Background

Obesity is associated with increased risks of morbidity and mortality, and globally it represents a considerable burden for healthcare systems. The prevalence of obesity in children and adolescents has risen dramatically in recent decades across most western countries and several low-income countries.1 Although some recent data suggest prevalence may have stabilised in the USA, UK and Sweden,2-4 it is too soon to know whether this is a true abating of the epidemic and levels remain high in all of these countries. A life course perspective is essential in the study of obesity, as emphasised by the MRC obesity research strategy published in 2010. There are many unanswered questions in the field of life course influences on obesity, which it is important to address in order to further the understanding of the aetiology of obesity and therefore to inform the design of prevention and treatment initiatives. Although we know that overweight tends to track across the life course5 and that childhood overweight and obesity is associated with adverse health outcomes in adulthood6, few studies have had sufficiently detailed longitudinal data to model individual trajectories of adiposity across childhood and adolescence and therefore to examine in detail the nature of adiposity changes across childhood and adolescence and how these changes relate to the determinants and health consequences of obesity. There is evidence to suggest that obesity-related cardiovascular risk factors such as blood pressure and lipids track across the life course within individuals7, 8, and that the associations of these risk factors with atherosclerosis in adulthood are similar regardless of whether the risk factors are measured in childhood or in adulthood at the same time as assessment of atherosclerosis.9 Again, however, few studies have had sufficient data to model the development of these obesity-related cardiovascular risk factors across childhood and adolescence and to relate them to patterns of adiposity change. Longitudinal models of behavioural risk factors for obesity such as diet and physical activity are also relatively rare in the literature, and the interplay between multiple genetic, lifestyle, and socioeconomic risk factors for obesity and longitudinal trajectories of adiposity has not been well explored. It would be of great public health interest to use longitudinal models of changes in adiposity, obesity-related cardiovascular risk factors, and lifestyle risk factors for obesity to identify whether there are subgroups of the population who could be identified in early life as at risk of extreme obesity and associated cardiovascular risk in adolescence. Furthermore, little is known about why some obese individuals remain metabolically healthy; patterns of change in adiposity, cardiovascular risk factors and lifestyle factors could be explored in such individuals to explore the issue of 'protective phenotypes' in obesity - identified as a priority area in the MRC obesity research strategy. I propose to address these areas of life course epidemiology of obesity using longitudinal modelling of adiposity, its determinants and its health consequences across childhood and adolescence using data from prospective cohort studies.

Aim 1: To characterise patterns of change in adiposity and its associated cardiovascular risk factors across childhood and adolescence

Specific research questions:

1. How does adiposity change across childhood and adolescence, as measured by BMI (0-18 years), waist circumference (7-18 years) and directly determined total body fat mass (9-18 years)?

2. Do changes in BMI, waist circumference and directly determined fat mass from age 9 to 18 mirror each other or are there periods where change between these differs? For example does timing of change in waist differ from that in general adiposity (BMI or total fat mass) around the time of puberty?

3. How does blood pressure change from age 7-18 years?

4. How do lipids, insulin, and inflammatory markers change from age 9-18 years?

Aim 2: To identify lifestyle and socioeconomic determinants of adiposity changes across childhood and adolescence

Specific research questions:

1. What are the joint effects of trajectories of physical activity (measured using accelerometers) and diet (measured using food frequency questionnaires and diet diaries) from ages 7-13 years on changes in adiposity from ages 7-18 years?

2. What are the associations of changes in socioeconomic position (e.g. differences between grandparental, parental and own education) on changes in adiposity from birth to 18 years?

3. Are there socioeconomic differences in trajectories of physical activity (measured by accelerometer) and diet (measured by diet diaries and food frequency questionnaires) across childhood and adolescence? Are these inequalities of consistent magnitude across childhood and adolescence and to what extent do they mediate socioeconomic inequalities in adiposity trajectories?

Aim 3: To identify how changes in adiposity relate to changes in cardiovascular risk factors in childhood and adolescence

1. How do changes in BMI (birth to 18), waist circumference (ages 9-18), and fat mass (ages 9-18) relate to changes in blood pressure (ages 7-18), insulin, lipids and inflammatory markers (ages 9-18) in childhood and adolescence?

2. Do increases in markers of adiposity precede increases in cardiovascular risk factors (blood pressure, insulin, lipids and inflammatory markers) and if so what is the lag period between adiposity change and risk factor change for each risk factor?

Aim 4: To develop predictive models of extreme obesity and of metabolic resilience to obesity

Specific research questions:

1. Are there specific subgroups of children/adolescents with particularly adverse changes in adiposity, cardiovascular risk factors and lifestyle factors who could be identified in early life as at risk of extreme obesity?

2. Are there some overweight adolescents who remain more metabolically healthy than other overweight adolescents? If so, what characterises these individuals with respect to individual trajectories of adiposity, cardiovascular risk factors and lifestyle factors across childhood and adolescence?

Statistical analysis:

As part of my current role, I have modelled trajectories of height, weight, and ponderal index/BMI from birth to ten years using random effects linear spline models (in application put references to your papers here). In order to extend the models for BMI to age 18 and to model trajectories of waist circumference from ages 7-18 and DXA-determined total body fat mass from ages 9-18, a different modelling strategy will be required owing to individual variation in the timing of puberty onset. A number of potential modelling strategies could be used for these changes, with each having different strengths and limitations. I therefore propose to model BMI, waist circumference and fat mass from their earliest age of measurement (birth, 7 and 9 respectively) through to age 18 in several different ways. I will compare how trajectories derived by these different approaches associate with potential determinants and consequences. Trajectories of blood pressure, lipids, insulin and inflammatory markers will be modelled using random effects linear spline models in a similar way to the models I have already constructed for height, weight and adiposity to age 10. Trajectories of physical activity are currently being modelled by a colleague (Alex Griffiths); these will be available for use by the start of this fellowship. I will analyse longitudinal diet patterns in multilevel models using repeat scores from principal component analysis of dietary patterns using data from diet diaries and food frequency questionnaires. Associations between adiposity and cardiovascular risk factor trajectories and their determinants (physical activity, diet, etc) will be modelled using multivariate models. The relationship between adiposity trajectories and cardiovascular risk factor trajectories will be modelled in a joint multivariate model. Predictive models for 1) extreme obesity in adolescence and 2) metabolic resilience to obesity will be developed and internally validated. Accuracy of prediction will be assessed by discrimination using area under the receiver operator curves (AUROC) and calibration by comparing observed to predicted risk by tenths of the predicted risk score.

Relation to existing ALSPAC approved projects:

Aims 1-3 are already have approval under the ALSPAC exec approval for the health strand of the ESRC large grant (strand leader: Debbie Lawlor). The aims of that strand were somewhat ambitious and it will not be possible to complete all of them during the period of the grant and hence this application sensibly extends the aspect of that grant related to child obesity. The additional analyses proposed for this fellowship build on work that I have already completed in my position as RA on the ESRC large grant. Aim 1 is also already partly covered by the MRC life course grant for which Kate Tilling is PI and which already has ALSPAC exec approval. Only Aim 4 is completely novel and not covered by the existing approval.

Plan for completing work:

1. LH, DL and KT will agree specific objectives and analysis methods

2. LH will put together datasets and conduct analyses (LH already has approved access to ALSPAC built datasets)

3. LH will draft initial manuscripts

4. LH, DL, KT and other co-authors will be involved in study design, and critical evaluation of manuscripts as appropriate for each paper

All derived variables from the trajectories work will be returned to the exec. with appropriate documentation so that they can be widely used by others (as we have already done with the growth data from birth to 10).