Abstract
Background: Although the behavioral changes with progressive dementia are seen to increasingly depend on the environmental context until late stage disease, measurement has not reflected this interaction in real time to allow examination of antecedents to disruptive behavior.
Objectives: To develop and evaluate the psychometric properties of the Environment-Behavior Interaction Code (EBIC) for use in dementia care research with either sequential or non-sequential analyses of behavioral data.
Method: Development of the computer-based (sequential event format) EBIC provided an observational coding system to classify all behavior and environmental context in real time, so that the probability of social environmental antecedents to resident disruptive behavior could be estimated. A checklist (interval format) EBIC, based on the same behavioral taxonomy, was developed for clinical outcome research. A total of 158 elderly residents of dementia care units were purposively selected from three large long-term care facilities for the psychometric study components.
Results: Psychometric results indicated significant (p < 0.01) known-groups validity for the disruptive behavior construct, which was defined as a composite of aversive, harmful, and high intensity neutral behavior. Interrater agreement for the event format of the EBIC was estimated by average kappa (0.65) and percentage agreement (78%). For the interval format, the mean interrater kappa was 0.80 with 96% agreement. Stability of the event format using a 2-week retest interval ranged from r = .50 (positive behavior) to r = 0.73 (negative behavior, defined as aversive + harmful). On replication with a new sample, stability was higher for positive (r = 0.92) and negative (r = 0.95) components, and for composite scores of nondisruptive (positive + low intensity neutral; r = 0.65) and disruptive (r = 0.85) behavior.
Conclusion: This research provided support for the reliability and validity of both event and interval EBIC formats. Measurement using the EBIC taxonomy has applicability to dementia care research questions that call for either sequential analysis of social interactions or nonsequential analysis of behavioral outcomes in intervention studies.