Search

Background and objectives: Sub-Saharan Africa (SSA) has an increasing burden of noncommunicable diseases (NCDs). Comprehensive and context-specific evidence on the major modifiable risk factors for NCDs, such as physical activity, is required to help inform strategies to address this burden. Accurate measurement of physical activity and other health indicators will be important to achieving these objectives. This thesis aims to contribute to the current understanding of physical activity in SSA, and the development of frameworks to objectively assess health indicators, including physical activity, in free-living populations. Methods: Using existing large-scale population data from SSA (N=29606), I examined the prevalence and determinants of physical activity. The association between physical activity and key intermediate risk factors for NCDs was also assessed. Furthermore, I conducted a feasibility study (N=32) to assess the usability and validity of two commercial wrist-worn heart rate monitors as potential tools in the measurement of physical activity and related dimensions in free-living populations. Results: Across these diverse sub-Saharan African populations, there was substantial variation in the prevalence of NCD risk factors, including insufficient physical activity. Women, older individuals, those with higher levels of education, overweight and obese individuals, the unemployed, and urban residents were more likely to be insufficiently active. Furthermore, there was strong evidence of sociodemographic differences in the domains within which physical activity is undertaken, with potential inequities observed for participation in leisure-time physical activity. A weak but statistically significant association was observed between physical activity and BMI, and physical activity and diastolic blood pressure. Finally, the commercial wrist-worn heart rate monitors examined demonstrated acceptable coverage, with minimal data loss over the study period, and comparable estimates for population-level estimates. However, there was evidence of systematic differences in individual heart rate measurements, potentially limiting their use in active populations. Conclusions: This work provides key evidence on physical activity, its distribution and determinants, in 10 diverse populations in SSA. It also highlights opportunities for reducing the obesity and hypertension burden in these populations, and hence informs future directions for research and strategies to address the increasing NCD burden. Lastly, it provides the basis for further validation studies of commercial wearable devices in free-living populations and highlights the need for improvements to device algorithms.

STUDY 1. INTRODUCTION. Insufficient physical activity is a major risk factor for developing type 2 diabetes. Using isotemporal substitution models, the influence of replacing modest durations of sedentary time with physical activity on diabetes risk scores can be studied. The aims of this study were to examine the relationship between diabetes risk scores, sedentary time and physical activity measured using wrist worn accelerometry, and to model the changes in risk scores by reallocating movement behaviours from lower to a higher intensity. METHODS. Data from 251 (93 males; aged 56.7 ± 8.8) participants from a mixed ethnicity cohort from Leicestershire, UK were selected for analysis. The relationship between diabetes risk (using the Leicester Diabetes Risk Assessment Score), physical activity and sedentary time was identified using multiple linear regressions and isotemporal substitution analysis. Models were calculated for main effects and also adjusted for peak oxygen uptake (VO2) and accelerometer wear time. RESULTS. Both unadjusted and adjusted models revealed that diabetes risk was inversely related to sedentary time, and positively related to light and moderate to vigorous physical activity (MVPA) (p < 0.0005). Unadjusted, the replacement of sedentary time with 10 minutes of either light or MVPA resulted in a reduction in diabetes risk score of -0.22 and -0.54, respectively. There was an eight to nine times greater reduction in risk for the same MVPA replacement models when the least active participants were compared to the pooled analysis (3.601 unadjusted). CONCLUSION. Diabetes risk is associated with sedentary time and physical activity estimated from wrist worn accelerometry. The replacement of sedentary time with MVPA is most beneficial for the least active individuals. STUDY 2. INTRODUCTION. Most associations between physical behaviours and health are assessed using intensity and duration based estimations; however, individuals accrue physical activity in differing ways and behavioural profiles have been linked with varying cardiometabolic risk factors. The frequency or regularity of behaviour may hold additional relationships with health, but have not been extensively explored. Accelerometers provide researchers with a large stream of raw data to analyse. The aim of this paper was to calculate a novel method of behavioural regularity called sample entropy from wrist worn accelerometry and to ascertain whether there are associations with cardiometabolic risk factors in adults. METHODS. Data from 290 (107 males; aged 57.0 ± 8.8) participants from a mixed ethnicity cohort from Leicestershire, UK were selected for analysis. Entropy scores were calculated using 60-second count data within MATLAB. The relationship between entropy scores, physical activity, sedentary time and cardiometabolic risk factors was identified using multiple linear regressions. Models were calculated for main effects and also adjusted for age, sex, accelerometer wear time and body mass index (BMI). RESULTS. Sample entropy scores were significantly related to high-density lipoprotein (HDL) cholesterol (b = 0.148, p = 0.042), triglycerides (b = -0.293, p = 0.042) and glycated haemoglobin (HbA1c) (b = -0.225, p = 0.006), even after adjustment for confounding variables. Traditional intensity estimates of physical activity were not associated; however, the frequency of breaks in sedentary time were significantly related to entropy scores (b = 0.004, p = 0.002). CONCLUSION. Using a novel measure of signal complexity, associations have been revealed with cardiometabolic risk factors; however further analysis in a larger, more diverse dataset is required to ascertain the utility of this technique within behavioural research and if so, what constitutes typical/average levels of entropy within a population. STUDY 3. INTRODUCTION. Acute physiological changes such as reductions in postprandial glucose excursions have been demonstrated within experimental studies that have compared being physically active to sedentary conditions. However, for this information to be truly useful, the coupling of behaviour and glucose data in a free-living environment needs to be achieved. The aim of the study was to ascertain if there is a relationship between objectively measured physical activity, sedentary time and glucose variability using glucose monitoring in an adult population. METHODS. Data from 29 participants recruited from a mixed gender sample from Leicestershire, UK were selected for analysis. Physical activity, sedentary time and interstitial glucose was measured continuously over 14 days using an accelerometer and the Freestyle Libre flash glucose monitor. Daily time (minutes) spent sedentary, and in light activity and moderate to vigorous physical activity (MVPA) were regressed against glycaemic variability indices including daily mean (average) glucose, standard deviation and mean amplitude of glycaemic excursions (MAGE). Generalised Estimating Equations were calculated between behaviour and glycaemic variability variables. Models were calculated for main effects and also adjusted for age, gender and accelerometer wear time. RESULTS. Physical activity and sedentary time were associated with measures of glucose variability, however low fitness individuals showed a stronger relationship between MVPA and MAGE (MAGE: whole sample b = -0.002, low fitness b = -0.012. Additionally, after adjustment for covariates, sedentary time was positively associated with a higher daily mean glucose (b = 0.001, p = 0.001) and MAGE (b = 0.002, p < 0.0005) for the low fitness group. MVPA was negatively associated with mean glucose (b = -0.004, p < 0.0005) and MAGE (b = -0.012, p < 0.0005); however, standard deviation of glucose was not associated with behaviour of any intensity. The magnitudes of the relationships were small, although participants were non-diabetics and exhibited relatively good glucose control i.e. minimal fluctuations in daily glucose variability. CONCLUSION. This study shows that sedentary time, physical activity and glucose variability are related. Despite supporting the previous laboratory research, it is uncertain whether any changes in glucose will reliably occur in all individuals. MVPA confers the largest reductions in glucose variability indices, yet as one of the few studies to couple behaviour and glucose data, more research is needed on larger and more diverse samples.

The increasing evidence of associations between sedentary behaviour and low levels of physical activity in adults and both immediate and long term health implications is of public health concern. There is a need to further our understanding of adult's health behaviours, to facilitate the development of behaviour change strategies promoting healthy behaviours. This thesis provides four independent but interlinked studies focusing on adult s sedentary behaviour and physical activity in the context of measurement and behaviour change. Chapter 1 provides an introduction to the thesis where the scene is set for the placement of the studies in this thesis in the field of sedentary behaviour, physical activity, and measurement methods. Chapter 2 describes a systematic review of the relationship between sedentary behaviour and physical activity in adults. This systematic review is of primary importance as it was instrumental in shaping and informing the direction of the research described in later chapters. Chapter 3 describes a laboratory study investigating the measurement of energy expenditure during common sitting and standing tasks and also examines the 1.5 MET definition of sedentary behaviour. This study provides evidence that the 1.5 MET threshold for sedentary behaviours seems reasonable however some sitting-based activities may be classified as non-sedentary in people of differing weight status. This study raised some important questions on the validity of objective measurement devices for differentiating between sitting and standing postures. Thus, Chapter 4 of this thesis describes a laboratory study investigating the validity of the ActiGraph inclinometer algorithms for differentiating between sitting and standing postures. Chapter 5 is an intervention investigating sedentary behavior and physical activity compensation outside working hours in a sample of office workers exposed to sit-to-stand desks in the workplace. This thesis found that light physical activity, especially standing, could be one of the most efficient and feasible behaviours to replace sedentary behaviour. Such findings add considerably to the existing literature. Targeting such facets of adults behaviour and specially office workers holds great potential for behaviour change strategies.