Methods Two consecutive population-based diabetes surveys among Chinese adults aged 20-74 years were conducted in 2002 (n = 1986) and 2006 (n = 4336). Demographic and anthropometric measures were collected following similar procedures. Standard 2-h 75-g oral glucose tolerance tests (OGTTs) were performed to diagnose diabetes in both surveys. Fasting capillary plasma glucose (FCG) and glycated haemoglobin (HbA1c) were also measured together with the OGTTs on the same day of the 2006 survey. Beta coefficients estimated using logistic regression analysis derived from data of the 2002 survey were used to develop the risk assessment algorithm. The performance of the algorithm was validated in the study population of the 2006 survey.
Results Of all the variables tested, waist circumference, age and family history of diabetes were significant predictors of diabetes and were used to construct the risk assessment score. The score, ranging from 3 to 32, performed well when applied to the study population of the 2006 survey. The area under the receiver operating characteristic curve was 67.3% (95% CI, 64.9-69.7%) for the score, while it was 76.3% (73.5-79.0%) for FCG alone and 67.8% (64.9-70.8%) for HbA1c alone. At a cut-off point of 14, the sensitivity and specificity of the risk score were 84.2% (81.0-87.5%) and 39.8% (38.2-41.3%).
Conclusions The risk score based on age, waist circumference and family history of diabetes is efficient as a layperson-oriented diabetes screening tool for health promotion and for population-based screening programmes.