Accuracy of bedside diagnosis in stroke.

The clinical diagnosis of the type of acute cerebrovascular diseases is often considered unreliable, although this has not been validated prospectively in representative patients. The accuracy of bedside diagnostics was, therefore, tested in 206 patients consecutively admitted to the Stroke Unit of the Serafimerlasarettet in Stockholm. Bedside diagnosis turned out to be correct in 69%. In 24% the diagnoses were altered after hospital investigation and in the remaining 7% no defined preliminary and/or final diagnosis could be made. When the diagnoses were considered "fairly certain" they were accurate in 87%, compared to 53% when regarded as only "probable". The diagnostic accuracy improved during the period studied. Sensitivity in identifying hemorrhages was much lower (39%) than for cerebral infarctions (83%). It is suggested that new investigational methods should be compared with what can be accomplished with bedside methods alone.