I read with interest the report by Cooper et al., who described the association between sleep and academic performance. They used Actiwatch and the Actiware-Sleep software (Mini Mitter Co., Inc., Bend, OR, USA) was used to calculate several sleep variables.
Harada et al. reported a cross-sectional study that sleep duration by sleep diary and actigraphy showed a significantly inverse relationship with plasma fasting glucose. They also used Actiwatch and the Actiware-Sleep software, but the appropriate cut-off value is considered to be different from person to person. As expected, the low- and high-sensitivity settings yielded different values on sleep such as sleep durations. The lower sensitivity setting requires a greater number of movements for labelling the subject as the wakeful state than the high-sensitivity setting. In other words, a longer sleep duration is recorded when the low-sensitivity setting is adopted, as compared with the high-sensitivity setting. Unfortunately, there is no description on cut-off point in papers presented by Cooper et al. and Harada et al.
Kushida et al. found the best overall accuracy and ability to detect wakefulness using Actiwatch and sleep polysomnography, when the high-sensitivity setting was used. But validation of the setting should be changed by the target subjects. Kilkus et al. reported no association between sleep duration measured by Actiwatch and eating behaviour factors by using a medium-sensitivity setting of 40.
There is no definite evidence to determine the best sensitivity for children reported by Cooper et al. although they quote references that Actiwatch is a valid and reliable measure of sleep in children. Furthermore, they measured sleep by using Actiwatch for only two consecutive nights, which should also be paid with caution.
The author previously monitored sleep duration of 15 female residents of a nursing home with mean age of 85 years old. The mean sleep duration (standard deviation) of residents calculated using the high-sensitivity setting were 418 min (55), and when the low-sensitivity was adopted, the mean sleep duration (standard deviation) was 474 min (44). The cut-off value of sensitivity for Actiwatch seems important to make final conclusion of the study conducted by Cooper et al. I recommend them to conduct validation study of Actiwatch using sleep polysomnography as a gold standard.