Identifying predictors of early childhood caries among Australian children using sequential modelling: Findings from the VicGen birth cohort study

Objectives: The aim of this birth cohort study was to identify concurrent associations between early childhood caries and putative risk and protective factors. Methods: Data were collected in seven waves over five years. The study outcome measure, d(3-6) mfs, was modelled in a set of sequential negative binomial regressions that introduced the variables in steps starting from health determinants most distal to the child and ending with the more proximal ones. The goodness of fit of each model at each step was tested using the quasi-likelihood under independence model criterion (QIC). A final model included all significant factors identified in the sequential modelling. Bacterial composition of the child's saliva was determined by 16S RNA gene sequencing. Results: Overall, 467 children (48.6 % female) participated, of whom 419 (89.7 %) had at least one follow-up visit after baseline. Of the 419 children included in the analyses, 133 (31.7 %) had their saliva samples sequenced for microbiomic determination. Independent protectors of surface cavitation included water fluoridation, and older age of mothers. Risk for d(3-6) mfs was significantly higher among children whose mothers were current smokers (IRR 3.29, 95 % CI 1.09-9.88, p = 0.034), children who went to bed with a bottle (IRR 2.67, 95-6.88, p = 0.041) and whose saliva sample sequencing over time showed higher percentages of Streptococcus mutans (IRR 1.39, 95 % CI 1.11-1.74, p = 0.005). Model fit was mostly improved by child's proximal variables. Household and mother covariates did not substantially improve model fit. Conclusion: This analysis highlights the relevance and importance of child-proximal risk factors in childhood dental cavitation. Clinical significance: The study findings inform clinical decision making for the management of early childhood caries at both the individual and population level. At an individual and family level these risk factors should be incorporated into caries risk assessment tools for more precise identification of risk and evidence-informed interventions by health professionals.