Factors associated with agreement between self-perception and clinical evaluation of dental treatment needs in adults in Brazil and Minas Gerais

This study sought to describe the agreement between self-perception and clinical evaluation of dental treatment needs in adults and analyze associated factors. The sample comprised adult individuals who took part in SBBrazil 2010 and SBMinas Gerais 2012. The study’s outcome was agreement between self-perception and clinical evaluation of dental treatment needs. We used multiple Poisson regression in order to determine the factors associated with the outcome. Agreement between self-perception and clinical evaluation was 78.8% in Brazil and 73.8% in Minas Gerais. Clinical and self-reported oral health conditions that affect function and quality of life were associated with a higher agreement, while a recent visit to the dentist was associated with a lower agreement. Identifying associated factors may enable the development of questionnaires that favor correct self-perception regarding treatment needs. Diagnostic Self Evaluation; Dental Health Surveys; Oral Health Correspondence A. R. Nascimento Rua Doresópolis 591, Bloco 3, apto. 402, Belo Horizonte, MG 30190-002, Brasil. alexrnbr@gmail.com 1 Centro de Pesquisas René Rachou, Fundação Oswaldo Cruz, Belo Horizonte, Brasil. 2 Secretaria de Estado de Saúde de Minas Gerais, Belo Horizonte, Brasil. 3 Instituto de Ciências Exatas, Universidade Federal de Minas Gerais, Belo Horizonte, Brasil. Nascimento AR et al. 2 Cad. Saúde Pública, Rio de Janeiro, 32(10):e00039115, out, 2016 Introduction Self-perception of oral health conditions, collected either through questionnaires or interviews, is a simpler, more economical instrument than dental exams for use in large-scale epidemiological surveys 1,2,3,4,5. Different studies have described the prevalence of self-perception of dental treatment needs and associated factors 6,7,8,9,10,11 and emphasize the importance of using these indicators to complement clinical indicators 8 in evaluating dental needs, as there are differences between the two 11,12,13. However, lack of agreement between these measures makes self-reports less sensible when they are the only measure used in population surveys to identify demands, when clinical exams performed by trained professionals are not feasible. Thus, we need to know the percentage of agreement between self-perception and clinical assessment, estimating how close self-reports are to professional evaluations and studying factors associated with this agreement. This would be helpful in formulating more sensible questions and constructing strategies seeking to enhance individuals’ self-perception of their clinical conditions, such as dental caries, for example, which, if untreated, may evolve to irreversible outcomes such as dental loss. Agreement percentages may be higher than 70% 14,15,16 or lower than 50% 15,16,17,18 depending on the condition being studied, clinical criteria, target population and the question used in the survey. Tervonen 14 found a 76% agreement for quality of total prostheses and 77% for partial prostheses. Costa et al. 16 found a similar percentage (76.9%) for the agreement between satisfactory quality according to professional evaluations and user satisfaction with total prostheses, however, agreement on unsatisfactory prostheses was 39%. Tervonen & Knuuttila 15 found an 86% agreement between examiner and examinee on the existence of zero to two decayed teeth, and for three or more caries agreement was around 30%. The same authors also found a 40.7% agreement on the need to replace prostheses. In a group of elderly individuals in England, 42% of individuals who had some dental treatment need, according to professional evaluation, agreed with the diagnosis 17. Among German youth, adults, and elderly, there was a 13% agreement on the need for prosthesis treatment and an 18% agreement on the lack thereof 18. The literature on factors associated with agreement is sparse and restricted to evaluations of dental prosthesis needs. Colussi et al. 19 found a 64% agreement regarding need for prosthesis use among elderly individuals and concluded that their underestimation when compared with professional evaluations – that is, when the individuals in the study failed to notice a need reported by the dentist – was associated with sociodemographic variables (male sex, living in a rural area and being over 70 years of age) and service use (not participating in elderly groups and not having sought dental services over much of their lives). In an analysis of the SBBrazil 2010 survey data on the need for total prosthesis use or replacement in the sample of elderly individuals, Nascimento et al. 20 found a 69.2% agreement between self-perception and clinical evaluation by a dentist. Factors associated with higher odds of agreement were sociodemographic (between 70 and 74 years of age, living in the South, Southeast or Central regions of Brazil and having one to four or nine or more years of schooling) and oral health condition variables (last dentist appointment three or more years earlier and already having a total prosthesis). Given this absence in the literature, this article seeks to describe the prevalence of, and analyze factors associated with, the agreement between self-perception and clinical evaluation of dental treatment needs in adults in Brazil and Minas Gerais.


Introduction
Self-perception of oral health conditions, collected either through questionnaires or interviews, is a simpler, more economical instrument than dental exams for use in large-scale epidemiological surveys 1,2,3,4,5 .
Different studies have described the prevalence of self-perception of dental treatment needs and associated factors 6,7,8,9,10,11 and emphasize the importance of using these indicators to complement clinical indicators 8 in evaluating dental needs, as there are differences between the two 11,12,13 .However, lack of agreement between these measures makes self-reports less sensible when they are the only measure used in population surveys to identify demands, when clinical exams performed by trained professionals are not feasible.Thus, we need to know the percentage of agreement between self-perception and clinical assessment, estimating how close self-reports are to professional evaluations and studying factors associated with this agreement.This would be helpful in formulating more sensible questions and constructing strategies seeking to enhance individuals' self-perception of their clinical conditions, such as dental caries, for example, which, if untreated, may evolve to irreversible outcomes such as dental loss.
Agreement percentages may be higher than 70% 14,15,16 or lower than 50% 15,16,17,18 depending on the condition being studied, clinical criteria, target population and the question used in the survey.
Tervonen 14 found a 76% agreement for quality of total prostheses and 77% for partial prostheses.Costa et al. 16 found a similar percentage (76.9%)for the agreement between satisfactory quality according to professional evaluations and user satisfaction with total prostheses, however, agreement on unsatisfactory prostheses was 39%.
Tervonen & Knuuttila 15 found an 86% agreement between examiner and examinee on the existence of zero to two decayed teeth, and for three or more caries agreement was around 30%.The same authors also found a 40.7% agreement on the need to replace prostheses.
In a group of elderly individuals in England, 42% of individuals who had some dental treatment need, according to professional evaluation, agreed with the diagnosis 17 .Among German youth, adults, and elderly, there was a 13% agreement on the need for prosthesis treatment and an 18% agreement on the lack thereof 18 .
The literature on factors associated with agreement is sparse and restricted to evaluations of dental prosthesis needs.Colussi et al. 19 found a 64% agreement regarding need for prosthesis use among elderly individuals and concluded that their underestimation when compared with professional evaluations -that is, when the individuals in the study failed to notice a need reported by the dentist -was associated with sociodemographic variables (male sex, living in a rural area and being over 70 years of age) and service use (not participating in elderly groups and not having sought dental services over much of their lives).
In an analysis of the SBBrazil 2010 survey data on the need for total prosthesis use or replacement in the sample of elderly individuals, Nascimento et al. 20 found a 69.2% agreement between self-perception and clinical evaluation by a dentist.Factors associated with higher odds of agreement were sociodemographic (between 70 and 74 years of age, living in the South, Southeast or Central regions of Brazil and having one to four or nine or more years of schooling) and oral health condition variables (last dentist appointment three or more years earlier and already having a total prosthesis).
Given this absence in the literature, this article seeks to describe the prevalence of, and analyze factors associated with, the agreement between self-perception and clinical evaluation of dental treatment needs in adults in Brazil and Minas Gerais.

Methods
This study used data from the Brazilian National Oral Health Survey (SBBrazil 2010) and the Survey on Oral Health Conditions of the Minas Gerais Population (SBMinas Gerais 2012).SBBrazil 2010 was the most recent national epidemiological survey on oral health.Its sample was planned to be representative of the country as a whole, each of the five regions, cities not located in the Metropolitan Regions, and the capitals and Federal District 21,22 .
The SBMinas Gerais 2012 was carried out in 2012 in order to obtain representative data for Minas Gerais State.This survey is representative of the state as a whole and each of the two domains of cities not located in the Metropolitan Regions, defined according to the Allocation Factor, an indicator built based on the Health Needs Index and the Economic Size Index.The Allocation Factor classifies Minas Gerais cities in quartiles.Cities classified in Group 1 have lower relative need of financial resources and those of Group 4 have higher need.For the SBMinas Gerais 2012, cities not located in the Metropolitan Regions were grouped into two domains, one formed by Groups 1 and 2 and the other, by Groups 3 and 4 23 .
This study used the SBBrazil 2010 sample of adults, comprising 9,564 individuals 21 .For the SBMinas Gerais 2012 analysis, the sample comprised all 1,182 adults who were examined 23 .
Self-perception of dental treatment needs was assessed through the following question that was part of both surveys and asked of all participants: "do you believe that you currently need dental treatment?".There were three possible answers (no; yes; did not know/would not answer).We only included the first two in our analysis.
Clinical exams for determining oral conditions were carried out in accordance with criteria established by the World Health Organization 24 .Individuals were classified as having dental treatment needs by the clinical evaluation when at least one of the following conditions was present: crown or root caries, calcified dental plaque or periodontal pockets, need for dental prosthesis use or replacement.
The study's outcome was the agreement between self-perception and clinical evaluation of dental treatment needs.This variable took on the following values: (1) agreement: dentist and examinee both considered there was need for treatment, or both considered this need was absent; (2) disagreement: there was a clinical need, but the individual did not perceive it (underestimation), or there was no clinical need, but the individual declared needing treatment (overestimation).We chose to categorize the outcome in two categories due the low percentage of overestimation, which would harm analytical precision if the outcome were used as a multinomial variable.
Explanatory variables were: sex (female and male); age group (35 to 39 years and 40 to 44 years); schooling (0 to 4 years, 5 to 8 years, 9 years or more); last visit to an oral health service (not recent -those who reported never having been to a dentist or having done so over a year previously -and recent -those who reported having been to a dentist in the previous year); reason for last dentist appointment (prevention/check up; pain; extraction/treatment/other); number of remaining teeth (0 to 20 teeth and 21 or more); use of some dental prosthesis (no and yes); presence of caries (no and yes); presence of periodontal alteration (no and yes); need for prosthesis (no and yes); reporting dental pain in the previous six months (no and yes); and reporting impact on daily life due to oral conditions (no and yes for each of the Oral Impact on Daily Profile (OIDP) 25 : difficulty when eating or ingesting liquids, discomfort when brushing teeth, nervousness or irritation due to teeth, impediment to going out or having fun due to teeth, stopped practicing sport due to teeth, had difficulty speaking due to teeth, felt embarrassed speaking or smiling due to teeth, teeth get in the way of work tasks, stopped sleeping or slept badly due to teeth).
We used a bivariate analysis to verify possible associations between explanatory variables and the outcome, using a Pearson chi-square test with Rao-Scott correction 26 .
In order to determine the factors associated with the outcome, we adjusted Poisson regression models, calculating the prevalence ratio (PR) and respective 95% confidence intervals (95%CI) 27 .Variables were included in the following order: sociodemographic, health service use, clinical oral health measures (number of remaining teeth, current use of a dental prosthesis, presence of caries, presence of periodontal alteration, need for prosthesis), and self-reported impacts resulting from oral conditions (reporting dental pain in the previous six months and dimensions from the OIDP).
Sociodemographic and service use variables were used in the final models to control for the other variables, so that we could identify the influence of clinical and self-reported oral health conditions on the outcome.We included clinical measures and self-reported conditions in the models one by one.Only those that were significant at the 5% level remained in the final models.
We used the software R version 3.0.0(The R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org) for the data analysis.Given the complex design of the SB-Brazil 2010 and SBMinas Gerais surveys, we included the sampling parameters when calculating point estimations and respective confidence intervals, using the Survey package.This guaranteed a robust estimation of variance in the Poisson models.
The SBBrasil 2010 survey was submitted to the Health Ministry's Ethical Review Board and was approved and registered with the National Committee for Ethics in Research of the National Health Council under the number 15,498, on January 7, 2010 21 .
The SBMinas Gerais survey was submitted to the Ethical Review Board of the Pontifical Catholic University of Minas Gerais and was approved on March 28, 2012, and registered under the number 9,173.

Results
Of the total of adults examined in the SBBrazil 2010, 9,358 had valid data for self-perception of dental treatment needs.In SBMinas Gerais, there were 1,162 individuals.
Most of the adults in both surveys were female, aged between 35 and 39 years, had nine or more years of schooling, had their last dentist appointment over one year previously, did not report dental pain in the six preceding months, had 21 or more teeth and did not use dental prostheses.The description of the study population's characteristics can be found in Table 1.
We present the percentages of agreement, underestimation and overestimation between self-perception and clinical evaluation in Table 2. Results were similar for SBBrazil 2010 and SBMinas Gerais 2012, with high agreement and higher underestimation than overestimation.
The bivariate analyses of the outcome and explanatory variables are presented in Table 3 (SBBrazil 2010) and Table 4 (SBMinas Gerais 2012).Of the sociodemographic variables, only schooling was associated with the outcome in both bivariate analyses.Time elapsed since the last dentist appointment and reason for last dentist appointment were also associated with the outcome in both surveys, as did the presence of clinical and self-reported conditions.
Table 5 presents the final models of the Poisson Regressions, for SBBrazil 2010 and SBMinas Gerais 2012 separately, with respective Prevalence Ratios, 95%CI and p-values.There was a negative association with a recent dentist appointment in both samples.On the other hand, there was a positive association with the reason for the last appointment only for SBBrazil.In both surveys, clinical conditions and self-reported discomfort were positively associated with agreement between self-perception and clinical evaluation of treatment needs.

Discussion
Agreement between self-perception and clinical evaluation of dental treatment needs was higher than 70% in both surveys.Other studies have described agreement on quality of dental prostheses (76% for total prostheses 14 and 77% for partial prostheses 14 , 76.9% for satisfactory total prostheses 16 and 39% for unsatisfactory total prostheses 16 ), on the need for treatment among elderly individuals (42%) 17 or on the need for prostheses among elderly individuals (64% 19 and 69.2% 20 ).However, considering the current literature, this is the first study to analyze agreement on dental treatment needs in adults as the outcome, identifying associated factors.
Underestimation was higher than overestimation in both samples, reaching around 20% of individuals.Underestimation of treatment needs may lead individuals not to seek dental care, since self-perception is one of the factors that influence use of oral health service 8,28 .
Sociodemographic variables were not associated with agreement in the final model.Only schooling was significantly associated in the bivariate analysis and in the initial regression models.Individuals with nine or more years of schooling agreed less with the clinical evaluation.However, this association was not present in the final model, losing significance with the inclusion of oral health condition variables.These findings indicate that sociodemographic differences among individuals in this study do not directly influence their capacity to perceive the presence or absence of oral problems in agreement with clinical evaluations.Therefore, educational strategies seeking to increase this agreement may be directed at heterogeneous groups of adults in a population.Studies on agreement regarding prosthesis needs among the elderly have found an association with demographic characteristics, revealing a difference in the factors that influence agreement depending on the condition being studied and the target population.
Individuals who had a recent visit to the dentist agreed less with the clinical evaluation, however, the reasons supporting this result are not clear.A study with young adults found that those who had been to the dentist less than one year previously had lower odds of reporting a need for treatment 6 .A possible hypothesis is that those individuals considered they no longer needed treatment, despite having unresolved clinical needs.This hypothesis is reinforced by the manner in which the question was asked, because it did not specify the number of appointments or procedures, or the effectiveness of treatments or resolution of all needs present at the time of the appointment.
These results show the importance of informing individuals who have recently been to the dentist about the need for continued care until all treatment demands are met.
The reason for the last dental appointment was also associated with agreement, though only in the Brazilian sample.Individuals who sought an oral health service due to oral problems agreed more with clinical evaluations than those who sought preventive care.Likewise, clinical oral health conditions were associated with agreement.The existence of carious lesions and the need for prostheses remained in the final models for both samples.Periodontal disease had an association in the SBBrazil.Studies addressing self-perception of treatment needs show that the presence of oral problems is associated with selfreporting of those needs 6,29 .The presence of 21 or more teeth was associated with agreement in the Brazilian sample.Underestimation by adults with fewer than 20 teeth may lead to an aggravation of morbidities, which, if left untreated, tend to worsen, possibly leading to dental loss.
Reports of pain and impact on daily activities were associated with agreement between self-perception and clinical evaluation.Feeling ashamed when smiling or speaking remained in the final models for both surveys, indicating that aesthetic concerns are an important factor to be considered.Other OIDP components were associated with the outcome.Discomfort when brushing teeth and difficulty when eating or pain when ingesting liquids remained in the final SB-Brazil model.On the other hand, in the SBMinas Gerais survey, there was an association with nervousness or irritation due to teeth.Despite this difference between the two samples, in both, variables remaining in the final models were the ones that affect individuals' quality of life.
The literature on self-perception of treatment needs points in a similar direction.Oral conditions that are apparent to individuals are associated with self-reporting treatment need 12,30 .Bedos et al. 31 found that pain predisposes individuals to perceiving the need for treatment; other authors found a similar association 6,7,9,10,29 .Reporting some types of impact on daily activities is also associated with the self-perception of treatment needs 7,10,11 .
The fact that we used two large oral health surveys carried out recently enables us to compare  results in two different contexts.In the final models, similar significant associations were verified in SBBrazil 2010 and SBMinas Gerais 2012, showing the stability of the outcome in the national and state samples.However, answers obtained through questionnaires always have some degree of imprecision, both because of the way respondents understand questions and because of how questions are phrased.The question used in the two surveys does not specify whether "dental treatment" refers exclusively to carious lesions or to any type of dental intervention.Thus, we cannot determine which types of treatments respondents considered when formulating their answers.For this reason, we chose to compare self-perception with any clinical need present at the time of the exam.
This article shows that agreement between self-perception and clinical evaluation of dental treatment needs was high in both surveys we analyzed.Clinical and self-reported oral health conditions that affect function and quality of life were associated with a higher agreement, which may enable the development of instruments that reinforce individuals' correct evaluation of their treatment needs, since questionnaires that combine several questions or answer options tend to have high values of validity 32,33 .
The results we found also draw attention to the need for developing educational strategies to enlighten the population on the need for continuing dental treatments, as well as on the chronic and asymptomatic nature of some oral health harms, which may be treated in order to avoid future aggravation, if they are underestimated by the individual.Prevalence ratio (PR) and 95% confidence intervals (95%CI) for the agreement between self-perception and clinical evaluation of treatment needs in adults.

Table 1
Description of adults according to sociodemographic characteristics, service use, clinical and self-reported oral health conditions.SBBrazil 2010 and SBMinas

Table 2
Percentages of agreement, underestimation and overestimation between self-perception and clinical evaluation.SBBrazil 2010 and SBMinas Gerais 2012.

Table 3
Bivariate analysis of the agreement between self-perception and clinical evaluation of need for dental treatment in adults.

Table 4
Bivariate analysis of the agreement between self-perception and clinical evaluation of need for dental treatment in adults.
A. R. Nascimento, F. B. Andrade and C. C. César contributed to the study elaboration, data analysis and interpretation, writing and revision of the text, and final approval of the article.