Abstract

This study aims to develop a risk score to identify people at high risk for type 2 diabetes mellitus (T2DM) in the population of Chonburi, Thailand, and to compare this risk score with 2 previous predictive risk scores for T2DM. Data from 7,284 individuals aged ≥ 20 years were collected from the Thai population, using a cross-sectional analytical study method. A screening algorithm was developed based on the first half and validated in the second half of the study population. Logistic regression analysis was used to determine the risk factors for T2DM, by performing a predictive model in which only significant factors were included. Afterwards, our predictive model was compared with the other 2 predictive models, where risk scores were derived from Thai adults. Our results showed that significant predictive variables were age, BMI, hypertension, history of diabetes in parents or siblings, and marital status. A cutoff score of 9 out of 17 produced the optimal sums of sensitivity (74%) and specificity (97%). The area under the receiver-operating characteristics curve (AUC) was 0.969. Our predictive model had a higher AUC when compared to the other 2 models. When the risk score was applied, the predictive model selected 425 subjects who should undergo further testing for diagnosing diabetes and 3,259 subjects who should not. A simple T2DM risk score, based on a set of variables, can be used for the investigation of early intervention to delay or prevent T2DM in Thailand.