OBJECTIVE

The aim of the study was to estimate inter-observer and test-retest reliability of the Neuropsychiatric Inventory Nursing Home version (NPI-NH) and the Cohen-Mansfield Agitation Inventory (CMAI), and to establish their Reliable Change Index (RCI). Reliable Change methodology is a practical method for estimating the least change acquired in outcome measures.

METHODS

Nursing home physicians and certified nurses assessed 105 patients with dementia (in five nursing homes) at baseline and after 2 weeks. Spearman rank correlations were calculated and Reliable Change Difference Scores (S(diff) (80)).

RESULTS

NPI-NH inter-observer correlations ranged 0.14-0.70. NPI-NH test-retest correlations ranged 0.23-0.80. CMAI inter-observer correlations ranged -0.10 to 0.72. CMAI test-retest correlations ranged 0.32-1.00 (CMAI total score, ρ=0.89). S(diff) (80) for NPI-NH items ranged 1.7-5.0. A change of 11 points on the NPI-NH total score can be considered a true behavioral change. S(diff) (80) for CMAI total score was 8 and factor analysis based sub-scale scores physically aggressive behavior, physically non-aggressive behavior, and verbally agitated behavior were 3, 6, and 4, respectively.

CONCLUSION

Reliability estimates and RCI for the NPI-NH were modest, seriously challenging its reliability and sensitivity to change over time. NPI-NH may only be useful for monitoring behavioral changes in individual patients with dementia, when symptoms are moderate to severe, or when effect sizes are large. Reliability of the CMAI was good, supporting its usefulness in clinical practice. Poor inter-observer agreement on behavioral observations poses a real challenge in nursing homes. Reliable scales are needed that include unambiguously formulated items.