Risk prediction tools (PREDICT - 2)

Globally, 382 million people have diabetes. Of these, 175 million are unaware of their condition and over 80% are living in low- and middle-income countries. One important strategy for tackling the diabetes burden is to screen for undiagnosed diabetes and for the future risk of developing diabetes.

Diabetes risk scores are a cheap and simple way of assessing an individual’s risk of having undiagnosed diabetes and their risk of future diabetes. With limited resources in low- and middle-income countries, diabetes risk scores can be a simple and cost effective way to identify people with undiagnosed diabetes or at risk of developing diabetes. However, most currently available diabetes risk scores only work well in populations in which the risk scores were developed. Many low- and middle-income countries do not have the data required to develop diabetes risk prediction scores for their populations.

The risk prediction tools for identifying people at high risk of developing type 2 diabetes (PREDICT-2) project, an initiative of the International Diabetes Federation, has been formed to establish and validate a methodology for adapting diabetes risk prediction scores for populations with locally available demographic data. This will allow countries without longitudinal data to develop their own country specific diabetes prediction score based on a set of instructions from PREDICT-2 and local diabetes risk factor variables that are easily obtainable within their countries. The PREDICT-2 dataset currently includes eight longitudinal and 14 cross-sectional studies from ten countries. Caucasians comprise 60% of participants and 75% of the remaining participants are mostly migrants or descendents of migrants. For this project to be globally relevant, PREDICT-2 will require data from other ethnic populations to modify, develop and validate risk scores that will be suitable for use in most populations.

The PREDICT-2 analysis team is, therefore, calling for researchers to assist with the project through the contribution of datasets. Researchers interested in the PREDICT-2 project should contact Dr. Crystal Lee at crystal.lee@sydney.edu.au.