BMI trajectories in middle to old age: methodological developments, missing data and cost-effectiveness

The UK's aging population and increasing obesity prevalence are two of the most significant public health challenges. Currently, more than 3 million people aged 80 years and over live in the UK, and this figure is projected to exceed 8 million by 2050 at increasing costs to the Department of Work and Pensions and the National Health Service. Obesity rates have almost trebled in the last 30 years at a cost of over £3 billion a year and, if forecasting is correct, over half our population will be obese by 2050. Significant obesity related co-morbidities (e.g. diabetes, osteoarthritis, cardiovascular disease), coupled with multiple health problems associated with aging (e.g. Alzheimer's disease, arthritis and cancer), places older people at significant risk. While there is a focus on obesity research in young people and the general adult population, there is relatively little research into how obesity develops into old age and the consequences of obesity in older adults.

It is useful to predict BMI trajectories in patients with different characteristics in order to inform public health policies. If predictions can be made for a patient's future BMI, then these long-term estimates can be used in economic models in order to estimate how cost-effective potential interventions, actions or policies are expected to be. For policy purposes, it is critical to understand how these effects develop over time, and longitudinal analyses are relied upon to provide these long-term predictions. A number of cost-effectiveness analyses have used BMI trajectories, including models for diabetes.

Previous research has investigated mean BMI trajectories and has shown that they differ as patients go from middle age to older age. However, it is well known that obesity prevalence and BMI differ between different social groups and that an individual's environment can have a large influence on their weight. This project will identify different groups of individuals who are likely to experience different types of BMI trajectories.

Medical literature suggests that there are certain types of individual who are more likely to see a drop in BMI as they enter old age, known as the 'obesity paradox' because BMI drops but body fat percentage increases. In this group of people, BMI might be misleading and measures of muscle mass (such as grip strength) could be an important in identifying individuals with similar BMI trajectories but different risks of obesity. These different trajectories will also be linked with long term health outcomes, including a range of diseases and death.

Missing data can be a particular problem when researching the elderly due to illness, memory loss and access to individuals in care homes. This could mean that in a sample of individuals over the age of 50 years, more elderly individuals are under-represented. In order to account for the potential underrepresentation of certain individuals, analysis should account for missing data and this study will investigate the influence that missing data might have on estimated BMI trajectories.

The study will use data from the English Longitudinal Study of Aging (ELSA) and the analysis described above will provide more detailed evidence for health care professionals helping to identify those most at risk of morbidities and mortality by the BMI trajectories that they are expected to follow. The results will also be used to inform cost-effectiveness analyses, which can help policy makers to determine the most cost-effective interventions to help improve the health of future generations as they enter old age. I will use a diabetes prevention model as a case study, to illustrate how potential diabetes prevention interventions might influence patients with different characteristics and to demonstrate the impact that missing data could have on policy if it remained unaccounted for.

Medical literature suggests that certain individuals lose muscle mass as they get older, while others continue to show a steady increase in BMI. Those who lose muscle mass experience a drop in BMI but an increase in body fat percentage (known as the obesity paradox) and could be at increased risk of comorbidities or mortality which is left unidentified due to their lower BMI. This study will use a range of growth models to estimate mean trajectories of BMI as individual's transition from middle to old age. Using latent class growth modelling, it will be possible to identify different types of individuals who are more or less likely to experience different BMI trajectories, allowing individuals at risk to be targeted by health professionals and policy makers earlier.

Multiple BMI trajectories identified in this study will be linked with comorbidities and mortality in order to determine which BMI trajectories are a cause for concern. This will be particularly important if the different trajectories have very different outcomes. Again, this will allow at risk individuals to be targeted earlier with the intention of leading to better health outcomes.

Missing data will be accounted for in a number of ways. These will include complete case analysis and imputation, which have been used previously in the literature as well as pattern mixture and selection models. These methods will be compared and additional methods development will be considered if appropriate.

Results from the analysis described above will be used to update the existing School for Public Health Research (SPHR) diabetes prevention model. This will allow optimal subgroups of individuals to be identified for which interventions will be most cost-effective as well as producing unbiased estimates after accounting for missing data.

This project will benefit UK as well as other countries with aging populations by informing public debate and policy decisions associated with BMI trajectories into older age. The research implemented in this project will help inform policy makers about the most effective way to reduce the risks of poor health in later life by suggesting key indicators of at risk demographics and BMI trajectories. In order to maximise the impact of the project, we will aim to build strong links across a range of key stakeholders including health care institutions, local authorities, the government and public health bodies, whilst also involving charitable organisations, and where applicable informing the general public.

A summary of interested parties I expect to benefit from work from this project is outlined below.

The research implemented in this project will be valuable to institutions and individuals directly involved in formulating and implementing policies which seek to reduce the risks of poor health in later life by suggesting key indicators of at risk demographics and BMI trajectories. The estimated BMI trajectories as well as the case study investigating the cost-effectiveness of diabetes prevention interventions, will allow policy makers such as the National Institute for Health and Care Excellence (NICE), Public Health England (PHE) and the Department of Health (DH) to update their guidance surrounding obesity and diabetes prevention and allow them to determine which groups of people and individuals with which characteristics should be targeted to ensure that they are as effective as possible. Similarly, policy makers interested in other health care or disease areas could benefit from implementing the newly estimated BMI trajectories found in this study, in their own cost-effectiveness analysis. The case study will update the School for Public Health Research (SPHR) diabetes prevention model. This model has been used for decision making by a number of potential stakeholders including Public Health England, NHS England, the Department of Health and local commissioners highlighting the wide spread interest for research of this type and the breadth of impact that this project could have.

As a result of improved evidence of BMI trajectories being fed into policy decision making, the health of the public will benefit. Improved policies in diabetes prevention, as well as any other disease areas which are related to BMI trajectories and other obesity related comorbidities, could help to prevent ill-health in the population. Similarly, the public will benefit from better value for money in the health care system paid for by the tax payer and improved economic outcomes if the research leads to improvements in productivity due to the prevention of obesity and related diseases.

The research might also be of interest to a variety of charitable and third sector institutions, such as Age UK. The findings of this research project will provide a knowledge exchange with these institutions, resulting in improved understanding within these institutions of how to identify people who are more likely to experience risky BMI trajectories which are linked with poor health outcomes.

Further details of how interested parties might benefit from this work can be found in the attached pathways to impact document.