•  
  •  
 

COMPARING DIARY LOGS TO AUTOMATED ALGORITHMS AND VISUAL INSPECTION FOR TOTAL-BEDTIME ESTIMATION IN POSTPARTUM WOMEN

Abstract

Bethany Grace Rand1, Samantha F. Ehrlich1, Assiamira Ferrara2, Scott E. Crouter, FACSM1, Matthew Buman, FACSM3, Monique M. Hedderson2, Susan D. Brown4. 1The University of Tennessee Knoxville, Knoxville, TN. 2Kaiser Permanente Northern California, Oakland, CA. 3Arizona State University, Phoenix, AZ. 4University of California, Davis, Davis, CA.

BACKGROUND: Automated methods for identifying in-bed and out-of-bed times to accurately identify sleep windows using wearable physical activity monitors with 24-hour wear protocols have the potential to reduce researcher and participant burden in analyses of sleep time. Traditionally, diary logs are used to estimate the window between in-bed and out-of-bed, going to bed and waking for the day, i.e., total bedtime. This exploratory analysis compared the van Hees and Tracy bedtime algorithms, and visual inspection (i.e., manual inspection of the actigraphy tracings by a trained observer), to participant diaries for the identification of total sleep time in postpartum women. METHODS: Participants (N= 5) came from the PETALS-2 study, and wore an ActiGraph wGT3X-BT device, initialized at 30 Hz, on the non-dominant wrist for 7-days at 6 months postpartum. Participants logged daily to-bed and wake times in a paper diary. The first day was dropped and the remaining days’ times averaged for individual-level mean in-bed time, out-of-bed time, and total-bedtime; these were then compared across methods. Group mean differences (i.e., diary minus estimate) in in-bed times, out-of-bed times, and total-bed-time were calculated and equivalence testing used to compare the group mean total-bed-time from van Hees, Tracy, and visual inspection to the diary (reference), with an equivalence region of ± 30 min. RESULTS: The van Hees and Tracy algorithms and visual inspection estimated in-bed times were, on average, 13.8 min (SD 36.7), 3.6 min (29.0), and 13.2 min (36.9) later, respectively, than the diary. The van Hees and Tracy algorithms estimated out-of-bed times were 36.8 min (59.4) and 23.4 min (40.0) earlier than the diary, respectively, while visual inspection’s out-of-bed time estimate was 0.8 min (13.6) later than the diary. Average total-bedtime for the van Hees [432 min (SD 45.2)], Tracy [454 min (47.5)], and visual inspection [466 min (41.7)] were not equivalent to the diary [480 min (75.2)]. CONCLUSION: The van Hees and Tracy bedtime algorithms, and visual inspection, may not adequately capture total-bedtime in postpartum women, as defined by traditional diary logs. Future analyses will compare these methods in a larger sample from the PETALS-2 study.

This document is currently not available here.

Share

COinS