Prevalence and Associated Factors of Self-Reported Gingival Bleeding: A Multicenter Study in France

Gingival bleeding (GB) is a common sign of gingival inflammation which indicates the presence of periodontal diseases. This cross-sectional multicenter survey aimed to assess the prevalence of self-reported gingival bleeding (SRGB) in French adults and identify the main associated factors. A questionnaire-based interview was randomly proposed to 794 individuals in four French cities (Nancy, Montpellier, Paris, and Rennes). Subjects were recruited in preventive medicine centers (50%), railway stations, and malls (50%). The questionnaire comprised 25 items: SRGB characteristics, socioeconomic variables, oral hygiene habits, use of drugs, and anxiety level. The overall prevalence of SRGB was 63.2% [59.8%; 66.6%], with 58.7% bleeding after toothbrushing and 4.5% spontaneous bleeding. Males reported significantly lower SRGB prevalence than females (p = 0.04). The distribution of SRGB frequency was inversely proportional to age (p < 0.0001). No association between drug use and SRGB was found. The people interviewed in the preventive medicine centers reported the highest frequency of SRGB (p < 0.0001). In the multivariate logistic model, SRGB was significantly related to occupation, smoking status, brushing frequency, and anxiety level. In conclusion, SRGB was prevalent in more than half of the sample and was mainly associated with age, toothbrushing frequency, and anxiety level. Thus, providing information to patients about the importance of this oral manifestation may play an important role in preventing periodontal diseases.


Introduction
Gingival bleeding (GB) is one of the first clinical manifestations of periodontal diseases such as gingivitis and periodontitis. This bleeding is the sign of gingivitis, which is considered a reversible form of periodontal disease when only the gingiva is affected but may turn into periodontitis for certain patients if left untreated. Severe periodontitis is the sixth most prevalent disease in the world, and it is the main cause of disability-adjusted life years among oral conditions because of the centers. The cities (Paris, Rennes, Nancy and Montpellier) were selected because they are located in four different parts of France (north, west, east, and south) according to the French National Institute for Statistics and Economic Studies [17] and because they all had a dental school.

Interview
In each city, one student from the dental school conducted face-to-face interviews lasting about 5 min in the three locations (preventive medicine center, railway station, and mall).
The questionnaire consisted of 25 items, including socioeconomic status, tobacco use, toothbrushing habits, anxiety level, and type of bleeding. The primary endpoint was the prevalence of GB reported by the subjects who took part in the survey. More details are provided in Appendix A.
The ethics committee of the Nancy CHRU Hospital gave a favorable opinion regarding this interview (Saisine No. 282).

Statistical Methods
The calculation of the sample size is detailed in Appendix B. Percentages and proportions were drawn from the different items of the questionnaire, and the GB prevalence was calculated accordingly. Chi-square tests were performed in bivariate analyses between GB and its potential risk factors. A stepwise multiple logistic regression (respectively, a stepwise multinomial logistic regression) was used to determine factors associated with GB (there were three modalities: spontaneous GB, toothbrushing GB, and absence of GB) among covariates, with a p-value less than 0.20 in the bivariate analyses. All analyses were performed with the statistical software SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA) and the significance level was set at 0.05.

Results
In total, 794 questionnaires (200 in Montpellier, 200 in Nancy, 200 in Paris, and 194 in Rennes) were completed correctly. Table 1 presents the demographics and information on toothbrushing in the interviewed population by center. The distribution of age, smoking status, educational level, and occupation of the interviewees was statistically different between preventive medicine (PM) centers and railway stations or malls (p = 0.039, p = 0.003, p = 0.0001, and p = 0.001, respectively). People interviewed in PM centers presented a greater number of smokers than those in the other places (i.e., railway stations and malls). No difference between the sex of the interviewees in PM and other centers was observed. More details are provided in Appendix C. A full description of the characteristics of people included in the four French cities is given in Table S1. Notes: Cig/day: cigarettes per day; low level: primary education (8 years or less); medium level: some secondary education (9-11 years); high level: completed secondary education (12 years or more); PAA: platelet antiaggregant; AVK: anti-vitamin K; NSAID: nonsteroidal anti-inflammatory drug; * chi-square test.

Self-Reported Prevalence of Gingival Bleeding
The overall prevalence of SRGB was 63.2% [59.8%; 66.6%], with 4.5% reporting spontaneous bleeding and 58.7% reporting bleeding after toothbrushing. Among those reporting GB, 22.1% reported bleeding appearing less than once a week, and 19.9% reported bleeding appearing more than once a week ( Table 2). Regardless of how often bleeding appeared (less than once per month, more than once per month, less than once per week, or more than once per week), SRGB was mainly found after toothbrushing rather than spontaneously (88.2% vs. 11.8%, 96.1% vs. 3.9%, 95.5% vs. 4.5%, and 91.0% vs. 9.0%, respectively) ( Figure 1). Among women reporting GB, only 18.7% (52/278) mentioned hormonal gingival bleeding (during pregnancy, menstrual bleeding, and the use of contraceptive pills) ( Table 2). The description of SRGB frequency in the different cities is available in Appendix D. Note: * Self-reported gingival bleeding of the people interviewed in all the centers of the four cities; ** hormonal bleeding during pregnancy, menstrual bleeding, and the use of contraceptive pills. Regarding the overall prevalence of SRGB, the report of bleeding was statistically different depending on the location of the interview; regardless of the city, the highest prevalence was observed in PM centers (p < 0.0001) ( Table 3). By contrast, no significant difference was observed regarding the distribution of the frequency of SRGB (less than once per month, more than once per month, less than once per week, or more than once per week) in all centers (Table 3). Among all participants, 59.6% of men and 66.5% of women reported GB. This difference of SRGB prevalence was significant (p = 0.04) ( Table 4), but after stratification by age, no significant difference was observed in regards to sex (data not shown). Regarding the overall SRGB data, the distribution of its prevalence was inversely proportional to age (p < 0.0001) for both spontaneous and toothbrushing-associated bleeding (Table 4, Figure 2).

Relationship between GB and Socioeconomic, Medical, and Dental Variables
Regarding the socioeconomic status of the participants, the distribution of the occupational status (but not of the educational level) was significantly associated with SRGB (p < 0.0001) ( Table 4). Retired people and categories such as teachers, liberal professions, and executives presented the lowest frequency of SRGB. No association was observed between SRGB and medications, except for anxiolytics and antidepressants (p = 0.067). No significant differences were observed between smokers and nonsmokers regarding SRGB, even if the percentage of SRGB was lower in people smoking more than 10 cigarettes per day. By contrast, the distribution of anxiety levels was significantly associated with SRGB (p = 0.0004). Those with high or severe anxiety presented the highest frequency of SRGB.

Relationship between Self-Reported GB and Toothbrushing Variables
Only the frequency of toothbrushing was statistically correlated with the prevalence of gingival bleeding (p < 0.025) (Table S2). More details about these variables are available in Appendix E.

Factors Associated with Self-Reported GB
Independent variables with a p-value < 0.20 in the bivariate analyses were likely to enter the multivariate model. The age variable was withdrawn from the model because of its correlation with the occupation variable.
Factors associated with SRGB were occupation, smoking status, toothbrushing frequency, and anxiety level. The occupations classes none, unemployed, student, other or manual worker, employee, artisan, retailer, and administrative (p < 0.0001); toothbrushing less than two times per day (p = 0.0047); and a high/severe anxiety level (p = 0.0031) were associated with a higher risk of GB. In contrast, smoking was associated with a lower risk of SRGB (p = 0.0012) ( Table 5). The multinomial logistic regression model, including three modalities for GB (spontaneous SRGB, toothbrushing SRGB, absence of SRGB), confirmed the results of the multiple logistic regression model, with differences detected only between toothbrushing SRGB and absence of SRGB, due to fewer events of spontaneous SRGB (Table 6).

Discussion
This study is the first self-evaluation of gingival bleeding in France. Among responders, the overall prevalence of SRGB was 63.2%, regardless of age and circumstances of onset. In the literature, a wide range of SRGB prevalence has been published. In different studies of self-reported periodontal health, from 6 to 78% of responders reported GB [18,19]. Pinelli et al. evaluated self-perceived oral health conditions and concluded that among the 200 Brazilians interviewed, 53.5% declared GB after toothbrushing and 6% declared spontaneous GB, which is in accordance with our results. Furthermore, Genco et al. found similar results for the question "bleeding gums in the past"; 53% of interviewees in the Erie County Study answered positively [20]. The SRGB frequency we found is in accordance with the results of the French clinical epidemiologic study evaluating GB with Community Periodontal Index of Treatment Needs (CPITN), which underlines the fact that gingival inflammation has not decreased in the French population for 23 years [8,9]. The large diversity in results might be explained by the difference in length of the questionnaires. Indeed, people respond less easily to long questionnaires than short ones [21,22]. Moreover, the many different ways of collecting data that explore gingival health [12,23], the different methodologies of evaluating gingival health (swollen or bleeding gingiva, gingival disease or inflammation) [24,25], and population-characteristic differences [26,27] make the comparison of results difficult.
Apart from the Gilbert and Nuttall study with the highest score of SRGB (78% of British adult patients in dental hospitals), all the papers described lower percentages of SRGB than our results [19]. In the Dietrich cohort, 44% of people reported GB after toothbrushing [24]. The same percentage was reported in the Yamamoto cohort of men aged 50 to 59 [26]. The lowest percentages of SRGB were observed in the Saka-Herrán, Lintula, Kim, and Taylor studies: 32% of the Spanish interviewed, 24% of the 31-year-old Finnish, 11.3% of selected Koreans having GB after toothbrushing, and 17% of Americans interviewed presented GB in the previous week, respectively [14,23,28,29]. The differences in age groups, gender ratios, educational levels, occupations, and the percentages of smokers included in the cohorts could explain the discrepancies with the results of our study.
Previous epidemiological studies have demonstrated the age dependence and gender differences of gingival bleeding data [8,9,28,30]. Our results found a higher prevalence of GB among females and are in line with those of other studies [8,28,31]. The higher SRGB prevalence observed in females could be attributed to the facts that (i) women are more aware of their oral health and (ii) women are more susceptible to gingival inflammation exacerbations due to hormonal fluctuations [32,33]. In contrast to our results, Kim et al. observed similar SRGB prevalence between males and females, but no information regarding smoking was available and the subjects of the survey were slightly older [29]. Our study showed a significant decrease in SRGB frequency with age; a lower percentage was reported by people aged 60 and over, which is in accordance with the conclusions of Kim [29]. Two French clinical epidemiologic studies evaluated GB with the CPITN and showed similar results for each age group evaluated [8,9]. By contrast, Buhlin et al. observed an inverse association of SRGB according to age [34]. Elderly people have a higher number of missing teeth than those younger, but most importantly, they seem to be less aware of their gingival bleeding, as shown in the study of Romano et al. [31]. Ebersole et al. demonstrated the presence of age-associated alterations in innate immune function within the periodontium, which could affect not only the initiation but also the resolution of inflammation [35]. This phenomenon may modulate the GB of aging people.
The relationship between socioeconomic status (SES) and general health, including oral health and involving different mechanisms, has often been explored [36]. Individuals with low SES are more likely to be affected by oral disease [37]. Although SRGB prevalence was significantly higher among unemployed people in the bivariate analysis, we could not highlight any difference in SRGB prevalence according to occupational status in the multivariate model after adjusting for the other covariates. Since occupational status is only a limited aspect of SES, this result would need further analyses, such as those incorporating data about household income, to be confirmed. The prevalence of SRGB was statistically different depending on the centers. Interviewees in preventive medicine centers presented the highest frequency of SRGB, probably because they were more aware of their periodontal health than the others. Indeed, these interviewees were present in these centers either because they were having a health check-up and/or because they were included in national health surveys on different diseases.
Many authors have reported that medications may cause GB through an effect on vascular permeability, coagulation, or platelet function [38,39]. A French pharmacoepidemiologic study analyzed GB as a possible serious adverse drug reaction and showed that GB represented only 0.09% of all reported adverse drug reactions and is more commonly associated with an increased international normalized ratio (INR), thrombocytopenia, or hematuria [7]. In our study, no significant association was found between SRGB and any of the drugs evaluated, but it is interesting to note that few participants reported taking medications. Therefore, few or no instances of GB reported by the participants of our sample could be attributed to adverse drug reactions.
The threshold value of smoking more than 10 cigarettes per day that was used in the present study is commonly adopted as a risk factor for periodontal disease, and it remains in the 2017 classification system of periodontal diseases [2,40]. We found that SRGB prevalence was associated with smoking status (logistic model), with a lower prevalence in the group smoking more than 10 cigarettes per day. This is in line with the fact that smokers show reduced signs of gingival inflammation and lower GB prevalence compared to nonsmokers; the gingival vasoconstriction induced by the action of nicotine could suppress the inflammatory response that accounts for impaired wound healing [41][42][43]. Nevertheless, certain authors found that BoP was more prevalent in smokers [44,45], but noticed that plaque levels were higher among smokers, and assumed that the vasoconstrictive properties of nicotine may only contribute to a delay in observed GB. In the latter study, smoking status was a significant factor for GB after adjusting for plaque and periodontal probing depth. In the present study, plaque levels could not be measured, but the frequency of toothbrushing was taken into account, as it is thought to be linked to plaque accumulation. Only the frequency of toothbrushing (but neither the type of toothbrush nor the technique) was significantly associated with SRGB. This result is in accordance with previous studies showing that less frequent toothbrushing was significantly associated with a higher frequency of SRGB [28].
Stress and anxiety are known predisposing factors for periodontal disease [46], including necrotizing ulcerative gingivitis and GB [47]. Patients with higher anxiety scores showed significantly more bleeding on probing than patients with lower scores [5]. Stressors and anxious responses to psychological stressors may modulate the immune response to bacteria and influence the progression of gingivitis and periodontitis [48]. Interestingly, smoking (OR = 0.26) and stress (OR = 1.78) were significantly associated with the GB [49]. Academic stress may lead to increased plaque levels and gingival index in students [50,51]. However, in the case of self-assessed GB, anxiety may lead to an overestimation of most symptoms, including GB. Thus, a higher prevalence of SRGB found among anxious subjects may partly derive from this bias in the assessment process.
The representability of the sample at a national level is relatively satisfactory since the distributions according to the main sociodemographic factors are comparable between the general French population and the selected samples drawn from the four French cities. The outcome variables are based on self-reporting and not on a clinical examination by a practitioner. Systematic reviews evaluating the validity of a questionnaire and/or self-reported measures for periodontal diseases, including gingivitis screening, have concluded that self-perceived pathologies were somewhat more specific (SP) than sensitive (SN) [11,12]. The same results were recently found by Ueno et al. and Saka-Herran et al. [14,52], with higher specificity than sensitivity, even if sensitivity was remarkably high in the latter study. For a given item, sensitivity and specificity depend on the cut-off used to define the bleeding and on how the gingival bleeding is evaluated, which is specific to each study. For example, Buhlin et al. found an SN of 42% and an SP of 76% for the question "Do your gums usually bleed?", but the cut-off point for gingivitis as regards BoP was set at 30% for the middle-aged group and 50% for the older-aged group, which is not the actual definition of gingivitis as reported in the new classification of 2018 [34]. "Bleeding during toothbrushing" reported by patients in the Kallio study was significantly correlated with the percentage of sites bleeding on probing, showing an agreement of 74%, with a kappa of 0.27, a specificity of 72%, and a sensitivity of 24% [30]. Gilbert and Nuttall presented similar conclusions regarding the comparison of a self-reported and clinical evaluation of GB: a high SP (86-88%) and low SN (19-35%) were found, showing that many people with clinically detected bleeding were unaware of any bleeding after toothbrushing [19]. Nevertheless, the self-evaluation of gingival bleeding can be easily accomplished by anyone and is likely to be a useful diagnostic indicator of gingivitis. This self-evaluation could be an interesting tool for detecting periodontal diseases, as GB is one of the primary symptoms of periodontitis, and may not only facilitate the implementation of lower-cost epidemiological surveys but also improve and reinforce oral hygiene. Indeed, as shown in the Walsh et al. study, a test group having carried out self-assessment of GB with toothbrushing and toothpicks showed a significant reduction of bleeding on probing sites after three months in comparison with a control group [53].

Conclusions
Until now, there have been no available data at a national level in France regarding SRGB. More than half of the interviewed population (63.2%) experienced GB at least once, spontaneously or after brushing. In the multinomial regression, many parameters were found to be associated with SRGB: nonsmoking young people with high anxiety and toothbrushing twice per day or less reported GB more often. Detecting, monitoring, and managing GB, which is one of the early signs of gingivitis, could be a primary preventive strategy for limiting the development of periodontal diseases such as periodontitis. Therefore, patient awareness regarding the impact of GB may play an important role in public health dentistry.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/17/22/8563/s1, Table S1. Comparison of the characteristics of people from each 4 cities, Table S2: Prevalence of self-reported gingival bleeding according to the type of toothbrush and the toothbrushing frequency. of respondents differed significantly between the four regions (Table S1). However, the region was not associated with the outcome, i.e. gingival bleeding, and then, may not be considered as a confounding factor. So, no adjustment was performed on the region in the multivariable analysis to identify risk factors of SRGB.

Appendix E Relations between Self-Reported GB and the Type and Frequency of Toothbrushing
Neither the type of toothbrush nor the technique of toothbrushing have been significantly associated with self-reported GB. Only the frequency of toothbrushing is statistically correlated, and after mixing in the same category 'less than twice a day' and 'twice a day' (sometimes people state that they brush their teeth twice a day, even when actually it is less often), the bleeding prevalence was significantly associated with the brushing frequency (p < 0.008) (Data not shown).