Decomposing Socioeconomic Inequality in Early Childhood Caries Among 3 to 5-Year-Old Children in China

Abstract

Introduction and aims

Early childhood caries (ECC) is a widespread oral disease that harms children's health in China. Although previous studies have linked ECC prevalence to socioeconomic status, few have measured the degree of socioeconomic inequality. This study aimed to evaluate the socioeconomic inequality of ECC in children aged 3 to 5 years in China and identify the contributor to the inequality.

Methods

We extracted data on 3 to 5-year-old children from the fourth National Oral Health Survey. We measured the inequality of ECC by the average household income per capita. We used the average household income per capita to measure the inequality of ECC. To describe inequality both qualitatively and quantitatively, we used the following methods: concentration curve, Erreygers-corrected concentration index, relative index of inequality and slope index of inequality. We also applied a decomposition based on the probit model to identify the factors that contributed to inequality.

Results

The prevalence of ECC in Chinese preschool children was 63.11% (95% CIs: 60.54%, 65.61%). The negative value of the Erreygers-corrected concentration index (−0.0459; 95% CIs: −0.0594, −0.0324), slope index of inequality (−0.0674; 95% CIs: −0.0876, −0.0471) and the positive value of relative index of inequality (0.7484; 95% CIs: 0.6856, 0.8169) all indicated that ECC prevalence was higher among children from low-income families. The main factors contributing to inequality were average household income, parents’ educational level and living areas.

Conclusion

There is a pro-poor inequality in ECC among 3 to 5-year-old children in China.

Clinical Relevance

To improve oral health equality, policymakers should focus more on children from low-income families, with less educated parents and living in rural areas.

Introduction

Early childhood caries (ECC) is a term that describes one or more decayed, missing or filled primary teeth in children younger than 6 years old.¹ ECC is a preventable disease that affects different populations unevenly, causing significant economic burden and quality-of-life problems and posing a new challenge to global public health^2,3 Dental caries may be a lifelong disease once children suffer from it at a very young age.⁴ Deciduous dental caries can impair pronunciation, mastication and aesthetics, lower the quality of life, and increase the risk of dental caries in permanent teeth.5, 6, 7 However, unlike life-threatening conditions such as diabetes, cardiovascular disease and obesity, the severity of dental caries has not yet received enough attention.^4,8,9

Health inequality, which is used to assess health inequity indirectly, refers to the measurable differences in health outcomes among groups of people with different socioeconomic status (SES), based on factors such as social, economic, demographic or geographic characteristics. Prevalence or the average level of health in a population, does not capture the distribution of health among its members. By analysing health inequality, one can gain insights into the health status of different subgroups. Monitoring health inequality is a crucial way to inform policies, programs and practices that aim to promote health equity, a shared goal across the globe. Inequality seems to be an obscure and ambiguous concept but can be measured using a series of statistical methods. The concentration index, relative slope of inequality and slope index of inequality are the most common complex measurements for subgroups with a natural ordering. The advantages of these measures include that the calculation can take account of the population size and that they specifically reflect socioeconomic gradients in health.¹⁰ They can make a qualitative and quantitative analysis about inequality and offer more information on health policy-making.

Deciduous dental caries are unequal across the world.11, 12, 13, 14 Previous studies have shown that untreated dental caries are more common in less developed countries¹⁵ and have identified the causes and risk factors of dental caries.¹⁶ However, there is a large variation in the prevalence and incidence of ECC among and within countries.^2,17 Therefore, it is important to understand the reasons for this variation.

The study has 2 objectives: (1) to evaluate the socioeconomic inequality in ECC among children aged 3 to 5 years in China and (2) to identify the determinants of the socioeconomic inequality in ECC.

Material and methods

Study population and data

This study used secondary data from the fourth National Oral Health Survey, a cross-sectional study conducted in 2015 across 31 provinces, autonomous regions, and municipalities in China. The study adopted a multistage, stratified, equal volume, and random sampling method. Oral health examinations and questionnaires followed the standards of the World Health Organization (WHO).¹⁸ Details of the sampling and research methods are available in the previous study.¹⁹ Ethical approval (Approval No: 2014-003) for the study was received from the Ethics Committee of the Chinese Stomatological Association and written informed consent was obtained from the parents of each participant. All methods were carried out in accordance with relevant guidelines and regulations of the Ethics approval under the Declarations of Helsinki.

We included all the participants aged 3 to 5 years old (N = 40,360). Given that population distribution is extremely uneven and sex ratio varies in different regions in China, the equal volume sampling cannot reflect the real distribution and characteristics of the whole population. Therefore, we weighted the samples following poststratification by province, living area (urban or rural) and sex, according to the sixth National Census in 2010.²⁰

Variables

We considered dental caries experience (caries prevalence) as a binary outcome variable (0: dmft = 0; 1: dmft > 0). Table 1 present the content of the questionnaire in detail. We used annual household income per capita, obtained from the family scale and annual household income, as the measurement of SES.

Table 1.

Descriptive characteristics of participants (N = 40315).

Characteristics	Weighted proportion (95% CIs)	Weighted ECC prevalence (95% CIs)
ECC
dmft = 0	36.89 (34.39-39.46)	——
dmft > 0	63.11 (60.54-65.61)	——
Sex
Male	54.42 (54.10-54.74)	63.17 (62.30, 64.03)
Female	45.58 (45.26-45.90)	63.04 (62.18, 63.90)
Area
Urban	41.40 (34.85-48.26)	60.34 (59.51, 61.17)
Rural	58.60 (51.74-65.15)	65.07 (64.20, 65.93)
Geographical region
West	28.43 (19.67-39.19)	62.60 (61.58, 63.61)
Middle	34.30 (23.88-46.50)	57.89 (56.72, 59.05)
East	37.26 (26.95-48.89)	68.31 (67.35, 69.28)
Parental education attainment*
Low	49.35 (44.64-54.06)	65.12 (64.25, 65.99)
Medium	24.89 (22.57-27.36)	63.36(62.10, 64.62)
High	25.76 (22.47-29.35)	59.02 (57.85, 60.19)
Income
1st quintile (poorest)	20.00 (16.19-24.45)	65.61 (64.24, 66.98)
2nd quintile	20.00 (18.10-22.03)	64.67 (63.25, 66.08)
3rd quintile	22.22 (19.94-24.68)	62.32 (60.94, 63.70)
4th quintile	17.78 (15.79-19.97)	63.05 (61.62, 64.48)
5th quintile (richest)	20.00 (16.99-23.39)	60.00 (58.77, 61.23)
Feeding way
Exclusively breastfed	43.32 (40.56-46.12)	66.21 (65.29, 67.13)
Predominantly breastfed	20.57 (19.19-22.03)	65.29 (63.94, 66.64)
Predominantly formula-fed	12.32 (11.11-13.64)	53.92 (52.14, 55.70)
Exclusively formula-fed	7.57 (6.89-8.30)	59.19 (56.99, 61.38)
Mixed-fed (50/50)	16.22 (14.94-17.59)	60.88 (59.37, 62.40)
Sweet frequency†
High	47.61 (45.05-50.19)	64.11 (63.23, 64.99)
Medium	36.27 (34.59-37.99)	63.43 (62.42, 64.45)
Low	16.11 (14.75-17.58)	59.45 (57.88, 61.02)
Sweet frequency before sleep
Usually	8.66 (7.92-9.46)	69.03 (67.08, 70.99)
Occasionally	56.62 (54.17-59.05)	65.84 (65.03, 66.66)
Never	34.71 (32.47-37.03)	57.17 (56.13, 58.21)
Toothbrushing frequency
Less than twice a day	82.71 (80.16-84.99)	63.04 (62.35, 63.72)
Twice a day or more	17.29 (15.01-19.84)	63.46 (62.09, 64.83)
Dental service utilization in the past 12 months
No	86.83 (85.45-88.10)	60.16 (59.49, 60.83)
Yes	13.17 (11.90-14.55)	82.58 (81.33, 83.84)
Knowledge‡
Low	36.92 (34.82-39.08)	64.76 (63.76, 65.77)
High	63.08 (60.92-65.18)	62.14 (61.37, 62.92)
Attitude§
Negative	12.49 (11.05-14.09)	66.03 (64.31, 67.74)
Positive	87.51 (85.91-88.95)	62.69 (62.04, 63.35)
Caregiver-rated general health
Very poor/ poor	2.59 (2.15-3.11)	66.95 (63.30, 70.60)
Fair	28.06 (25.70-30.54)	64.19 (63.02, 65.36)
Good/ very good	69.35 (66.50-72.06)	62.53 (61.80, 63.26)
Caregiver-rated oral health
Very poor/poor	10.37 (9.59-11.21)	95.28 (94.48, 96.08)
Fair	35.54 (34.00-37.12)	72.48 (71.51, 73.45)
Good/ very good	54.08 (52.17-55.98)	50.78 (49.92, 51.64)

^⁎Parental education attainment is classified into 3 groups: low (junior high school and below); medium (medium-technical secondary school and senior high school); high (junior college and above).

^†Sweet frequency is subdivided into 3 categories: high (no less than once a day); medium (less than once a day and more than once a week); low (less than once a week).

^‡Knowledge is divided into 2 levels: low (correct answers <4 questions in 8); high(correct answers ≥4 questions in 8).

^§Attitude is divided into 2 levels: Low (correct answers <3 questions in 6); High(correct answers ≥3 questions in 6).

We used expectation maximization (EM) to impute the missing values in the income data, which were missing or misclassified due to the sensitive nature of income reporting. We then analysed the data using quintiles of annual household income per capita.

Measurement of inequality

We used the slope index of inequality (SII) to measure the magnitude of absolute inequality and the relative index of inequality (RII) and concentration index to measure the magnitude of relative inequality. We also performed decomposition analysis to examine how various factors contributed to the inequality among children.²¹

Calculation of SII and RII

SII and RII are regression-based measures that account for the entire population.¹⁰ The weighted sample was ranked from the poorest (rank 0) to the richest (rank 1) subgroup based on the annual household income per capita. Then, using a logistic model, caries experience was regressed on the midpoint value for each subgroup, given the binary outcome and the predicted values of caries prevalence were computed for the 2 extremes, ${\nu }_0$ and ${\nu }_1$.²² SII is the difference between the predicted values at rank 1 and rank 0, ${\nu }_1-{\nu }_0$. RII is the ratio of the predicted values at rank 1 to rank 0, ${\nu }_1/{\nu }_0$, ranging from 0 to 1. Thus, positive SII values indicate that health outcomes are more common in the most advantaged population. Moreover, RII >1 implies higher prevalence in the most advantaged subgroup.

Concentration curve and calculation of concentration index

The participants were ranked by their annual household income per capita. Then, the concentration curve was plotted by showing the cumulative percentage of ECC in the population (Y-axis) versus the cumulative population (X-axis) from the poorest to the richest. A 45-degree line (the equality line) would indicate an equal distribution of ECC. However, if the concentration curve lies between the equality line and the X- and Y-axes, it means there exists inequality. The concentration index (CI) is twice the area between the concentration curve and the equality line. The prevalence of ECC is pro-poor when the concentration curve lies above the equality line and the CI is negative. Positive values indicate pro-rich inequality with the concentration curve below the equality line. CI is defined as follows:

$$CI=cov\left(y,r\right),$$

where $\overline{y}$ is the mean of the health outcome variables, ECC and r is the fractional rank of individuals in the personal income distribution.

To measure inequality in ECC prevalence, which is a dichotomous variable, the Erreygers-corrected concentration index (EI) is preferred over the conventional concentration index, which has an arbitrary value.²³ Furthermore, the natural log of the annual household income per capita, rather than the income quintiles, was used to calculate EI. The formula for EI is:

$$EI=\frac{4\overline{y}}{y_{max}-y_{min}}CI,$$

where $y_{max}$ and $y_{min}$ represent the upper and lower limits of caries prevalence, respectively. Therefore, it can be simplified as:

$$EI=8cov\left(y,r\right),$$

where y is the health outcome variable, i.e. the presence of ECC and r is the fractional rank of individuals in the income distribution.

Decomposition of the concentration index

We used a probit model to decompose EI. We mainly classified the determinants into 3 sources: (1) socioeconomic factors, (2) oral health knowledge, attitude and practice and (3) caregiver-perceived health status. We included income as a determinant in the decomposition regression to avoid overestimating the contribution of other factors to the measured inequality.²⁴ We excluded individuals with missing information for any of these variables because no methods have been proposed for applying multiple imputation estimates in decomposition analyses. To decompose EI, we can write it as follows:

$$\text{EI}=4*\left(\sum _k\left({\beta }_k^m{\overline{x}}_k\right)C_k+G{C}_{\epsilon }\right),$$

where ${\overline{x}}_k$ is the mean of the determinants included in the decomposition analysis, ${\beta }_k^m$ is the coefficient of the determinants, $C_k$ is the concentration index of the determinants and $G{C}_{\epsilon }$ is the generalized concentration index of the error term$\epsilon$.

The decomposition analysis results show the elasticity, concentration index and relative contribution of each determinant. Elasticity measures how the dependent variable changes when the determinants change by 1 unit. A negative (positive) elasticity means that the explanatory variable reduces (increases) the ECC prevalence. The concentration index of any determinant indicates how it is distributed by socioeconomic rank. Each contribution indicates how much the determinant's variation across income levels explains the relationship between income and ECC prevalence. A positive contribution rate means that the variable increases inequality and vice versa.

We conducted a sensitivity analysis to assess the decomposition model's robustness by changing the reference groups.²⁵

All analyses accounted for the multistage sampling design and sample weights using Stata 15 (STATA Corporation).

Results

After excluding 45 participants with missing values, we included 40,315 preschool children aged 3 to 5 years in the analysis. The prevalence of ECC was 63.11% (95% CIs: 60.54%-65.61%). Table 1 shows that the poorest subgroup had the highest prevalence of 65.61% (95% CIs: 64.24%-66.98%), while the richest subgroup had the lowest prevalence of 60.00% (95% CIs: 58.77%-61.23%). In general, higher prevalence was found among children from rural areas and eastern region. Children whose caregivers had a lower education level, low knowledge and negative attitude had a higher prevalence. There were more children affected by dental caries who were exclusively or predominantly breastfed in the first 6 months. More children suffered from dental caries who had high sweet frequency in daily life and before sleep and had dental attendance experience in the past 12 months. Also, children who were rated as poor/ very poor in general and oral health had a higher prevalence. Table 1 presents the descriptive statistics of all variables and the prevalence for each subgroup.

Figure 1 shows the graph for SII, which declines from the poorest subgroup to the richest subgroup, indicating that ECC is more prevalent among the poorer quintiles. Figure 2 shows the concentration curve of ECC, which lies above the equality line, indicating pro-poor inequality. Table 2 presents the results of absolute (SII) and relative (RII and EI) inequality among the participants. The prevalence difference (SII) between the richest and poorest groups was −0.0674 (95% CIs −0.0876, −0.0471). The corresponding ratio (RII) was 0.7484 (95% CIs 0.6856, 0.8169). The value of EI was −0.0459 (95%CIs −0.0594, −0.0324). All inequality indices showed a significant concentration of ECC among the disadvantaged children.

Fig. 1.

Income-based inequality in ECC (The SII value is the difference between the extreme values of this distribution).

Fig. 2.

Concentration curve of ECC in preschool children in China (N = 40315)

Table 2.

Income-related inequality indices in ECC prevalence among the preschool children in China in 2015 (N = 40315).

Index	Estimate (95% CIs)	P
SII	−0.0674 (−0.0876, −0.0471)	<0.001
RII	0.7484 (0.6856, 0.8169)	<0.001
EI	−0.0459 (−0.0594, −0.0324)	<0.001

Table 3 shows the decomposition of socioeconomic inequality in ECC prevalence. Income (71.61%), parental education level (56.14%) and living areas (34.41%) were the main factors that increased inequality. Other factors that also contributed to the observed pro-poor were caregiver-rated oral health (18.24%), oral health-related knowledge (11.01%) and attitude (9.98%) and feeding method (3.39%). Alternatively, geographic region (−34.52%), dental service utilization in the past 12 months (−32.70%) and caregiver-rated overall health (−30.45%) largely reduced the inequality. Other variables that had a negative impact on inequality were toothbrushing frequency (−6.37%), sweet consumption before sleep (−4.34%), sweet consumption (−2.06%) and sex (−0.02%). Low consumption of sweets and never having sweets before bedtime were more prevalent among the poorer population. Conversely, dental service utilization, adequate knowledge, positive attitude, good/very good general health and oral health as rated by the caregiver were more prevalent among the richer population. These determinants accounted for 94.33% of the total inequality. The results were consistent in the sensitivity analysis (see Supplementary Table 1).

Table 3.

Decomposition of socioeconomic inequality in ECC among the preschool children in China in 2015 (N = 40315).

Variables	Elasticity	Concentration index	Contribution	%
Socioeconomic factors
Sex
Male	Reference
Female	-0.0216	-0.0005	<0.0001	-0.02%
Area
Urban	Reference
Rural	0.3138	-0.0503	-0.0158	34.41%
Geographical region				-34.52%
West	Reference
Middle	-0.0509	0.0174	-0.0009	1.92%
East	0.1042	0.1606	0.0167	-36.45%
Parental education attainment				56.14%
Low	Reference
Medium	-0.0180	0.1020	-0.0018	4.01%
High	-0.0560	0.4277	-0.0239	52.14%
Income	-0.6628	0.0496	-0.0329	71.61%
Practice, knowledge and attitude
Feeding way				3.39%
Exclusively breastfed	Reference
Predominantly breastfed	-0.0090	0.0207	-0.0002	0.41%
Predominantly formula-fed	-0.0528	-0.0181	0.0010	-2.08%
Exclusively formula-fed	-0.0174	-0.0142	0.0002	-0.54%
Mixed-fed (50/50)	-0.0373	0.0689	-0.0026	5.60%
Sweet frequency				-2.06%
High	Reference
Medium	0.0101	0.0398	0.0004	-0.87%
Low	-0.0073	-0.0753	0.0005	-1.19%
Sweet frequency before sleep				-4.34%
Usually	Reference
Hardly	-0.0162	0.0219	-0.0004	0.77%
Never	-0.0836	-0.0281	0.0023	-5.11%
Toothbrushing frequency
Less than twice a day	Reference
Twice a day or more	0.0863	0.0339	0.0029	-6.37%
Dental service utilization in the past 12 months
No	Reference
Yes	0.0839	0.1790	0.0150	-32.70%
Knowledge
Low	Reference
High	-0.0613	0.0824	-0.0051	11.01%
Attitude
Negative	Reference
Positive	-0.1209	0.0379	-0.0046	9.98%
Caregiver-rated health status
Caregiver-rated general health				-30.45%
Very poor/poor	Reference
Fair	0.0001	-0.1018	<0.0001	0.01%
Good/very good	0.2779	0.0503	0.0140	-30.46%
Caregiver-rated oral health				18.24%
Very poor/poor	Reference
Fair	-0.5271	-0.0275	0.0145	-31.57%
Good/very good	-1.1432	0.0200	-0.0229	49.81%
Residual			-0.0016	5.67%

Discussion

The WHO has emphasised that the well-being of society as a whole is at risk if some subgroups disproportionately bear more disease burden all the time.¹⁰ The study found a pro-poor inequality in ECC prevalence on both relative and absolute scales, which matched the trend seen in both developed and developing countries.11, 12, 13, 14^,26 A decomposition analysis was conducted to reveal the factors that influence socioeconomic inequality.^27,28

Household income and parental education level were the main contribution factors of pro-poor inequality. The 3rd NOHE also showed that children's caries experience varied greatly by household income.²⁹ This trend is consistent with global reports.^26,30,31 SES could affect children's oral health through parental behaviours, knowledge and attitudes.³² People with higher SES may have better access to oral health information and dental services. Moreover, many studies have indicated that low-income families tend to consume cheaper and energy-dense foods, such as added sugars, which can increase the risk of dental caries.³³ A study in China has found an “inverted U-shaped” relationship between household income and consumption of sugary beverages.³⁴ Moreover, poor and vulnerable populations have limited financial and psychological resources, which prevent them from providing adequate health care for their children.³³ Therefore, the authorities should pay more attention to the oral health of low-SES children who are at risk. School-based oral health programmes can be effective in enhancing oral health literacy and facilitating access to preventive oral health care for disadvantaged preschool children in underserved areas.³⁵

Residential area is another major factor that contributes to inequality. Human resources for oral health (HROH) are distributed unevenly between urban and rural areas, as shown by domestic and international studies.36, 37, 38, 39, 40 Studies conducted in a northern province and Shanghai in China found that there are more oral health institutions and dental human resources in urban areas.^36,40 According to the China Health Statistical Yearbook, it is reported that there were 945 stomatological hospitals in 2015, of which 702 hospitals were located in urban regions.⁴¹ Per capita disposable income of urban residents was about 2.73 times that of rural residents in 2015.⁴² Though rural children in China have a higher treatment need than urban children,⁴³ residents from rural areas have much lower spatial accessibility and poorer affordability of oral healthcare resources than urban areas. The rural population may face difficulty in accessing HROH, which reduces their chances of getting restorative or preventive care. Without proper treatment, the oral condition may deteriorate and lead to more inequality in ECC. Researchers have suggested that the national conditions demand a more rational dental workforce system to address the oral health challenges of remote rural areas.^36,40,44 Dental assistants, who require less training time, could follow the model of dental therapists in other countries who provide both oral health promotion programmes and regular dental care (such as prevention and restoration).⁴⁵ Moreover, financial incentives are necessary to motivate oral health care professionals to serve the underserved population.

China has undergone remarkable urbanization at a record-breaking pace since it implemented the reform and opening-up policy in 1978. Data from the National Bureau of Statistics show that the urbanization rate of China's population rose from 17.92% in 1978 to 65.22% in 2022.⁴² However, urbanization in developing countries also has negative impacts, despite its benefits for economic growth, job creation and infrastructure and health care improvement.46, 47, 48 As urbanization develops, dietary habits have shifted to a Western-style diet that contains more sugary foods.^49,50 The survey shows that people in the eastern region of China, where urbanization is higher, consume more sugar-sweetened beverages.^51,52 This implies that the population's health-related behaviour level may not grow as fast as urbanization. This may be the reason for the higher prevalence of ECC in the eastern region. The dietary pattern that reduced socioeconomic inequality in ECC was harmful to oral health. This pattern will worsen health inequality in the future if it persists. In China, the dietary guidelines and the National Nutrition Programme (2017-2030)⁵³ recommend that the daily intake of added sugar should be no more than 50 g and ideally less than 25 g. To address inequality in the long term, it is vital to implement “upstream” measures to curb the consumption of added sugar, such as imposing higher taxes on sugar and using nutrition labels.

The rich subgroup used dental services more than others. However, the positive elasticity showed that preschool children who had visited a dentist in the last 12 months had higher ECC prevalence. According to Andersen's behavioural model,⁵⁴ studies have suggested that need factors (such as parents’ rating of oral health, caries experience and dental pain) are the main reasons for using dental services for both children and adults in China.^55,56 Other studies in China have also reported this passive dental-visit pattern.^57,58 The limited coverage of dental care and the low awareness and poor attitude of parents may contribute to the pro-rich and passive pattern of dental visits.^59,60 Problem-oriented dental visits can reduce inequality, but they are not good for oral health. A better way to improve both general and oral health is to combine early dental visits with immunization as part of paediatric primary health care.⁶¹ Moreover, oral health examinations in kindergartens and communities can help children get more access to early and timely dental care.

Oral diseases have caused significant economic and quality-of-life losses, but global health policy has not prioritized them.^62,63 Oral health is often seen as separate to children's general health.⁶⁴ Like the rest of the world, Chinese parents have overlooked their children's oral health. The positive elasticity suggested that children whose parents rated their general health as good/very good had higher ECC prevalence. Preschool children lack the awareness and ability to take care of their oral health by themselves. Thus, children's poor oral health may result from caregiver neglect. The World Health Assembly only adopted a resolution in 2021 to include oral health in the noncommunicable disease agenda and to incorporate oral health care interventions into universal health coverage programs.⁶⁵ To raise public awareness of oral health, the Chinese authority has integrated oral health into the National Program for Chronic Disease Control and Prevention (2017-2025) and Healthy China 2030 and launched an Oral Health Action Plan (2019-2025).66, 67, 68

The study used a proper sampling method to include a large number of children aged 3 to 5 years and proper weighting method to make the results sufficiently representative. Another strength of this survey is that it is the first to examine the socioeconomic inequality of ECC among Chinese children using absolute and relative methods and to apply the decomposition model simultaneously, providing information for inequality monitoring and policy-making.

This study has some limitations that should be acknowledged. First, the findings are based on a cross-sectional survey, which does not allow for causal inference. Thus, the results should be interpreted with caution. Second, recall bias cannot be avoided due to the caregiver-reported questionnaires. Third, the data of 3 to 5-year-olds in the National Census were unavailable, so the data of 1 to 4-year-olds were used as a proxy, which may introduce some inaccuracy. Fourth, there were some missing values for annual household income and family size. However, to account for the importance of annual household income per capita as an indicator of SES, EM imputation was used to fill in the missing values and maximize the data utilization. Finally, children who did not attend kindergarten were excluded from the study, particularly those living in remote and poor areas. Therefore, the extent of inequality may have been underestimated.

Conclusions

To conclude, we found that ECC is more prevalent among poorer preschool children in China (pro-poor inequality) and we provided information for policymakers on how to reduce this inequality. The authorities should focus on the upstream interventions of ECC and allocate resources more effectively and efficiently to help more children. More research is needed to monitor oral health inequalities and their changes after the related policies are implemented.