Systemic, behavioural, and psychosocial determinants of incident dental caries among Chinese schoolchildren: a prospective cohort study using structural equation modeling

Abstract

Background

This study investigated the relevant factors influencing the caries increment among schoolchildren in Liaoning Province and constructed a structural equation model to analyze the complex interrelationships among these factors.

Methods

Baseline data of 7,904 schoolchildren were collected through clinical oral examinations and questionnaire surveys in 2024 and oral examinations for the cohort were collected in 2025.

Results

The prevalence of dental caries was 31.1% in 2024, with a mean DMFT/dmft of 0.84 ± 1.57, which increased to 50.6% and 1.63 ± 2.24, respectively, in 2025. A total of 2,865 individuals showed caries increment (defined as △DMFT/dmft ≥ 1), corresponding to an incidence of 36.2%. The results of the multivariate logistic regression analysis indicated that socioeconomic region, grade, gender, weight status, and deliberate exclusion from or isolation in group activities were significantly associated with the caries increment. Structural Equation Modeling analysis (SEM) converged to identify socioeconomic factors as a factor associated with caries incidence, showing both direct associations and indirect associations mediated by poorer dietary habits, although the observational one-year design precludes strong causal clclaims. Notably, the model also indicated a protective association with higher weight status, a finding that merits further investigation.

Conclusions

While conventional factors such as socioeconomic status, dietary habits, and weight status are well-established influencers of caries increment in schoolchildren aged 9–12, our findings highlight that psychological factors, such as experiences of social isolation, constitute a significant determinant. This dimension merits particular attention in both future research and preventive strategies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28526-3.

Background

Adolescent oral health has garnered significant public health attention worldwide [1, 2]. The age group of 9–12 years, marked by the mixed dentition stage, is pivotal for establishing lifelong oral health behaviors and awareness [3]. Oral health not only influences mastication and nutrient intake, but also profoundly affects physiological development and psychological well-being [4, 5]. Dental caries are a major public health concern and the primary cause of orofacial pain in schoolchildren [6]. Its consequences extend beyond the oral cavity and adversely affect the child’s quality of life, academic achievement, and overall family functioning, consequently generating significant economic and caregiving burdens. Thus, investigating region-specific determinants of dental caries is essential to develop tailored and efficacious preventive interventions.

Epidemiological data have highlighted substantial disparities in the prevalence of caries among schoolchildren aged 9–12 years across different regions. In North America, the prevalence ranges from 15.8% in the United States [7] to 55.1% in Canada [8], reaching as high as 88.5% in Mexico [9]. European figures show considerable variation, ranging from a relatively low 11.6% in Belgium [10] to 40.1% in Germany [11]. Similarly, in Africa, the rates varied from 48.2% in Libya [12] to 68.79% in São Tomé [13]. Disparities in the prevalence of caries are also evident across Asia. Southeast Asian nations have reported rates of 31.7% in Singapore [14], 59.4% in Vietnam [15], and 61% in Indonesia [16]. However, the prevalence in East Asia is notably lower, as exemplified by Japan’s rate of 25.7% [17]. These differences underscore the significant influence of regional factors on oral health outcomes.

Current research on the determinants of caries has focused predominantly on oral hygiene and dietary habits [18]. In contrast, investigations into the roles of systemic factors, such as obesity and psychological stress, in schoolchildren aged 9–12 years remain limited. Therefore, analysing the association between these multilevel factors and caries risk in specific populations is of considerable scientific and public health importance.

In this study, a prospective cohort survey was conducted to investigate the effects of various factors on the development of newly developed caries. A baseline survey, including clinical oral examinations and questionnaires, was conducted in 2024 among schoolchildren aged 9–12 years in Liaoning Province, China. The same cohort was followed up in 2025 using an identical protocol. This study aimed to analyze the association between new-onset caries and a range of variables, including socioeconomic status, dietary habits, oral health behaviours, systemic health, and psychosocial factors.

Methods

Sample selection

A multistage stratified sampling strategy was used to select a representative sample of schoolchildren aged 9–12 years in the Liaoning Province. The sampling flowchart is shown in Fig. 1. In the first stage, 14 cities were selected. In the second stage, students from urban and rural areas were randomly chosen within each city, with the monitoring site serving as the sampling unit. In the third stage, fourth and fifth-grade students were proportionally sampled from each site, maintaining a 1:1 sex ratio. The specific demographic composition of the participants is summarised in Supplementary material 3.

Fig. 1.

Study flowchart

Sample size calculation

The sample size for this prospective longitudinal study was estimated using the standard formula for cohort studies [19]. The formula incorporates the following parameters: p₁ and p₀ represent the anticipated incidence rates in the exposed and unexposed groups, respectively; p̄ denotes the average incidence; q = 1-p̄; α indicates the probability of a Type I error; β represents the probability of a Type II error; Z₁-α/₂and Z_β are the standard normal variates corresponding to α and β. With a significance level (α) of 0.05 (Z₁‐α/₂ = 1.96) and statistical power (1-β) of 80% (Z_β = 0.842), the parameters were set. As psychological factors were expected to yield the smallest effect size among all the exposures considered, this scenario required the largest sample size. Therefore, the calculation was based on incidence rates associated with psychological factors to ensure adequate power for all planned analyses. Using reference values from previous studies, the incidence rates were set at p₁ = 18.3% [20] for the exposed group and p₀ = 21.2% [21] for the unexposed group. The calculation yielded the minimum required sample size of 5,915 participants.

Furthermore, the initial sample size was increased by 20% to account for potential data loss because of participant attrition and incomplete or invalid questionnaires. This adjustment raised the minimum required sample size to 7,394. The final sample obtained in this study met these requirements.

Data collection

Before the examination, informed consent was obtained from all participants, and a prospective longitudinal study was conducted with baseline data collection involving clinical oral examinations and questionnaires administered in December 2024. A follow-up examination of the same cohort was conducted in December 2025 to identify the factors associated with incident dental caries. The number of participating schoolchildren is shown in Fig. 2. The final sample included in the study comprised 7,904 participants. Oral clinical examinations were conducted at schools by a total of 28 examiners, with two senior dentists selected from each region to ensure reliability of the examinations. Dental caries were examined and diagnosed in all teeth according to the World Health Organization criteria [19]. All clinical examiners were certified according to the standardised protocols of the Oral Health Status Monitoring Program for Key Populations (2021), which includes qualifying examinations in both theory and practice. Data reliability was determined using kappa values. The kappa value for dental caries should exceed 0.8.

Fig. 2.

Participants’ enrollment chart

Trained personnel measured body weight and height using standardised protocols. Overweight and obesity were assessed using sex- and age-specific body mass index (BMI) reference tables from the Chinese national standard ‘Screening for Overweight and Obesity in School-Age Children and Adolescents’ (WS/T 586–2018). Participants were classified as overweight if their BMI was greater than or equal to the overweight cutoff, but less than the obesity cutoff for their age and sex groups. Those with a BMI exceeding the obesity cutoff were classified as obese.

A standardised questionnaire on frequently encountered health issues among students and health-influencing factors was administered. All questionnaire administrators underwent comprehensive training prior to data collection. The overall Cronbach’s α coefficient for all the items in the questionnaire was greater than 0.80, which confirmed the internal consistency reliability. Spearman correlation analyses for all the items and factors revealed that the internal consistency coefficient α for each factor was greater than the coefficients between this factor and other factors. The questionnaire encompassed several key domains: demographic characteristics (e.g., sex, socioeconomic status, region, and grade), dietary habits, oral health habits, and psychological factors (Supplementary material 1). For the regional analysis, each city was assigned a district code based on its socioeconomic development level. Specifically, the cities were categorised into three distinct socioeconomic levels according to their gross domestic product.

Data analysis

Data analyses were performed using IBM SPSS Statistics for Windows version 20 (IBM Corp., Armonk, NY, USA).The specific grouping details are shown in Supplementary material 2. Continuous variables with a normal distribution are presented as mean ± standard deviation, whereas non-normally distributed data are summarised as median and interquartile range. Categorical variables are expressed as proportions (%). The dependent variable was the caries increment, which encompassed both caries-free individuals transitioning to a caries-active status and those with an increased Decayed, Missing, Filled Teeth (DMFT/dmft) index. This comprehensive definition allows a more holistic analysis of the factors influencing the development of caries.

Univariate analysis and multivariate logistic regression

In the analytical phase, non-parametric rank-sum tests and chi-square tests were employed to assess the associations between caries occurrence and influencing factors, including sex, economic region, monitoring site, grade level, dietary habits, oral health behaviours, and psychological status among schoolchildren aged 9–12 years. Statistical significance was defined as a two-tailed P < 0.05. In addition to variables that were significant in the univariate analysis, we also included several potential confounders that have been strongly associated with dental caries in previous studies, including urban versus rural residence, toothbrushing habits, use of fluoride toothpaste, and frequency of sugary beverage consumption. All these variables were entered together into the multivariable regression analysis. Odds ratios (OR) and 95% confidence intervals (CI) were calculated to evaluate the relationship between changes in caries status and relevant predictors.

Structural equation modeling analysis

Data analysis was conducted using IBM SPSS Amos version 26.0 (IBM Corp., Armonk, NY, USA) to construct a structural equation model (SEM) based on a directed acyclic graph (DAG; Fig. 3). The DAG was developed based on evidence from the literature and assumed directional relationships [22–25]. Structural equation modeling (SEM) was performed using Amos (an SPSS plug-in) to examine the causal pathways between caries increment occurrence and latent variables derived from the measured domains, which pertained to oral health behaviors and included the following variables: dietary habits (breakfast consumption and vegetable intake), socioeconomic factors (urban-rural division and local economic level), psychological factors (experience of social isolation), use of fluoridated toothpaste, and weight status. Model fit was assessed using multiple commonly adopted indices in SEM analysis. The model was deemed suitable for further analysis only when it met the recommended thresholds for goodness-of-fit indices [26].

Fig. 3.

Directed acyclic graph specifying the hypothesized causal structure for the structural equation model

Results

Characteristics of the study population

A total of 7,904 schoolchildren (aged 9–12 years) were recruited at baseline from 14 cities in the Liaoning Province. The study population included 3,847 boys and 4,057 girls; 3,604 fourth-grade students and 4,300 fifth-grade students. All enrolled children underwent clinical examinations and fully completed the questionnaire, resulting in a 100% response rate. Further details of this prospective longitudinal study are presented in Supplementary material 3.

The prevalence of dental caries was 31.1% in 2024, with a mean DMFT/dmft of 0.84 ± 1.57, which increased to 50.6% and 1.63 ± 2.24, respectively, in 2025. A total of 2,865 individuals showed caries increment (defined as △DMFT/dmft ≥ 1), corresponding to an incidence of 36.2% (Table 1). Among individuals who were caries-free at baseline, the incidence of new caries during the follow-up period was 28.4%, whereas among those with pre-existing caries at baseline, the incidence of new caries was 53.7% (Table 1).

Table 1.

The prevalence of caries among schoolchildren aged 9 to 12 years old in Liaoning Province

Year	N	Number of schoolchildren with dental caries	DMFT / dmft
			M ± SD	Median	IQR	0	1	2	3	≥4
2024	7904	2456 (31.1%)	0.84 ± 1.57	0	1	5448(68.9%)	566(7.2%)	899(11.4%)	355(4.5%)	636(8.0%)
2025	7904	4003 (50.6%)	1.63 ± 2.24	1	3	3901(49.4%)	878(11.1%)	1084(13.7%)	561(7.1%)	1480(18.7%)
New caries		2865 (36.2%)	0.79 ± 1.39	0	1	0	1223(15.5%)	850(10.8%)	340(4.3%)	452(5.7%)

			DMFT / dmft at baseline
			0	1	2		3		≥ 4	≥ 1
The number of newly schoolchildren with dental caries			1547(28.4%)	330(58.3%)	444(49.4%)		220(62.0%)		324(50.9%)	1318(53.7%)

Factors influencing caries increment in studied schoolchildren: based on the socioeconomic factors

The influence of socioeconomic factors on caries increment is summarised in Table 2. The caries increment varied across demographic and socioeconomic strata. Geographically, the rates were 36.7% and 35.9% in rural and urban areas, respectively. When stratified by socioeconomic regions, the incidence rates were 38.7% (low income), 32.0% (middle income), and 36.4% (high income). Among schoolchildren, the incidence was higher in fourth graders (39.2%) than in fifth graders (33.7%) and in girls (38.3%) than in boys (34.1%). By weight status, the incidence rates were 39.5%, 33.6%, and 33.8% in the normal-weight, underweight, and overweight groups, respectively.

Table 2.

Socioeconomic factors, oral health, dietary habits, and psychological factors among schoolchildren in Liaoning Province

Variable		N (△DMFT/dmft ≥ 1)	Total	X2	P
Socioeconomic factors of schoolchildren
Socioeconomic status	Low-income regions	1560 (38.7%)	4036(51.1%)	28.35	0.000
	Middle-income regions	740 (32.0%)	2314(29.3%)
	High-income regions	565 (36.4%)	1554(19.6%)
Distribution	Rural	1283 (36.7%)	3496(44.2%)	0.553	0.457
	Urban	1582 (35.9%)	4408(55.8%)
Grade level	Fourth grade	1414 (39.2%)	3604(45.6%)	25.573	0.000
	Fifth grade	1451 (33.7%)	4300(54.4%)
Gender	Female	1552 (38.3%)	4057(51.3%)	14.535	0.000
	Male	1313 (34.1%)	3847(48.7%)
Weight status	Normal	1340 (39.5%)	3393(42.9%)	27.103	0.000
	Underweight	75 (33.6%)	223(2.8%)
	Overweight	1450 (33.8%)	4288(54.3%)
Oral health and dietary habits of schoolchildren
Toothbrushing habit	Twice daily	2121(36.9%)	5750(72.7%)	3.798	0.150
	Once daily in the morning	448(34.3%)	1305(16.5%)
	Once daily in the evening	296(34.9%)	849(10.8%)
Use of fluoride toothpaste when brushing	No	532 (36.4%)	1463(18.5%)	0.182	0.913
	Unsure	1440 (36%)	3997(50.6%)
	Yes	893 (36.5%)	2444(30.9%)
Frequency of sugary consumption/week	Never	1239 (35.6%)	3485(44.1%)	1.303	0.254
	≥ Once / week	1626 (36.8%)	4419(55.9%)
Frequency of fried food consumption /week	Never	849 (35.3%)	2403(30.4%)	1.262	0.532
	< Once/ day	1979 (36.7%)	5399(68.3%)
	≥ Once/ day	37 (36.3%)	102(1.3%)
Frequency of vegetable consumption /week	Never	56 (32.4%)	173(2.2%)	2.836	0.418
	< Once/ day	359 (36.8%)	976(12.3%)
	Once/ day	1115 (37.1%)	3006(38.1%)
	Twice / day	1335 (35.6%)	3749(47.4%)
Daily breakfast consumption	Never	26 (45.6%)	57(0.8%)	3.615	0.164
	Occasionally	247 (38.4%)	644(8.1%)
	Daily	2592 (36%)	7203(91.1%)
Psychological factors of schoolchildren
Experience of serious injury in the past 12 months	No	2827 (36.3%)	7794(98.6%)	0.140	0.708
	Yes	38 (34.5%)	110(1.4%)
Subjected to malicious teasing	Never	2670 (36.4%)	7334(92.8%)	1.482	0.477
	Sometimes	185 (33.9%)	545(6.9%)
	Frequently	10 (40%)	25(0.3%)
Deliberately excluded from group activities or isolated	Never	2631(35.2%)	7470(94.5%)	62.227	0.00
	Sometimes	224 (53.7%)	417(5.3%)
	Frequently	10 (58.8%)	17(0.2%)
Threatened or intimidated	Never	2828 (36.2%)	7805(98.7%)	0.100	0.951
	Sometimes	35 (37.6%)	93(1.2%)
	Frequently	2 (33.3%)	6(0.1%)
Physically assaulted	Never	2796 (36.2%)	7726(97.8%)	2.541	0.281
	Sometimes	64 (37.6%)	170(2.2%)
	Frequently	5 (62.5%)	8(0.0%)
Teased due to physical defects or appearance	Never	2811 (36.3%)	7739(97.9%)	0.938	0.626
	Sometimes	51 (32.9%)	155(2.0%)
	Frequently	3 (30%)	10(0.1%)
Subjected to scolding or physical punishment by parents in the past 30 days	No	2522 (36.1%)	6989(88.4%)	0.687	0.407
	Yes	343 (37.5%)	915(11.6%)

Caries increment was significantly associated with socioeconomic region, grade level, sex, and weight status. The occurrence of caries increment was significantly associated with socioeconomic region, geographic distribution, grade level, sex, and weight status. Specifically, schoolchildren from middle-income regions developed significantly fewer caries increment than those from low- or high-income regions. Fifth-grade students exhibited a significantly lower incidence of caries increment than fourth-grade students. Boys were less affected than girls, and underweight schoolchildren showed significantly fewer caries increment than normal-weight and overweight children.

Factors influencing caries increment in studied schoolchildren: based on oral health and dietary habits

Table 2 shows that 91.1% of the schoolchildren consumed breakfast daily, 47.4% consumed vegetables at least twice per day, 30.4% never consumed fried foods, 44.1% never consumed sugar-sweetened beverages, 30.9% used fluoride toothpaste for brushing, and 72.7% brushed their teeth twice daily.

As shown in Table 2, the caries increment varied according to oral hygiene and dietary habits. Regarding brushing frequency, schoolchildren who brushed once daily in the morning showed a new caries incidence of 34.3%, compared with 34.9% among those who brushed once daily in the evening, and 36.9% among those who brushed twice daily. The use of fluoride toothpaste was associated with a 36.5% incidence rate of new caries. Several dietary factors were also associated with the incidence of new caries: 36.8% for sugar consummation more than once per week, 36.3% for fried food consummation more than once daily, and 35.6% for vegetable consummation more than twice daily. Notably, schoolchildren who ate breakfast daily had a new caries incidence rate of 36.0%. Oral health and dietary habits were not significantly associated with caries increment incidence.

Factors influencing caries increment in studied schoolchildren: based on the psychological factors

As shown in Fig. 4, several psychological factors were scored using a scale where “never” = 0, “sometimes” = 1, and “often” = 2. The results indicated that “subjected to malicious teasing” and “deliberately excluded from group activities or isolated” were the most frequently reported items.

Fig. 4.

Assessment of psychological factors in schoolchildren

The influences of psychological factors are presented in Table 2. The caries increment showed significant variation when stratified according to psychological factors. Schoolchildren who had experienced serious injuries in the past 12 months had a caries incidence of 34.5%. Those subjected to frequent malicious teasing recorded a rate of 40.0%, whereas those who were frequently excluded from group activities or isolated teasing exhibited the highest incidence of 58.5%. Other adverse experiences were associated with the following rates: 37.6% for those who were sometimes threatened, 37.6% for those who were physically assaulted, and 32.9% for those who were frequently teased. Additionally, schoolchildren subjected to parental scolding or physical punishment in the past 30 days had an incidence of 37.5%. Notably, frequent social isolation or deliberate exclusion from group activities was associated with the most pronounced increase in the incidence of caries increment among all factors examined.

Multivariate regression analysis of caries increment in studied schoolchildren

Multivariate regression analysis was conducted to determine the factors influencing caries increment, and the results are presented in Table 3. Socioeconomic region, grade, sex, weight status, and deliberate exclusion from group activities or isolation were associated with caries increment. The results were summarised as follows: middle-income regions compared with low-income regions (OR: 0.755; 95% CI: 0.677–0.842); the fifth grade compared with fourth grade (OR: 0.788; 95% CI: 0.718–0.865); boys compared with girls (OR: 0.869; 95% CI: 0.790–0.955); and the overweight compared with normal-weight (OR: 0.816; 95% CI: 0.741–0.898) were less likely to experience caries increment. Furthermore, schoolchildren who were sometimes isolated or deliberately excluded from group activities compared with never excluded (OR: 2.139; 95% CI: 1.752–2.612) and schoolchildren who were frequently isolated or deliberately excluded from group activities compared to never (OR: 2.683; 95% CI: 1.014–7.1) were more likely to experience caries increment.

Table 3.

Multivariate regression analysis of incident dental caries among schoolchildren in Liaoning Province

Variable		B	S.E.	Waldχ2	df	P	OR	95% C.I.
								L	U
Socioeconomic Regions	Low-income regions			25.68	2	0
	Middle-income regions	-0.281	0.056	25.678	1	0	0.755	0.677	0.842
	High-income regions	-0.094	0.062	2.282	1	0.131	0.91	0.805	1.028
Grade	Fifth grade (Reference: Grade 4)	-0.238	0.047	25.298	1	0	0.788	0.718	0.865
Gender	Male (Reference: Female)	-0.141	0.048	8.449	1	0.004	0.869	0.790	0.955
Weight status	Normal			17.437	2	0
	Underweight	-0.193	0.147	1.713	1	0.191	0.825	0.618	1.101
	Overweight	-0.204	0.049	17.119	1	0	0.816	0.741	0.898
Deliberately excluded from group activities or isolated	Never			59.403	2	0
	Sometimes	0.76	0.102	55.79	1	0	2.139	1.752	2.612
	Frequently	0.987	0.497	3.949	1	0.047	2.683	1.014	7.1

In addition to variables that were significant in the univariate analysis, all other confounding variables were entered together into the multivariable regression analysis, including urban versus rural residence, toothbrushing habits, use of fluoride toothpaste, and frequency of sugary beverage consumption

Results and comparison of SEM

The coding of all variables included in the SEM is presented in the Supplementary material 2. Parameter estimation was performed using the default maximum likelihood (ML) method. Given the potential non-normality of the data, bootstrapping (2,000 resamples; 95% bias-corrected confidence intervals) was conducted to obtain robust standard errors and significance tests. The following model fit thresholds were used: χ²/df < 3, RMSEA < 0.08, GFI > 0.9, NFI > 0.9, and CFI > 0.9. The fit indices of the final SEM model were χ²/df = 0.667, RMSEA = 0.000, GFI = 1.000, NFI = 0.977, and CFI = 1.000, indicating an excellent model fit.

As depicted in Fig. 5, the conceptual framework incorporated six primary constructs: socioeconomic variables, weight status, use of fluoride toothpaste, dietary behaviours, deliberate exclusion from group activities or isolation, and dental caries outcomes. The path analysis revealed that higher frequency of social isolation was significantly associated with poorer dietary habits (β = -0.074, standard error [SE] = 0.036, P = 0.037). Conversely, higher family socioeconomic status was significantly associated with better dietary habits (β = 0.455, SE = 0.133, P < 0.001), better adherence to fluoride toothpaste use (β = 0.262, SE = 0.148, P < 0.001) and a lower risk of caries increment (β = -0.140, SE = 0.124, P = 0.012).

Fig. 5.

Structural equation modeling with standardized coefficients. Paths in bold: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

As depicted in Table 4, socioeconomic factors exerted a significant positive total effect on dietary habits (β = 0.453, SE = 0.132, BC 95% CI: 0.233–0.767), with a similarly positive direct effect (β = 0.455, SE = 0.133, BC 95% CI: 0.233–0.773), indicating that schoolchildren from higher socioeconomic backgrounds had healthier dietary patterns. Similarly, a significant positive total effect was observed regarding fluoride toothpaste use (β = 0.245, SE = 0.064, BC 95% CI: 0.149–0.404), which was also driven by a positive direct effect (β = 0.262, SE = 0.148, BC 95% CI: 0.128–0.682), suggesting better oral health prevention in higher socioeconomic families. Conversely, social isolation showed a significant negative total effect on dietary habits (β = -0.074, SE = 0.036, BC 95% CI: -0.149 to -0.009), with an identical direct effect, demonstrating that more frequently isolated schoolchildren had poorer diets and confirming its direct detrimental influence.

Table 4.

Total effect, direct effect and indirect of the variables of model

Variable				Estimate	SE	Percentile 95%CI		Bias-corrected percentile 95%CI		P
						Lower	Upper	Lower	Upper
Dietary Habits	←	Socioeconomic factors	Total	0.453	0.132	0.213	0.748	0.233	0.767	0.001
			direct	0.455	0.133	0.217	0.753	0.233	0.773	0.001
			indirect	-0.003	0.005	-0.016	0.002	-0.017	0.002	0.363
Use of Fluoride Toothpaste	←	Socioeconomic factors	Total	0.245	0.064	0.144	0.396	0.149	0.404	0.001
			direct	0.262	0.148	0.133	0.707	0.128	0.682	0.001
			indirect	-0.017	0.100	-0.321	0.019	-0.348	0.018	0.539
Dietary Habits	←	Social Isolation	Total	-0.074	0.036	-0.145	-0.006	-0.149	-0.009	0.037
			direct	-0.074	0.036	-0.145	-0.006	-0.149	-0.009	0.037
			indirect	0	0	0	0	0	0	0.433
New dental caries  development	←	Social Isolation	Total	0.019	0.011	-0.002	0.042	-0.002	0.042	0.075
			direct	0.020	0.018	-0.002	0.064	-0.003	0.060	0.078
			indirect	-0.001	0.011	-0.030	0.006	-0.040	0.005	0.875
New dental caries  development	←	Weight Status	Total	-0.046	0.017	-0.067	-0.006	-0.073	-0.012	0.037
			direct	-0.046	0.020	-0.066	0.002	-0.073	-0.007	0.052
			indirect	0	0.004	-0.009	0.001	-0.011	0.001	0.461
New dental caries  development	←	Socioeconomic factors	Total	-0.131	0.050	-0.071	-0.008	-0.254	-0.050	0.003
			direct	-0.140	0.124	-0.071	0.002	-0.482	-0.023	0.012
			indirect	0.009	0.090	-0.031	0.256	-0.033	0.226	0.699

The SEM identified distinct pathways for the incidence of caries increment. Socioeconomic factors were identified as directly associated with lower caries increment, exhibiting a significant positive total effect (β =-0.131, SE = 0.050, BC 95% CI: -0.254 to -0.050) and a congruent direct effect (β = -0.140, SE = 0.124, BC 95% CI: -0.482 to -0.023), indicating that higher socioeconomic backgrounds reduced caries incidence. Besides, weight status demonstrated significant protective total effects (β = -0.046, SE = 0.017, BC 95% CI: -0.073 to -0.012). However, the non-significance of its direct and indirect effects within the model suggests that its overall influence is likely mediated by other unspecified pathways not captured in this model specification, warranting further investigation.

Discussion

This prospective cohort study was conducted to examine the determinants of the caries increment among schoolchildren aged 9–12 years in Liaoning Province. To elucidate the complex interplay between these factors, we constructed and validated an SEM that allowed us to delineate the direct and indirect pathways through which socioeconomic, behavioral, and psychological variables influenced oral health outcomes. SEM is a multivariate statistical method that can evaluate a network of relationships between observed and latent variables and measure the overall model fit [26]. In this study, ΔDMFT/dmft ≥ 1 was used as the outcome indicator, which encompasses two scenarios: (i) participants who were caries-free at baseline developed one or more decayed teeth (incident caries); and (ii) participants who already had caries at baseline experienced an increase in the number of decayed teeth (caries progression). Combining incident cases (from sound to decayed) and progressive cases (from fewer to more decayed teeth) does not allow us to distinguish risk factors for caries onset from those for caries progression. However, this approach effectively increases the sample size and enables a comprehensive assessment of risk factors associated with an increase in decayed teeth. Importantly, the oral environments of these two scenarios are fundamentally different. The first appearance of a carious lesion indicates a transition of the oral environment from a globally reversible state to a locally irreversible breakdown [27]. Once a cavity is formed, the oral microbiota shifts from a diverse symbiotic community to a locally dominant cariogenic flora, creating retention sites where the cleaning effect of saliva is markedly reduced [28]. When a patient already has one or more cavities and the number of decayed teeth continues to increase, this signals a deterioration from a local breakdown to a whole-mouth cariogenic environment, characterised by dominance of cariogenic bacteria, failure of salivary defense, and self-sufficiency of endogenous substrates. Breaking this vicious cycle requires restoring all existing cavities combined with modulation of the whole-mouth microbiota [28].

In addition to oral health outcomes, this study recorded an overweight prevalence of 33.8% in the pediatric cohort. This rate is considerably higher than the national average (15.1% for overweight and 10.7% for obese) reported in the 2017 Physical Activity and Fitness in China survey [29]. This disparity is consistent with the recognised geographical gradient in childhood obesity across China, wherein northern regions such as Liaoning exhibit higher rates than southern provinces [30, 31]. Both multivariate regression and SEM indicated a significant inverse association between overweight status and caries increment, which is consistent with previous similar reports [24]. However, this exploratory association should be interpreted cautiously, as it may reflect residual confounding or socioeconomic patterning rather than a direct biological effect of overweight status. This inverse relationship in the Chinese context may be partly attributed to socioeconomic factors. Prior studies have demonstrated that overweight and obese children in China are more likely to come from economically advantaged families with higher parental education, which is also a factor associated with better oral health resources and preventive care. The increased caries increment experience was more commonly seen in schoolchildren who often consumed sugared beverages and desserts, and who had night eating behaviour; a previous study was also in accordance with these findings [32]. Therefore, warning labels regarding sugar intake are required. It is also important for clinicians to analyse schoolchildren’s dietary behaviours and provide advice regarding dietary intake in treatment plans. Although numerous studies have demonstrated that fluoride use and dietary habits influence the development of dental caries [33], neither factor demonstrated a significant association with caries in our results, as indicated by both the multivariable analysis and the SEM. We attribute this finding to the following considerations. First, there are various methods of fluoride use, whereas our study only addressed the use of fluoride toothpaste, which may not fully capture the caries-preventive effect of fluoride. Second, although dietary habits in this study encompassed multiple factors, neither individual dietary factors nor the composite measure demonstrated a significant effect. It is possible that dietary habits influence caries through other mediating variables not examined in this study [34], warranting further research.

This study revealed a significant association between socioeconomic factors and incidence of dental caries. Specifically, schoolchildren from middle-income families exhibited a significantly lower rate of new-onset caries. The SEM further demonstrated that socioeconomic status was directly associated with both dietary habits and fluoride use. Moreover, it directly influences the development of new caries, indicating its role as an independent predictor of caries. Previous studies have demonstrated that parents with high educational background levels were associated with better self-evaluation of oral health, reducing the likelihood of dental pain [6] and dental caries [35]. The observed lower caries increment among middle-income schoolchildren compared to their high- and low-income peers presents a compelling public health phenomenon. Most previous studies have indicated that low-income levels are associated with poor oral health habits, ultimately leading to an increased prevalence of dental caries [36–38], whereas high-income levels often serve as a protective factor [39]. This differs somewhat from our findings; however, we believe that this pattern can be attributed to a unique convergence of socioeconomic and behavioral factors. Middle-income families typically possess sufficient resources to afford preventive care, such as fluoride toothpaste and dental sealants, yet often maintain more traditional, less-processed diets than their high-income counterparts, who may have greater access to sugary snacks and beverages. Critically, these families often demonstrate high health literacy, enabling consistent supervision of schoolchildren’s oral hygiene and enforcement of dietary rules, a balance that is sometimes challenged by time constraints in wealthier households and resource limitations in poorer households. Furthermore, middle-income schoolchildren may experience a protective psychosocial environment with moderate levels of parental expectations and stress, which indirectly supports stable health habits, including those related to oral health. This “Optimal state” underscores that oral health disparities are not merely economic, but represent complex behavioural adaptations within specific social contexts, highlighting the need for nuanced, targeted public health interventions.

A noteworthy innovation of this study is the incorporation of psychological factors into the analytical framework, which revealed their significant impact on the development of caries increment. Both regression analysis and structural equation modeling (SEM) were conducted. Regression analysis demonstrated that social isolation was directly associated with both dietary habits and caries increment. In the SEM, although the direct path between social isolation and caries increment did not reach statistical significance, there was evidence of an indirect or partial association, potentially mediated by dietary habits or other behavioral factors. Accumulating evidence shows that psychological well-being is a critical determinant of the development of caries increment among schoolchildren. Recent studies have elucidated both the direct and indirect pathways of this association. For instance, specific adverse childhood experiences, such as bullying or exposure to violence, have been significantly linked to dental fear, which in turn mediates an increased caries experience, creating a vicious cycle in which fear leads to dental avoidance and deterioration of oral health [23, 40]. Furthermore, studies have indicated that psychosocial and behavioural variables can be associated with the severity of early childhood caries, and that early childhood caries can be associated with a negative impact on the well-being and ability to perform daily activities of preschoolers and their families [41]. These findings collectively highlight that social isolation is integral to the caries process, emphasizing the need for integrated interventions that address both psychosocial well-being and clinical dental care in schoolchildren.

This study has several strengths. First, while previous investigations regarding caries determinants have predominantly emphasised localised oral factors, our study primarily focused on systemic contributors, thereby introducing a novel perspective for caries prevention and management. Second, the longitudinal design, encompassing 7,904 participants, provided robust evidence supporting the reliability of our conclusions.

This study had some limitations. First, as students aged 9–12 years will gradually transition from primary to middle school in the coming years, a high rate of loss to follow-up is expected, which prevented a long-term follow-up in this study. Future research could be conducted starting around 6 years to extend the follow-up period. Second, the dental examination results for students aged 9–12 years did not analyse primary and permanent teeth separately. This aspect will be refined in future studies to enhance the accuracy of our findings. Meanwhile, in our study of dental caries among children aged 9–12 years, we observed that some carious primary teeth exfoliated due to physiological replacement, leading to an apparent reduction in the number of decayed teeth at follow-up. This issue is unavoidable in caries research involving the mixed dentition stage. It also represents an important reason why primary and permanent teeth will be analysed separately in our future studies.

Conclusions

This prospective cohort study demonstrates that caries increment in schoolchildren aged 9–12 years in Liaoning Province is associated with a complex interplay of socioeconomic, behavioural, and psychological factors. The lowest caries increment observed in middle-income children highlights a protective ‘optimal state’ shaped by balanced preventive resources, dietary habits, and health literacy. The inverse association between overweight status and caries increment likely reflects socioeconomic patterning rather than a direct biological effect. SEM showed that Socioeconomic factors were directly associated with caries incidence, alongside indirect associations mediated through poorer dietary habits. Social isolation indirectly influenced caries development through dietary behaviours. These findings advocate for integrated, multi-level interventions that extend beyond traditional oral health measures to address psychosocial determinants.

Abbreviations

BMI

Body mass index

CI

Confidence interval

DAG

Directed acyclic graph

DMFT/dmft

Decayed, Missing, and Filled Teeth index for permanent/primary teeth

GFI

Goodness-of-fit index

ML

Maximum likelihood

NFI

Normed fit index

OR

Odds ratio

RMSEA

Root mean square error of approximation

SE

Standard error

SEM

Structural equation modeling

Authors’ contributions

LL and JL designed this study. JL, WZ, KZ, SZ, LC, and QG performed the experiments, and analysed and interpreted the data. WZ and JL wrote the original draft of the manuscript. LL and MW reviewed and revised the manuscript.

Funding

This work was supported by the Liaoning Provincial Natural Science Foundation of China (No. 2025-MSLH-773).

Data availability

The data that supports the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Informed consent to participate was obtained from all participants, and this study adhered to the tenets of the Declaration of Helsinki and was approved by the Medical Ethics Committee of the Liaoning Centre for Disease Prevention and Control (approval number: LNCDCEC-2024-013). The institute is the Liaoning Centre for Disease Prevention and Control.

Consent for publication

Written informed consent was obtained from the participants before administering oral examinations and questionnaires. All data published here are under the consent for publication.

Competing interests

The authors declare no competing interests.