Preliminary Modeling of Children’s Caries Risk in Medical Settings

Objectives: Development of a validated caries risk assessment tool is essential in targeting preventive/referral strategies in medical settings. Objective: Develop a caries risk tool for use in pediatric medical settings.
Methods: To date, 592 primary caregiver (PCG)-child pairs (out of 1326 recruited) have been followed for 36 months after their first birthday to assess the caries predictive validity of a 52-item questionnaire. Children received caries examinations using the ICDAS criteria at 12±3 months (baseline), 30±3 months (80% retention) and 48±3 months of age (ongoing final exams). In preliminary analyses, caries at age 4 (dmft; d=ICDAS>3) was assessed and tested for associations with questionnaire items using generalized estimating equation models applied to logistic regression. Multivariable analysis used stepwise model selection, with a limit of 10 items selected.
Results: Children were 48% female, Hispanic-12%/white-44%/black-31%/other-2%/multi-racial-11%, 55% enrolled in Medicaid, 94% living in urban communities, and 69% had fluoride varnish during the 36-month study period. 24% of children had cavitated lesion experience by age 4. The accuracy of the preliminary multivariable models for caries prediction varied (c-statistic=0.79 using only baseline responses, 0.85 if using responses from all visits). The preliminary model using baseline responses included the following significant (p<0.001) variables (odds ratio): child participating in public assistance programs (2.4), being Hispanic (1.5), going to sleep daily while nursing or drinking something other than water (1.4), PCG cleaning the pacifier with juice/soda/honey or sweet drink (3.7), sharing/tasting food daily with child using same spoon/fork/glass (2.4), PCG brushing their teeth less than daily (1.4), PCG’s gums bleeding daily when brushing (2.2), PCG having cavities/fillings/extractions in last 2 years (2.3), having no adult employed in the household (1.6), and PCG never getting dental check-ups (1.8).
Conclusions: Preliminary 10-item models using partial final caries exam data show good agreement with cavitated caries lesion experience and may be useful for identifying high caries risk children.