Skip to main content
. 2024 Sep 12;10:20552076241277150. doi: 10.1177/20552076241277150

Table 3.

Descriptive statistics of sleep-related variables from the criterion measure (polysomnography) and the Garmin Vivofit 4 monitor.

Polysomnography Garmin Garmin difference MAE
Bedtime (hours)a 22:14:40 ± 0.58 23:01:18 ± 1.16 0.84 ± 1.60 0.95 ± 1.53
Risetime (hours)a 05:59:45 ± 0.13 05:39:55 ± 0.47 −0.34 ± 0.70 0.47 ± 0.62
Sleep onset latency (minutes)b 22.8 ± 20.0 77.4 ± 100.9 54.6 ± 95.1 57.2 ± 93.6
Total sleep time (hours) 7.03 ± 0.78 6.49 ± 1.47 −0.55 ± 1.21 0.86 ± 1.01
Wake after sleep onset (minutes) 43.2 ± 37.9 5.2 ± 9.3 −38.0 ± 38.8 39.0 ± 37.7
Sleep efficiency (%)b 86.4 ± 8.5 83.1 ± 19.0 −3.3 ± 13.8 8.9 ± 10.9
Sleep stages (REM Lightc)
 Light sleep (%) 71.1 ± 10.3 50.4 ± 13.8 −20.7 ± 15.9 21.2 ± 15.2
 Deep sleep (%) 28.9 ± 10.3 49.6 ± 13.8 20.7 ± 15.9 21.2 ± 15.2
Sleep stages (REM Deepc)
 Light sleep (%) 53.7 ± 9.7 50.4 ± 13.8 −3.3 ± 14.9 12.2 ± 8.9
 Deep sleep (%) 46.3 ± 9.8 49.6 ± 13.8 3.3 ± 14.9 12.1 ± 8.9

Values are mean ± SD.

a

Summary values are circular mean ± SD, where SD was calculated using the mean shorter distance method (see paulhibbing.com/daytime); circular operations were not needed for paired calculations (mean difference and MAE), which is why the difference of circular means differs slightly from the values in the mean difference colum.n

b

N of 24 rather than 25 because Garmin predicted bedtime before “lights out” for one participant, precluding calculation for these variables.

c

Garmin monitor reported time in light and deep sleep. Polysomnography reported time in the REM stage and three non-REM stages (N1–N3). Polysomnography stages were coded in two ways, the first (REM Light) defining REM and N1–N2 as light sleep versus N3 as deep, while the second (REM Deep) defined only N1–N2 as light sleep and both N3 and REM as deep.

MAE: mean absolute error; REM: rapid eye movement.