Jessica R Lunsford-Avery1, Jack D Edinger2, Andrew D Krystal3. 1. Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, 2608 Erwin Road Suite 300, Durham, NC 27710, USA. Electronic address: Jessica.r.avery@duke.edu. 2. Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, 2608 Erwin Road Suite 300, Durham, NC 27710, USA; Department of Medicine, National Jewish Health, 1400 Jackson St, Denver, CO 80206, USA. Electronic address: EdingerJ@NJHealth.org. 3. Department of Psychiatry and Behavioral Sciences, Duke University School of Medicine, 2608 Erwin Road Suite 300, Durham, NC 27710, USA; Departments of Psychiatry and Neurology, University of California San Francisco School of Medicine, 400 Parnassus Ave, San Francisco, CA 94143, USA. Electronic address: Andrew.Krystal@ucsf.edu.
Abstract
OBJECTIVE: To determine the best of commonly used methods for computing the rate of decline in non-rapid eye movement (NREM) sleep EEG delta power overnight (Delta Decline) in terms of vulnerability to missing data and to evaluate whether this rate is slower in insomnia patients than healthy controls (HC). METHODS: Fifty-one insomnia patients and 53 HC underwent 6 nights of polysomnography. Four methods for estimating Delta Decline were compared (exponential and linear best-fit functions using NREM (1) episode mean, (2) peak, and (3) total delta power and (4) delta power for all available NREM epochs). The best method was applied to compare groups on linear and exponential rates of Delta Decline. RESULTS: Best-fit models using all available NREM epochs were significantly less vulnerable to deviation due to missing data than other methods. Insomnia patients displayed significantly slower linear and exponential Delta Decline than HC. CONCLUSIONS: Computing Delta Decline using all available NREM epochs was the best of the methods studied for minimizing the effects of missing data. Insomnia patients display slower Delta Decline, which is not explained by differences in total sleep time or wake after sleep onset. SIGNIFICANCE: This study supports using all available NREM epochs in Delta Decline computation and suggests a slower rate in insomnia.
OBJECTIVE: To determine the best of commonly used methods for computing the rate of decline in non-rapid eye movement (NREM) sleep EEG delta power overnight (Delta Decline) in terms of vulnerability to missing data and to evaluate whether this rate is slower in insomnia patients than healthy controls (HC). METHODS: Fifty-one insomnia patients and 53 HC underwent 6 nights of polysomnography. Four methods for estimating Delta Decline were compared (exponential and linear best-fit functions using NREM (1) episode mean, (2) peak, and (3) total delta power and (4) delta power for all available NREM epochs). The best method was applied to compare groups on linear and exponential rates of Delta Decline. RESULTS: Best-fit models using all available NREM epochs were significantly less vulnerable to deviation due to missing data than other methods. Insomnia patients displayed significantly slower linear and exponential Delta Decline than HC. CONCLUSIONS: Computing Delta Decline using all available NREM epochs was the best of the methods studied for minimizing the effects of missing data. Insomnia patients display slower Delta Decline, which is not explained by differences in total sleep time or wake after sleep onset. SIGNIFICANCE: This study supports using all available NREM epochs in Delta Decline computation and suggests a slower rate in insomnia.