Available online 19 June 2013
Various programs have been developed for caries risk assessment (CRA). Nevertheless, scientific evidence on their validity is lacking. This study aimed to compare the validity of 4 CRA programs (CAT, CAMBRA, Cariogram, and NUS-CRA) in predicting early childhood caries.
A total of 544 children aged 3 years underwent oral examination and biological tests (saliva flow rate, salivary buffering capacity and abundance of cariogenic bacteria mutans Streptococci and Lactobacilli). Their parents completed a questionnaire. Children's caries risk was predicted using the 4 study programs without biological tests (screening mode) and with biological tests (comprehensive mode). After 12 months, caries increment in 485 (89%) children was recorded and compared with the baseline risk predictions.
Reasoning-based programs (CAT and CAMBRA screening) had high sensitivity (≥93.8%) but low specificity (≦43.6%) in predicting caries in children. CAMBRA comprehensive assessment reached a better balance (sensitivity/specificity of 83.7%/62.9%). Algorithm-based programs (Cariogram and NUS-CRA) generated better predictions. The sensitivity/specificity of NUS-CRA screening and comprehensive models were 73.6%/84.7% and 78.1%/85.3%, respectively, higher than those of the Cariogram screening (62.9%/77.9%) and comprehensive assessment (64.6%/78.5%). NUS-CRA comprehensive model met the criteria for a useful CRA tool (sensitivity + specificity≥160%), while its screening model approached that target.
Our results supported algorithm-based approach of caries risk modelling and the usefulness of NUS-CRA in identifying children susceptible to caries.
This prospective study provided evidence for practitioners to choose tools for assessing children's caries risk, so that prevention measures can be tailored and treatment plan can be optimized.