Sample and study design
A cross-sectional study was conducted using secondary data from the 2010 National Oral Health Survey (SBBrasil 2010) developed by the Brazilian Health Ministry in the five regions of the country . The geographic division of the country into regions was determined by the Brazilian Institute of Geography and Statistics and is adopted in epidemiological studies with a national scope. These regions were part of the sampling plan with the 27 state capitals (including the Federal District), totaling 32 domains formed by 177 municipalities (the 27 capitals + 30 municipalities in each region). The sample was obtained through a random selection of municipalities and census sectors, characterizing multi-stage cluster sampling with probability proportional to population size . Detailed information on the sample planning can be found elsewhere [19,20].
The participants of the present study were dentate adults between 35 and 44 years of age who did not wear any type of fixed or removable denture. Edentulous individuals and those who wore dentures were excluded because the system adopted for the classification of reduced dentition, which was that proposed by NGUYEN et al. (2011) , is based on only natural remaining teeth. Moreover, this exclusion criterion was chosen considering the unavailability of information on satisfaction with one's dentures and therefore the impossibility of controlling for the effect of this variable on self-reported perceptions of denture need.
Data collection involved oral examinations and interviews using questionnaires addressing demographic and socioeconomic characteristics as well as perceptions regarding oral health. The field teams were formed by examiners who had undergone calibration exercises (Kappa >0.65) and trained annotators . The oral examinations were performed following the guidelines of the World Health Organization (WHO) manual for epidemiological studies , using the Decayed, Missing and Filled Teeth (DMFT) index, Community Periodontal Index and clinical attachment loss for the determination of the dental and periodontal status, respectively. The total number of teeth was determined by the number of teeth present, excluding codes 4 and 5 (missing) and 8 (unerupted) of the DMFT index. A posterior occluding pair (POP) was defined as a pair of antagonist posterior teeth on each side of the mouth, such as the pair formed by teeth 16 and 46, for example.
The self-perceived need for CD was determined by the answer to the following question: “In your opinion, do you need complete dentures or need to exchange your current dentures?” , for which the response options were "yes", "no" or "does not know/did not answer". As denture wearers were excluded from the study, the response referred to the self-perceived need for complete dentures. This criterion eliminated those who reported wanting to exchange their CD.
Assessment of dentition
A dentition classification system was employed for the definition of reduced dentitions. This system consists of five hierarchical levels based on the functionality of the teeth in terms of esthetics and occlusion [21,23]. A functional dentition (FDClassV) [21,23] was recorded when the dentition had all five of the following levels: Level I – at least one natural tooth in the maxilla and mandible; Level II – at least 10 teeth in each arch to enable nine to 10 pairs of opposing teeth; Level III – all 12 anterior teeth present; Level IV – at least three pre-molar POPs; and Level V – at least one molar POP bilaterally.
Based on this system, the following categories were defined:
a) 10 teeth in each arch, <12 anterior teeth, with or without ≥3 premolar POPs, with or without ≥1 molar POP bilaterally (reference category)
b) Functional dentition (FDClassV)
c) 10 teeth in each arch, 12 anterior teeth, ≥3 premolar POPs, without ≥1 molar POP bilaterally
d) 10 teeth in each arch, 12 anterior teeth, without ≥3 premolar POPs, with or without ≥1 molar POP bilaterally
e) Less than 10 teeth in each arch, 12 anterior teeth, with or without ≥3 premolar POPs, with or without ≥1 molar POP bilaterally
f) Less than 10 teeth in each arch, <12 anterior teeth, ≥3 premolar POPs, with or without ≥1 molar POP bilaterally
g) Less than 10 teeth in each arch, <12 anterior teeth, without ≥3 premolar POPs, ≥1 molar POP bilaterally
h) Less than 10 teeth in each arch, <12 anterior teeth, without ≥3 premolar POPs, without ≥1 molar POP bilaterally
i) Less than one tooth in each arch
The reference category was adopted based on the concept of "20 well-distributed teeth", with at least 10 teeth in each arch, which is believed to ensure minimum functionality. This category is also based only on the number of teeth, without considering tooth type and occlusal functions and was used for the purposes of comparison with other dentitions considering all or some of the functionality criteria defined by NGUYEN et al. (2011) . This comparison enables the identification of specific criteria from which an association with a self-perceived need for CD may be found.
Assessment of oral impacts
Oral impacts were assessed using nine items of the Oral Impacts on Daily Performance (OIDP) scale, which is widely used for the assessment of OHRQoL . The OIDP addresses problems caused by the teeth in the previous six months regarding the following aspects of daily living: eating, cleaning the teeth, affected mood, having fun, practicing physical activities, speaking, smiling without embarrassment, working and sleeping. Each item has dichotomous response options (0: absence of impact; 1: presence of impact). Oral impact was recorded for any individual who reported that problems with the teeth affected one or more aspects of daily living (OIDP >1).
The covariables were demographic characteristics (sex and self-declared skin color [white, black, brown, yellow/indigenous]), socioeconomic characteristics (income and schooling), perceptions regarding oral health and symptoms (satisfaction with teeth/mouth and reports of dental pain in the previous six months) and clinical variables (gingival bleeding, calculus, shallow periodontal pocket [4-5 mm] and deep pocket [6 mm]). Income was the sum of monthly family earnings determined in Brazilian currency and converted into American dollars (mean exchange rate in 2010: R$1.76 = US$1.00) in the following categories: ≤ US $ 284, US $ 285 to US $ 852, US $ 853 to US $ 2557 and > US $ 2557). Schooling was categorized based on complete years of study (≤ four years, five to eight years, nine to 11 years, and 12 or more years). Satisfaction with the teeth/mouth was assessed using the following question: “Regarding your teeth, would you say you are very satisfied, satisfied, neither satisfied nor dissatisfied, dissatisfied or very dissatisfied?"  The response options were grouped into satisfied (very satisfied + satisfied) and dissatisfied (neither satisfied nor dissatisfied + dissatisfied + very dissatisfied).
Descriptive analysis was performed to characterize the sample and obtain the frequency of adults according to the categories of the variables investigated. The chi-square test was used to compare denture wearers and non-wearers. Crude and adjusted logistic regression models were then used to estimate the association between dentition and the self-perceived need for CD. The variables maintained in the final multiple model were those for which the association with the response variable was statistically significant (p ≤0.05) or that contributed to the fit of the model and had importance recognized in the literature (sex and schooling). The goodness of fit of the model was evaluated using the Hosmer-Lemeshow test. Variables that remained in the final model were used in a structural equation model to evaluate direct and indirect associations between a reduced dentition and the self-perceived need for complete dentures.
Structural equation modeling consists of a measurement model that establishes how latent constructs are measured and a structural model used to analyze associations between variables. In the present study, the latent variable (represented by a circle) was obtained through the nine OIDP items using confirmatory factor analysis. The parameters were estimated using weighted least squares and variance estimates. We estimated the total effects, which are composed of both direct effects (a direct path from one variable to another [e.g., dentition → self-perceived denture need]) and indirect effects (path mediated by other variables [e.g., income → self-perceived denture need via oral impacts]). The estimates of the parameters of direct and indirect associations and 95% confidence intervals (CI) were determined using the bootstrapping method with 1500 iterations. The goodness of fit of the model was evaluated using the root mean square error of approximation (RMSEA), the comparative fit index (CFI) and the goodness-of-fit index (GFI). RMSEA <0.05 indicates a strong fit, 0.5 to 0.08 indicates a reasonable fit and >0.1 indicates an inadequate fit. CFI and GFI of 1.0 indicate a complete fit of the model and CFI and GFI >0.95 indicate a good fit [25,26].
All analyses were performed using Stata® 15.0 (StataCorp, College Station, Texas, USA) and Mplus® 8.3 (Muthén & Muthén, Los Angeles, California, USA), considering complex sampling and sample weights.
The SBBrasil 2010 project was conducted in accordance with the standards stipulated in the Declaration of Helsinki and received approval from the National Human Research Ethics Committee (certificate number: 15.498, on January 7th, 2010).