Diagnosing narcolepsy: Validity and reliability of new diagnostic criteria

Background and purpose: No gold standard currently exists for the diagnosis of narcolepsy. Conventional diagnostic criteria are unwieldy and arbitrary. Clearly defined criteria for case selection are needed to compare the results of different studies. Methods: We developed new clinical and neurophysiologic criteria for narcolepsy using four categories, based on the degree of diagnostic certainty. Category A is Definite Narcolepsy, Category B is Probable Narcolepsy (Laboratory Supported; two subgroups B1 and B2) and Category C is Probable Narcolepsy (Clinical). We assessed the charts of 69 new or established patients with narcolepsy seen consecutively in the Mayo Sleep Disorders Center and classified them according to our system. The human leucocyte antigen (HLA) DQB1*0602 status was determined for each patient as an indirect measure of validity. Two investigators independently assessed 30 charts to assess interrater reliability. We assessed additional 337 charts of patients with other hypersomnolence states to assess the specificity of our definitions. Results: Seventy-four percent were positive for HLA DQB1*0602 (including 85% of those with cataplexy). Only 33% of those without cataplexy were HLA DQB1*0602 positive. The two investigators agreed on the classification of 29/30 patients (0.97 reliability). None of the 337 additional patients fulfilled criteria for narcolepsy, and specifically not for cataplexy. Conclusions: We conclude that our new research diagnostic criteria for narcolepsy possess high interrater reliability and appear valid descriptors of the syndrome, based on HLA typing. They may be useful in providing consistent criteria to compare different research studies.