Advancing the measurement of sedentary behaviour : classifying posture and physical (in-)activity

Abstract

Sedentary behaviour, defined by a sitting body posture with minimal-intensity physical activity, is an emergent public health topic. The time spent sedentary is associated with the incidence of non-communicable chronic diseases such as type 2 diabetes and cardiovascular disease and significantly shortens life-expectancy in a dose-response relationship. Office workers are at particular risk of developing diseases related to sedentary behaviour due to their excessive sedentary work. Even though thigh-worn posture sensors are recommended to measure sedentary behaviour, the vast majority of the evidence was collected with waist-worn physical activity sensors, and we still lack a method to measure the posture and the physical activity component of sedentary behaviour simultaneously.

This thesis aims to advance the measurement of sedentary behaviour in an office context by developing new device-based methods to measure both components simultaneously, and by validating and subsequently applying the most promising method to measure the actual amount of sedentary behaviour in the daily life of office workers. The method development showed that it is possible to measure both components of sedentary behaviour with only one sensor, preferably worn on the thigh or waist. While an accelerometer is sufficient for the thigh, an inertial-measurement-unit is preferable for the waist due to a significantly improved posture classification. The method validation subsequently confirmed that waist-worn physical activity sensors, the prevailing choice to measure sedentary behaviour, measure minimal-intensity physical activity. Furthermore, the study uncovered a serious postural dependency causing a systematic overestimation of minimal-intensity physical activity while sitting compared to standing. The subsequent method application considered the posture dependency and combined a thigh-worn posture sensor with a waist-worn physical activity sensor to POPAI, the Posture and Physical Activity Index. POPAI has a sensitivity of 92.5% and a specificity of 91.9% to measure sedentary behaviour and classified 45.0% of the office workers wake-time sedentary. The posture sensor alone overestimated sedentary time by 30.3%, and the physical activity sensor alone overestimated sedentary time by 22.5%. The difference can be explained by active sitting (2.0 hours per day) and inactive standing (1.8 hours per day), both of which are much more common than previously thought.

This thesis confirms the recommendation to use a thigh-worn accelerometer to measure sedentary behaviour and adds the information that such a sensor is also able to measure physical (in-)activity in sitting. Thus, there is no need to approximate sedentary behaviour with sitting, nor is there a need to approximate it with inactivity. In fact, these approximations lead to inaccurate and imprecise results substantially overestimating sedentary behaviour. Due to the predominant use of physical activity sensors to measure sedentary behaviour, recommendations to limit sedentary behaviour should address a limitation of the time spent inactive rather than the time spent sitting. If it turns out that sitting matters, one could expect a much stronger relationship between sedentary behaviour measured with a combined method such as POPAI and detrimental health effects.