Chronic Disease Epidemiology Section

Diabetes is increasing and early detection and treatment of complications is key to preventing the disease from getting worse.  Researchers at City St George’s and Moorfields Eye Hospital, UCL, Homerton Healthcare NHS Foundation Trust, Kingston University, UK and University of Washington and National Eye Institute National Institutes of Health (NEI NIH), USA have received a Wellcome Collaborative Award.

The researchers have used automated retinal image analysis systems, which harness artificial intelligence, to analyse images of the back of the eye. This has allowed them to develop a model that accurately predicts cardiovascular disease from retinal images in community settings (which performs as well as established risk prediction models), but without the need for a blood test or blood pressure measurement.  UK people with diabetes are offered annual eye screening which involves taking a digital image of the retina.

The researchers will apply these methods to images from one of the largest Diabetic Eye Screening Programmes in the country, to see if retinal image feature assessment in combination with measures of diabetic control can predict complications for an individual better, to allow optimised review intervals and targeted prevention.

The intersection of tuberculosis (TB) disease with non-communicable diseases (NCDs), including diabetes mellitus, pulmonary disease, and cardiovascular disease, has emerged as a critical public health concern. Rapidly expanding NCD epidemics threaten TB control in low- and middle-income countries, where preventing and treating TB disease remains a great burden.

However, to date, the notion that TB disease may increase the risk of chronic NCDs has not been well explored. Using two large healthcare databases, we are examining the extent that TB increases the risk of post-TB diabetes, pulmonary disease, and cardiovascular disease.

Funder NIH R21 award Dec 2020-Nov 2023.

Investigators: Julia Critchley, Iain Carey, Elizabeth Limb, Derek Cook, Tess Harris, Steve DeWilde,

Collaborators: Steve DeWilde, Umar Chaudhry

Dr Hudda and colleagues have developed a novel prediction algorithm which accurately predicts lean and fat mass in children and adolescents using only readily available information on weight, height, age, sex and ethnicity (Hudda et al. BMJ 2019). This algorithm has been shown to provide fat mass estimates as accurate as those provided by established methods of bioimpedance and DXA (Hudda et al. Int J Obes 2020) and has now been extensively validated in 19 childhood settings outside of the UK (Hudda et al. BMJ 2022).

Ongoing research includes the use of this algorithm to assess the short- and long-term consequences of high childhood body fat levels as well as the development of body fat cut-offs which are indicative of the risk of chronic disease development.