Participants and public involvement
Community dwelling older adults aged between 60-90 years of age were recruited from a senior activity centre located in the central west district of Singapore and National University Polyclinic (NUP), Jurong. The inclusion criteria were: (1) aged between 60-90 years, (2) eligible for dental care benefits, (3) can understand/communicate in Mandarin/Hokkien, English and Malay.
Informed Consent
Written informed consent was obtained from all participants after the objectives and procedures of the research were fully explained to them by a member of the research team. Participants were informed that they could withdraw from the research at any time without giving any reasons. Participants with mild/moderate dementia were asked to invite their legally appointed representatives to the consent-taking process and data collection. The research team ascertained that the person deciding on behalf of the participant with dementia, was acting in the best interest of the participant and considered the participant’s concerns and desires.
Dental clinical examination
The dental clinical examination was performed by calibrated examiners. The dental examination included recording details of: (1) oral pathology (if any), (2) periodontal status (calculus, probing depth, loss of attachment and bleeding), (3) coronal status (active decay, filled, missing due to decay), (4) root status (active decay, filled and missing), (5) tooth wear (loss of enamel, dentine, pulp exposure or exposure of secondary dentine), (6) number and distribution of occlusal contacts (specifically of posterior teeth) and (7) prosthodontic status (presence/absence of denture, denture fit, retention, stability, extension and hygiene).
Collection of saliva sample
Alongside the clinical examination of the subject's dentition, saliva samples were also collected to record stimulated salivary flow rate and buffering capacity of saliva. Subjects were instructed to chew on paraffin wax for five minutes and spit out their stimulated residual saliva into a sterilised container. The volume of stimulated saliva was recorded and the salivary flow per minute was subsequently calculated.
Other data
Data on potential covariates were obtained from in-person interviews and included age, gender, ethnicity, education, marital status, oral health related factors and oral health attitude. Housing type has been considered as a social determinant of health (22). In Singapore, there is a mixture of public rental housing together with owner-occupied public housing within each residential precinct. For the needy, heavily subsidized public rental housing is available from the Housing Development Board (HDB). In 2017/2018, Average monthly household expenditure by type of dwelling was S$3,956 for people staying in HDB, S$7,963 for those staying in condominium or apartments and S$10,500 for individuals residing in landed property (23). In our study, we have collected information related to ‘housing type’ of study participants as a proxy of socio-economic status (SES).
Statistical analysis
Decayed and filled root surfaces (DFRS) was calculated by adding the number of decayed and filled root surfaces from a maximum of 32 teeth. In addition, we computed pseudo-R-squared coefficients to summarize the predictive power of the regression models. This type of measure describes the power of the explanatory variables (oral health determinants) to predict the dependent variable (24). The main outcome of this study was DFRS of older adults which was categorised as DFRS≤ 3 and DFRS > 3 (it was categorised based on Mean DFRS (SD) of sample population which was 3(3.5)). During statistical analysis DFRS ≤3 was considered as a reference group. To identify correlation between DFRS and other host factors including: age, gender, ethnicity, education, marital status, housing type (proxy of SES), dental visits, self-perceived oral health, recession, coronal caries experience, denture wear, oral health attitude and volume of saliva, we first performed bivariate analysis. After bivariate analysis, we conducted multivariate analysis to predict potential factors associated with DFRS of study participants.
For final prediction model, we performed a backward stepwise model building approach that eliminates and includes variables into the model with p-value>0.09 and p-value<0.05 respectively. P-value <0.05 was considered significant. All statistical analyses were performed using SPSS version 24.