Childhood tooth decay is a considerable public health issue. Nearly one quarter (23.4%) of 5-year-old children in England in 2019 had experience of dental decay – that same year 37,406 children were admitted to hospital for extraction of decayed teeth under general anaesthetic. Dental decay in childhood is clearly associated with missed school days, lower self confidence, dental anxieties and poor oral health in adulthood. This project aims to use the extensive data describing early childhood found in the Born in Bradford birth cohort, to identify predictors of bad oral health outcomes. The findings demonstrate a relationship between educational and socioeconomic factors and poor oral health that have yet to be demonstrated in England.
Data and methods
The data used in the study can be divided into two groups: The early-life descriptive features are taken from the Born in Bradford birth cohort, and the oral health outcome variables are taken from 3 dental datasets.
Born in Bradford is a birth cohort following 13,776 pregnancies to birth and through early life. The data used in this study is taken from the “mother baseline survey”, socioeconomic/sociodemographic data taken during pregnancy, and the “education records” dataset, educational assessment data collected once children have entered the education system.
The dental outcomes are taken from 3 separate datasets:
- Bradford General Anaesthetics: a data capture of 1082 hospital admissions for dental extractions/restorations over a period covering the Born in Bradford birth cohort.
- PHE Oral Health Survey of 5-year-old children: oral health examinations of 5-year-old children in England in 2019.
- PLATOON Study: Oral health examinations from a case-control study of young children from the Born in Bradford cohort with/without experience of dental extractions.
The dental data provide detailed information about oral health, including tooth-level decay categories (shown in the figure below) and d3mft scores, or decayed/missing/filled teeth.
The analysis was primarily descriptive, and focussed on modelling oral health outcomes using simple binary logistic regression models. Experience of general anaesthetics and any dental decay (no decayed teeth vs one or more decayed teeth) were modelled using the early-life descriptive features from the Born in Bradford datasets. Statistically significant predictors of poor oral health outcomes were identified and used to estimate their total and adjusted effects on the odds of experiencing adverse oral health outcomes.
General anaesthetic outcomes proved a difficult target to predict using the data from the Born in Bradford cohort. The overall proportion of children with experience of hospital admission for dental decay was very low, and few of the descriptive variables were able to meaningfully separate the cohort into groups that were significantly more likely to have experienced a general anaesthetic. It was suggested that hospital admission for dental decay might be too rare an outcome to meaningfully highlight predictors of bad oral health, and that many children in high risk groups may not be engaging with oral health services that would otherwise result in their referral for dental extraction.
Using examinations of oral health to establish a decay/no decay outcome proved to be far more successful. The relationships between the descriptive variables and decay experience were much more significant, and showed a much greater differentiation in the risk of decay experience when grouping children by different observations – particularly so given the much smaller samples for the decay datasets. It is likely this is because the socioeconomic/ sociodemographic links between the best oral health outcomes (no decay) and others are far stronger than the worst outcomes (requiring extractions under general anaesthetic), and that visual observations of a random sample of children are less prone to bias than selecting a subset of children that have been given a specific treatment for dental decay.
Among the statistically significant predictors of decay outcomes were the educational attainment variables. Children at a lower than expected level of attainment experience higher levels of decay, and children achieving higher levels of attainment are far less likely to experience decay. This relationship persists when adjusting for other socioeconomic factors that have known associations with dental decay. This link between school performance and oral health has been demonstrated in the wider literature, though it has yet to be shown in a UK setting.
Value of the research
The results of this study reinforce the body of literature that demonstrates links between educational learning and development goals and oral health. This is a novel finding in a UK setting, and is the basis of a proposed publication. Results also confirm the relationship between low socioeconomic status and poor oral health that has been well established in the literature, and that the educational attainment link persists when adjusting for socioeconomic status.
Initial experiments also highlighted the success of using a decay/no-decay target to predict oral health outcomes using economic/demographic predictors. These results make a strong case that collecting oral health observations for a wider sample of the Born in Bradford cohort would be invaluable.
- Children in England at lower levels of educational attainment show higher levels of dental decay, and those at higher levels of attainment show lower levels of decay.
- It is challenging to predict general anaesthetic outcomes using socioeconomic/sociodemographic variables – decay/no-decay outcomes are much easier to predict.
- Additional oral health data for a wider subset of the Born in Bradford cohort would be invaluable.
- Health informatics
Sam Relins, Data Scientist Intern, Leeds Institute for Data Analytics
Dr Peter Day, Consultant in Paediatric Dentistry, Leeds Teaching Hospitals NHS Trust
Dr Ning Lu, Senior Research Fellow, Bradford Institute for Health Research
Dr Erin Giles, Clinical Lecturer, University of Leeds
Dr Kara Gray-Burrows, Lecturer, University of Leeds
Bradford Institute for Health Research