The objective of this research was to establish and validate a predictive model for the early identification of severe caries among 12-year-olds residing in Guizhou Province, China. A cross-sectional study, including oral examination and questionnaire, was conducted involving 2,437 12-year-olds in 33 secondary schools across 11 districts (counties) in 9 cities. The participants were randomly allocated to the training and validation set in a 6:4 ratio. Severe caries was defined as DMFT ≥ 3. Crucial variables for nomogram development were determined by multivariate logistic regression. Performance evaluation of the predictive model included the area under the receiver operating characteristic curve (AUC), confusion matrix, calibration curve, and decision curve analysis (DCA). Meanwhile, the model was further evaluated using the validation set. Among the 2,437 children, 384 (15.8%) were found to have severe caries. Significant risk factors included ‘female’ (odds ratio OR = 2.319; 95% confidence interval CI: 1.825–2.960), ‘ethnic minorities’ (OR = 1.708; 95% CI: 1.337–2.184), ‘Class III economic region’ (OR = 1.992; 95% CI: 1.487–2.690), ‘only child’ (OR = 1.993; 95% CI: 1.438–2.738), ‘self-evaluation of dental and oral conditions (very poor)’ (OR = 3.097; 95% CI: 1.724–5.419), and ‘dentist visit’ (OR = 1.746; 95% CI: 1.375–2.213). The AUCs for the training and validation sets were 0.711 (95% CI: 0.676–0.745) and 0.701 (95% CI: 0.657–0.744), indicating good discriminatory ability. The calibration curves indicated a high level of concordance between predicted risks and actual incidence rates. DCA demonstrated a net clinical benefit of the predictive model across threshold probabilities ranging from 6% to 71% in the training set and 5% to 83% in the validation set. By utilizing easily accessible variables, this model provides a simple and practical approach for the preliminary screening of severe caries among children in resource-limited areas.
Xu et al. (Mon,) studied this question.