Estimates and multivariable risk assessment of gingival recession in the population of adults from Porto Alegre, Brazil

Transcription

1 J Clin Periodontol 2014; 41: doi: /jcpe Estimates and multivariable risk assessment of gingival recession in the population of adults from Porto Alegre, Brazil Fernando S. Rios 1, Ricardo S. A. Costa 1, Mauricio S. Moura 2, Juliana J. Jardim 2, Marisa Maltz 2 and Alex N. Haas 1 1 Faculty of Dentistry, Periodontology, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; 2 Faculty of Dentistry, Preventive and Social Dentistry, Federal University of Rio Grande do Sul, Porto Alegre, Brazil Rios FS, Costa RSA, Moura MS, Jardim JJ, Maltz M, Haas AN. Estimates and multivariable risk assessment of gingival recession in the population of adults from Porto Alegre, Brazil. J Clin Periodontol 2014; 41: doi: / jcpe Abstract Aim: To describe estimates and determine risk indicators of gingival recession in the urban population of Porto Alegre, Brazil. Materials and Methods: A representative sample of 1023 adults 35 years and older was obtained using a multistage probability sampling strategy. A structured questionnaire was applied and a clinical examination was performed at four sites of all teeth present. Complex survey commands were used for the estimation of gingival recession (GR) and during risk assessment. Results: GR 1 mm affected 99.7% of subjects. The percentage of subjects with 1 tooth with GR 3 mm and 5 mm was 75.4% and 40.7%, respectively. 67.6%, 27.8% and 9.5% of teeth per subject showed GR 1 mm, 3 mm and 5 mm, respectively. Older age, male gender, smoking exposure, poor selfreported oral hygiene, history of periodontal treatment and high percentage of calculus were significant risk indicators for GR found after multivariable risk assessment. When buccal GR was analysed separately, only gender, age, smoking and high education were significant risk indicators for GR, whereas variables related to oral hygiene were not associated. Conclusion: GR is highly prevalent in this Brazilian population. Preventive strategies for GR may target a variety of socio-demographic, behavioural and clinical risk indicators. Key words: Brazil; epidemiology; gingival recession; prevalence; risk indicators Accepted for publication 19 August 2014 Gingival recession (GR) is characterized by the apical migration of the gingiva resulting in the exposure of the root surface (Smith 1997, Kassab & Cohen 2003, Du et al. 2009). GR Conflict of interest and source of funding The authors declare no conflict of interest associated with the present study has not been considered a major concern in the periodontal epidemiology contrasting its aesthetic consequences, functional disabilities, and increased risk of root caries (Kularatne & Ekanayake 2007). Since the classic study by L oe et al. (1992) in Sri Lanka and Norway, few population-based studies have been conducted. In this regard, only three surveys assessed GR as the primary outcome using representative samples, one in the United States (Albandar & Kingman 1999), one in Brazil (Susin et al. 2004) and another in France (Sarfati et al. 2010). Other population-based studies only assessed GR as a secondary outcome and reported very limited estimates of the condition (Thomson et al. 2000, Dye et al. 2007, Holtfreter et al. 2009). The remaining literature refers to studies with convenience samples (van Palenstein

2 Estimates of gingival recession 1099 Helderman et al. 1998, Slutzkey & Levin 2008, Toker & Ozdemir 2009, Minaya-Sanchez et al. 2012). According to data derived from the three available population studies, GR affects a significant proportion of subjects and teeth per subject. Using data from NHANES III, Albandar & Kingman (1999) found that 58% of the adult population in the United States presented GR 1 mm. Higher estimates of buccal GR were observed in France, reaching 84.6% of the subjects (Sarfati et al. 2010). In Brazil, three national surveys were conducted since 1980 and none of them assessed GR (Brasil 1986, 2004, 2011). The epidemiology of GR was first assessed in Brazil by Susin et al. (2004) using a representative sample from Porto Alegre observing GR 3 mm in more than half of the individuals. Many socio-demographic, behavioural and clinical risk indicators have been associated with GR. Nevertheless, the majority of the studies failed to account for confounding (L oe et al. 1992, Brown et al. 1996, van Palenstein Helderman et al. 1998, Albandar & Kingman 1999, Slutzkey & Levin 2008, Toker & Ozdemir 2009), with very few studies applying multivariable models (Susin et al. 2004, Sarfati et al. 2010). Overall, GR has been associated with traditional risk factors for destructive periodontal disease, including older age, male gender, race and smoking (L oe et al. 1992, Albandar & Kingman 1999, Susin et al. 2004, Sarfati et al. 2010). However, there is lack of epidemiological evidence to support the association between GR and other important factors such as history of periodontal and orthodontic treatments, self-reported oral hygiene practices, and gingival inflammation. The aim of the present study was to assess the prevalence, extent, severity and risk indicators of gingival recession in a representative sample of adult and elderly residents of a capital city located in south Brazil. Materials and Methods Study population This cross-sectional observational population-based study was designed by the Caries-Perio Collaboration Group from the Federal University of Rio Grande do Sul, Porto Alegre, Brazil. The study assessed a variety of oral outcomes including dental caries (coronal and root), dental erosion, dental hypersensitivity, gingivitis, gingival recession and tooth loss. According to census data from 2003, the target population comprised 591,297 inhabitants of both genders aged 35 years and older living in the city of Porto Alegre located in southern Brazil. The study was conducted between June 2011 and June Ethical considerations The study protocol was reviewed and approved by the Research Ethics Committee, Federal University of Rio Grande do Sul, Porto Alegre, Brazil. Before the interview, the participants read and signed an informed consent. Sample size Considering that this is a multi-disciplinary study involving various oral outcomes, the sample size was estimated using the worst-case scenario considering a prevalence of 50% for any oral condition recorded. It was also estimated that the multistage sampling used in the present study would yield approximately 50% inefficiency. A standard formula for prevalence estimation was used adjusting the sample size for the above mentioned design effect. Considering precision of 4% and 95% confidence interval, it was estimated that the required sample size was 940 individuals. Sampling procedures This study employed a multistage probability sampling strategy (Fig. 1) based on information provided by governmental agencies (IBGE 2004). Detailed information is provided in a previous publication (Costa et al. 2014). In the first stage, the city was divided into 86 neighbourhoods comprising the primary sampling units (PSUs). PSUs were stratified into two strata of high and low income. PSUs were randomly selected proportionally to the number of PSUs in each stratum. The second stage consisted of a random selection of sectors proportional to the total number of sectors in each PSU. Sectors were defined as map areas comprising approximately 300 households each. Forty-eight of the 373 eligible sectors were selected. The third stage consisted of selecting households consecutively according to the sector starting point until the sector sample size was reached. The number of individuals to be selected within each sector was estimated based on the proportional distribution of the sample size according to the number of individuals 35 years and older living in each sector. All household members 35 years and older were considered eligible for the study. Individuals were excluded if they presented with any mental or systemic health condition that did not allow them to perform the interview or the clinical examination. Places such as nursing homes and commercial establishments were excluded. Study sample A total of 1600 individuals were eligible for the study. Of these, 375 (23.4%) did not participate in the study. In total, 1225 individuals comprised the whole sample of the study. Among those, 1023 (83.5%) were dentate and were included in the present study. The mean age was 52.6 years (standard deviation 11.8) ranging from 35 to 95 years. Of these, 398 (38.9%) individuals were males and 625 (61.1%) were females. Table 1 depicts the sample characteristics. Response rate Data and reasons for non-response are provided in Fig. 1. A short version of the questionnaire was answered by 219 (58.4%) of the non-respondents that were compared to individuals included in the study sample. Non-respondents were slightly older than respondents ( versus , p = 0.001). The percentages of individuals with high education and socio-economic status were statistically higher in non-respondents compared to respondents. There were no significant differences between respondents and non-respondents in regard to gender and self-reported mean number of lost teeth.

3 1100 Rios et al. Interview Participants were interviewed using a structured questionnaire containing questions regarding socio-demographic variables, oral hygiene habits, self-perceived oral health, access to dental services, medical history and behavioural factors. Three trained and calibrated interviewers conducted the interview. Clinical examination All permanent fully erupted teeth were examined by two periodontists using a manual periodontal probe (PCP10-SE, Hu-Friedy Mfg. Co. Inc., Chicago, IL, USA). Gingivitis, supragingival calculus and GR were assessed at four sites per tooth at mesiobuccal, mid-buccal, distobuccal and mid-lingual surfaces. Gingivitis was assessed using the gingival bleeding index (Ainamo & Bay 1975). Supragingival calculus was assessed using the periodontal probe. Gingival recession was defined as the distance from the cement-enamel junction (CEJ) to the free gingival margin. If the CEJ was located apical to the gingival margin, this assessment was given a negative sign. Fig. 1. Flowchart of sampling strategy and response rate. Statistical analyses accounted for non-response using the inverse probability weighting strategy (Hernan et al. 2004). A non-response weight variable was generated for each sector, considering the eligible and actual numbers of included individuals and the distributions of age, gender, and education. Fieldwork procedures A research team (two examiners and one assistant) conducted the fieldwork for this study. One researcher visited each selected sector 1 day before the start of data collection to invite residents to participate. Residents were not included only after the third attempt of invitation. Interviews and clinical examinations were conducted inside the household. Examinations were conducted using three portable devices: a medical headlight, a portable compressor and a bendable chair. Reproducibility Reliability of the questionnaire was assessed during the fieldwork using the test-retest approach in 50 participants. A set of key-questions was used to assess the reproducibility of the questionnaire, and the Kappa coefficients varied from 0.91 to Intra and inter-examiner reproducibility of GR was assessed before the start of the study and during the fieldwork. Before the start of the study, duplicate measures of GR were conducted in a total of 16 patients (1231 sites) from the Periodontal Department. During the fieldwork, 42 participants (2896 sites) allowed to perform the duplicate measurements. Examiners reliability was assessed by the Intra-class Correlation Coefficient (ICC) and by weighted Kappa (absolute agreement considered for GR records within 1 mm). Initial intra-examiner reliability revealed weighted Kappa values of 0.97 and 0.98, and the inter-examiner value was During the fieldwork, intra and inter-examiner weighted Kappa values were 0.98, 0.99 and 0.91, respectively. ICC values for means of GR ranged between 0.96 and 0.99.

4 Estimates of gingival recession 1101 Table 1. Distribution of study individuals according to the presence and absence of gingival recession 3 mm No GR GR 3 mmin 1 tooth Whole sample Statistical analyses Prevalence of GR was defined by the percentage of dentate individuals presenting at least one tooth with GR of various thresholds. Extent of GR was defined as the percentage of teeth within each dentate individual n % n % n % Age years years years years Gender Male Female Education Low Middle High Socio-economic status Low Middle High Gingivitis Low ( 22%) High (>22%) Calculus 0 19% % % Brushing frequency 1/day /day /day Proximal cleaning Never /day /day Brushing movement Horizontal Vertical Circular All Smoking exposure Never-smokers Moderate smokers Heavy smokers Dental visits None Irregular Regular Periodontal treatment No Yes Orthodontic treatment No Yes Total presenting GR of various thresholds. Mean GR was used to estimate the severity of the condition. Separate analyses were performed considering all tooth sites and only buccal sites. A sampling weight variable was computed using census information provided by IBGE. Complex survey commands were used in all analyses to account for cluster correlations expected for the multistage sampling strategy used in the study. Pair-wise comparisons of crude estimates were carried out using the Wald test. The significance level was set at 5%. Data analyses were performed using a statistical package (Stata 10 for Macintosh, STATA Corp., College Station, USA). Educational level was defined according to years of education into low ( 4 years), middle (5 10 years) and high ( 11 years). Socioeconomic status was categorized using cut-off points adapted from the CCEB classification (ABEP 2013) that considers the amount of consumer goods and the educational level of the head of the family as follows: low ( 20 points), middle (21 26 points) and high ( 27 points). Self-reported tooth brushing movement was assessed as horizontal, vertical, circular and the combination of all. Tooth brushing frequency was categorized into 1 time/day, 2 times/day and 3 times/ day. The frequency of inter-proximal cleaning was categorized into never, 1/day and 2/day. The total number of packs of cigarettes consumed in a lifetime (packyears) was calculated for each individual by multiplying the number of cigarettes consumed per day by the years of habit, and dividing by 20. Smoking exposure was categorized into never-smokers (0 packyears), moderate smokers (<20 packyears) and heavy smokers ( 20 packyears). Individuals were categorized as having regular dental care if they reported going to the dentist for prevention with a frequency of 1 times/year. Dental visits only for emergencies were classified as irregular dental care. No dental care was defined when individuals reported no dental visits during the last three years. Self-reported histories of periodontal and orthodontic treatments were assessed dichotomously. Gingivitis was dichotomized into low and high using the median (22%) of the percentage of bleeding sites. Individuals were arbitrarily categorized into three groups according to the percentage of sites with supragingival calculus using an approximation of

5 1102 Rios et al. the distribution of subjects into quintiles and the amount of calculus considered to be clinically relevant [0 19% (~1st quintile), 20 39% (~2nd quintile) and 40% (~3rd, 4th and 5th quintile)]. Survey binary logistic regression models were fitted to assess risk indicators for GR. Two outcomes were used in separate models for GR 3 mm and 5 mm in at least one tooth. Univariable models were fitted for each independent variable and those presenting p values <0.25 were entered in the multivariable model. Maintenance of variables in the final model was determine by a combination of p values <0.05 and analyses of effect modification (Hosmer & Lemeshow 2000). Collinearity between independent variables was assessed before modelling and none was observed. Also, no interactions were found during model fitting. Goodness-of-fit (GOF) was assessed by the Archer and Lemeshow GOF test for survey logistic regression that takes into account the sampling weights and design (Archer & Lemeshow 2006). The severity of GR was analysed using the individual mean GR, and risk indicators for severity of GR were determined fitting survey multivariable linear regression models considering all sites and buccal sites only using the same strategy described above for model fitting. Assumptions of the linear regression models were evaluated by the distribution of residuals. Multicollinearity and interactions were not identified during model fitting. Results Gingival recession 1 mm was a universal finding in this population (Table 2). Additionally, 75.4%, 40.7% and 12.5% of the individuals presented at least one tooth with GR 3 mm, 5 mm and 7 mm, respectively. A high percentage of teeth (67.6%) with GR 1 mm was observed, whereas the percentages of teeth affected by GR 3 mm, 5 mm and 7 mm were lower. The overall mean GR for all sites was 1.49 mm. The prevalence, extent and severity of GR increased with increasing age. Prevalence, extent and severity estimates of GR were lower when only buccal sites were considered, mainly for higher thresholds of GR. The prevalence of moderate to severe GR was significantly lower in individuals reporting higher frequency of tooth brushing and interproximal cleaning (Table 3). Individuals with higher percentages of calculus presented significantly higher prevalence of GR. The extent of GR was significantly lower for individuals with higher frequencies of tooth brushing and interproximal cleaning, and was significantly higher for individuals with higher percentages of calculus, irrespective of the threshold of GR. The prevalence and extent of GR was also higher in smokers and males compared to never-smokers and females, respectively (Fig. 2). Mandibular central incisors, second premolars and maxillary first molars presented the highest frequencies of GR 1 mm and 3 mm (Supplementary Figure S1). The maxillary central and lateral incisors had the lowest frequency of GR. Table 2. Prevalence (percentage of subjects), extent (percentage of teeth) and severity (mean) of gingival recession according to age strata considering all sites and only buccal sites Age (years) Total Estimate SE Estimate SE Estimate SE Estimate SE Estimate SE All sites Prevalence (% subjects) 1 mm mm mm mm Extent (% teeth) 1 mm mm mm mm Severity (mm) Only buccal sites Prevalence (% subjects) 1 mm mm mm mm Extent (% teeth) 1 mm mm mm mm Severity (mm) SE, standard error.

6 Estimates of gingival recession 1103 Table 3. Prevalence (percentage of subjects) and extent (percentage of teeth) of different thresholds of gingival recession according to oral hygiene variables Gingival Recession Brushing frequency p* Proximal cleaning p* Calculus p* 1/day 2/day 3/day Never 1/day 2/day 0 19% 20 39% 40% % SE % SE % SE % SE % SE % SE % SE % SE % SE Prevalence 1 mm mm < mm < mm <0.001 Extent 1 mm < < mm < mm < mm SE, standard error. * Wald test. Statistically significant associations were observed for gender, age, education, smoking and calculus for the two thresholds of GR in the univariable models (Supplementary Table S1). Additionally, individuals who reported tooth brushing using circular movement had significantly lower chances of having GR 3 mm than horizontal movement. Regular and irregular dental visits were associated with higher chances of GR 5 mm compared to the absence of dental care. Better oral hygiene practices represented by higher frequencies of self-reported tooth brushing and interproximal cleaning were significantly associated with lower chances of GR 5 mm. When only buccal sites were considered, the univariable models followed a similar pattern of associations. In the multivariable logistic regression models, male gender, older age and smoking exposure were found to be statistically significant risk indicators for GR 3 mm in all sites and only buccal sites (Table 4). When GR 5 mm was modelled considering all sites, gender and age were also significant risk indicators, added to a protective effect of higher frequency of tooth brushing. Moreover, dental visits and positive history of periodontal treatment were significantly associated with higher chances of gingival recession. When only buccal sites were analysed, high education was significantly associated with higher chances of GR 5 mm compared to low education, together with gender and age. The overall mean GR was 1.49 mm (SE: 0.06) and increased linearly with age [0.29 mm (SE: 0.10) among years of age to 2.55 (SE: 0.14) among individuals 60 years and older]. Moreover, mean GR was statistically higher among males, older individuals, individuals of high education, smokers, individuals with regular dental visits and with higher percentage of sites with gingivitis when considering all sites in the multivariable analysis (Table 5). Individuals reporting interproximal cleaning ( 1/day or 2/day) had lower chances of having GR than those who never cleaned the interproximal area. With regard to buccal sites, proximal cleaning, moderate smoking and gingivitis

7 1104 Rios et al. Fig. 2. Prevalence (percentage of subjects) and extent (percentage of teeth) of gingival recession of 3 mm or more according to smoking habit and gender. were not significantly associated with the severity of GR, whereas male gender, older age, high education, heavy smoking and regular dental visits remained as risk indicators. Discussion The present population-based study demonstrated that the prevalence of GR was high among adults 35 years and older living in a capital city from Brazil. Older age, male gender, cigarette smoking, lower frequency of tooth brushing, higher percentage of calculus and access to dental and periodontal care were found to be significantly associated with GR using multivariable risk assessment. When buccal gingival recession was assessed separately, only age, smoking, gender and high education were found to be significant risk indicators. GR has not been considered a major concern in the periodontal epidemiology, in contrast to the numerous interventional studies that evaluate aesthetic approaches (Chambrone et al. 2010). Only three population-based studies assessed GR as the primary outcome (Albandar & Kingman 1999, Susin et al. 2004, Sarfati et al. 2010). Susin and co-workers conducted one of these studies first assessing the epidemiology of GR in the metropolitan area of Porto Alegre (Susin et al. 2004). Although their study has deeply Table 4. Multivariable logistic regression models of risk indicators for gingival recession 3 mm and 5 mm considering all sites and only buccal sites separately All sites Buccal sites 3 mm 5 mm 3 mm 5 mm OR 95% CI OR 95% CI OR 95% CI OR 95% CI Age (years) 1.08** ** ** ** Gender Females 1 (Ref) 1 (Ref) 1 (Ref) 1 (Ref) Males 2.06* ** ** ** Education Low 1 (Ref) Middle High 1.85** Brushing frequency 1/day 1 (Ref) 2/day 0.48* /day 0.60* Smoking exposure Never smokers 1 (Ref) 1 (Ref) Moderate 1.79* Heavy 2.07* * Dental visits None 1 (Ref) Irregular 1.80* Regular 1.93* Periodontal treatment No 1 (Ref) Yes 1.77* Calculus 0 19% 1 (Ref) 20 39% % 1.88* OR, odds ratio; 95%CI, 95% confidence interval; Ref: reference category. *p < 0.05, **p < 0.01.

8 Estimates of gingival recession 1105 Table 5. Multivariable linear regression models of risk indicators for mean gingival recession considering all sites and only buccal sites separately All sites Buccal sites Beta 95% CI p Beta 95% CI p Age (years) < <0.001 Gender Females 0 (Ref) 0 (Ref) Males Education Low 0 (Ref) 0 (Ref) Middle High < Proximal cleaning Never 0 (Ref) NI 1/day /day Smoking exposure Never smokers 0 (Ref) 0 (Ref) Moderate Heavy < Dental visits None 0 (Ref) 0 (Ref) Irregular Regular Gingivitis Low ( 22%) 0 (Ref) NI High (>22%) %CI, 95% confidence interval; Ref, reference category; NI, not included in the model. evaluated the epidemiology of GR, some issues were not explored such as the association of GR with important behavioural and clinical variables, as well as risk indicators for GR in buccal sites. Besides a 10-year period has passed between this and Susin s study, overall similar estimates of GR were found. In comparison to the findings from the United States (Albandar & Kingman 1999), the prevalence and extent of GR observed in this study seem to be higher. Nevertheless, the differences in the estimates of GR may be explained, at least in part, by different examination protocols used in the two studies (Susin et al. 2005, Albandar 2011). It has been demonstrated that age is considered a strong risk indicator for GR (Albandar & Kingman 1999, Susin et al. 2004, Holtfreter et al. 2009, Sarfati et al. 2010), and the findings from this study corroborate previous observations. These findings as a whole may be explained by the cumulative experience of periodontal tissue loss and exposure to risk factors during an individual s life, although the casual effect of ageing on periodontal attachment loss may not be discarded. In the present study, men had higher estimates of GR compared to women. This finding is consistent with the literature (Albandar & Kingman 1999, Susin et al. 2004, Sarfati et al. 2010). Moreover, male gender remained a significant risk indicator in all multivariable risk models in this study. Sarfati et al. (2010) also found in multivariable risk models that buccal GR was higher in males than females. Contrarily, gender was not a significant risk indicator for GR in the study by Susin et al. (2004). Comparisons to other studies cannot be performed because of the absence of multivariable risk assessment. Smoking was strongly associated with GR in two previous studies with representative samples (Susin et al. 2004, Sarfati et al. 2010), while others did not assess the effect of smoking on GR (Brown et al. 1996, Albandar & Kingman 1999, Thomson et al. 2000, Holtfreter et al. 2009). Studies with non-representative samples had conflicting results (Toker & Ozdemir 2009, Minaya- Sanchez et al. 2012). In this study, smoking was associated with higher odds of GR 3 mm and with higher mean GR, providing additional evidence for the detrimental effect of smoking in the marginal periodontal tissues. GR was associated with low selfreported frequencies of daily oral hygiene and with high percentages of supra gingival calculus in this study. Other studies have also observed that GR is associated with poor oral hygiene (L oe et al. 1992, Susin et al. 2004, Sarfati et al. 2010, Minaya- Sanchez et al. 2012). Susin et al. (2004) found that individuals with higher percentage of calculus had six times higher odds of GR. In France (Sarfati et al. 2010), plaque and gingivitis were correlated with GR in buccal sites. Similarly, individuals with higher percentage of gingival bleeding had more severe GR in this study. In this study, regular dental visits and self-reported history of periodontal treatment were associated with higher chances of GR and higher mean GR. These findings may be explained by the fact that dental/periodontal care may result in the resolution of periodontal inflammation. The recession of the gingival margin after periodontal therapy has been demonstrated by clinical investigations (Claffey et al. 2004), but this is the first study to demonstrate this association on a population basis. However, it is noteworthy to remember that the cross-sectional design of this study does not allow any conclusion regarding the temporality of these associations. Various factors have been associated with the occurrence of GR in buccal surfaces, including traumatic tooth brushing (Litonjua et al. 2003), hardness of tooth brushing s tuff (Greggianin et al. 2013), bone dehiscences (Kassab & Cohen 2003) and orthodontic treatment (Renkema et al. 2013). In general, clinical (Joshipura et al. 1994, Serino et al. 1994, Checchi et al. 1999, Daprile et al. 2007, Greggianin et al. 2013) and histological (Hallmon et al. 1986) evidences indicate a positive correlation between high standards of oral hygiene, traumatic oral hygiene and buccal GR. Nevertheless, little is known about the association between oral hygiene and GR on a population basis. In the study by L oe et al. (1992), it was clearly observed that two distinct populations demonstrated GR related to

9 1106 Rios et al. poor (Sri Lanka) and good (Norway) oral hygiene practices. Contrarily, in France (Sarfati et al. 2010), it was observed that buccal GR was associated with traditional risk indicators for periodontitis, including poor oral hygiene. In this study, variables related to oral hygiene were not associated with buccal GR. It is noteworthy that high educational level was significantly associated with higher mean and odds of buccal GR. These findings suggest that different risk indicators may be associated with buccal GR in comparison to other surfaces. However, further investigations are needed to better elucidate the epidemiological association between oral hygiene and GR on tooth and site levels. Important bias on estimates of periodontal attachment loss and probing depth is present when partial recording protocols (PRP) are used (Susin et al. 2005, Kingman et al. 2008). Nevertheless, there is no direct evidence that estimates of GR are affected when PRP are used. It may be speculated that some bias may be present in this study, but this is probably of low magnitude since all present teeth were examined to assess GR at four of the six sites assessed in a fullmouth protocol. It can be concluded that GR is highly prevalent in this Brazilian population. Older age, male gender, smoking and calculus were risk indicators for GR. Furthermore, our findings provide epidemiological evidence supporting access to dental/ periodontal care and lower frequency of self-reported oral hygiene practices as risk indicators for GR. Acknowledgements This study was funded by the National Coordination of Post- Graduate Education (CAPES) and the Federal University of Rio Grande do Sul (PROPESQ). References ABEP (2013) Brazilian Association of Research Companies - Criterio de classificacß~ao econ^omica Brasil (CCEB). Ainamo, J. & Bay, I. (1975) Problems and proposals for recording gingivitis and plaque. International Dental Journal 25, Albandar, J. M. (2011) Underestimation of periodontitis in NHANES surveys. Journal of Periodontology 82, Albandar, J. M. & Kingman, A. (1999) Gingival recession, gingival bleeding, and dental calculus in adults 30 years of age and older in the United States, Journal of Periodontology 70, Archer, K. J. & Lemeshow, S. (2006) Goodnessof-fit test for a logistic regression model fitted using survey sample data. The Stata Journal 6, Brasil (1986) Levantamento Epidemiologico em Saude Bucal: Brasil, zona urbana. Ministerio da Saude, pp Brasılia: Ministerio da Saude, Divis~ao de Saude Bucal. Brasil (2004) Projeto SB Brasil 2003: condicß~oes de saude bucal da populacß~ao brasileira : resultados principais. Ministerio da Saude, pp. 68. Brasılia: Ministerio da Saude, Divis~ao de Saude Bucal. Brasil (2011) Projeto SB Brasil 2003: condicß~oes de saude bucal da populacß~ao brasileira 2010: resultados principais. Ministerio da Saude. Brasılia: Ministerio da Saude, Divis~ao de Saude Bucal. Brown, L. J., Brunelle, J. A. & Kingman, A. (1996) Periodontal status in the United States, : prevalence, extent, and demographic variation. Journal of Dental Research, 75(Spec No), Chambrone, L., Sukekava, F., Araujo, M. G., Pustiglioni, F. E., Chambrone, L. A. & Lima, L. A. (2010) Root-coverage procedures for the treatment of localized recession-type defects: a Cochrane systematic review. Journal of Periodontology 81, Checchi, L., Daprile, G., Gatto, M. R. & Pelliccioni, G. A. (1999) Gingival recession and toothbrushing in an Italian School of Dentistry: a pilot study. Journal of Clinical Periodontology 26, Claffey, N., Polyzois, I. & Ziaka, P. (2004) An overview of nonsurgical and surgical therapy. Periodontology , Costa, R. S., Rios, F. S., Moura, M. S., Jardim, J. J., Maltz, M. & Haas, A. N. (2014) Prevalence and risk indicators of dentin hypersensitivity in adult and elderly populations from Porto Alegre, Brazil. Journal of Periodontology 85, Daprile, G., Gatto, M. R. & Checchi, L. (2007) The evolution of buccal gingival recessions in a student population: a 5-year follow-up. Journal of Periodontology 78, Du, M., Jiang, H., Tai, B., Zhou, Y., Wu, B. & Bian, Z. (2009) Root caries patterns and risk factors of middle-aged and elderly people in China. Community Dentistry and Oral Epidemiology 37, Dye, B. A., Tan, S., Smith, V., Lewis, B. G., Barker, L. K., Thornton-Evans, G., Eke, P. I., Beltran-Aguilar, E. D., Horowitz, A. M. & Li, C. H. (2007) Trends in oral health status: United States, and Vital and Health Statistics Series 11, Greggianin, B. F., Oliveira, S. C., Haas, A. N. & Oppermann, R. V. (2013) The incidence of gingival fissures associated with toothbrushing: crossover 28-day randomized trial. Journal of Clinical Periodontology 40, Hallmon, W. W., Waldrop, T. C., Houston, G. D. & Hawkins, B. F. (1986) Flossing clefts. Clinical and histologic observations. Journal of Periodontology 57, Hernan, M. A., Hernandez-Dıaz, S. & Robins, J. M. (2004) A structural approach to selection bias. Epidemiology 15, Holtfreter, B., Schwahn, C., Biffar, R. & Kocher, T. (2009) Epidemiology of periodontal diseases in the Study of Health in Pomerania. Journal of Clinical Periodontology 36, Hosmer, D. & Lemeshow, S. (2000) Applied logistic regression. New York: John Wiley & Sons. IBGE (2004) Resultados da amostra do Censo Demografico 2000 Malha municipal digital do Brasil: situacß~ao Rio de Janeiro: IBGE. Joshipura, K. J., Kent, R. L. & DePaola, P. F. (1994) Gingival recession: intra-oral distribution and associated factors. Journal of Periodontology 65, Kassab, M. M. & Cohen, R. E. (2003) The etiology and prevalence of gingival recession. Journal of the American Dental Association 134, Kingman, A., Susin, C. & Albandar, J. M. (2008) Effect of partial recording protocols on severity estimates of periodontal disease. Journal of Clinical Periodontology 35, Kularatne, S. & Ekanayake, L. (2007) Root surface caries in older individuals from Sri Lanka. Caries Research 41, Litonjua, L. A., Andreana, S., Bush, P. J. & Cohen, R. E. (2003) Toothbrushing and gingival recession. International Dental Journal 53, L oe, H., Anerud, A. & Boysen, H. (1992) The natural history of periodontal disease in man: prevalence, severity, and extent of gingival recession. Journal of Periodontology 63, Minaya-Sanchez, M., Medina-Solıs, C. E., Vallejos-Sanchez, A. A., Marquez-Corona, M. D., Pontigo-Loyola, A. P., Islas-Granillo, H. & Maupome, G. (2012) Gingival recession and associated factors in a homogeneous Mexican adult male population: a cross-sectional study. Medicina Oral, Patologıa Oral y Cirugıa Bucal 17, e807 e813. van Palenstein Helderman, W. H., Lembariti, B. S., van der Weijden, G. A. & van t Hof, M. A. (1998) Gingival recession and its association with calculus in subjects deprived of prophylactic dental care. Journal of Clinical Periodontology 25, Renkema, A. M., Fudalej, P. S., Renkema, A., Kiekens, R. & Katsaros, C. (2013) Development of labial gingival recessions in orthodontically treated patients. American Journal of Orthodontics and Dentofacial Orthopedics 143, Sarfati, A., Bourgeois, D., Katsahian, S., Mora, F. & Bouchard, P. (2010) Risk assessment for buccal gingival recession defects in an adult population. Journal of Periodontology 81, Serino, G., Wennstr om, J. L., Lindhe, J. & Eneroth, L. (1994) The prevalence and distribution of gingival recession in subjects with a high standard of oral hygiene. Journal of Clinical Periodontology 21, Slutzkey, S. & Levin, L. (2008) Gingival recession in young adults: occurrence, severity, and relationship to past orthodontic treatment and oral piercing. American Journal of Orthodontics and Dentofacial Orthopedics 134, Smith, R. G. (1997) Gingival recession. Reappraisal of an enigmatic condition and a new index for monitoring. Journal of Clinical Periodontology 24,

10 Estimates of gingival recession 1107 Susin, C., Haas, A. N., Oppermann, R. V., Haugejorden, O. & Albandar, J. M. (2004) Gingival recession: epidemiology and risk indicators in a representative urban Brazilian population. Journal of Periodontology 75, Susin, C., Kingman, A. & Albandar, J. M. (2005) Effect of partial recording protocols on estimates of prevalence of periodontal disease. Journal of Periodontology 76, Thomson, W. M., Hashim, R. & Pack, A. R. (2000) The prevalence and intraoral distribution of periodontal attachment loss in a birth cohort of 26-year-olds. Journal of Periodontology 71, Toker, H. & Ozdemir, H. (2009) Gingival recession: epidemiology and risk indicators in a university dental hospital in Turkey. International Journal of Dental Hygiene 7, Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. Intra-oral distribution of gingival recession 1 mm and 3 mm in maxillary and mandibular teeth (1 to 8 indicate tooth numbers from central incisors to third molars). Table S1. Univariable logistic regression models of risk indicators for gingival recession 3 mm and 5 mm considering all sites and buccal sites separately. Address: Alex Nogueira Haas Rua Ramiro Barcelos Porto Alegre-RS Brazil alexnhaas@gmail.com Clinical Relevance Scientific rationale for the study: Gingival recession has not been extensively assessed in the periodontal epidemiology, besides its well-known aesthetic and functional disabilities, being the most important risk factor for root caries. Principal findings: Gingival recession affected significant proportions of the population and was associated with traditional risk factors for periodontal diseases. Contrarily, variables related to poor oral hygiene were not associated with buccal recession, which was related to demographics, smoking and education. Practical implications: Gingival recession should be considered a major concern in this and similar populations. Preventive strategies aiming to reduce gingival recession may target socio-demographic, behavioural and clinical risk indicators.