0630 Comparing Accuracy, Sensitivity, and Specificity of Actigraphy to Polysomnography in Cognitively Normal Older Adults with Obstructive Sleep Apnea
Year: 2025
Type: conference-abstract
Abstract: Introduction There is continued research interest in determining the concordance between polysomnography (PSG) and actigraphy, particularly in individuals with Obstructive Sleep Apnea (OSA). We investigated the accuracy, sensitivity, and specificity of actigraphy-measured sleep vs. PSG in a cohort of older adults with and without OSA. Methods PSG for each participant was recorded overnight at bedside. During a separate period of seven consecutive nights, actigraphy was collected for each participant at home using wrist-worn actigraphs on the non-dominant hand. Sleep time was validated with participant-provided sleep logs. Statistical analyses, including t-tests, chi-square tests, and sensitivity-specificity assessments, were performed. Lin’s concordance correlation coefficient (CCC) and reduced major axis regression (RMAR) were utilized as measures of agreement and proportional bias, respectively. Results The study included 151 community-dwelling older adults (ages 47-84) evaluated at the NYU Alzheimer’s Disease Research Center and enrolled in studies investigating sleep health and aging. Of the 151 participants, 108 (71.5%) were white, 43 (28.5%) were Black/African American, and 71 (47.0%) had OSA (defined as apnea hypopnea index [AHI] > 5%). Among mean PSG-measured sleep parameters, Sleep Efficiency was 81.2% (SD 11.81%), Total Sleep Time (TST) 373.4 (76.06) minutes, Sleep Latency 12.4 (22.34) minutes, and Wake Time 72.6 (45.4) minutes. Sensitivity, specificity, and accuracy of actigraphy compared to PSG were low: 0.646, 0.220, and 0.500, respectively. Among participants with OSA, sensitivity was slightly lower (0.634), and specificity slightly higher (0.234), with similar accuracy (0.500). The largest difference among measures was in TST in both OSA and non-OSA participants: actigraphy overestimated TST by 26.23 (100.13) minutes overall, and 16.47 (102.47) minutes in OSA participants. Lin’s CCC was low (< 0.9) for all actigraphy-measured parameters compared to PSG, regardless of OSA status. In addition, RMAR slopes (≠ 1) suggest proportional bias across all measures, while RMAR intercepts (≠ 0) for all measures show fixed bias, regardless of participant OSA status. Conclusion In this study population, there was low accuracy and concordance of actigraphy measured sleep parameters compared to PSG, with both proportional and fixed bias across measures, regardless of OSA status. This suggests that OSA status is not a factor in PSG/Actigraphy concordance. Support (if any)
Source: SLEEP
Language: English
Topic: Sleep and related disorders
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W4410482081