Objective measurement of posture allocation and sedentary behaviours in the pre-school child: a validation study.
MSc(R) thesis, University of Glasgow.
Full text available as:
Sedentary behaviours contribute to energy imbalance in young children. Time spent sitting may be an important component of sedentary behaviour but validated measures of posture and posture transitions in the pre-school child are lacking. Accelerometer based posture detection systems validated in the adult literature have often shown excellent agreement with the gold standard of direct observation in controlled environments, but their potential use for the young child is likely to be limited by weight and the need to use multiple sensor sites. Single unit sensors are a potentially more practical alternative that may be suitable for use in research involving young children. This thesis describes the validation of two single unit accelerometer based monitors for their ability to each measure posture and postural transition objectively: the activPAL (PAL Technologies, UK) and the DynaPort MicroMod MoveMonitor 1.2 (McRoberts, NL). It also compares sedentary behaviours as detected by the activPAL and DynaPort monitors with conventional accelerometry using the ActiGraph. The activPAL and DynaPort MoveMonitor algorithms for posture and activity identification in comparison to the gold standard of direct observation have been validated in adults. Neither has previously been validated in young children.
A validation study of the activPAL and DynaPort MicroMod MoveMonitor involving 30 pre-school children is described. The study took place in each child’s usual nursery environment. Children were videoed for one hour undertaking usual activities in nursery while wearing an activPAL and DynaPort MicroMod. In addition, children also wore an ActiGraph accelerometer. The ActiGraph does not measure posture but is well established in physical activity research in childhood. It provided objective information about activity intensity (in particular sedentary behaviour) for each child during the observation period. Video (gold standard) was analyzed on a second-by-second basis and compared with monitor output.
From direct video observation, the proportion of time spent during the one hour of video recording was sit/lie 46%; stand 35%; and walk 16%. The remaining 3% of time was spent in non-sit/lie/upright postures (e.g. crawl, crouch, kneel up) although transitions involving these contributed disproportionately to total posture transitions. The number of sit-stand posture transitions on direct observation was not associated with time spent sedentary. The overall proportion of time detected as ‘sit/lie’ was 42% and 32% as detected by activPAL and DynaPort respectively. Similarly, for activPAL and DynaPort detected ‘walk’, this was 16% and 15% respectively.
Overall sensitivity for time detected as activPAL ‘sit/lie’ was 87%, specificity 97% and positive predictive value 96%. DynaPort MicroMod sensitivity for ‘sit’ was lower but specificity remained high (91%). There was poor correlation between activPAL ‘sit/lie’ and ActiGraph-defined sedentary behaviour (<1100 counts per minute), r = 0.16. However, there was good correlation (r=0.87) between activPAL [‘sit/lie’ + ‘stand’] and ActiGraph defined sedentary behaviour.
The validation results for the activPAL were similar to those described in the adult literature and although those for the DynaPort monitor were less good, both show promise as measurement tools in this age group. Single unit accelerometers capable of detecting posture may have a role in the evaluation of sedentary behaviour in young children, beyond the capabilities of currently used objective monitors such as the ActiGraph. However the role of (and importance of objectively capturing) posture transitions, including non-sit/lie/upright postures requires further investigation. Ultimately, knowledge of posture and postural transitions may provide a better understanding of movement and activity in young children. This potential to help evaluate sedentary behaviours is of interest for childhood obesity research.
Actions (login required)