Sage Journals: Discover world-class research

Abstract

Objectives:

Socioeconomic status (SES) has a significant effect on the burden of early childhood caries (ECC), yet addressing SES disparities remains challenging. This study aimed to identify and quantify the most impactful mediator linking SES effect to the occurrence of ECC using advanced causal mediation analysis, to inform targeted interventions that reduce SES-related disparities in ECC.

Methods:

Data were drawn from the Study of Mothers’ and Infants’ Life Events, a cohort of 2,182 mother–child dyads recruited from Adelaide’s 3 largest public hospitals (2013–2014). Baseline questionnaires captured family SES, while ECC clinical indicators were assessed at age 5 y. Three mediation pathways linking SES and ECC were examined including dietary intake represented by free sugar intake (FSI); oral hygiene practices, including toothbrushing habits and plaque presence; and dental visiting patterns. Mediation effects were quantified as natural indirect effects (NIE) using causal mediation analysis based on the counterfactual framework, with validation via 5-fold cross-validation to ensure robustness.

Results:

FSI was the only pathway with a significant mediation effect. Each 1-standard-deviation increase or decrease in SES was associated with a corresponding 6% reduction or increase in ECC risk at age 5 y through the mediating effect of FSI at age 2 y. The NIE of FSI accounted for 52% of SES’s total effect on ECC. In contrast, oral hygiene and dental visiting patterns showed no significant mediation effects.

Conclusions:

Reducing early childhood FSI could mitigate half of SES-related disparities in ECC. Targeted interventions focusing on FSI reduction hold promise for lowering ECC risk, with future research needed to evaluate their effectiveness.

Knowledge Transfer Statement:

Clinicians and policymakers should prioritize nutrition education and sugar reduction initiatives as key components of early prevention in pediatric dentistry, especially for low-SES populations. Incorporating FSI screening into routine visits enables medical and dental practitioners to educate caregivers on the impact of free sugars on dental health.

Keywords

causal mediation analysis (CMA)socioeconomic factors dental caries free sugar intake toothbrushing dental visit

Introduction

Early childhood caries (ECC) is defined as the presence of 1 or more carious lesions, white spot lesions, tooth loss due to caries, or filled surfaces in the primary teeth of a child younger than 6 y (World Health Organization [WHO] 2019). A systematic review found that nearly half of children worldwide are affected by ECC, making it a global public health challenge (Uribe et al. 2021). In Australia, the National Child Oral Health Study reported no significant improvement in ECC rates since 2000 (Do and Spencer 2016).

Socioeconomic status (SES), defined as one’s access to social and economic resources (Antonoplis 2023), is widely recognized as a root cause of ECC, influencing risk at the individual, family, and community levels (Yousaf et al. 2022). In the Australian context, SES-related disparities are particularly pronounced. For example, a large study in Victoria found that among 144,938 children younger than 6 y attending public dental services (2009–2019), Indigenous children, those from non–English-speaking households, and those from low-income families experienced a significantly earlier onset of caries (Singh et al. 2025).

While Do et al. (2023) and others have highlighted structural, policy, and systemic contributors to oral health inequalities, our study focuses specifically on the pathways through which family- and community-level SES factors influence ECC outcomes, partly due to the nature of our available data and the aim to explore modifiable, proximal influences on ECC.

The dietary pathway reflects broader structural conditions in which families with limited financial resources may face barriers to accessing fresh, nutrient-rich foods, often due to cost, availability, or time constraints. These conditions can increase reliance on more affordable, energy-dense options that are higher in free sugars, key risk factors for dental caries (Chi et al. 2015).

The oral hygiene pathway acknowledges that daily routines such as twice-daily toothbrushing (TB) can be affected by competing demands, limited oral health education, and broader caregiving challenges, factors often shaped by socioeconomic context (Kyan et al. 2022).

The dental care pathway highlights that families facing financial or structural barriers (e.g., transportation, appointment availability, or past negative experiences with health systems) may have reduced access to preventive dental visits (Harris et al. 2017), although the specific impact of visit frequency on ECC remains debated (Amarasena et al. 2023).

However, studies exploring the potential mediating effects between SES and ECC are limited and have predominantly used traditional mediation analyses (TMA) (Dao, Do, Stormon, Dhanapriyanka, et al. 2025). TMA typically relies on conventional regression models to estimate indirect effects (Rijnhart et al. 2021). However, TMA assumes no confounding and linear relationships between variables, limiting its ability to capture complex interactions (Rijnhart et al. 2021).

Causal mediation analysis (CMA) has gained increasing popularity over the past decade, as it leverages statistical methods by applying a counterfactual framework, rigorously accounting for confounding and effectively addressing nonlinearities (Rijnhart et al. 2021). CMA offers a more nuanced understanding of the pathway and allows for unbiased quantification of the extent to which a particular exposure influences an outcome via mediators (Rijnhart et al. 2021). Despite these advantages, CMA has been rarely applied in the dental field to inform the most influential mediator for ECC-targeted interventions (Abanto et al. 2023).

This study used CMA to test the hypotheses that SES contributes to the occurrence of ECC through 3 key mediators: (1) dietary patterns, (2) oral hygiene patterns, and (3) dental visiting patterns. The goal was to identify the most influential mediation effect linking SES to ECC to guide targeted interventions aimed at reducing SES-related disparities in ECC.

Methods

Study Design and Data Source

The study was an observational study, applying CMA and adhering to the STROBE guidelines. It examined the effect of SES as the exposure variable to predict ECC as the outcome while considering the mediating role of the 3 groups of mediators. Data were collected from the Study of Mothers’ and Infants’ Life Events Affecting Oral Health (SMILE), a birth cohort study, funded by NHMRC Project grant 1046219 (2013–2016). The study was conducted with the approval of ethical research committees (Do et al. 2014).

Details of the SMILE participants’ recruitment are described elsewhere (Do et al. 2020). In brief, all newborns who were born and alived at the 3 main public hospital in Adelaide, South Australia, between July 2013 and August 2014 were eligible for inclusion, regardless of birth weight and gestational age. A total of 2,182 mother–infant dyads were recruited within the first 48 h after delivery and were followed up at 3 and 6 mo and at ages 1, 2, and 5 y. Written informed consent was obtained at each stage. For the purpose of this study, data on baseline SES, dietary intake, and oral hygiene practices at age 2 y as well as dental visit patterns during the first 5 y were assessed. ECC and plaque presence were clinically examined at ages 2 and 5 y.

Measurements

Exposure: SES

Although SES reflects one’s access to social, economic, and cultural resources, it is shaped by broader structural forces, including racialized systems of inequality (Williams et al. 2019). Understanding health disparities such as ECC thus requires an intersectional lens that considers how SES intersects with race, ethnicity, and place (O’Brien et al. 2020).

The methodology for creating the SES composite to be exposure in this study is detailed elsewhere (Dao, Do, Stormon, Nguyen, et al. 2025). In brief, with the intersectional lens, at baseline, SMILE included 21 SES-related items adapted from the Australian National Child Oral Health Study (Do et al. 2016), covering conventional SES indicators, household income, parental education, employment, and occupation, as well as markers of racialization, such as Aboriginal and Torres Strait Islander status, English as a first language, and parental country of birth. Of these, 11 items were grouped into 7 child/family indicators and 10 into 5 parental indicators, yielding 12 SES variables used to construct the exposure (Appendix Table 1 and Table 2).

In total, 5 SES composites were developed from 12 indicators using 3 approaches: decision tree analysis (DTA), principal component analysis (PCA), and a hybrid DTA–PCA method. DTA excluded 2 indicators, subjective financial attitude and number of parents, as irrelevant. The most robust composite emerged from the hybrid approach, incorporating 5 indicators (family education, employment, occupation, income, and health insurance) that significantly loaded on the first PCA component. Each was weighted by its factor loading to construct a comprehensive, multidimensional SES measure.

Outcome: ECC

ECC was evaluated through oral epidemiological examinations conducted by trained dental examiners following the standardized protocol developed for the 2012–2014 National Child Oral Health Study (Do and Spencer 2016). The clinical examinations were conducted when the children turned 2 and 5 y. For each child, any tooth surface affected by decay (cavitated or noncavitated), missing due to caries, or filled was recorded and assigned a single score. These were summed to generate a dmfs score (decayed, missing, and filled surfaces) in accordance with established methodology (Do and Spencer 2016).

Mediators

Three mediator groups were examined: dietary patterns, represented by free sugar intake (FSI); oral hygiene patterns, measured through TB practices and plaque presence; and dental visit patterns, assessed based on the age at first dental visit and the total number of visits during the first 5 y.

Dietary Patterns

FSI is defined as adding any type of sugar in any form to food and beverages. These include sugar-sweetened beverages (fruit-based and milk-based sweetened drinks and 100% fruit juices) as well as confectionery, cakes, biscuits, sweetened cereals, sweet desserts, sucrose, honey, syrups, and preserves. This does not include natural sugars found in fruits, vegetables, or milk (WHO 2015).

At age 1 y, dietary intake was assessed through a single 24-h recall and 2 additional nonconsecutive days of intake recorded in a 3-d food diary. At age 2 y, the 3-d food diary was replaced by a 99-item Food Frequency Questionnaire (SMILE-FFQ), with its repeatability and validity assessed against 24-h recalls. At age 5 y, the 99-item FFQ was administered again, with modifications for age appropriateness. FSI quantity was estimated from 3 rounds of dietary assessments, converted into grams per day, and analyzed as a continuous variable (Do et al. 2020). Given FSI’s highly right-skewed distribution, outliers above the 99th percentile were excluded. The remaining data were log-transformed to approximate a normal distribution for modeling purposes.

Oral Hygiene Patterns

TB practices were reported by parents or caregivers, detailing how often they cleaned their child’s gums at age 1 y and how frequently they brushed their child’s teeth daily at ages 2 and 5 y. Information on toothpaste use, including type and brand, was collected to classify it as fluoride or nonfluoride. The TB variable was generated as a binary variable, with standard TB defined as TB with toothpaste at least twice per day at age 1 y and brushing at least twice daily with fluoridated toothpaste at ages 2 and 5 y; otherwise, it was classified as unstandard TB.

The presence of plaque was assessed during the oral clinical examinations at ages 2 and 5 y and was categorized into 4 levels: 0, no plaque; 1, thin plaque detected with dental probing; 2, visible plaque without probing; and 3, abundant plaque (Do et al. 2020). Given that plaque serves as a proxy for oral hygiene behavior, it was recoded into a binary variable. The first 2 levels, described as “invisible plaque to the naked eye,” were interpreted as indicative of adequate TB, while the latter 2 levels, “visible plaque,” were taken to reflect poor oral hygiene practices (Do et al. 2020).

Dental Visit Patterns

Parents were asked to list each dental visit during their child’s first 5 years. The child’s age at their first dental visit was recorded in months and used as the “age at first visit” variable (FDV), while the total number of dental visits in the first 5 y was summed and used as the “total number of dental visits in the first 5 y” variable (DV during first 5 y ) (Do et al. 2020).

Confounders

Assumed causal model and assumptions

Figure 1 describes the study direct acyclic graphic, visualizing the pathway from SES to ECC at age 5 y via mediators, split into 2 parts. The first part represents the effect of the SES on the mediator corresponding to the “mediation model,” while the second part shows the effect of the mediator on ECC, adjusted for covariate SES, corresponding to the “outcome model.” Both the mediation and outcome models were examined while controlling potential confounders, which were identified based on the disjunctive cause criterion (VanderWeele 2019).

Figure 1.

Direct acyclic graphic (DAG) visualizing pathway from socioeconomic status to early childhood caries at age 5 y via mediators.

In the mediation model, earlier measurements of the mediator were considered confounders (Keogh et al. 2018). Specifically, if the mediator was FSI at age 2 y, then FSI at age 1 y was a confounder. For TB and plaque presence as mediators, earlier measures of each were treated as confounders. If the age of the first visit was the mediator, the total number of visits in the first 5 y was considered a confounder, and vice versa. For the outcome model, the confounders included FSI at ages 1, 2, and 5 y; TB at ages 1, 2, and 5 y; plaque at ages 2 and 5 y; FDV; DV during the first 5 y; and ECC at age 2 y (Wang et al. 2017).

Statistical Analysis

For the CMA, the outcome was modeled as either binary, to assess whether SES influences the occurrence of ECC via mediators, or as count data to capture both occurrence and severity. To address the research question, whether SES contributes to the occurrence of ECC via mediators, the dmfs score was dichotomized as no caries (dmfs = 0) versus any caries experience (dmfs > 0). In addition, a sensitivity analysis was conducted using a zero-inflated count model for ECC.

To assess potential mediation, we used the mediate command in Stata 18, based on the potential outcomes framework (StataCorp 2024) (Figure 2). Sequential regression models were fitted: a probit model for the binary outcome (ECC at age 5 y) and a linear regression for the continuous mediator (log-transformed FSI at age 2 y). This approach estimates the natural direct effect (NDE), natural indirect effect (NIE), and the proportion mediated, assuming no unmeasured confounding and no exposure–mediator interaction.

Figure 2.

Counterfactual framework.

Three regressions, the mediation, the outcome, and the counterfactual models, were fit using the completed observations (Rijnhart et al. 2021). The mediator and outcome models corresponded to equations 1 and 2, while the counterfactual models serve equation 3a and 4a in which exposure and mediator were held at a defined threshold (Rijnhart et al. 2021). All models included the child’s age, mother’s age, and child’s sex as preexposures in addition to the identified potential confounders to satisfy CMA’s sequential ignorability assumption (Imai et al. 2010). Given 10 putative mediators, each of the 3 models was 10 times fit, respectively. The presence of a mediator was confirmed if all the models of equations 1, 2, and 4a met the statistical threshold of P ≤ 0.05.

Specifically, the NDE was calculated by estimating the difference or risk ratio between ECC outcomes when exposure (SES) changes from the exposed state (X = 1) to the unexposed state (X = 0), while the mediator (M) is held at the value associated with X = 0 (equations 3a and 3b, respectively). The NIE is the difference or ratio between ECC outcomes when X is held at the exposed state (X = 1), while the mediator (M) shifts from the value under exposure (1) to the value under unexposure (0) (equations 4a and 4b, respectively). Finally, the NIE proportion was calculated by dividing the NIE by the total effect of SES on ECC at age 5 y, which is the sum of NIE and NDE (equation 5):

M = i_{2} + a X + d C + ε_{2}

(1)

Y = i_{3} + c^{'} X + b M + d C + ε_{3}

(2)

\begin{array}{l} N D E = Y [X (1), M (0)] \\ - Y [X (0), M (0)] \end{array}

(3a)

N D E^{R R} = \frac{Y [X (1), M (0)]}{Y [X (0), M (0)]}

(3b)

\begin{array}{l} N I E = Y [X (1), M (1)] \\ - Y [X (1), M (0)] \end{array}

(4a)

N I E^{R R} = \frac{Y [X (1), M (1)]}{Y [X (1), M (0)]}

(4b)

NIE proportion = \frac{N I E}{N I E + N D E}

(5)

Notations used across the equations: X: exposure; M: mediator; Y: outcome; C: confounder; i: intercept; ε residual; c: total effect of X on Y; a: effects of X on M; b: indirect effects of X on Y through M; c′: direct effect of X on Y, controlling for M; d: effects of C on Y and M; NDE: natural direct effect; NIE: natural indirect effect; RR: risk ratio.

Validation analysis

To examine the consistency and generalizability of the estimated NIE, 10-fold cross-validation was conducted (Kuhn and Johnson 2013) after CMA was implemented in the entire data set. Specifically, the SMILE data set was randomly divided into approximately 10 equal folds in which each fold is used once as the test set, while the remaining 9 folds were used for training. For each iteration, the model is trained on 9 folds and tested on the remaining fold.

Results

Table 1 presents the distributions of study variables across the waves. Of 2,182 observations at baseline, the response rates decreased across waves since participants had the right to decline. The distribution of the mother’s age and child’s sex showed minimal differences among those remaining across years 1, 2, and 5. By contrast, significant differences were observed in SES among those remaining across 3 waves.

Table 1.

Distributions of Study Variables.

Variables and Type of Statistics	Year 1	Year 2	Year 5
Child’s sex
n	1,394	1,007	830
Male (%)	53.7^a	53.5^b	54.2
Mother’s age
n	1,300	996	830
Mean of SES score	30.4	30.6^b	30.7^c
SD (min; max)	5.1 (15; 5)	4.94 (15; 5)	5.00 (15; 5)
SES
n	1,878	955	808
Mean of SES score	0.2^a	0.5^b	0.6^c
SD (min; max)	1.45 (−4.2; 3.2)	1.3 (−3.9; 3.2)	1.29 (−4.1; 3.1)
FSI
n	1,170	1,057	733
Mean of FSI in g/d	8.8	32.2	44.2
Median (min; max)	5.8 (0.0; 183.4)	23.0 (0.3; 540.5)	31.9 (2.1; 560.0)
Toothbrushing
n	1,120	1,157	817
% of standard toothbrushing	19.8	30.5	52.0
Age of first dental visit asked at age 5 y
n	NA	NA	546
Mean age in months at first dental visit	NA	NA	31.3
Median (min; max)	NA	NA	24 (2; 7)
% of children with first dental visit at 12 mo			6.4
Number of dental visits, first 5 y
n	NA	NA	82
Mean	NA	NA	1.8
Median (min; max)	NA	NA	1.0 (0; 9.0)
Plaque index
n	NA	1,041	866
% of children with no observed or thin plaque identified by probing	NA	97.0	70.0
% of children with some visible plaque or abundance of plaque		3.0	30.0
Early childhood caries
n	NA	1,040	879
Mean decayed, missing, and filled tooth surfaces	NA	0.5	1.7
Median (min; max)	NA	0.0 (0;072)	0 (0.0; 60)
%	NA	10.6	32.5

FSI, free sugar intake; n, number of observations; NA, not applicable; SES, socioeconomic status.

Distribution at baseline.

Distribution at baseline for participants remaining until year 2.

Distribution at baseline for participants remaining until year 5.

The SES distribution exhibited slight left skewness, with mean scores of participants remaining at years 2 and 5 similar but much higher than those interviewed at year 1 (Appendix Fig 1.1). The distribution of ECC at ages 2 and 5 y was highly right skewed (Appendix Fig 1.2 and 1.3), with median scores mostly at 0, but the prevalence increased from 10.6% at age 2 y to 32.5% at age 5 y.

The distributions of FSI were highly right skewed across 3 y (Appendix Fig 2.1; 2.2; 2.3), approaching the normal after being log-transformed (Appendix Fig 2.4; 2.5; 2.6). The FSI average quantity increased across 3 surveys from 8.77 mg/d at age 1 y to 44.24 mg/d at age 5 y. The percentage of children having standard TB increased from 20.00% at age 1 y to 31.00% at age 2 y and reached 52.00% at age 5 y. Plaque presence showed a significant increase in the level of visible plaque, from 2.98% in year 2 to 30.03% in year 5. The average age for the first dental visit was 31 months, with an average of 1.77 visits by age 5 y.

Generally, SES and ECC at age 5 y showed a significant negative association (Appendix Fig. 3), and mediators (FSI at age 2 and 5 y) were significantly associated with both SES and ECC at age 5 y. Other mediators were significantly associated with only either SES or the ECC at age 5 y, with oral hygiene patterns showing the weakest association (Appendix Table 3). There were neither correlations between TB at age 1 y and plaque at age 2 y nor between TB at age 1 y and plaque at age 5 y (Appendix Fig 4.1, 4.4). In addition, the correlation between the child’s age at the first dental visit and the number of dental visits showed a negative association (Appendix Fig 5).

Table 2 shows the results of the CMA, with the first section displaying the results of the outcome, mediation, and counterfactual models across the 10 putative mediators. In the outcome models, 4 of the 10 mediators, FSI at age 2 y, plaque at age 5 y, FDV, and DV during the first 5 y, showed significant effects on ECC at age 5 y. In the mediation models, SES showed effects on only 2 mediators, FSI at ages 1 and 2 y. In the counterfactual models, only FSI at age 2 y demonstrated a significant indirect effect. Overall, only FSI at age 2 y consistently displayed a significant mediating role across all 3 models, linking SES to ECC at age 5 y. In the sensitivity analysis, where ECC was treated as a zero-inflated count outcome, the mediating effect of FSI at age 2 y was not statistically significant but remained in the same direction as in the binary outcome model.

Table 2.

Causal Mediation Analysis.

Socioeconomic Status Impact on Early Childhood Caries via 10 Putative Mediators.

	Outcome Model^a			Mediation Model^b			Counterfactual Models^c
	Outcome Model^a			Mediation Model^b			NIE			NDE
	Coef	95% CI		Coef	95% CI		Coef	95% CI		Coef	95% CI
FSI_1Y^d	0.32	−0.01	0.65	−0.11	−0.18	−0.05	0.00	−0.01	0.01	−0.02	−0.09	0.05
FSI_2Y^d	0.52	0.01	1.03	−0.11	−0.17	−0.06	−0.02	−0.03	−0.01	−0.02	−0.09	0.06
FSI_5Y^d	−0.32	−0.87	0.24	0.01	−0.03	0.05	0.00	0.00	0.00	−0.02	−0.09	0.05
TB_1Y	−0.42	−1.31	0.47	−0.11	−0.30	0.09	0.00	0.00	0.01	0.01	−0.07	0.08
TB_2Y	0.14	−0.58	0.86	0.03	−0.15	0.21	0.00	0.00	0.01	0.01	−0.07	0.08
TB_5Y	0.62	−0.03	1.27	0.03	−0.15	0.21	0.00	0.00	0.00	0.01	−0.07	0.08
Plaque_2Y	−0.16	−0.78	0.46	0.00	−0.13	0.13	0.00	0.00	0.00	0.01	−0.06	0.08
Plaque_5Y	1.02	0.48	1.55	−0.04	−0.12	0.04	−0.01	−0.01	0.00	0.01	−0.06	0.08
FDV	0.05	0.02	0.08	−0.77	−1.83	0.29	−0.01	−0.02	0.01	−0.01	−0.09	0.07
DV_F5Y	0.35	0.13	0.58	0.03	−0.02	0.09	0.01	0.00	0.03	0.01	−0.07	0.09

Socioeconomic Status Impact on Early Childhood Caries via Age 2 y Free Sugar Intake Estimated with Entire Dataset and 10-Fold Cross-Validation.

Model	n	Risk Ratio	95% CI		Proportion^e	95% CI
Entire data	433	0.94	0.90	0.99	0.52	−0.64	1.69
10-fold cross-validation
Average	389.7	0.95	0.90	0.99	0.55	−1.07	2.18

CI, confidence interval; Coef, effect of the regression models; DV_F5Y, number of dental visits during first 5 y of life; FDV, age in months of first dental visit; FSI_iY, free sugar intake at age i; NDE, natural direct effect; NIE, natural indirect effect; Plaque_iY, plaque index at age i; TB_iY: toothbrushing at age i.

Model of mediator predicting outcomes adjusted for SES and other covariates.

Model of exposure predicting mediators.

Model based on the outcome and mediation models, conditioning for exposure and mediator.

Log-transformed free sugar intake at age i.

Proportion of the indirect effect compared with the total effect = NIE/(NIE + NDE).

Bold figures indicate statistically significant effects.

The second section presents the effect of SES on ECC at age 5 y through the mediator FSI at age 2 y (Fig. 3), measured as the risk ratio and validated using 10-fold cross-validation (Appendix Table 4). In the CMA for the entire SMILE data set, the results indicate a statistically significant negative association between SES and ECC at age 5 y via FSI at age 2 y. Each unit deviation of SES above or below the mean was associated with a 6% decrease or increase, respectively, in the risk of ECC, mediated by FSI at age 2 y. This effect accounted for 52% of the total effect but was not statistically significant. The average estimates from 10-fold cross-validation were consistent with those obtained from the entire dataset.

Figure 3.

Effect of socioeconomic status on early childhood caries at age 5 y via free sugar intake at age 2 y.

Discussion

This study explored the indirect effects of SES on ECC through 10 putative mediators belonging to 3 main patterns: dietary intake, oral hygiene, and dental visits. Among these factors, only FSI at age 2 y consistently demonstrated a mediating effect of FSI at age 2 y on the occurrence of dental caries at age 5 y.

The findings underscore the importance of early childhood FSI as a mediator linking SES to ECC. This aligns with Abanto’s finding that sugar consumption at age 2 y mediates the association between prolonged breastfeeding and ECC (Abanto et al. 2023). The result also conforms with evidence indicating that excessive FSI in early childhood is a major cause for dental caries (Mahboobi et al. 2021), as children aged 1 and 2 y who consumed free sugar constituting more than 10% of their energy intake were nearly twice as likely to develop ECC compared with those consuming less than 5% (Devenish et al. 2020).

The results of the primary analysis, which focused on ECC occurrence (binary outcome), directly addressed the central research question. Sensitivity analyses using zero-inflated count models suggested that FSI may influence the onset of ECC but not its severity. These findings reinforce the conceptual justification for modeling ECC as a binary outcome in this study.

Importantly, the results highlight early-life sugar consumption as a key behavioral pathway through which SES may influence ECC. This is particularly important for Australia, where high levels of sugar consumption are common (Devenish et al. 2019) and a sugar tax has not yet been implemented (Allen and Allen 2020). Comprehensive sugar reduction initiatives are needed, including sugar-sweetened beverage tax, already implemented in more than 50 countries (Backholer et al. 2016); restrictions on the marketing and sale of sugary foods (Kraak et al. 2016); clear front-of-pack labeling requirements (Villalobos et al. 2020); and school nutrition standards (Reyes et al. 2020). These strategies also align with broader efforts to address commercial determinants of health (Chung et al. 2022).

While FSI at age 2 y consistently emerged as a mediator, other pathways may explain the remaining indirect effect of SES on ECC. Oral hygiene and dental visits did not mediate this association, likely due to measurement limitations and contextual constraints. Reported brushing practices correlated poorly with plaque levels, possibly due to unmeasured factors such as technique and duration (Hwang et al. 2023), obscuring potential mediation (Schuler et al. 2024). For dental visits, limited variability, and late timing, 94% of children had their first visit at about 31 months, creating a ceiling effect that hindered detection of mediation (Simkovic et al. 2019). These late, often reactive visits reflect structural barriers disproportionately affecting disadvantaged groups (Sanguida et al. 2019). The mediating role of dental visits may emerge only in settings with equitable access to timely, preventive care. Future research should explore additional mediators such as parental oral health literacy, dietary diversity, and access to preventive services.

This study has several strengths, including its cohort design, the application of the counterfactual framework, and the validation technique, all of which enhance the unbiased and generalizable estimation of the natural indirect effect. To our knowledge, this is only the second study, following Abanto’s, to employ an advanced causal inference technique to identify and quantify the mediator’s effects on dental caries (Abanto et al. 2023; Dao, Do, Stormon, Dhanapriyanka, et al. 2025). In addition, this study used SMILE data, which captured added sugar intake from all food sources, addressing a limitation of previous studies that relied solely on sugar-sweetened beverages (Yan et al. 2022). In addition,, while SES was measured as fixed information at baseline, other time-varying confounders were included, such as ECC at age 2 y, FSI, TB, plaque presence at multiple time points, age at first dental visit, and total dental visits by age 5 y. This approach allowed us to capture changes over time in these key variables.

The study has some limitations, including the reliance on recalled information of 3 dietary assessments, which may not fully represent the overall early childhood FSI pattern. In addition, the SMILE study is subject to attrition bias, with SES differences emerging across follow-ups; participants remaining at year 5 had notably higher SES than those at year 1 did. Given the higher attrition among participants with lower SES, it is possible that the true caries burden was underestimated, potentially leading to weaker observed associations. However, this was mitigated by recruiting exclusively from public hospitals, which serve more socioeconomically disadvantaged populations (Australian Institute of Health and Welfare 2025), and by oversampling mothers from the lower SES group (Do et al. 2020). These strategies enhance the sample’s representativeness, supported by comparable caries outcomes at age 5 y between SMILE and the broader South Australian population (dmfs: 1.34 vs. 1.40; caries prevalence: 23.4% vs. 25.3%) (Do and Spencer 2016). Furthermore, although key confounders were taken into account, unmeasured factors such as the child’s general health or variations in Community Water Fluoridation CWF coverage may still introduce bias. Given the high fluoridation coverage in Adelaide, this risk is likely reduced but not excluded.

While our findings emphasize early dietary intervention as a key strategy to mitigate SES-related disparities in ECC, we acknowledge that this recommendation must be interpreted within the broader context of structural inequality. SES is not a neutral or isolated construct; rather, it is often shaped by racialized systems of stratification that influence access to health-promoting resources from early life (Williams et al. 2019; O’Brien et al. 2020). Thus, although dietary guidance is crucial, it must be implemented alongside broader structural interventions that address the underlying social determinants of health, including racism, cultural marginalization, and systemic inequities in access to care. Future research and policy efforts should consider how these intersecting factors influence health behaviors and outcomes to design interventions that are not only effective but also equitable.

In conclusion, this study identifies early childhood FSI as an important mediator in the relationship between SES and ECC, emphasizing the importance of early dietary interventions targeting FSI as a primary strategy to mitigate SES-related disparities in ECC. To further inform policy, future studies can potentially investigate the effectiveness of interventions targeting FSI on dental caries in reducing SES-related ECC disparities.

Author Contributions

A.T.M. Dao, contributed to conception and design, data acquisition, analysis, and interpretation, drafted and critically revised manuscript; L.G. Do, D.H. Ha, contributed to conception and design, data acquisition and interpretation, critically revised the manuscript; N. Stormon, contributed to conception and design, data interpretation, critically revised the manuscript; H.V. Nguyen, contributed to conception and design, data analysis and interpretation, critically revised the manuscript. All authors gave final approval and agreed to be accountable for all aspects of the work.

Supplemental Material

sj-docx-1-jct-10.1177_23800844251365536 – Supplemental material for Key Mediators Reducing Socioeconomic Inequality in Early Childhood Caries

Supplemental material, sj-docx-1-jct-10.1177_23800844251365536 for Key Mediators Reducing Socioeconomic Inequality in Early Childhood Caries by A.T.M. Dao, L.G. Do, N. Stormon, H.V. Nguyen and D.H. Ha in JDR Clinical & Translational Research

Footnotes

Acknowledgements

We would like to express our gratitude to the SMILE research team for their efforts in securing the study grant and coordinating and implementing the surveys. We also greatly acknowledge the contribution of the South Australian Dental Service and Dr. Vu for allowing us to use their dental clinics to examine participants during phase 5 of the study. Our sincere thanks go to the SMILE participants for providing valuable information and taking part in the clinical examinations. In addition, we would like to thank the Oral Health Centre (OHC) at the University of Queensland for their technical support with this manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study used data from the Study of Mothers’ and Infants’ Life Events Affecting Oral Health (SMILE), which was supported by the NHMRC Project grant APP1161581. The authors declare no conflict of interest related to this article.

ORCID iDs

A.T.M. Dao

L.G. Do

N. Stormon

A supplemental appendix to this article is available online.

References

Abanto

Maruyama

Pinheiro

Matijasevich

Antunes

JLF

Bönecker

Cardoso

Grp

MI-BS

. 2023. Prolonged breastfeeding, sugar consumption and dental caries at 2 years of age: a birth cohort study. Community Dent Oral Epidemiol. 51(3):575–582.

Allen

WMK

Allen

. 2020. Should Australia tax sugar-sweetened beverages? J Paediatr Child Health. 56(1):8–15.

Amarasena

Luzzi

Brennan

2023. Effect of different frequencies of dental visits on dental caries and periodontal disease: a scoping review. Int J Environ Res Public Health. 20(19):6858. https://doi.org/10.3390/ijerph20196858.

Antonoplis

2023. Studying socioeconomic status: conceptual problems and an alternative path forward. Perspect Psychol Sci. 18(2):275–292.

Australian Institute of Health and Welfare. 2025. Patient demographics. Canberra: Australian Institute of Health and Welfare; [accessed 2025 May 23]. https://www.aihw.gov.au/hospitals/topics/admitted-patient-care/patient-demographics?utm_source=chatgpt.com.

Backholer

Sarink

Beauchamp

Keating

Loh

Ball

Martin

Peeters

2016. The impact of a tax on sugar-sweetened beverages according to socio-economic position: a systematic review of the evidence. Public Health Nutr. 19(17):3070–3084.

Chi

Dinh

da Fonseca

Scott

Carle

. 2015. Dietary research to reduce children’s oral health disparities: an exploratory cross-sectional analysis of socioeconomic status, food insecurity, and fast-food consumption. J Acad Nutr Diet. 115(10):1599–1604.

Chung

Westerman

Martin

Friel

2022. The commercial determinants of unhealthy diets. Public Health Res Pract. 32(3):3232221. https://doi.org/10.17061/phrp3232221.

Dao

ATM

Stormon

Dhanapriyanka

. 2025a. Causal analyses in longitudinal observational studies in oral health: a scoping review. Community Dent Oral Epidemiol. 53(1):7–16.

10.

Dao

ATM

Stormon

Nguyen

. 2025. Enhancing socioeconomic status prediction for cavities: a hybrid method. J Dent Res. 104(9):947–954.

11.

Devenish

Golley

Mukhtar

Begley

Scott

. 2019. Free sugars intake, sources and determinants of high consumption among Australian 2-year-olds in the SMILE cohort. Nutrients. 11(1):161. https://doi.org/10.3390/nu11010161.

12.

Devenish

Mukhtar

Begley

Spencer

Thomson

Scott

. 2020. Early childhood feeding practices and dental caries among Australian preschoolers. Am J Clin Nutr. 111(4):821–828.

13.

Bell

Devenish

Golley

Leary

Manton

Thomson

Scott

Spencer

. 2020. Study of mothers’ and infants’ life events affecting oral health (smile) birth cohort study: cohort profile. BMJ Open. 10(10):e041185. https://doi.org/10.1136/bmjopen-2020-041185.

14.

Scott

Thomson

et al. 2014. Common risk factor approach to address socioeconomic inequality in the oral health of preschool children – a prospective cohort study. BMC Public Health 14,429. https://doi.org/10.1186/1471-2458-14-429.

15.

Song

Spencer

. 2023. Socioecological determinants of child oral health—a scoping review. Community Dent Oral Epidemiol. 51(5):1024–1036.

16.

Spencer

. 2016. Oral health of Australian children: the National Child Oral Health Study 2012–14. Adelaide (Australia): University of Adelaide Press.

17.

Harris

Raison

Christian

Bakare

Okwundu

Burnside

2017. Interventions for improving adults’ use of primary oral health care services. Cochrane Database Syst Rev. 2017(8):CD012771. https://doi.org/10.1002/14651858.CD012771.

18.

Hwang

C-H

Song

H-J

Jung

M-J

Choi

Y-J

Hwang

. 2023. Effect of toothbrushing application for kids on dental plaque removal and interest in toothbrushing of preschool children. J Dent Hyg Sci. 23(3):208–215.

19.

Imai

Keele

Tingley

2010. A general approach to causal mediation analysis. Psychol Methods. 15(4):309–334.

20.

Keogh

Daniel

VanderWeele

Vansteelandt

2018. Analysis of longitudinal studies with repeated outcome measures: adjusting for time-dependent confounding using conventional methods. Am J Epidemiol. 187(5):1085–1092.

21.

Kraak

Vandevijvere

Sacks

Brinsden

Hawkes

Barquera

Lobstein

Swinburn

. 2016. Progress achieved in restricting the marketing of high-fat, sugary and salty food and beverage products to children. Bull World Health Organ. 94(7):540–548.

22.

Kyan

Takakura

Kamiya

Kinjo

Nakasone

2022. Socioeconomic inequalities in toothbrushing behaviours in young children: a children’s lifestyle survey in a representative population of a city, Okinawa prefecture, Japan. Eur Arch Paediatr Dent. 23(6):969–977.

23.

Mahboobi

Pakdaman

Yazdani

Azadbakht

Montazeri

2021. Dietary free sugar and dental caries in children: a systematic review on longitudinal studies. Health Promot Perspect. 11(3):271–280.

24.

Kuhn

Johnson

2013. Applied predictive modeling. New York: Springer.

25.

O’Brien

Neman

Seltzer

Evans

Venkataramani

2020. Structural racism, economic opportunity and racial health disparities: evidence from U.S. counties. SSM Popul Health. 11:100564. https://doi.org/10.1016/j.ssmph.2020.100564.

26.

Reyes

Smith Taillie

Popkin

Kanter

Vandevijvere

Corvalán

2020. Changes in the amount of nutrient of packaged foods and beverages after the initial implementation of the Chilean law of food labelling and advertising: a nonexperimental prospective study. PLoS Med. 17(7):e1003220. https://doi.org/10.1371/journal.pmed.1003220.

27.

Rijnhart

JJM

Lamp

Valente

MacKinnon

Twisk

JWR

Heymans

. 2021. Mediation analysis methods used in observational research: a scoping review and recommendations. BMC Med Res Methodol. 21(1):226. https://doi.org/10.1186/s12874-021-01426-3.

28.

Sanguida

Vinothini

Prathima

Santhadevy

Premlal

Kavitha

2019. Age and reasons for first dental visit and knowledge and attitude of parents toward dental procedures for Puducherry children aged 0-9 years. J Pharm Bioallied Sci. 11(suppl 2):S413–419.

29.

Schuler

Coffman

Stuart

Nguyen

Vegetabile

McCaffrey

. 2024. Practical challenges in mediation analysis: a guide for applied researchers. Health Serv Outcomes Res Methodol. 25(1):57–84.

30.

Simkovic

Traeuble

Hutson

. 2019. Robustness of statistical methods when measure is affected by ceiling and/or floor effect. PLoS One. 14(8):e0220889. https://doi.org/10.1371/journal.pone.0220889.

31.

Singh

Hegde

Silva

Whelan

Karahalios

Manton

Goldfeld

English

Dashper

2025. Social inequities in early childhood caries in Australia: a population-based study on statewide public dental services data. Int J Paediatr Dent. 35(1):186–193.

32.

StataCorp. 2024. Causal mediation analysis. College Station (TX): StataCorp; [accessed 2024 Oct 8]. https://www.Stata.Com/new-in-stata/causal-mediation-analysis/.

33.

Uribe

Innes

Maldupa

2021. The global prevalence of early childhood caries: a systematic review with meta-analysis using the WHO diagnostic criteria. Int J Paediatr Dent. 31(6):817–830.

34.

VanderWeele

TJ.

2019. Principles of confounder selection. Eur J Epidemiol. 34(3):211–219.

35.

Villalobos Dintrans

Rodriguez

Clingham-David

Pizarro

. 2020. Implementing a food labeling and marketing law in Chile. Health Syst Reform. 6(1):e1753159. https://doi.org/10.1080/23288604.2020.1753159.

36.

Wang

Cheng

Yuan

Hong

2017. Association between socio-economic status and dental caries in elderly people in Sichuan Province, China: a cross-sectional study. BMJ Open. 7(9):e016557. https://doi.org/10.1136/bmjopen-2017-016557.

37.

World Health Organization. 2015. Guideline: sugars intake for adults and children. Geneva (Switzerland): World Health Organization.

38.

World Health Organization. 2019. Ending childhood dental caries: WHO implementation manual. Geneva (Switzerland): World Health Organization.

39.

Williams

Lawrence

Davis

2019. Understanding how discrimination can affect health. Health Serv Res. 54(suppl 2):1374–1388.

40.

Yan

Chan

Louie

JCY

. 2022. Evidence does not support the reduction of added sugar intake from all food sources. Nat Rev Cardiol. 19(12):845. https://doi.org/10.1038/s41569-022-00791-w.

41.

Yousaf

Aslam

Saeed

Sarfraz

Cherrez-Ojeda

2022. Individual, family, and socioeconomic contributors to dental caries in children from low-and middle-income countries. Int J Environ Res Public Health. 19(12):7114. https://doi.org/10.3390/ijerph19127114.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.50 MB