A Predictive Model for Primary Care Providers to Identify Children at Greatest Risk for Early Childhood Caries.

Purpose: The purpose of this study was to create an early childhood caries (ECC) risk-screening tool that fits into the primary care provider (PCP) well-child workflow. Methods: Integrated health records were employed to develop a predictive model for infants/toddlers at ECC risk; 2,009 patients with 12-, 15-, or 18-month well-child visits and at least one dental visit were used to develop a predictive model for ECC risk at the first dental visit. Independent model validation used 880 18- to 48-month-olds at their first dental appointment after at least one well-child visit. Results: Age at the first dental visit strongly predicted caries risk (odds ratio for one-year increase in age equals 2.11; 95 percent confidence interval equals 1.80 to 2.47). Three factors predicted high-caries risk: breast feeding status, preferred language not English, and no-show rates for pediatric clinic visits greater than 20 percent. All three non-age risk factors in well-child exams prior to 18 months predicted 42 percent probability of having caries if present for the first dental visit at 18 months. If that child was not seen until four years of age for the first dental visit, the probability of high caries risk increased to 83 percent. Model performance for independent validation was very close to expected performance. Conclusions: Existing clinical documentation plus a validated predictive model enables an effective caries risk assessment within well-child visits.