| Given the subjective nature of pain, obtaining valid pain assessment among dementia patients with limited ability to communicate can be challenging. The American Geriatrics Society (AGS, 2002) and a recent expert consensus statement on pain assessment (Hadjistavropoulos et al., 2007) encourage the incorporation of six different behavioural domains when conducting pain assessments among seniors with dementia. Although several behavioural pain measures have been developed specifically for older adults with dementia, data regarding the measures' psychometric properties are limited. There is also paucity of research investigating these measures under the same conditions using the same assessors and the same sample of participants. Symptoms of delirium can mimic symptoms of pain and it is unclear if this overlap has an effect on observational pain assessment measures. The purpose of this study was to investigate observational pain assessment measure performance under two different pain conditions, as well as performance after the removal of delirium-related items. This was accomplished within the framework of the Communications Model of Pain (Hadjistavropoulos & Craig, 2002) that describes factors that can complicate the expression and the effective decoding of pain messages. Although it was expected that all measures would differentiate between pain states, it was also hypothesized that measures that included all six AGS behavioural domains would outperform other measures even after removal of delirium-related items. Following all appropriate ethics clearances, patients with severe dementia were filmed during potentially painful routine care procedures (i.e., a flu vaccination [n = 100], and routine discomforting movements such those involved in some physiotherapy sessions [n = 109]) as well as at rest. Participants' reactions were carefully coded using six observational pain assessment measures. Consistent with the hypotheses, all six measures were able to differentiate between pain and baseline states in their original format. Reliability and validity varied across measures. Most measures continued to differentiate between pain and baseline states following the removal ' of delirium-related items. Overall, the PACSLAC (Fuchs-Lacelle & Hadjistavropoulos, 2004), the only measure under examination that comprehensively covers all of the six AGS behavioural assessment domains, contributed variance to pain detection over above the variance accounted for by each of the other measures under investigation. These results provide much needed psychometric information regarding newly developed pain measures. Moreover, from a theoretical standpoint, inclusion of all six AGS domains resulted in more consistent and valid results particularly with regard to delirium-related items. Finally, the results suggest that acute phasic pain cues do not differ substantially from cues displayed during movement-related exacerbations of musculoskeletal pain. |
