Agreement analysis of subjective and objective sleep quality in adolescents with mood disorders

Accurate assessment of sleep quality is crucial for understanding sleep problems and their impact on health. This study analyzed the agreement between subjective sleep assessments and objective sleep monitoring in adolescents with mood disorders, aiming to provide a reliable methodological foundation for related research.

Adolescents with mood disorders were recruited from psychiatric outpatient clinics of three domestic hospitals. A consensus sleep diary and an actigraph (activity tracker) were used to monitor sleep for 14 days. The differences between subjective and objective measurements were compared using the Wilcoxon signed-rank test, while Bland-Altman plots and intraclass correlation coefficients (ICC), described their agreement. Spearman's rank correlation coefficients were used to assess their correlation, and mixed-effects models analyzed factors influencing the differences between subjective and objective measurements.

Significant differences were observed between subjective and objective measures for the number of awakenings after sleep onset (NWAK), waking after sleep onset (WASO), total sleep time (TST), and wake-up time (P<0.001); however, the difference in sleep onset time was not statistically significant (P=0.283). Subjective and objective measurements of sleep onset (ICC=0.821, r=0.838) and wake-up time (ICC=0.821, r=0.836) demonstrated good agreement and correlation; TST showed moderate agreement (ICC=0.640) and correlation (r=0.682); NWAK (ICC=0.210, r=0.276) and WASO (ICC=0.358, r=0.365) exhibited poor and correlation. In the Bland-Altman plots, most data points for sleep onset, wake-up time, and TST were uniformly distributed within the 95% limits of agreement (LoA). The differences between subjective and objective measurements for WASO and NWAK increased with higher average values. However, the 95% LoA of the differences between subjective and objective measurements of all the above indicators exceeded the acceptable ranges of the corresponding indicators, indicating poor agreement. Baseline depression levels were associated with the differences between subjective and objective measurements of NWAK (β=0.034, P<0.05), TST (β=2.617, P<0.01), and sleep onset (β=1.454, P<0.05), while sleep quality scores were associated with the difference in WASO (β=0.051, P<0.01).

Considerable discrepancies remain between subjective sleep diaries and sleep monitoring data from actigraphy (activity tracker). Future research should further investigate factors influencing the discrepancies between subjective and objective measurements and refine measurement methods to obtain more reliable sleep data.