Sociodemographic Predictors of Child-Parent Agreement on Oral Health

Oral health is vital for children's overall well-being. Parents play a critical role in shaping children's oral health through preventive care behaviors and treatment adherence. Although prior studies have focused on parents as reliable proxies for reporting their children's oral health, notable discrepancies between parental reports and children's perspectives reveal gaps in knowledge, attitudes, and practices. These differences are often shaped by parental sociodemographic factors, including age, gender, education, income, and ethnicity. This study aimed to evaluate child-parent agreement on self-reported oral health measures across four key domains: professional care, hygiene practices, periodontal health, and aesthetic concerns, and inform interventions to enhance communication and pediatric oral health.

This cross-sectional study included 306 child-parent dyads from 12 Los Angeles County schools. Surveys were adapted using the NIH PROMIS (Patient-Reported Outcome Measurement Information System) framework, and included seven shared questions to assess alignment in perceptions and behaviors. Bivariate and multivariate logistic regression models were conducted to analyze the associations between child-parent agreement and parental age, gender, income, education, and ethnicity. Machine learning models (Random Forest, XGBoost) were applied to evaluate the predictive performance of these variables.

Child-parent agreement on oral health behaviours and perceptions was significantly associated with sociodemographic factors. In multivariate analysis, non-Hispanic parents showed higher agreement with their children regarding cavity history (p = 0.003). Parents from households with an annual income over $80 000 showed higher agreement on cavity presence (p = 0.022), brushing before sleep (p = 0.007), and gum bleeding while brushing (p = 0.048), but lower agreement on flossing (p = 0.009). Female parents exhibited greater agreement with children on brushing behaviour (p = 0.046) and gum bleeding while brushing (p = 0.001). Parents with completed college education had significantly lower agreement regarding gum bleeding while flossing (p = 0.002), whereas high school graduates showed lower agreement across several behaviours. Machine learning models (Random Forest and XGBoost) demonstrated similar predictive performance in modelling agreement (accuracy range: 0.417-0.738; CI: 0.29-0.83), with DENTIST and TEETH2 yielding the highest predictive accuracies.

Parent sociodemographic factors are significantly associated with child-parent agreement on self-reported child oral health. Tailored interventions, culturally sensitive strategies, and improved survey tools are essential for enhancing the accuracy of outcome reporting.