Inter-rater reliability of the STEP protocol

An inter-rater reliability test of the Stepwise Comparative Status Analysis (STEP) is presented. The STEP is a protocol for the clinical examination of patients with dementia, within the scope of a neuropsychiatric investigation. It combines psychiatric and neurologic bedside examination methods. The analysis is made in three steps where primary, observable symptom variables are successively aggregated via compound variables to the final determination of one of seven possible dominant regional brain syndromes (global, frontal, subcortical, parietal, frontosubcortical, frontoparietal, other), here also called complex variables. In the present study, two senior physicians assessed 50 patients independently and simultaneously. None of the patients was known to both physicians. In 42 patients (84%), the same dominant brain syndrome was determined by the two clinicians. The probability (P value) of this (or better) agreement was calculated at 2.0 x 10(-12). Kappa coefficients were calculated as a measure of assessment agreement regarding the 50 STEP variables. For 20 variables, the coefficient was 0.75 or above, indicating excellent agreement; for 22 variables, the coefficient was below 0.75 and above 0.40, indicating moderate agreement; and for 4 variables, the value was 0.40 or below, indicating poor agreement. Kappa calculations regarding the assessments of four variables were either not possible or were considered inappropriate.