmacol 2025;35(8):471–478
STUDY TYPE: Retrospective, matched cohort study

Anorexia nervosa in kids and teens is hard to treat once cognitive rigidity and food avoidance stall family-based therapy. Antipsychotics have been tried, but results are mixed.

Researchers reviewed charts from a comprehensive pediatric eating disorder program and identified 42 patients ages 11–18 who started on aripiprazole for eating disorder cognitions. They were matched 1:2 with 84 controls. Most participants were female with the restricting subtype; the aripiprazole group was somewhat more medically ill at baseline. The medication was typically started midway through treatment.

Patients treated with aripiprazole showed greater improvement in food-avoidant behaviors, with scores improving by about 3.5 points versus less than 1 point in controls on the 42-point, 14-item Ease of Eating Scale. Weight outcomes also favored aripiprazole: Patients gained about 4.8 kg versus 3.1 kg in controls, and 43% reached target weight by discharge compared with 30% of controls. Time to target weight was similar between both groups. Dosing was low: The average starting dose was 1.9 mg/day, with discharge doses around 2.8 mg/day. Side effects were mild, most commonly sedation and dizziness, and over half reported no side effects at all.

CARLAT TAKE
This uncontrolled study isn’t definitive, but it may be clinically useful. In youth with anorexia who are stuck (rigid, distressed, and unable to complete meals), low-dose aripiprazole may loosen cognitive rigidity enough to help therapy and weight restoration move forward. Doses stayed well below typical antipsychotic ranges and were mostly well tolerated. No medication has proven benefit for anorexia nervosa, and D2 blockade carries its own risks in this population. We wouldn’t use this routinely, but for patients who aren’t progressing despite solid family-based treatment, a brief, cautious trial at 1–2 mg/day may be reasonable with thoughtful monitoring.