Behaviours that challenge in people with intellectual disability

Behaviours that challenge (BtC), such as self-injurious behaviour (SIB), aggression and destruction of property (DoP), are common in people with intellectual disability (ID). These can be persistent with significant negative impacts, thus, delivery of early interventions is critical. A meta-analysis was conducted to establish the efficacy of interventions for topographies of BtC (SIB, aggression, DoP, irritability, total BtC). Interventions for the reduction or prevention of BtC in people with ID evaluated utilising a randomised control trial design were included. Intervention, methodological and participant characteristics were examined as potential moderators. Small to moderate intervention effects were identified across topographies of BtC. Study design, pharmacological medication type and duration of intervention were associated with aggression outcomes. IQ scores were associated with irritability outcomes. Critically, no significant differences were found for non-pharmacological versus pharmacological interventions. To inform delivery of preventative interventions, correct identification of individuals at risk of BtC is required. This was the focus of the empirical paper. Prediction models of risk for topographies of BtC (SIB, aggression, DoP, any BtC) were generated utilising known risk markers and machine learning approaches. The models showed fair to good recall and precision. Application of these models to longitudinal data indicated good ability to detect persistence and incidence of behaviour over 12 months. The results of the meta-analysis and empirical paper have important implications that may guide the provision of interventions for BtC in clinical practice. Specifically, non-pharmacological interventions are efficacious in reducing BtC and should be considered as the primary response to BtC given their greater long-term efficacy and lower side effects relative to pharmacological approaches. Further, the delivery of interventions may be informed by prediction models of risk. This would allow services to move towards a proactive rather than reactive healthcare system for individual with ID showing or at risk of showing BtC.