Sleep

Self-reported sleep was collected in NNPAS 2023 and NATSINPAS 2023 (non-remote). The types of self-report questions were different for respondents based on their age. Directly measured sleep data from accelerometers was also collected in NNPAS 2023 and NATSINPAS 2023.

The types of self-reported sleep questions varied by age in NNPAS 2023 and NATSINPAS 2023. The following table shows self-reported sleep topics by age for each survey:

Self-report sleep duration was collected for all respondents 2 years and over. It was calculated from the time the respondent reported going to bed and turning the lights out to go to sleep the night prior to the interview and the time they reported waking up on the day of the interview. This approach was used because it can be difficult for respondents to report the actual time that they fell asleep. The time the respondent reported going to bed and turning the lights out to go to sleep is treated as the time they went to sleep. 

Respondents were directed to exclude time spent reading or watching television in bed, as well as any time they may have woken throughout the night.

A parent or guardian was asked these questions on behalf of children aged 2–14 years old. Young people aged 15–17 years could respond on their own behalf with consent from a parent or guardian.

Quality of sleep was only collected in NNPAS 2023. The respondent was asked to rate their quality of the sleep on the night prior to interview on a scale of very poor to very good. If a person was assisting or responding on behalf of another person (18 years and over), quality of sleep was asked only if that person was present during the interview. A parent or guardian rated the sleep quality when responding on behalf of a child (2–17 years). Young people aged 15–17 years could respond on their own behalf with consent from a parent or guardian.

Sedentary screen activity in the hour prior to going to bed to go to sleep was collected for pre-school and school-aged children in NNPAS 2023. The use of audio devices or devices designed to assist with sleep were excluded.

Whether screen-based devices were used in the bedroom the day prior to interview and the type of screen-based device the child or young person had in the bedroom was collected in NNPAS 2023.

The total time children aged 2–4 years, and 5 year-old children not attending school, spent napping on the day prior to the interview was collected. 

The NNPAS 2023 and NATSINPAS 2023 collected data from accelerometers to produce modelled estimates of sleep.

Sleep estimates derived from accelerometer data provide an objective measure of sleep timing and duration. Unlike self‑reported sleep measures, they are not affected by recall bias, memory or respondent interpretation. 

The use of wrist‑worn devices allows sleep and physical activity to be measured simultaneously using a single instrument. This enables analysis of sleep, inactivity, and physical activity across multiple consecutive days.

While accelerometer‑derived sleep measures do not directly assess sleep quality, they provide a cost-effective approach for analysing statistics like sleep timing consistency and sleep efficiency.

The Heuristic algorithm looking at Distribution of Change in Z‑Angle (HDCZA)[1] was used to estimate the main "sleep window" period. This method was originally developed for the UK Biobank, which also did not collect a sleep diary from participants[2]. 

HDCZA identifies periods likely to be sleep by examining the change in the z-axis angle of the acceleration vector. When the angle shows long periods of minimal change, the algorithm classifies this as sustained inactivity, which is used as an indicator of sleep. The longest sustained inactivity bout is set as the main sleep period time window (SPT-window). Within this window, the total amount of sustained inactivity (i.e. time spent asleep) can be calculated. Activity in this window contributes to wakefulness estimates. 

For some participants, it was clear that the HDCZA algorithm was unable to correctly identify the sleep period. This was particularly for those with inconsistent sleep, longer wake windows in the middle of the night and light sleepers. In these instances, estimates of “time went to bed” (or got out of bed) were provided as inputs to the algorithm to assist it in determining the sleep period. These values are provided on the ‘Accelerometer - Sleep Level’ dataset. See NNPAS 2023 and NATSINPAS 2023 microdata.

Sleep estimates derived from accelerometer data are modelled based on detected movement patterns and periods of inactivity, rather than direct physiological measures of sleep. As a result, these estimates are subject to several limitations that should be considered when interpreting results.

Accelerometers do not measure brain activity or other physiological indicators of sleep. Periods of low movement may be inferred as sleep, which means wakefulness (such as lying still in bed) may be misclassified as sleep. Estimates do not distinguish between sleep stages and may be less accurate for people with atypical sleep patterns or restless sleep.

As wrist‑worn devices measure arm movement, sleep estimates may be influenced by behaviours such as sleeping positions, arm restraint, device removal, or non‑wear during sleep.

Sleep estimates depend on the algorithms and thresholds used to define sleep and wake periods. Differences in modelling approaches or parameter choices may affect comparability across studies or over time.

1. van Hees, V., Sabia, S., Jones, S., Wood, A., Anderson, K., Kivimäki, M., Frayling, T., Pack, A., Bucan, M., Trenell, M., Mazzotti, D., Gehrman, P., Singh-Manoux, B., Weedon, M. 2018. Estimating sleep parameters using an accelerometer without sleep diary. Scientific Reports, 8(1). https://doi.org/10.1038/s41598-018-31266-z
2. Doherty, A., Jackson, D., Hammerla, N., Plötz, T., Olivier, P., Granat, M. H., et al. 2017. Large scale population assessment of physical activity using wrist worn accelerometers: The UK Biobank study. PLoS ONE, 12(2). https://doi.org/10.1371/journal.pone.0169649