Pain assessment in profound cognitive impaired children using the Checklist Pain Behavior; is item reduction valid?

There are both commonalities and idiosyncratic features in the reaction of pain children with profound cognitive impairment (CI), and that there is no evidence to suggest that idiosyncratic behavior is more characteristic of this population than of any other population. The main objective of this study was to identify whether the 23-item version of the Checklist Pain Behavior could be reduced to 10 items. Previous research demonstrated that only these 10 items discriminated between absence and presence of pain. Second, we wanted to explore the underlying structure of these 10 selected items including its performance. Data of 477 observations in 73 children were used. All these children were video-taped while they were admitted to the Sophia Children's Hospital for surgery, twice before and five times after surgery. These video-tapes were scored by an independent observer. A visual analogue scale (VAS) by a researcher was used to assess the presence of pain. We tested whether the underlying structure was unidimensional, and whether it had differential qualities between pain and no pain, and to which degree. Using a modern psychometric method, i.e., Mokken scaling model, we unraveled the interdependency of the pain response in CI-children, in that the structure turned out to be unidimensional. In addition, these behaviors could be hierarchically ordered in terms of frequency of occurrences. Finally, these behaviors had to a high degree the potentialities to estimate the likelihood of occurrence of pain.