Development Of An Item Bank For A Computer Adaptive Measure Of Functional Cognition For Stroke

Velozo CA, Duncan PW, Wang J, Donovan N, Heaton S, Kendall D, Kwon S, Singletary F, et al .. Development Of An Item Bank For A Computer Adaptive Measure Of Functional Cognition For Stroke. Poster session presented at: International Society for Pharmacoeconomics and Outcomes Research Annual European Congress; 2009 May 16; Orlando, FL.

The prerequisite for monitoring the effectiveness of pharmaceutical interventions for stroke-related cognitive deficits is the development of reliable and valid cognitive measures. Existing measures of post-stroke cognition: 1) are few in number, 2) lack precision and efficiency for stroke, and 3) fail to reflect cognitive functioning that is related to everyday activities. Applied or functional cognition was defined as the ability to accomplish everyday activities that rely heavily on cognitive abilities. The purpose of this study was to establish an item bank for Computer Adaptive Measure of Functional Cognition for Stroke (CAMFC-S). Methods. 244 functional cognition items across 10 domains (Language, Reading/Writing, Numeric Calculation, Limb Praxis, Visuospatial Function, Social Use of Language, Emotional Function, Attention, Memory, and Executive Function) were administered to 128 individuals with stroke 49 acute (7-21 days post onset) and 79 chronic (83-372 days post onset). Rasch and classical test analyses were performed on128 self-report ratings and 124 caregiver ratings. Neuropsychological/ functional testing was performed on random selection of 62 patients. Results. Confirmatory factor analysis of each domain supported both unidimensionality and hypothesized multidimensional structures for 5 of 10 domains with 1 domain only supporting a multidimensional structure. The All Item measure and separate Domain measures showed excellent internal consistency (high person separation reliability and high Cronbach's alphas). The All Item measure and Domain measures (except Limb Praxis) produced expected item-difficulty hierarchical orders. The All Item measure showed excellent person separation and except for Limb Praxis and Numeric Calculation (patient self-report), Domain measures showed good person separation. While the All Item measure showed no floor or ceiling effects 5 of 10 Domain measures showed ceiling effects. The Domain measures showed fair-moderate correlations with analogous neuropsychological/functional tests. Conclusion: With the exception of Limb Praxis, the findings support developing an All item and Domain measure computer adaptive tests of functional cognition.