Assessing early to late stage dementia: the TSI and BANS-S scales in the nursing-home.

Background The traditional assessment tools for dementia, such as the MMSE, have only limited ability to follow subjects with severe dementia because they show a floor effect. Specific observational and performance-based instruments were recently developed. Objectives To directly compare an observational scale to a performance-based instrument in moderate to severe dementia. Methods We compared a slightly modified version of the performance-based Test for Severe Impairment (mTSI) to the observer-based Bedford Alzheimer Nursing Severity Scale (BANS-S). Both scales were administered, together with the Mini-Mental State Examination (MMSE) and the Clinical Dementia Rating scale (CDR), to a nursing-home sample of 130 women suffering from different types of dementia (CDR range:1–4; MMSE range:0–18), defined according to DSM-IV criteria. Mean age was 86.9 ± 7.3 years and mean education was 2.7 ± 1.1 years. Results The BANS-S could be applied to all patients, the mTSI to 87 subjects (66.9%). Mean mTSI score decreased progressively from CDR stage 2 to CDR stage 4, whereas no difference was detectable between CDR stages 1 and 2. By contrast, the BANS-S was not significantly different for CDR stages 1 to 3, and the mean BANS-S score worsened only in CDR stage 4. Results were similar for AD and non AD dementia. Both scales were independent from age and education and their test–retest and inter-rater reliabilities were satisfactory. Conclusion The mTSI looks promising in the moderate-to-severe range, whereas the BANS-S seems more useful in the very late stage of dementia. However, neither scale was optimal and additional instruments should be tested in future studies. Copyright © 2005 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]