| Sign In to gain access to subscriptions and/or personal tools. |
The Effects of Social Class and Dental Attendance on Oral Health
1 King’s College London Dental Institute, Weston Education Centre, Off 262 Cutcombe Road, Denmark Hill, London SE5 9RJ, UK; Correspondence: * corresponding author, nora.donaldson{at}kcl.ac.uk
The relationship between socio-economic status (SES) and oral health is well-established. We investigated whether the association between SES and the number of sound teeth in adults is explained by dental attendance patterns, in turn determined by the effect of SES on barriers to dental attendance. Data on 3817 participants from the 1998 Adult Dental Health Survey in the UK were analyzed. Using structural equation modeling, we found a model with 4 factors (aging, SES, attendance-profile, and barriers-to-dental-attendance) providing an adequate fit to the covariance matrix of the 9 covariates. The final model suggests that the association between SES and the number of sound teeth in adults in the UK is partially explained by the pathway [SES  barriers-to-dental-attendance dental-attendance-profile number-of-sound-teeth]. A direct relationship, SES number-of-sound-teeth, is also significant.
Key Words: social class • structural equation modeling • mediating relationships
It is widely accepted that there are socio-economic inequalities in oral health (Locker, 2000; Watt, 2007). A socio-economic gradient is found in a range of clinical and self-reported oral health outcomes (Enjary et al., 2006; Jamieson and Thomson, 2006; Lopez et al., 2006; Sanders et al., 2006b). A growing body of empirical research (Sisson, 2007), together with theoretical modeling of the social determinants of oral health (Petersen, 2005; Brunner and Marmot, 2006), suggests that the socio-economic gradient in oral health may be related to social, environmental, and political factors, which act through material, behavioral, and psychosocial pathways. Dental attendance may be a factor contributing to the socio-economic gradient in oral health. Regular dental attendance is more prevalent in high-socio-economic groups (Hjern et al., 2001; Jamieson and Thomson, 2006) and is associated with better oral health outcomes, after adjustment for socio-economic status (SES) (Unell et al., 1999; Petersen et al., 2004; Dye and Selwitz, 2005; Sanders et al., 2006a; see APPENDIX). Attitudes and perceptions influence dental attendance patterns, including anxiety, cost (of dental care) concerns, value placed on restored teeth, and beliefs regarding the importance of regular dental attendance (Abrahamsson et al., 2001; Bagewitz et al., 2002; Riley et al., 2006). Positive attitudes and perceptions about dental attendance, which are associated with better oral health (Unell et al., 1999; Bagewitz et al., 2002; Riley et al., 2006), tend to be held by high-socio-economic groups (Gilbert, 2005; Armfield et al., 2006; Riley et al., 2006). Only one study has investigated the contribution of dental attendance to the socio-economic gradient in oral health (Sanders et al., 2006a). The findings differed according to oral health outcome. The gradient in oral-health-related quality of life (OHRQoL) was significantly attenuated by dental attendance, whereas the gradient in self-reported missing teeth was not. The identification of causal pathways for oral health inequalities is essential for informing dental public health policy. In this paper, we use structural equation modeling (SEM) to investigate whether the relationship between SES and the number of sound teeth in adults is explained by dental attendance patterns or barriers to dental attendance.
Data The data (Table  ) were taken from the 1998 Adult Dental Health Survey, in which a representative sample of 3817 adults in the United Kingdom (UK) had a dental examination and were interviewed about topics including personal perceptions and attitudes to dental treatment, patterns of dental attendance, and socio-demographic data (Kelly et al., 2000). The main outcome variable was the number of sound teeth. Including restored sound teeth, this measure is more sensitive than the sum of decayed, missing, and filled teeth (DMFT) in identifying social and behavioral risk factors for oral health (Marcenes and Sheiham, 1993). We analyzed 9 explanatory variables (Table). Socio-demographic and behavioral variables included age, gender, head of household’s social class (social class), weekly-household-income (income) (pounds), and regularity in visiting the dentist (regular). Social class was a three-class re-coding of the six-class Registrar General ranking, with higher values indicating higher SES. Members of the Armed Forces who held a professional or technical qualification would be included in that category, but the 13 people classified as "Armed Forces" were included in the "unskilled/partly-skilled" category, because we believed that this would reflect the typical educational background of individuals when joining the armed forces. Higher values of regular indicated less regular attendance: 1 = regular check-up/2 = occasional check-up/3 = only if trouble with teeth (Table). Personal perceptions and attitudes to dental treatment included importance given to visiting the dentist regularly (importance), finding NHS dental treatment expensive (cost), wanting simple rather than fancy/intricate dental treatment (fancy), and feeling anxiety when visiting the dentist (anxiety). These were scored on an ordinal scale, with high values indicating positive perceptions/attitudes: 1 = definitely-feel-like-that, 2 = to-some-extent, 3 = notat-all.
There were no missing data for the dependent variable, number of sound teeth. The high proportions of missing data in cost (19%) and fancy (15%) reflect the fact that participants answered ’don’t know’ to statements on finding NHS dental treatment expensive (cost) and wanting simple rather than fancy/intricate dental treatment (fancy), respectively. There were missing data for social class (6%) and income (8%). Data were missing completely at random (MCAR) for fancy and missing at random (MAR) for income, social class, and cost. The analysis we present in this paper is based on the 2329 (61%) participants for whom complete data were available. We divided our sample of 3817 cases into two parts, to form an exploratory random sample and a confirmatory random sample (MacCallum and Austin, 2000). We formed the exploratory sample by taking a simple random sample of 2875 (75%) cases from the available sample of 3817 cases. The remaining 942 (25%) cases formed the confirmatory sample. Of the 2875 cases in the exploratory sample, 1754 had complete data. Of the 942 cases in the validation sample, 575 had complete data.
Statistical Models
Starting with the exploratory sample, we used SEM on the covariance matrix of the 9 variables to evaluate the goodness of fit of the theoretical model, and then, in an exploratory manner, to detect and correct for specification errors, until a model that fit the data well was achieved. This involved eliminating parameters with small t values (unless they had practical importance) and adding parameters with large modification indices (if they were theoretically sound). The SEM gave the analysis of our cross-sectional study many of the virtues of a longitudinal study, allowing for an exploration of the complex causal pathways involved in disease processes (Der, 2002; Singh-Manoux et al., 2002; Garson, 2004). The final model was cross-validated in the confirmatory sample. We used maximum likelihood (ML) estimation in the presence of some ordinal variables, because the ML method is robust to departures from normality, and the standard errors are underestimated only when the observed variables deviate really far from normality, which was not the case in our data. We nevertheless double-checked our findings using generalized least-squares (GLS) and robust standard errors, because these methods are a good choice when deviations from multivariate normality are severe, and the sample size is not very large (Finch et al., 1997). We used the EQS software (Bentler, 1989).
We present descriptive statistics for the overall sample (Table  ) and the path diagram of the original theoretical model (Fig. 1), after non-significant bi-directional relationships (including an inverse relationship between attendance and number of teeth) were removed. Although the distribution of the standardized residuals from this model was symmetrical around 0, the fit indices (Fig. 1) were poor. The chi-square was 982 (26 df). On further theoretical consideration, we recognized that importance may have a psychological element that merited moving it out of the barriers factor. The Lagrange-Multiplier Test (LMtest) signaled a strong correlation between importance and regular attendance, and a significant improvement to the fit was obtained when we allowed these 2 variables to load together onto a latent factor that we labeled ’attendance’. Also signaled were correlations between age and social class, between cost and income, and between regular attendance and anxiety; these correlations were judged to be theoretically reasonable.
These considerations led us to contemplate a model (Fig. 2
As we can see from the maximum likelihood estimators of the regression coefficients and their standard errors (Fig. 2 ), the most important determinants of the number of sound teeth in adulthood were aging (a latent factor highly associated with age), socio-econ (a latent factor highly associated with social and income), and attendance (a latent factor highly associated with regular attendance and importance). The latent factor barriers was determined by the latent factor socio-econ, hence showing an indirect effect on the number of sound teeth. A direct path was also found between the latent factors socio-econ and aging. From the estimated coefficients, we note that each level increase in aging is associated with a mean increase of 16 (SE = 0.26) yrs of age, yielding an average decrease of 3 (SE = 0.14) sound teeth. One level increase in attendance is associated with a 0.4 (SE = 0.03) increase in irregularity and a 0.44 (SE = 0.03) decrease in importance, yielding a mean decrease of 0.8 (SE = 0.12) sound teeth. One level increase in socio-econ is associated with a mean increase in the social class of 0.5 (SE = 0.03) and a mean increase in income of £242 (SE = £16), yielding a mean increase of 1.5 (SE = 0.18) sound teeth. Similar results were found with the use of GLS and robust standard errors; the scaled chi-square (20 df) = 131, and goodness of fit indices and the maximum likelihood estimators and their standard errors were essentially unchanged. The final model was unchanged when multiple imputation of the missing data was performed. The results were also confirmed when the model was applied to the validation sample.
The study suggests that the association between SES and the number of sound teeth in adults in the UK is partially explained by the pathway SES barriers-to-dental-attendance dental-attendance number-of-sound-teeth. Although the theoretical model showed a poor fit to the data, the essential pathways among SES, barriers to dental attendance, dental attendance, and oral health remained unchanged in the improved final model. The study also suggests that there is a persistent direct relationship between SES and the number of sound teeth in adults that is unexplained by barriers to dental attendance and dental attendance profile. The results of our study concur with the findings of Sanders et al.(2006a), that the socio-economic gradient in OHRQoL in Australian adults was significantly attenuated by dental attendance, but are in contrast to their finding that the gradient in self-reported missing teeth was not attenuated by dental attendance. This may reflect the different oral health outcome measures used in the respective studies. It may also be a consequence of variations in the funding, availability and use of dental services, and the relative impact of other social determinants of oral health inequalities, such as water fluoridation, cultural norms, and government policy on diet and smoking between the two countries (Burt, 2005; Watt, 2007). From a policy perspective, the results of our study suggest that reducing the barriers to regular dental attendance and promoting regular dental attendance for low-socio-economic groups may reduce oral health inequalities to some extent. One of the strategic directions of the World Health Organisation (WHO) oral health group is to encourage the development of oral health systems that are financially fair and improve oral health outcomes equitably (WHO, 2007). In the UK, NHS dental treatment is free of charge for the poorest members of the population, and the recent reform of NHS dentistry has reduced the cost of large and complex courses of dental treatment (Department of Health, 2006). Nonetheless, it could be argued that dental services are becoming increasing inequitable, due to the rising proportion of services provided privately (British Dental Association, 2007), a trend which may undermine attempts to reduce oral health inequalities in the UK. Paradoxically, from our results, we inferred that there is also a need to look beyond dental attendance when attempting to reduce oral health inequalities. While the model does not suggest alternative explanations for the socio-economic gradient in oral health, other research points to the material, behavioral/cultural, and psychosocial factors, which affect oral health throughout life (Sisson, 2007). There are calls for ’upstream’ public health action, including the use of legislation and fiscal measures, to create social environments which actively promote oral health (Watt, 2007; WHO, 2007). Such an approach, aimed at populations rather than individuals, is believed by some to be more likely to reduce oral health inequalities than traditional behavioral approaches, by addressing the social structures and processes which affect oral health through material, behavioral, and psychosocial pathways (Burt, 2005; Watt, 2007). The study has several strengths. First, the use of SEM enabled us to test a model more closely resembling real-life inter-relationships than would have been possible using multiple regression analysis, addressing the limitation in inferring causality in a cross-sectional (rather than longitudinal) study. Second, the sample was large and deemed to be representative of the UK population. This might have contributed to the feature that the considerable missing data in cost and fancy were at random and did not appear to affect the results significantly. Some caution should be exercised when interpreting our findings, due to limitations of the data. Questions on the regularity of dental attendance and attitudes toward dental attendance and treatment were not validated for this survey. However, they had been used successfully in an earlier UK Adult Dental Health Survey (Todd and Lader, 1991), and although there may be some misclassifications, this should have no major bearing on the outcome. There might be additional complex mechanisms at play not easily accounted for in our model. Teeth rarely simply fall out. Usually, a decision is made about extraction, and then a dentist removes the tooth. Our model does not study dentist-patient communication and ethnicity, since they were not recorded. Research on ethnic disparities (Smedley et al., 2002; Kressin et al., 2003; Kressin, 2005) suggests that clinical decisions about whether to root-treat or extract may be influenced by the individual’s SES and ethnicity, and this may determine the number of teeth regardless of the regularity of dental attendance. Further research is required to test the model for other oral health outcomes, such as OHRQoL, and to include other potential barriers to dental attendance, such as a lack of access to dental services. Our findings apply only to the UK population, reflecting the structure of dental services and socio-cultural norms in the UK. It would be useful to test the model in other populations with different oral health systems, including other potential determinants of oral health inequalities, such as psychosocial stress, environmental factors, and behavioral and cultural norms (Burt, 2005; Sisson, 2007; Watt, 2007). We conclude that the socio-economic gradient in the number of sound teeth in adults is partially explained by dental attendance, which in turn is determined by the effect of SES on barriers to regular dental attendance. Overcoming barriers to regular dental attendance for low-socio-economic groups may reduce oral health inequalities.
The study was funded by the King’s College Hospital NHS Trust and by the King’s College London Dental Institute. A preliminary report of this study was presented in the 2nd international meeting, "Methodological issues in oral health research", Ghent, Belgium, April 19–21, 2006.
A supplemental appendix to this article is published electronically only at http://jdr.iadrjournals.org/cgi/content/full/87/1/60/DC1. Received for publication September 12, 2006. Revision received September 17, 2007. Accepted for publication October 12, 2007.
Journal of Dental Research, Vol. 87, No. 1,
60-64 (2008)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
TOP
ABSTRACT
INTRODUCTION
