Evaluation of the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) for sensitivity and specificity in screening for cognitive impairment following stroke: a pilot study

Background: Up to 70% of stroke patients experience cognitive impairment in at least one cognitive domain. Guidelines currently recommend that stroke patients be screened as soon as is reasonably practicable for potential cognitive impairment. For a screening test to be diagnostically valid it needs to demonstrate adequate levels of sensitivity and specificity. Cognitive impairment can be identified globally or as an impairment in a specific cognitive domain. Research into commonly used screening tests for cognitive impairment has failed to identify a test with adequate levels of sensitivity and specificity for cognitive impairment in an acute stroke population. The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) has demonstrated adequate diagnostic validity in differing diagnostic groups and has shown to be appropriate for use with acute stroke patients. However, the RBANS has not yet been evaluated for sensitivity and specificity for cognitive impairment following stroke. Aims: The aim of this pilot study was to establish whether an extended study of the diagnostic validity of the RBANS as a sensitive and specific screening test of post stroke cognitive impairment was justified. The study objectives were to compare the RBANS scores with full neuropsychological test battery overall conclusions for cognitive impairment and non-impairment, to compare the RBANS with measures of impairment in specific cognitive domains, and to identify RBANS optimum cut-off scores for discrimination of cognitive impairment / non-impairment in an acute stroke population. Methods: This study used a cross sectional design. Stroke patients admitted to a large city hospital were considered as potential participants. Patients were excluded if they had language impairment as identified by the Sheffield Screening Test, were aged over 80 years old, had had a previous stroke or neurological impairment, and had hearing or sight difficulties that precluded them from completing cognitive testing. Recruited participants completed the RBANS and a ‘gold standard’ battery of neuropsychological tests. Comparison of the two tests were made to identify levels of sensitivity and specificity on global and domain specific cognitive impairment. Analysis was completed to identify RBANS optimum cut of scores for identifying global and specific cognitive impairment. Results: 40 participants were recruited. The RBANS demonstrated poor levels of sensitivity (52%) and good levels of specificity (100%) for global cognitive impairment when using the manual recommended cut off scores. Receiver Operating Characteristic curve analysis identified an optimum cut off score for RBANS Total scale of 102.5 that provided excellent sensitivity (100%) and adequate specificity (83%), and index scores that showed adequate levels of sensitivity and specificity to domain specific cognitive impairment with the exception of Attention. Discussion: It was tentatively concluded that the RBANS demonstrated acceptable diagnostic validity, though problems were highlighted with the Attention index and the use of a test of visual memory within the full battery that placed a heavy burden on motor skills. Recommendations were made for potential improvements to the study design and procedure, and it was suggested that further research into the evaluation of the RBANS as a sensitive and specific screening test of post stroke cognitive impairment justified and potentially feasible