Subjective-Device-Estimated Sleep Discrepancies in Late-Life Depression: Agreement and Associated Factors

This study examined the agreement between subjective sleep assessments and device-estimated sleep parameters in patients with late-life depression, and explored potential factors associated with the discrepancies between the two measures.

This cross-sectional study enrolled 94 patients with late-life depression. Device-estimated sleep parameters were obtained via portable sleep respiratory monitoring. Participants also reported the previous night's total sleep time (TST), sleeponset latency (SOL), and sleep efficiency (SE) using a self-designed form. Agreement between subjective and device-estimated measures was evaluated with Bland-Altman analysis. Multivariate regression was conducted to identify factors associated with subjective-device-estimated sleep discrepancies.

Compared with device-estimated measures, subjective reports differed significantly across all three sleep parameters (Wilcoxon signed-rank test: TST, Z = -6.94, p < 0.001; SOL, Z = -2.98, p = 0.003; SE, Z = -5.02, p < 0.001). These results indicate a pronounced subjective-device-estimated sleep discrepancy in late-life depression, with subjective underestimation of TST and SE and overestimation of SOL. Greater depressive symptom severity was associated with greater subjective underestimation of TST (p = 0.037), whereas later age at first onset was associated with greater subjective underestimation of SE (p = 0.013). The association between age at first onset and subjective overestimation of SOL observed in the main analysis (p = 0.047) was no longer significant after additional inclusion of sex in the model (p = 0.054).

Patients with late-life depression exhibited subjective-device-estimated discrepancies in sleep parameters. Heavier depressive symptom burden was linked to greater underestimation of total sleep time, while later age at first onset was more closely related to underestimation of sleep efficiency. A later age at first onset showed a trend toward an association with overestimation of sleep onset latency, but this finding still requires further validation. In clinical practice, combining device-estimated sleep monitoring with subjective sleep assessment should be considered in order to obtain a more comprehensive understanding of patients' sleep status.