Agreement and correctness in adjusting antiepileptic drug treatment: A weed for rational drug treatment?

Purpose: To study interobserver variation in treatment decisions in a first follow-up contact after initiation of antiepileptic drug (AED) treatment. The results should aid us in the assessment of whether decision support can be of value in this situation. Methods: Data from patient records were used to construct 270 different test cases containing information about the course of the disease after initiation of drug treatment. The cases were presented to five neurologists from different general hospitals who previously agreed about the diagnosis and the initial treatment for these cases. They were asked to write a prescription for each test case. Results: All five neurologists agreed on a treatment decision in 21.9% of the 265 cases available for analysis. Each neurologist made a decision different from the decisions taken by all other neurologists in 14.0-19.6% of the cases. Kappa values for agreement among individual neurologists as well as for agreement between an individual and the group of his peers were low. In 82.6% of the cases, a majority of the neurologists agreed on a treatment decision. Comparing the decisions of individual neurologists with the majority decision reference (219 cases) showed a significant difference in correctness (range, 67.1-82.6%) among the neurologists. Conclusions: The fact that a majority decision could be reached in a considerable number of cases, as well as the variability in adjustment of an initiated drug treatment, leads us to the conclusion that decision support can contribute to a rational adjustment of drug treatment.