Does Cognitive Impairment and Agitation in Dementia Influence Intervention Effectiveness? Findings From a Cluster-RCT With the Therapeutic Robot, PARO.

To explore whether severity of cognitive impairment and agitation of older people with dementia predict outcomes in engagement, mood states, and agitation after a 10-week intervention with the robotic seal, PARO. Data from the PARO intervention-arm of a cluster-randomized controlled trial was used, which involved individual, nonfacilitated, 15-minute sessions with PARO 3 afternoons per week for 10 weeks. One hundred thirty-eight residents-aged ≥60 years, with dementia-from 9 long-term care facilities. A series of stepwise multiple linear regressions were conducted. Dependent variables were participants' levels of engagement, mood states, and agitation at week 10 [assessed by video observation and Cohen Mansfield Agitation Inventory-Short Form (CMAI-SF)]. Predictor variables were baseline levels of cognitive impairment [assessed by Rowland Universal Dementia Assessment Scale (RUDAS)] and agitation (CMAI-SF). Five models were produced. The strongest finding was that participants with more severe agitation at baseline had higher levels of agitation at week 10 (R = .82, P < .001). Predictors of positive response were less significant. Low levels of agitation at baseline predicted greater positive behavioral engagement with PARO (R = .054, P = .009) and fewer observed instances of agitation (R = .033, P = .045) at week 10, whereas greater visual engagement was predicted by both lower levels of agitation and cognitive impairment (R = .082, P = .006). Less severe cognitive impairment predicted greater pleasure at week 10 (R = .067, P = .004). Participants with severe agitation had poor response to PARO. Lower levels of agitation and higher cognitive functioning were associated with better responses. In clinical practice, we recommend PARO should be restricted to people with low-moderate severity of agitation. Further research is needed to determine the optimal participant characteristics for response to PARO