Development and application of novel accelerometer metrics to describe physical activity and explore associations with chronic disease.

Background:

The overall benefits of physical activity for health are well documented. However, more precise measurement methods are needed to accurately determine dose-response associations and compare groups. As part of this there is a need to develop new metrics to assess physical activity using accelerometers, add to the bank of normative values available for population level translation of data and assess how physical activity is associated with health outcomes. This thesis further develops recently introduced accelerometer metrics that capture the 24 hour physical activity profile and applies these metrics to determine differences between groups and associations with markers of health. In addition to addressing these important gaps in the literature, a focus is placed on assessing specific groups, based on age, ethnicity and health status.

Aims:

1) To provide normative acceleration values that facilitate meaningful age and accelerometer device appropriate translation of accelerometer data.

2) To further develop novel methods for profiling and comparing accelerometer-assessed physical activity.

3) To apply these methods to large accelerometer datasets to investigate the aspects of physical activity that differ by age, ethnicity, and health status.

4) To develop understanding of associations between activity and cardiometabolic health through application of these methods.

Key findings:

1) When examining older adults aged 40-79 years (N = 105), who are able to walk unaided, accelerations measured at the wrist clearly differentiate between self-paced slow, normal, and brisk walking. Normative acceleration values of 140 mg, 210 mg and 350 mg for slow, normal and brisk walking, respectively, are appropriate for 40-69 year olds, largely irrespective of device. However, brisk walking is slower, and the associated acceleration lower (270 mg), for those aged 70-79 years.

2) The methods developed to illustrate and translate the 24 hour accelerometer metrics provide greater insight into how and where the physical activity profile differs between groups than traditional methods. For example, those with post-gestational diabetes (N = 267, age 35.3 ± 4.9 years) had similar overall physical activity (and MVPA) to office workers (N =697, age 44.7 ±10.4 years) 29.3 ±6.9 mg (108.4 ±44.4 minutes per day) and 27.4 ±7.3 mg (99.1 ±35.6 minutes per day) respectively; but differing intensity distributions, -2.62 ±0.16 and -2.55 ±0.21respectively (identified using these methods).

3) Within the subset of the UK Biobank dataset who undertook accelerometer assessed physical activity measures (N = 95,914); adults of black ethnicity (N = 817, overall physical activity 30.5 ±9.2 mg) were more physically active than those of white ethnicity (N = 94,359, overall physical activity 28.3 ±8.5 mg), who in turn were more physically active than those of South Asian ethnicity (N = 738, overall physical activity 28.1 ±8.0 mg). Age related decline in physical activity occurred at a later age (65+ years) in black and South Asian adults compared to white adults (55+ years).

4) Both the amount of activity and the intensity of activity were associated with cardiometabolic risk in healthy individuals (N = 399, age 43.0 ± 10.5 years), however, only the amount of activity was associated with cardiometabolic risk in those who had a chronic disease (N = 1,137, age 65.2 ±9.2 years). Healthy individuals who undertook at least 10 minutes brisk walking and those with a chronic disease, who undertook at least 30 minutes self-paced slow walking had lower cardiometabolic risk than those who did not.

Conclusions:

This thesis contributes to the development, application, and translation of accelerometer metrics for the assessment of physical activity. The utility and appropriateness of these methods for investigation of group differences and associations with health is exemplified, thus improving understanding of the relationship between physical activity and health in specific populations. Future research should use this information to inform interventions and provide population specific recommendations for physical activity in relation to health.