Statistical analyses

Children with incomplete data on anthropometric or dental assessments were excluded from the analyses. Listwise deletion was applied for missing information in the questionnaires. Continuous and categorical variables are presented as means (SD) and frequencies (%), respectively. At individual level, Linear regression models were used to examine the association between area deprivation (IMD 2010-continous independent variable) and anthropometric measures (outcome continuous variables: weight height, BMI, and waist circumference). The association between deprivation (IMD 2010-continous independent variable) and caries (outcome variable: dmft counts) was examined using Poisson regression models. The association between anthropometric variables (weight, height, BMI, waist circumference as independent continuous variables) and dental caries presence as indicated by dmft (dependent variable, binary response: yes or no), was examined using multiple logistic regression.

Generalised linear models (log-linear Poisson models) were used to investigate the impact of neighbourhood characteristics on obesity and caries rates. The initial model included all available covariates presented previously. Then starting with the least significant covariate (based on p-value) and using backwards elimination, the covariates were removed from the model one-by-one. Each time a covariate was removed, the model was refitted. This process was repeated until only significant covariates were left in the model (all with p-values> 0.05).

The impact of several demographic variables and health behaviours on the likelihood of overweight/obesity (binary response variable: yes or no) and caries (binary response variable: yes or no) was examined using logistic regression models. In these models, covariates were inserted one by one using stepwise regression (i.e. combining backward elimination and forward selection, allowing for covariates removed early in the process of backwards elimination to be reconsidered at a later stage).

Demographic factors (independent variables) included in the model were the parent’s age group (< 30 years –reference level- vs > 30 years) and education (up to secondary school–reference level-, Technical/College, University), the family’s total income (up to £25.599–reference level-, 26.000–36.399, 36.400 and above), and the gender of the child (reference level: female). Individual behaviours examined included were tooth brushing habits (once or less per day –reference level- vs ≥2 times per day), age when the child started having his/her teeth cleaned (< 1 year –reference level- vs > 1 year), presence of adult with children when teeth are being brushed (yes –reference level- vs no) and frequency of in-between meals food/drink consumption (using never as a reference). Logistic regression models with response ‘the presence/absence of dental decay’ and ‘being/not being overweight/obese’ (outcome variables) were fitted.

Differences in tooth brushing frequency (once or less per day –reference level- vs ≥2 times per day) were examined using a logistic regression model with IMD categories (IMD-1 most deprived–reference level-, 2, 3 least deprived), income (up to £25.599–reference level-, 26.000–36.399, 36.400 and above) and education (up to secondary school–reference level-, Technical/College, University) included as covariates in the model. Logistic regression was used to test differences in the response rate (for the questionnaires) between parents/guardians living in areas with different levels of deprivation (IMD 1–reference level-, 2 and 3).

The Statistical Package for the Social Sciences (SPPS, version 32) and the R software were used for the analyses. A p-value of less than 0.05 was considered to suggest statistical significance. The STROBE checklist has been used to report this study.

The present study has been approved by the Faculty of Health and Human Sciences Research Ethics Committee of Plymouth University (ref: 13/14–240).