Objective: To estimate the interobserver reliability (IR) of the minimal diagnostic criteria for the parasomnias provided in the International Classification of Sleep Disorders Revised (ICSD-R).
Methods: Fifty consecutive subjects underwent a structured interview by three doctors based on the ICSD-R minimal criteria for the diagnosis of 13 parasomnias at any time in life. IR was calculated by Kappa statistics and interpreted according to conventional categories.
Results: In the group of Arousal Disorders, IR was "substantial" (kappa 0.74) for confusional arousals, "slight" (kappa 0.36) for sleepwalking, "fair" (kappa -0.02) for sleep terrors. In the group of Wake-Sleep Transition Disorders, IR was "substantial" to "almost perfect", but "moderate" for sleep starts (kappa 0.41). In the group of Parasomnias usually associated with REM Sleep, IR was "substantial" (kappa 0.69) for sleep paralysis, "moderate" (kappa 0.46) for RBD, "fair" (kappa 0.25) for nightmares. In the group of Other Parasomnias, IR was "substantial" to "almost perfect" (kappa between 0.73 and 0.93).
Conclusions: When the clinical diagnosis of parasomnias is based on the ICSD-R: 1) the majority of Arousal Disorders and REM sleep parasomnias showed only a "fair" to "moderate" IR; 2) all of the other parasomnias showed a "substantial" to "almost perfect" IR. Nosological entities with unsatisfactory IR share complex motor phenomenology associated with a mental state difficult for the patient to define. The source of disagreement probably lies in the difficulty in interpreting patients' reports. For these parasomnias IR must be verified and possibly improved with training.