2.1

Population and sample

Participants were recruited from a population of care-dependent people living in randomly chosen residential aged care facilities (RACFs) in South-East Netherlands. Purposive sampling was applied, aimed at achieving adequate numbers of residents with regard to variables whose outcomes were known a-priori and that were expected to influence DSU or brushing frequency (e.g. prosthodontic status (dentate/edentate) and level of care-dependency) . Following instructions by the principal researcher, managers of RACFs asked residents aged 65 and over who were, according to the manager, sufficiently cognitively alert to participate. Details of the sample are described elsewhere .

2.2

Questionnaire and variables

Self-reported data were obtained using a questionnaire that was administered through a personal interview. Questions not part of a validated questionnaire were pre-tested for comprehensibility and wording by a panel that comprised three senior dental researchers and three care-dependent older people.

The distinction made between frailty-related variables and non-frailty-related variables was based on Gobbens’ definition of frailty: ‘a dynamic state affecting an individual who experiences losses in one or more domains of human functioning (physical, psychological, or social)’ . Hence, 11 out of the 18 predisposing factors (including general health), and 7 out of the 8 enabling factors were considered directly frailty-related (FR, see Sections below and Table 1 , left column). Other variables, such as age and oral health status variables, were considered not or only indirectly frailty-related.

2.3

Predisposing factors

Data were collected on age, gender, and socioeconomic status (SES). SES (high/middle/low) was determined on the basis of the highest level of either education (high/middle/low) or last-held occupation (ISCO-08 classification ).

The only oral health variable that was included in the list of predisposing factors was ‘prosthodontic status’, as assessed by calibrated dental students (more details are provided in the sections ‘Need factors’ and ‘Oral health outcomes’ below). Prosthodontic status comprised: dentulous people with at least one natural teeth and without a removable dental prosthesis (RDP), dentulous people with at least one natural tooth and one or more RDPs, and edentulous people with complete RDPs (CRDPs). Edentulous elders who did not wear CRDPs were excluded, since they do not brush dentures or, in general, use dental services.

Perceived general health was assessed using the question: ‘How would you rate your general health?’ (very bad/bad/moderate/good/very good). Physical and mental health were assessed through the validated SF-12 (Short Form) health survey . Answers to SF-12 questions were used to compute a physical component summary (PCS) score and a mental component summary (MCS) score (using mean Dutch population-based (70–79 age group) norm scores of 44.06 (PCS) and 49.50 (MCS) ). Higher scores indicate better health. Social support was assessed through the validated ENRICHD Social Support Index (ESSI) . The ESSI consists of seven items/questions scored on a 1–5 point Likert scale; higher scores reflect more social support. Health variables that were considered to be frailty-related were derived from single validated SF-36 or SF-12 questions. Original multi-level answers were dichotomized: suffering from chronic pain (6-level answers dichotomized into yes/no); feeling depressed (6-level into yes/no); energy level (6-level into low/normal or high); mobility (3-level into good/moderate or bad); dexterity (3-level into good/moderate or bad). With regard to care-dependency, we included participants with care-dependency level 1 (low dependency) through 6 (high dependency) according to the Dutch national care-dependency classification system . In this system, the intensity and type of care needed are regularly determined by a medical authority. People with care-dependency level 5 were excluded since this level comprises predominantly cognitive impairment.

We also included variables indicative of people’s beliefs and attitudes to dental health services and oral health, based on the following questions: ‘Do you believe that use of dental services helps to maintain or improve your oral health?’ (yes/no); ‘If you think back to the time before you became care-dependent, which statement applies best to you’: ‘My oral health is more/equally/less important to me now than before the onset of my care-dependency’; ‘Do you have fear of going to the dentist?’ (yes/no); ‘Are you dissatisfied with your current (or most recent) dentist?’ (yes/not applicable/no); and ‘Would you like to use dental services more often than you do now?’ (yes/no).

2.4

Enabling factors

Specific barriers with regard to DSU were based on the questions: ‘Are costs of DSU a barrier to you?’ (yes/no); ‘Do you have difficulty finding a dentist?’ (yes/not applicable/no); ‘Do you have difficulty arranging a dental visit?’ (yes/not applicable/no); and ‘Do you have difficulty going to the dentist?’ (yes/not applicable/no). With regard to social support, we constructed the variable ‘lack of social support to go to the dentist’, which was based on the question: ‘In case you need or would need someone’s help to go to the dentist, what describes your situation best’: lack of support = ‘I ask support but no one is able to help me’/‘There is no one whom I can/could ask to help me’/‘I do or would not dare to ask anyone’; no lack of support = ‘I do ask and I do get support’/‘I would ask and I would expect support’/‘I do or would not ask since dental visits are not important enough to me to ask support for’/‘I do or would not ask since I do not wish to go to the dentist’.

With regard to brushing behavior, barriers were based on three questions, the first being ‘Do you ever forget to brush your teeth or dentures?’((almost) never/sometimes or often). In case someone indicated that (s)he was assisted by a caregiver in oral hygiene practices, we included the same question regarding the caregiver: ‘Does your caregiver ever forget…?’ We also asked: ‘Do you find it hard to clean your teeth or dentures?’ ((almost) never/sometimes or often); and ‘Can you summon the effort to brush your teeth or dentures?’((almost) always/sometimes or (almost) never).

2.5

Need factors

Clinical data were obtained through clinical oral examination according to WHO criteria by final-year calibrated dental students (all κ’s > 0.82; overall κ = 0.87; agreement = 90.1%) and final-year calibrated dental hygiene students (all κ’s > 0.66; overall κ = 0.74; agreement = 84.4%). Data included presence of RDPs or CRDPs (see ‘Predisposing factors’ above) and clinically assessed treatment need (yes/no), which comprised any need for professional dental treatment, including reline, rebase or replacement of RDPs or CRDPs, and periodontal treatment. The variable ‘perceived oral treatment need’ was based on the question: ‘Do you think you would need any type of oral treatment at the moment?’ (yes/no). Furthermore, presence of oral health complaints (yes/no) was assessed through combining the answers to two questions: ‘Do you experience pain or discomfort caused by your natural teeth or gums?’ and ‘Do you experience pain or discomfort caused by your removable or fixed prostheses?’ (nearly) always/sometimes/never or hardly ever). If the answers to both questions were ‘never or hardly ever’, the presence of oral complaints was set to ‘no’; otherwise it was set to ‘yes’.

2.6

Oral health outcomes

Self-reported oral health outcomes included a question on self-perceived oral health: ‘How do you perceive your oral health?’ (very bad/bad/moderate/good/very good) and oral health-related quality of life (OHRQoL), which was measured using the validated Dutch version of the Geriatric Oral Health Assessment Index (GOHAI) . The GOHAI consists of 12 questions on experienced functional and psychosocial impacts of oral health, scored on a 1–5 point Likert scale with higher scores indicating better OHRQoL. Clinically assessed outcomes included number of natural teeth and presence of one or more carious teeth (yes/no).

2.7

Oral health care behavior variables

The variable ‘DSU frequency’ (DSU Frequency) referred to the self-reported number of visits to a dentist and/or oral hygienist in the past 3 years for all types of professional oral health care such as routine check-up and complaint-based visits, and visits for curative treatment. The variable ‘changed dental service use since the onset of care-dependency’ (DSU Change) referred to a change of the frequency of (all types of) dental visits. DSU Change was assessed through the question: ‘If you think back to the time before you became care-dependent, which statement applies best to you’: ‘I used dental services about as frequently/more frequently/less frequently than I do now’. The variable ‘brushing frequency’ (Brushing Frequency) referred to the frequency of brushing teeth and/or cleaning RDPs or CRDPs. The variable ‘changed brushing frequency since the onset of care-dependency’ (Brushing Change) referred to a change of the frequency of brushing of teeth and/or cleaning RDPs or CRDPs and was assessed through the question: ‘If you think back to the time before you became care-dependent, which statement applies best to you’: ‘I brushed my teeth and/or cleaned my dentures about as frequently/more frequently/less frequently than I do now’.

DSU Frequency outcomes were dichotomized into ‘higher’: for dentates ≥ 1 visit per year in the past 3 years, for edentates ≥ 1 visit in the past 3 years vs. ‘lower’: less frequently. DSU Change outcomes were dichotomized into equal or higher vs. lower DSU frequency since the onset of care-dependency. Brushing Frequency outcomes were dichotomized into ‘higher’: for dentates ≥ 2 times daily, for dentates with RDP(s) ≥ 2 times daily and ≥ 1 time daily cleaning the RDP(s); for edentates cleaning the CRDPs ≥ 1 time daily vs. ‘lower’: less frequent. Brushing Change outcomes were dichotomized into equal or higher vs. lower brushing frequency since the onset of care-dependency.

2.8

Statistical analysis

In order to assess bivariate associations between the (dependent) oral health care behavior variables DSU Frequency, DSU Change, Brushing Frequency and Brushing Change, and (independent) variables of the predisposing, enabling, and need domains and health outcomes, frequency analyses and univariate logistic regression analyses were performed.

Addressing confounding effects, the associations between dependent variables and independent variables from the predisposing, enabling, and need domains were further examined multivariately using binary logistic regression analysis. This resulted in a model for each dependent variable: DSU Frequency (Model 1), DSU Change (Model 2), Brushing Frequency (Model 3), and Brushing Change (Model 4). Maximal model size was determined for each model, based on the presumption that a minimum of five observations/participants for each independent variable (based on the least occurring outcome event) is required for outcomes with acceptable accuracy in binary logistic regression . For each dependent variable, the associated independent variables with highest significance levels (lowest p-values, as derived from the univariate logistic regression) were selected up to the number of variables that was allowed, based on the maximal model size.The thus-obtained sets of independent variables were entered in the binary logistic regression analysis in order to construct the four respective models. The predictive efficiency of the resulting models was assessed through calculating percentages of correctly predicted cases and through measurement of the area under the ROC curve (AUC) of plotted predicted values. SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) was used for all data analyses.

2.9

Ethics approval and consent to participate

The study was approved by the Medical Ethics Committee (CMO) of the Radboud University Nijmegen Medical Center (CMO ref. 2012/294). All participants gave informed consent in writing to participate in the study and to publish anonymized results.