Factors associated with the efficacy of a group intervention for anger in people with intellectual disabilities

There is a growing literature that suggests cognitive behavioural interventions aimed at reducing inappropriately expressed anger by people with intellectual disabilities are effective. However, interventions provide little information about which aspects of the individual or their treatment may be contributing to the overall efficacy of the approach.

A group intervention is compared with a waiting list control. A cross-sectional regressional method was used to explore the relationship between change in a provocation inventory score over the course of an intervention and a number of other variables.

Data was collected from 50 participants who attended a series of groups with the aim of reducing aggressive behaviour. This was compared with a waiting list control of 36. Outcome was measured by a provocation inventory, which was administered pre-, post-group, and at follow-up. Other variables were also measured including, receptive vocabulary, age, gender, experience of primary therapist and whether staff accompanied participants to the group or not.

A 2 x 2 split plot ANOVA identified a statistically significant treatment effect. However, analysis examining clinical significance was more equivocal with only 11 of the individuals in the intervention group showing reliable change as defined by Jacobson and Truax (1991). Regression analysis indicated that 17.5% of the variance in change of provocation inventory scores could be accounted for by the independent variables. Immediately on completion of the group, being accompanied by a member of staff and increased receptive language scores contributed significantly to the variance.

A reduction in provocation inventory scores immediately after the group was more likely to occur if the participant was accompanied by a member of staff who knew them and if they had a higher score on the test of receptive vocabulary. This pattern changed with time. The implications of the results will be discussed and the relatively large amount of variance that is not accounted for will also be considered.