Study objective: Recent efforts have been made to develop quantitative frequency, duration, and severity criteria for insomnia. The current study was conducted to test a range of frequency and severity criteria sets for discriminating primary insomnia sufferers from normal sleepers.
Participants: Seventy-two adults with primary insomnia and 88 age-matched normal sleepers.
Methods: Participants completed 14 consecutive nights of sleep logs to monitor their home sleep patterns. Receiver-operator characteristic curve analyses were used to compare a range of severity and frequency criteria sets for discriminating the insomnia and normal-sleeper groups. In addition, sensitivity and specificity tests were conducted for a range of wake-time severity cutoffs based on 2-week mean sleep-log data.
Results: Receiver-operator characteristic curve analyses showed that no 1 combination of severity and frequency criteria maximized sensitivity and specificity. Rather, the optimal frequency cutoff decreased as the severity criterion increased. Analyses of mean sleep-log data showed that an average sleep-onset latency or middle-of-the-night wake time (ie, time awake between sleep onset and final morning awakening) cutoff of 20 minutes or longer over 2 weeks of sleep-log monitoring appeared to best maximize sensitivity (94.4%) and specificity (79.6%) for insomnia classification.
Conclusions: The optimal quantitative insomnia criteria found herein differ from those previously proposed. Nonetheless, results suggest that quantitative criteria derived from sleep-log data may be useful for classification of primary insomnia.