Naltrexone in treatment of self injurious behavior: a clinical study.

Two adults with profound intellectual disability lived in a state hospital. Both hit themselves hard, every day.

Doctors gave each person a pill every day for several weeks. Sometimes the pill was naltrexone, sometimes it was a fake pill. Staff who handed out the pills did not know which was which. The pill type switched back and forth four times so the team could see if the drug really helped.

Self-injury stayed the same in both people. Rate, force, and daily count did not budge when naltrexone was in the blood.

The fake-pill phases looked identical to the drug phases. Doctors rated the drug as useless for these two clients.

Taras et al. (1993) tested a simple behavioral plan with two similar adults. When staff asked each person to play with toys for longer and longer times while self-injury was ignored, hitting dropped 68-a large share. Same design, same people, opposite result.

Stokes et al. (1980) and Kohler et al. (1985) both used quick, contingent consequences—water mist to the face or brief positive-practice drills—and saw large, fast drops in SIB. Again, same reversal method, clear benefit, unlike the flat line seen with naltrexone.

van der Miesen et al. (2024) pooled 115 single-case studies and found behavioral plans give huge reductions in self-injury (Tau-U = -0.90). The 1987 naltrexone result sits at the very bottom of that effect range, confirming the drug is a weak stand-alone tool.

This study is a red flag: a pill that looks promising on paper can do nothing in real life. Use it as a quick reference when medical teams suggest naltrexone first. Show them these flat data lines, then pivot to plans that work—differential reinforcement, brief response-cost, or contingent exercise. Your session time is better spent on behavioral tactics that have proven, large effects.