Understanding mid-life risk and inequalities associated with frailty development in later life: NHS linked records data and prospective cohort analysis from England

Background Frailty is the most problematic expression of population ageing.

It is a state of vulnerability to external stressors that in older people reduces their resilience and ability to deal with stress, illness, or injury.

Frail individuals are at greater risk of adverse events, such as falls, hospitalisations, institutionalisation, and death. The prevalence of frailty is expected to rise with an ageing population. Understanding its determinants is crucial for promoting healthy ageing. Existing prediction models for frailty development are limited and have methodological weaknesses.

To date, no reliable prediction model exists to accurately identify patients at high risk of becoming frail in later life. Aims To identify mid-life factors associated with frailty development in later life. To develop and validate clinical prediction models for long-term risk of frailty development in mid-life.

To investigate life expectancy and inequalities in life expectancy in all states of frailty.

I will meet these aims through four objectives: Systematic review and critical appraisal of existing mid-life individual- and area-level factors associated with frailty development in later life.

Scoping review to identify and investigate methods to evaluate and reduce algorithm bias (fairness) in a clinical prediction model. Develop and externally validate a clinical prediction model for predicting 20-year risk of frailty among older adults.

Estimate the life expectancy of people in each frailty state according to sex, ethnicity, and area-level socio-economic status in mid-life.

Methods I will conduct a systematic review to identify mid-life factors associated with frailty development in later life. I will also critically appraise existing prediction models for risk of frailty.

I will develop and validate prediction models using two large UK databases containing primary care electronic patient health records for more than 16 million current patients. I will evaluate different model-building approaches, including machine-learning methods. I will investigate methods to evaluate and reduce algorithm bias to ensure fairness in the developed prediction models.

I will use a longitudinal study that collects data from a representative sample of the English population aged 50 to enrich the models with additional social and psychosocial factors in mid-life. Expected impact I anticipate that the research will have a substantial impact within 3-years. It will improve our understanding of the risk and inequalities of frailty in older adults.

I will identify the mid-life factors that contribute to frailty development.

My findings will help clinicians to identify individuals at high risk of becoming frail in later life, so that they can create more targeted personalised interventions.

Establishing whether risk factors are associated with frailty development could help to drive new policies for healthcare, which could include tailored programmes for weight loss or increasing physical activity, for example.

Dissemination The results will be published in relevant peer-reviewed, open-access journals and presented at national and international conferences.

To maximise public engagement, plain language summaries will be produced and shared with my PPI advisory group, patient charities, university websites, and social media platforms.