The Sleep Disorders Questionnaire. I: Creation and multivariate structure of SDQ

The development of the Sleep Disorders Questionnaire (SDQ) from the Sleep Questionnaire and Assessment of Wakefulness (SQAW) of Stanford University is described in detail. The extraction of the best question items from the SQAW and their subsequent rewording in the SDQ to insure greater completion rates are described. Two item test-retest reliability studies are reported on 71 controls and on 130 sleep-disorder patients, which confirmed adequate reliability. To create multivariate scoring scales, SDQ was then given in a multicenter study to 519 persons, 435 of whom were sleep-disorder patients with full polysomnography. Canonical Discriminant Function Analysis was employed, which resulted in four clinical-diagnostic scales: SA for sleep apnea, NAR for narcolepsy, PSY for psychiatric sleep disorder and PLM for periodic limb movement disorder. Each was adjusted for male and female responses and transformed to a percentile using the observed distribution of raw scores. Using Receiver Operating Characteristics analysis, cutoff points were determined for each scale to maximize its sensitivity and specificity. Positive and negative predictive values were also calculated. The SA and NAR scales proved to be the most discriminating.