Development and validation of a Diabetes Risk Score for screening undiagnosed diabetes in Sri Lanka (SLDRISK)

Abstract

Background

Screening for undiagnosed diabetes is not widely undertaken due to the high costs and invasiveness of blood sampling. Simple non-invasive tools to identify high risk individuals can facilitate screening. The main objectives of this study are to develop and validate a risk score for screening undiagnosed diabetes among Sri Lankan adults and to compare its performance with the Cambridge Risk Score (CRS), the Indian Diabetes Risk Score (IDRS) and three other Asian risk scores.

Methods

Data were available from a representative sample of 4276 adults without diagnosed diabetes. In a jack-knife approach two thirds of the sample was used for the development of the risk score and the remainder for the validation. Age, waist circumference, BMI, hypertension, balanitis or vulvitis, family history of diabetes, gestational diabetes, physical activity and osmotic symptoms were significantly associated with undiagnosed diabetes (age most to osmotic symptoms least). Individual scores were generated for these factors using the beta coefficient values obtained in multiple logistic regression. A cut-off value of sum = 31 was determined by ROC curve analysis.

Results

The area under the ROC curve of the risk score for prevalent diabetes was 0.78 (CI 0.73-0.82). In the sample 36.3 % were above the cut-off of 31. A risk score above 31 gave a sensitivity, specificity, positive predictive value and negative predictive value of 77.9, 65.6, 9.4 and 98.3 % respectively. For Sri Lankans the AUC for the CRS and IDRS were 0.72 and 0.66 repectively.

Conclusions

This simple non-invasive screening tool can identify 80 % of undiagnosed diabetes by selecting 40 % of Sri Lankan adults for confirmatory blood investigations.