Clinical utility of ethnic-based diabetes risk scores for predicting undiagnosed type 2 diabetes mellitus among Chinese patients with impaired fasting glucose in primary care setting in Hong Kong

Abstract

Context: Risk stratification based on the presence of concurrent risk factors deems a cost-effective approach to identify those patients with impaired fasting glucose (IFG) at highest risk of diabetes to undergo further confirmatory testing in primary care setting. At present, two non-laboratory-based risk assessment algorithms for diabetes have been developed for Chinese. Objective: To evaluate the performance of Qingdao Diabetes Prevention Program Risk Score and New Chinese Diabetes Risk Score in predicting the presence of diabetes diagnosed by OGTT among Chinese patients with IFG. Design: Cross-sectional study Setting: Two primary care out-patient clinics in Hong Kong Patients: 464 Chinese patients with IFG at baseline (i.e. fasting blood glucose level between 5.6-6.9mmol/L documented in their record within 18 months before date of attendance, without known diagnosis of diabetes and not taking any anti-diabetic medications) Instruments: Two risk algorithms, Qingdao Diabetes Prevention Program Risk Score (Gao, 2010) and New Chinese Diabetes Risk Score (Zhou, 2013) Outcome measures: Diagnostic accuracy as assessed by the area under the receiver operating characteristics curve (AUC), and the sensitivity and specificity at the recommended optimal cutoff point. Results: Overall, 94 (20.2%) patients had undiagnosed diabetes. The AUC of the two above risk scores were 0.544 and 0.521 respectively, indicating the poor diagnostic accuracy for their use among our IFG patients to predict the presence of diabetes diagnosed by OGTT. At the optimal cutoff point, the sensitivity/specificity of risk scores for predicting undiagnosed DM was 91.5%/9.7% and 100%/3.0%, respectively. Conclusion: Although these two risk scores were derived from Chinese populations, they seemed more applicable to the general population rather than patients with IFG. A different algorithm should be developed to aid identification of those IFG patients with highest diabetic risk for further testing and to avoid unnecessary OGTT for the lower risk group.