Dementia risk prediction models

Charlotte Warren-Gash

15 April 2019

The Prime Minister’s Challenge on Dementia 2020 aimed to make England the world leader in dementia care, support, research and awareness by 2020. While substantial progress has been made on many of the commitments, dementia remains a stubbornly difficult condition to prevent and treat. Numbers of people living with dementia in the UK are projected to exceed one million by 2025, as the population grows and ages.

Rising awareness means that around two thirds of people with dementia are now likely to have a recorded diagnosis. But for a condition such as dementia, for which there are no effective treatments, the benefits, costs and potential harms of receiving an early diagnosis are far from clear. The UK National Screening Committee does not recommend population screening for dementia given the lack of accuracy of existing tests combined with the lack of treatments to slow or prevent dementia. These issues apply even more strongly in the context of dementia risk prediction.

Our report Dementia Risk Prediction Models: what do policymakers need to know? explores the recent proliferation of dementia risk prediction models and their implications for research, health and society.

Research is needed into the effects of giving people a risk score for dementia, and, crucially, into the effectiveness and cost-effectiveness of interventions to prevent dementia

Dementia risk prediction models use information about patients e.g. data on their health, lifestyle, age and perhaps genetic information or results of brain scans to predict the chance that particular groups of people will develop dementia in the future. However, unlike risk prediction tools for other conditions e.g. cardiovascular disease, which are successfully used in routine practice by general practitioners to identify people at high risk and offer tailored lifestyle advice or interventions, risk prediction models for dementia are currently limited to use in research.

This is entirely appropriate given the many uncertainties and challenges around dementia risk prediction in the general population. At present, the ability of different models to predict dementia varies greatly, reporting of models is not standardised and few are validated for use across different populations.

The natural history of dementia and dementia at-risk states is poorly understood. Research is needed into the effects of giving people a risk score for dementia, and, crucially, into the effectiveness and cost-effectiveness of interventions to prevent dementia. Other future research should focus on the ethical, legal and social aspects of dementia risk prediction as well as attempting to unpick the balance of costs and benefits to individuals and health systems.

It is essential that the momentum around personalised prevention does not overshadow what are likely to be the most effective approaches to preventing dementia, i.e. policies to promote brain health at all ages and stages of life.

However, the case for precision public health, in which data and analytics are integrated into existing models of health promotion to improve targeting and efficiency, is gathering pace. Matt Hancock, Secretary of State for Health & Social Care, promoted this approach in his recent speech Prevention is Better than Cure. The NHS Long Term Plan also encourages the use of digital tools to help the NHS become more proactive e.g. predicting and preventing events that would otherwise lead to hospital admission. It is essential that the momentum around personalised prevention does not overshadow what are likely to be the most effective approaches to preventing dementia, i.e. policies to promote brain health at all ages and stages of life.

More from us