0322 Measuring sleep-wake state discrepancy in a heterogeneous sample of cancer survivors with insomnia symptoms

Abstract

Abstract Introduction Sleep-wake state discrepancy is a common phenomenon identified among people with insomnia where greater sleep difficulties are self-reported in comparison with estimates obtained from objective assessment. This study provides the investigation into the sleep-wake state discrepancy and correlation between sleep diary (subjective) and actigraphy-derived (objective) sleep measures. Methods Participants included 136 cancer survivors with insomnia symptoms (M age = 63.8 ± 10.0; 55.9% female; 87.5% White) from baseline data in an ongoing clinical trial. Demographics, Insomnia Severity Index (ISI), 7-consecutive days of sleep diary and actigraphy data were obtained. Sleep measures included time in bed (TIB), total sleep time (TST), sleep onset latency (SOL), wake after sleep onset (WASO), and sleep efficiency (SE%). Mean bias was defined as the discrepancy between sleep diary and actigraphy-derived sleep measures. The agreement between sleep diary and actigraphy-derived sleep measures were graphically assessed using the Bland-Altman plot. Using the mixed linear model approach, the estimated bias and 95% limits of agreement (LOA) were computed. Further, the Pearson correlation coefficient and concordance correlation coefficient (CCC), computed via maximum likelihood methods, were obtained. Results Self-reported TST and SE were shorter than derived by actigraphy (TST: 6.8 min. [95%CI: -18.7, 5.13]; and SE%: 0.7% [95%CI: -3.0, 2.0], respectively). Self-reported TIB, SOL, and WASO were longer than derived by actigraphy (TIB: 8.6 min. [95%CI: 3.7, 13.5]; SOL: 14.8 min. [95%CI: 9.4, 20.2]; and WASO: 20.7 min. [95%CI: 9.4, 20.2], respectively). Moderate to high agreement and correlation were found between the sleep diary and actigraphy-derived TIB (CCC=0.78; r=0.73) and TST (CCC=0.58; r=0.51). In contrast, SOL (CCC=0.48; r=0.35), WASO (CCC=0.36; r=0.18), and SE% (CCC=0.39; r=0.22) showed only fair or poor agreement and correlation. Calculated Bland-Altman LOA between sleep diary and actigraphy derived measures were as follows: TIB (95%LOA: -121.5, 138.7), TST (95%LOA: -197.9, 184.3), SOL (95%LOA: -82.5, 112.1), WASO (95%LOA: -123.5, 164.8), and SE% (95%LOA: -0.37, 0.36). Conclusion Among a heterogeneous sample of cancer survivors with insomnia symptoms, average self-reported sleep duration and efficiency were shorter and self-reported TIB, SOL, and WASO were longer than objectively measured sleep measures. Agreement between two methods varied across different measures. Support (if any) NIH/NINR R01NR018215 (Dean), ClinicalTrials-NCT03810365