Accuracy and reliability of periodic sharp wave complexes in Creutzfeldt-Jakob disease
B. J. Steinhoff, S. Racker, G. Herrendorf, S. Poser, S. Grosche, I. Zerr, H. Kretzschmar and T. Weber
Department of Clinical Neurophysiology, Georg-August University, Gottingen, Germany.
OBJECTIVE: To assess the sensitivity, specificity, and interobserver
reliability of periodic sharp wave complexes in the electroencephalograms
of patients with Creutzfeldt-Jakob disease. DESIGN: Sixty-eight
electroencephalograms in 29 patients who had been suspected of having
Creutzfeldt-Jakob disease were reanalyzed by an investigator who was
unaware of the clinical data. The incidence of periodic sharp wave
complexes in neuropathologically confirmed Creutzfeldt-Jakob disease vs
progressive dementia other than Creutzfeldt-Jakob disease was assessed.
Blinded electroencephalogram analysis was performed by a second
investigator. The interobserver reliability was assessed by the kappa
value. SETTING: University hospital, base of the German National
Creutzfeldt-Jakob Disease Surveillance Study. PATIENTS: Fifteen patients
with neuropathologically confirmed Creutzfeldt-Jakob disease and 14
patients who had been suspected of having Creutzfeldt-Jakob disease because
of rapidly progressive dementia but in whom other dementias were diagnosed
by unblinded investigators based on clinical and electroencephalographic
criteria. MAIN OUTCOME MEASURE: Sensitivity and specificity of periodic
sharp wave complexes assessed by their incidence in Creutzfeldt-Jakob
disease vs other dementias. Interobserver reliability of periodic sharp
wave complexes was expressed by the kappa value. RESULTS: For periodic
sharp wave complexes, blinded electroencephalographic analysis resulted in
a sensitivity and a specificity of 67% and 86%, respectively. Interobserver
reliability was excellent (kappa = 0.95). CONCLUSION: This blinded
electroencephalographic study in Creutzfeldt-Jakob disease confirms the
high diagnostic value of electroencephalography, as previously reported by
open studies.