Objective
To compare different scoring parameter settings in actigraphy software for inferring sleep and wake bouts for validating analytical techniques outside of laboratory environments.
Design
To identify parameter settings that best identify napping during periods of wakefulness, we analyzed 137 days on which participants reported daytime napping, as compared with a random subset of 30 days when no naps were reported. To identify settings that identify periods of wakefulness during sleep, we used data from a subsample of women who reported discrete wake bouts while nursing at night.
Setting
Equatorial Tanzania in January to February 2016.
Participants
The Hadza—a non-industrial foraging population.
Measurements
Thirty-three subjects participated in the study for 393 observation days. Using the Bland-Altman technique to determine concordance, we analyzed reported events of daytime napping and nighttime wake bouts.
Results
Only 1 parameter setting could reliably detect reported naps (15-minute nap length, ≤50 counts). Moreover, of the 6 tested parameter settings to detect wake bouts, the setting where the sleep-wake algorithm was parameterized to detect 20 consecutive minutes throughout the designated sleep period did not overestimate or underestimate wake bouts, had the lowest mean difference, and did not significantly differ from reported wake-bout events.
Conclusion
We propose operational definitions for multiple dimensions of segmented sleep and conclude that actigraphy is an effective method for detecting segmented sleep in future cross-site comparative research. The implications of such work are far reaching, as sleep research in preindustrial and developing societies is documenting natural sleep-wake patterns in previously inaccessible environments.