[Interobserver variation in the diagnosis of stroke].

Background: The variability in the diagnostic observations could be a problem both in the clinical and in the epidemiological field. Scarce data exist in the literature on this phenomenon, relatives to the field of the cerebrovascular illness, in their nosologic, semiologic and etiologic aspects. We pretend to study the interobserver variability in the diagnosis of the stroke, classifying it as absentee, present or possible in front of a given case.
Methods: 1. A questionnaire of 99 clinical cases has been elaborated, in order to be subjected to the evaluation of 10 neurologists (3 "junior", residents, 3 "senior" residents, and 4 "staff"); 2. The categories are: a) not stroke either transient ischaemic attack; b) probable stroke or transient ischaemic attack; c) transient ischaemic attack and d) stroke. 3.
Statistical Analysis: by means of the statistic kappa of Fleiss (kappa) for several observers and several categories, determination of the standard error and 95% confidence intervals.
Results: In the diagnosis of the cerebrovascular illness: a) among all the observers: kappa = 0.49 (0.44-0.50); b) among staff: kappa = 0.51 (0.46-0.56); c) among residents: kappa = 0.47 (0.44-0.50). Global data in transient ischaemich attack diagnosis: kappa = 0.52 (0.49-0.55); in the diagnosis of the stroke: kappa = 0.57 (0.54-0.60); in the exclusion of stroke: kappa = 0.64 (0.62-0.67).
Conclusions: With Fleiss' criteria, an acceptable to good agreement exists in the diagnosis of the stroke in base to the mere description of the clinical picture. This interobserver agreement is better if it is tried to exclude the stroke, and it could increase with the training of the neurologist.