0765 Comparative Analysis of Sleep Measurement Methods: PSG vs. Actigraphy in Cognitively Normal Elderly Patients

Abstract Introduction Polysomnography (PSG) is the gold standard for sleep measurements, requiring individuals to sleep in a controlled environment. In contrast, actigraphy is more naturalistic, allowing at-home monitoring. While most studies comparing actigraphy to PSG have focused on healthy adults and individuals with sleep disorders, there has been limited research involving a well-characterized, healthy older adult population. Although actigraphy is convenient, it has been noted to differ from PSG in some aspects. Our study aims to assess the correlation between PSG and actigraphy in cognitively healthy older adults participating in NYU research on sleep, aging, and memory. Methods PSG recordings were conducted overnight at the bedside, while wrist-based data were collected on separate nights from individuals' homes. Actigraphs were worn on the non-dominant hand for seven consecutive days, validated by sleep logs. Statistical analyses, including t-tests, chi-square tests, and sensitivity-specificity assessments, were performed. Results Of the 151 subjects, 97 (64.2%) were female, 54 (35.8%) were male, 43 (28.5%) were Black/African American, and 108 (71.5%) were White. Mean age was 66 years (61-71). Mean Epworth Sleepiness Score was 5 (3-8) for women and 6 (3-9) for men. Total sleep time for PSG was 376 (332-418) with a mean difference with actigraphy of -67.4, 26, and 121 minutes (minimum, median, and maximum thresholds respectively, p< 0.0001 for all) Sleep efficiency (%) was higher in the actigraphy group (mean difference 11.2 [SD12.7]) and sleep latency was shorter in the actigraphy group (p< 0.001 for all). Sensitivity for short sleepers(< 5h) was 0.7, increasing to 0.9 in females and 0.86 in Black or African Americans. Specificity corresponded to 0.95, decreasing to 0.87 in males, and to 0.48 in whites, while increasing to 0.96 in black patients. Both sensitivity and specificity for long-sleepers(< 9h) corresponded to 0.5. Positive predictive value for short sleepers was 0.9, and 0.76 for long sleepers. Conclusion Overall, the study highlights the complexity of sleep measurements and the potential impact of conducting PSG and actigraphy on different nights. It underscores the need to consider nightly variability in sleep studies, the different diagnostic methods used, and the influence of sex and ethnicity on sleep patterns. Support (if any)