Recognising that self injurious behaviour in people with profound and multiple disabilities usually has important negative consequences for people concerned, this meta-analysis set out to examine the effectiveness of the treatments currently being offered.
The authors looked at single-case studies which investigated non-aversive, non-intrusive forms of reinforcement, combining these using hierarchical linear models.

They discovered that the average effect of treatment was relatively large and statistically significant. They also found significant variance between studies and participants. They suggest that there was some evidence for a moderator effect of sensory impairment.

They could find no statistically significant moderator effects of medication, motor impairment, setting, age, gender, matching of treatment with behavioural function and contingency.

Self-injurious behavior in people with profound intellectual disabilities: A meta-analysis of single-case studies, Denisa J et al in Research in Developmental Disabilities, 32, 3, 911-923