Night to night variability in subjective and objective sleep quality metrics in patients evaluated for chronic insomnia.


Subjective reports of sleep quality, routinely obtained with a sleep diary, are commonly relied upon to diagnose and prescribe sleeping aids for chronic insomnia. The reliability of these self-reports, the influence of prescription sleeping aids on the self-reports, and the degree to which these subjective measures correlate with electrophysiological measures of sleep is not well understood. This study is an initial attempt to investigate the night-to-night reliability in subjective and objective measures of sleep in patients referred to a sleep medicine physician for evaluation of chronic insomnia.


A retrospective analysis was conducted using 35 multi-night EEG records from patients being evaluated for chronic insomnia.  Sleep diaries provided subjective reports of sleep patterns and use of sleeping aids prior to bed.  At the time of each in-home study, questionnaire responses were obtained to calculate depression (PHQ-9), daytime somnolence (ESS), insomnia (ISI), and anxiety (GAD-7) scores, and assess medication use (Rx).  From the sleep diary data, sleep time used for the comparative analysis was computed by subtracting the clock time from lights off to lights on, plus time to fall asleep and wake after sleep onset.  Objective measures of sleep were obtained with the Sleep Profiler records acquired with the Sleep Profiler (Advanced Brain Monitoring, Carlsbad, CA ) for two-nights in the home.  Worn on the forehead, the device recorded EEG from three frontopolar sensor sites, obtained pulse rate by reflectance methodology, and recorded snoring with  an acoustic  microphone.  After rejecting epochs with poor quality EEG, the scoring software assigned a sleep stage to each 30-second epoch prior to manual review and editing by sleep center staff. Subjects were stratified into groups based on reported use of prescription sleeping aids.  Chi-squared and Bonferroni-corrected t-tests, which assumed equal variances, were used to identify group differences.  Bonferroni-corrected Pearson correlations were used to compare objective and subjective measures of sleep by and across nights and groups.


Group Differences: No group differences were observed in any of the demographic or subjectively reported symptoms (Table 1).  The objective and subjective measures were compared by night to evaluate differences between the medicated and non-medicated groups.  No differences were observed in the medicated vs. non-medicated groups when comparing the Night 1 vs. Night 2 objective and subjective measures.  When comparing the subjective vs. objective measures, the medication group exhibited significantly less  objective WASO on Night 2 compared to subjective WASO (means: 18 vs. 42, p < 0.01).  

Night-to-Night Reliability: In the non-medicated group, a strong bias toward decreased recording time on Night 2 was observed in both the objective and subjective measures.  Conversely, the medicated group showed longer recording times on Night 2.  Those taking sleeping aids showed strong night-to-night agreement in both the objective and subjective  recording times (r = 0.73 and 0.85).  

Significant between-night consistency in subjective sleep times was observed in those  taking sleeping aids (r = 0.71) but not for the non-medicated cohort.  

Objectively measured sleep efficiency was consistent across nights in the non-medicated patients (r = 0.78), but not in the medicated group.  Subjective reports of sleep efficiency  were not consistent.

Objectively measured sleep latency was reliable in both non-medicated and medicated  patients (r = 0.78 and 0.67); subjective reports of sleep

Objective vs. Subjective Agreement: Night-to-night consistency in objective vs. subjective recording times were strong across nights and by group (r > 0.85 for all comparisons).  
Objective vs. subjective sleep times were significantly correlated on both Nights 1 and 2 for the medicated group (p = 0.68 and 0.75), but only on Night 2 for the no-medicated cohort (r = 0.82).  
Objective vs. subjective sleep efficiencies were strongly associated in the non-medicated group and only on Night 2.  Objective vs. subjective sleep latencies were significant for the non-medicated group on both nights  (r = 0.62 and 0.91).  There was essentially no relationship between objective and subjective sleep latencies in the medicated group.  

Subjectively reported recording times were > 15-min longer than the objective measures on both nights for the non-medicated group.  Similar latency differences were only noted on Night 2 in the medicated group.  Subjectively reported sleep latencies were on average >15-min longer than the objective measures in both groups and across both nights.


• All patients reported consistent night-to-night WASO values, however there was no association between subjective and objective WASO measures.

• The time to sleep onset was over-estimated by patients in both groups and on both nights.

• Sleeping aids impacted subjective estimates of sleep onset to the degree that there was no agreement with objectively measured sleep latency.

• For those not taking sleeping aids, the estimation of total sleep time and sleep onset improved on Night 2.

• On Night 2, the medicated group was bias toward longer objective and subjective recording and sleep times, while the non-medicated group was bias toward shorter recording and sleep time times.