Temporal patterns of disruptive vocalization in elderly nursing home residents.

This study used computer-assisted real-time data collection procedures to study temporal patterns of disruptive vocalization (DV) in demented nursing home residents. Residents (N = 68) were observed from 8 a.m. to 8 p.m. and total mean duration of DV within hourly blocks was calculated. Analysis of the temporal distribution of DV exhibited by the group of 68 residents indicated a significant cubic trend. Exploratory cluster analysis uncovered three distinct temporal patterns of DV; although two of the patterns are generally supportive of sundowning. Cluster 1 showed a relatively constant pattern of low rate DV. Clusters 2 and 3 showed cyclic trends. Peaks in DV were observed during noontime and in the evening for Cluster 2. Cluster 3 showed peaks in DV in the early morning hours and late afternoon. MANOVA was employed to examine any differences among clusters in age, cognitive status, and ADL status. Univariate analyses indicated that cognitive status was the only factor that differed significantly among clusters, with members of Cluster 1 (constant low rate DV) manifesting higher cognitive status scores than Clusters 2 and 3. A step-wise discriminant function analysis also showed cognitive status to be the only statistically significant predictor of cluster membership. The authors' note that the scientific literature on agitation is fraught with contradictory findings and assert that inconsistencies may be due to differences across studies in problem definition, method of measurement, and sample size. The results are discussed in relation to these methodological issues and clinical implications of the findings. Copyright 2001 John Wiley & Sons, Ltd.