[Evaluation of disruptive behavioral disorders: how situation specific and temporally accurate are behavior observations?].

The stability of expert behavior ratings was analyzed in an unselected inpatient sample of 49 children admitted for disruptive behavior disorders. Expert behavior ratings were obtained for two highly standardized situations (play behavior on ward, on-task behavior during tests) at two time points (T1: 1 week and T2: 4 weeks after admission). These scores were compared with initial diagnostic assessments (clinical judgement) and objective behavior data (actometer) in the test situation. A comparison of the standardized behavior ratings over time showed that in both settings problem behavior was more pronounced at T2 than at T1. Actometer data indicating a higher activity level at T2 were significantly correlated with behavior ratings. There was only a weak association between expert behavior ratings and an initial diagnosis of conduct disorder; the correlation between an initial diagnosis of attention-deficit hyperactivity disorder and the behavior ratings was statistically significant. In general, T2 behavior ratings corresponded better to the initial clinical judgement. Implications for the use of multiple sources of information and of repeated measurements are discussed.