Socioeconomic and ethnic inequalities in oral health among children and adolescents living in England, Wales and Northern Ireland

Abstract Objectives Although adolescence is a sensitive developmental period in oral health, the social equalization hypothesis that suggests health inequalities attenuate in adolescence has not been examined. This study analyses whether the socioeconomic gap and ethnic disadvantage in oral health among children aged 5 reduces among adolescents aged 15. Methods Data from the cross‐sectional Children's Dental Health Survey 2013 were analysed, comprising of 8541 children aged 5, 8, 12 and 15 attending schools in England, Wales and Northern Ireland. Oral health indicators included decayed and filled teeth, plaque, gingivitis and periodontal health. Ethnicity was measured using the 2011 UK census ethnic categories. Socioeconomic position was measured by family, school and residential deprivation. Negative binomial and probit regression models estimated the levels of oral health by ethnicity and socioeconomic position, adjusted for demographic and tooth characteristics. Results The predicted rate of decayed teeth for White British/Irish children aged 5 was 1.54 (95%CI 1.30‐1.77). In contrast, the predicted rate for Indian and Pakistani children was about 2‐2.5 times higher. At age 15, ethnic differences had reduced considerably. Family deprivation was associated with higher levels of tooth decay among younger children but not among adolescents aged 15. The influence of residential deprivation on the rate of tooth decay and filled teeth was similar among younger and older children. Moreover, inequalities in poor periodontal health by residential deprivation was significantly greater among 15‐year‐old children compared to younger children. Conclusions This study found some evidence of smaller ethnic and family socioeconomic differences in oral health among British adolescents compared to younger children. However, substantial differences in oral health by residential deprivation remain among adolescents. Community levels of deprivation may be particularly important for the health of adolescents.


| INTRODUCTION
There are substantial social inequalities in oral health, with children from disadvantaged and ethnic minority backgrounds experiencing poorer oral health. 1,2 Furthermore, there is considerable evidence that socioeconomic health inequalities appear to reduce from childhood to adolescence leading to the hypothesis of social equalization during adolescence. [3][4][5][6] This hypothesis involves a change in the pattern of socioeconomic differences in health from 1 in childhood characterized by health inequalities to 1 in youth characterized by relative equality. 3 However, the social equalization hypothesis in oral health has not been explicitly examined in previous studies.
Despite being preventable, dental caries and periodontal diseases are the 2 most prevalent oral diseases contributing to the global burden of chronic diseases. 7,8 While the prevalence of caries among children and adolescents living in the UK has reduced considerably over the last 40 years, caries is now concentrated in a minority of the child population. 9 Children from disadvantaged family socioeconomic backgrounds have a higher risk of dental caries 1 and periodontal disease. 10 Oral health inequalities by ethnicity 11 and area deprivation 12 have also been reported. In the UK, there is a high level of caries experience among preschool and school children from Pakistani, Bangladeshi, Chinese and East European backgrounds, even after controlling for levels of socioeconomic deprivation. 2,13,14 Afro-Caribbean children generally had better or similar oral health than White children. 2 Socioeconomic position (SEP) and ethnicity appear to influence caries or periodontal disease by influencing oral health behaviours, 15 in particular, oral hygiene habits, diet, smoking and dental attendance. 16,17 Reductions in socioeconomic inequalities as children age are apparent for several major single health conditions such as accidents, injuries and mental health 3 and for longstanding illnesses, psychosocial well-being and obesity. 6 However, similar reductions in ethnic differences in health as children grow older have not been reported, even though ethnic minority children tend to live in poorer socioeconomic circumstances. There is also some evidence that socioeconomic inequalities in oral health are attenuated among older children compared to younger children. 18,19 This socioeconomic equalization in health during adolescence is suggested to arise when the influences of the family and home environment diminish, with school, peers and youth culture playing a larger role in children's lives. 3 Adolescence is a period when oral health-related behaviours are not as closely monitored by parents as during childhood, with a potentially larger role for the school and neighbourhood factors in influencing adolescent oral health and related behaviours. In addition, the transition from childhood to adolescence is a sensitive developmental period in oral health with the replacement of primary teeth by permanent teeth. A reduction in tooth decay in early adolescence may reflect the lower lifetime exposure of permanent teeth to oral health risk factors.
This study examines whether the association between ethnicity, SEP and oral health differs in childhood and mid-adolescence. We hypothesize that family SEP and ethnicity will be less influential for the oral health of adolescents compared to younger children living in

| Variables
We selected multiple clinical indicators to capture the different aspects of children's oral health that include the condition of their teeth and gums (further details in Supporting Information). At all ages, the analyses included both the primary and the permanent teeth because of the mixed dentition of the children. At age 5 and 8, 47% and 99% of the children had at least 1 permanent tooth erupted, out of which 4% at age 5 and 38% at age 8 had experienced caries.
Clinical oral health outcomes included the number of decayed teeth. In line with developments in epidemiological studies, the CDHS 2013 adopted the International Caries Detection and Assessment System for assessing first staging of caries process. 21 We therefore used the number of teeth affected at the clinical decay threshold, which includes untreated obvious decay (visual and cavitated decay into dentine) and visual and cavitated decay into enamel.
The number of filled teeth was used as an indicator of access to dental services and receipt of dental care. We also analysed an indicator of poor periodontal health by combining the presence of some gum inflammation (gingivitis), the presence of plaque or the presence of calculus in more than 1 sextant. Separate analyses of gingivitis and plaque are presented in Tables S7 and S8. Gingivitis is a reversible periodontal condition that is prevalent among children and adolescents 10 and is particularly associated with puberty or presence of mixed dentition. 22 Dental plaque is the main biological aetiological ROUXEL AND CHANDOLA | 427 factor for the development of dental caries and periodontal diseases. 23 Missing teeth were not analysed as a separate dependent variable, as the reason for missing teeth in the primary dentition was not recorded. Missing primary teeth could be due to decay or natural exfoliation, and hence, to allow comparison for all age groups, the decayed, missing and filled teeth (dmft/DMFT) index was not used to measure caries experience. In the permanent dentition, there were very few children with missing teeth due to decay (4.8%).
Ethnicity of the children was collected from school records, which used parents' reporting of family ethnic group when their child started at school, and was assessed using the 2011 UK census ethnic categories. 24 25 We excluded children from Chinese, other Asian, Arab, Gypsy and Irish travellers and "Other" unspecified ethnic groups due to small numbers or the heterogeneous nature of the "Other" category.
Socioeconomic variables included free school meals eligibility (at the child level): a statutory benefit available for children from families who received income related benefits and is a proxy for family level relative income deprivation; 20 deprived school: schools with more than 30% of children eligible for free school meals were Covariates included in the regression models were the child's sex, urban/rural residence and country. As the pattern of caries is age dependent, and the tooth eruption age distributions result in caries pattern differences that vary by age, the models also controlled for the number of primary and permanent teeth.

| Statistical analysis
Statistical analyses were undertaken in STATA using a design factor to take account of the complex sampling and weighting procedures.
The weights derived by the CDHS survey team explicitly takes into account the pattern of missingness by ethnicity and SEP. 20 From the sample of 9866 children who completed a dental examination, we excluded 1090 children due to item nonresponse on explanatory variables. The overall rate of missingness was 11.1%, of which ethnicity accounted for 5.7%; free school meals 4.1% and IMD 4.2%. We also excluded children from "Other" ethnic groups (n = 235). Therefore, the sample was reduced to 8541 children: 5-year-olds n = 2217 (26.4%); 8-year-olds n = 2083 (24.3%); 12-yearolds n = 2183 (25.5%) and 15-year-olds n = 2058 (24.8%).
The relative index of inequality (RII) was used to summarize the magnitude of inequalities in oral health between the IMD quintiles.
The RII is especially useful in this analysis because it allows us to compare the IMD inequalities across different children's ages, even though the rate of poor oral health differs by age. The RII was generated from the country-specific IMD quintiles using the subgroup option of the STATA program file "riigen." 26 We additionally tested for country differences in the association of RII with the different oral health outcomes by including an interaction term between the RII and country in the regression models predicting oral health (Table S9).
It is important to ensure, empirically, that the effects of family SEP on children oral health are estimated net of individual ethnic minority status. Previous research shows that ethnic minority status influences health independently of income, education and socioeconomic characteristics. 27 There was no indication of colinearity between ethnicity and the different SEP indicators, as the regression coefficients and standard errors were relatively stable with different models.
We used negative binomial regression to model the count variables (clinical decay and filled teeth). As the number of teeth varies between children, these models estimated the rate of tooth decay and fillings (per child) by including the log number of teeth as an offset.
We used probit regression to model the binary outcome variables (gingivitis, plaque, poor periodontal health). We reported the predicted probabilities and rates of each of the outcomes by the ethnicity and SEP categories holding all the other variables in the models at their means. We examined the equalization hypothesis by testing whether the interaction between SEP/ethnicity and age group was significant. The reference group for the measures of SEP and ethnicity was always the most advantaged group in all the analyses-White British and Irish ethnicity, not eligible for free school meals, not attending a deprived school and least deprived residential area.
We conducted a series of sensitivity analyses to assess the robustness of our findings (Tables S5-S10). Methodological differences and lack of data on ethnicity and socioeconomic positions in the previous CDHS surveys prevented trend analyses. For dental decay, we repeated the negative binomial regression analyses using the 2003 criteria (Table S6). We also conducted multilevel logistic analyses, analysing the presence of tooth decay at the tooth level (level 1) clustered within children (level 2) (Table S10).

| RESULTS
The distribution of all the variables by the 4 age-cohort samples is displayed in Table 1. For reference, the mean number of decayed, Turning to socioeconomic differences, poorer family SEP (free school meal eligibility) and greater area deprivation (IMD rank) were associated with higher rates of decay among children age 5, and 12, while at age 15 none of the SEP measures appeared to influence decay.
In terms of ethnic differences in filled teeth for children at age 5 ( In Tables 2-4 and S6-S8, we examined whether the differences in our selected oral health outcomes by SEP/ethnicity across children's age were statistically significant. The interaction term between ethnicity and age was significant for tooth decay (Table 2), filled teeth (Table 3) and plaque (Table S8), suggesting that the risk of poor oral health for ethnic minority children significantly reduces from age 5 to age 15. Figure S1 shows the predicted rate of tooth decay among children of White, other White, Indian and Pakistani ethnicity. The ethnic differences that are clearly shown among children aged 5 are no longer apparent among children aged 15.
The interaction term between free school meal and age was also statistically significant for tooth decay ( Bold text indicates a statistically significant difference with a p-value less than .05. a Survey weighted models include ethnicity, all socioeconomic variables, sex, country, urban/rural, and number of permanent and primary teeth.

| DISCUSSION
We found strong evidence of smaller ethnic differences in dental decay among British adolescents aged 15 compared to children aged 5, and this pattern was repeated for all the oral health measures.
Moreover, the socioeconomic gap (using free school meal eligibility) in dental decay among children was significantly smaller among children aged 15 compared to 5-year-old children. However, the association between higher levels of residential deprivation and higher levels of dental decay remained similar across all the age groups.
Furthermore, the association between residential deprivation and poor periodontal health was stronger among 15-year-old children compared to 5-year-old children.
The equalization hypothesis suggests that socioeconomic inequalities in health reduce as children age. 3 Our study on children and adolescent oral health showed a complex picture, which does not fully support this hypothesis. There was some evidence of equalization in terms of ethnic and family-based SEP differences, particularly in terms of dental decay, but differences by residential deprivation remained throughout childhood and adolescence for dental decay, and increased during adolescence for filled teeth and poor periodontal health.
Although the hypothesis of socioeconomic equalization in oral health during adolescence has not been explicitly examined in previous studies, there is some cross-sectional evidence of smaller ethnic differences in oral health among Danish adolescents 28  .47 (7) .003 (7) .72 (7) .18 (21) Free school meal Three indicators (presence of plaque or calculus in more than 1 sextant, gingivitis) were combined to produce an indicator of poor periodontal health. *P < .05; **P < .01; ***P < .001. environment more open to influences of peers and nonfamily members. 30 The school environment could contribute to the social equalization of the health of adolescents but creates at the same time new disparities, with probably long-lasting consequences. 4 In our study, residential deprivation predicted poorer oral health among adolescents, whereas among younger children, it was familybased SEP that predicted tooth decay. Children who live in more deprived areas are closer to fast food outlets than children in less deprived areas. 31  In the UK, marked socioeconomic and ethnic inequalities exist for the use of dental services. 38 All the major ethnic minority and disadvantaged socioeconomic groups are less likely to visit the dentist, and more likely to visit due to problems with their teeth. 25 In our study, we found that adjusting for multiple measures of socioeconomic disadvantage did not explain the poorer oral health among ethnic minority children aged 5 and 8. Cultural beliefs play a role in dental care seeking behaviour. Due to the transient nature of the primary teeth, some caregivers of young children feel their care are not as important as permanent teeth and some cultural and ethnic groups may not have a strong preventive oral health orientation. 39,40 The type and variety of foods consumed in the early years influence longer-term eating behaviours. 41 Although White mothers report that they introduce solid foods earlier than mothers of Asian origin, 42 they were more likely to consider the sugar content of their child's food and to avoid teeth-damaging foods, whereas Pakistani mothers have been found to be more likely to give sweetened drinks and foods at an early age. 42 The main limitation of this study is that the data came from a cross-sectional survey. We were not able to examine how socioeconomic and ethnic inequalities in oral health changed as the children grew into adolescents. Hence, any differences in oral health between adolescents and children may not be related to the lifecourse and ageing, but may reflect cohort differences. Due to small numbers of cases from certain ethnic groups, we had to drop these groups from the main analyses. We acknowledge the risks that more complex empirical patterns might be overlooked. This was explored further in the sensitivity analyses (Table S10)