Establishment and Validation of an Early Childhood Caries Prediction Model: a Multicentre Prospective Cohort Study

To establish a risk assessment model for early childhood caries and verify its credibility.

This 1-year multicentre cohort study comprised cohorts of 3-year-old children with and without caries recruited from kindergartens in Beijing, Shanghai, Sichuan, Hubei and Guangdong. Oral disease-related social and behavioural risk factors were collected from both children and their parents using questionnaires. Clinical examinations were performed according to World Health Organization standards, and the plaque's ability to produce acid, saliva pH value and saliva-buffering capacity were assessed. Logistic regression methods were employed to construct a risk-prediction model for caries incidence. Cross-validation methods were applied for the internal validation of the model, and the area under the curve (AUC) and calibration curves were used to evaluate the predictive performance of the constructed model.

A total of 4,381 children were included in the study and randomly allocated 1:1 to the development and validation sets. Using the new occurrence of decayed, missing and filled surfaces as the dependent variable, the predictive model demonstrated AUCs of 0.83 (95% confidence interval [CI] 0.81 to 0.84) in the development model and 0.80 (95% CI 0.78 to 0.82) in the validation model at 1 year. The training and validation sets exhibited high sensitivity (0.69 to 0.72) and specificity (0.73 to 0.81).

The predictive models demonstrated good discriminability with both the training and validation sets and were suitable for primary oral care clinicians and stomatological hospitals.