Awareness of Clenching and Underweight are Risk Factors for Onset of Crowding in Young Adults: A Prospective 3-Year Cohort Study

Bruxism is a parafunctional activity that can seriously affect quality of life. Although bruxism induces many problems in the oral and maxillofacial area, whether it contributes to the onset of malocclusion remains unclear. The purpose of this prospective cohort study was to investigate the association between the onset of malocclusion and awareness of clenching during the daytime in young adults. Among 1,092 Okayama University students who underwent normal occlusion at baseline, we analysed 238 who had undergone a dental examination and had complete data after 3 years (2013–2016). We also performed subgroup analysis to focus on the association between awake bruxism and the onset of crowding (n = 216). Odds ratios (ORs) were calculated using multivariate logistic regression analyses. The incidences of malocclusion and crowding were 53.8% and 44.5%, respectively. In multivariate logistic regression, awareness of clenching was a risk factor for crowding (OR: 3.63; 95% confidence interval [CI]: 1.08–12.17). Moreover, underweight (body mass index < 18.5 kg/m2) was related to the onset of malocclusion (OR: 2.34; 95%CI: 1.11–4.92) and crowding (OR: 2.52, 95%CI: 1.25–5.76). These results suggest that awareness of clenching during the daytime and underweight are risk factors for the onset of crowding in young adults.

In a previous cross-sectional study, the prevalence of malocclusion (crowding) was significantly associated with awareness of clenching in university students [22]. Since malocclusion provides functional and aesthetic disturbances, and may lead to psychological stress [23], prevention or control are important for dental clinicians. Confirmation of causal relationships in a prospective cohort study is therefore necessary. Based on our previous study involving young adults [22], we hypothesized that clenching during the daytime as awake bruxism would be a risk factor for malocclusion. We configured the null hypothesis in which clenching does not induce malocclusion. The purpose of this prospective cohort study was to investigate the association between the onset of malocclusion and awareness of clenching during the daytime in young adults. We could clarify a part of association between clenching and malocclusion.

Study Population
The present study used a prospective cohort design. We estimated the sample size based on previous studies using SamplePower version 3.0 statistical software (IBM, Tokyo, Japan). For logistic regression analysis, this software computed power for a test of the null hypothesis in which the event rate in the two groups was identical. According to a previous study [22], the minimum sample size required in both positive and negative groups to detect significant differences in the awareness of clenching during the daytime was 55 with event rate (0.38 and 0.04), 80% power and a two-sided significance level of 5%. Assuming a follow-up rate of 26.2% [24], the planned minimum sample size was 209 participants.
We obtained data from first-year students who had undergone both a general health and oral examination at the Health Service Center of Okayama University in April 2013 (baseline). The inclusion criteria at baseline were Japanese students 18 or 19 years of age who did not show malocclusion, did not have experience receiving orthodontic treatment, and provided complete data in their questionnaires. Before graduation, students volunteered to receive both the general health and oral examination in April 2016 (follow-up). We excluded students who did not undergo the oral examination at follow-up, began orthodontic treatment during a residence, or provided incomplete data in their questionnaires at follow-up.

Ethical Procedures and Informed Consent
All study protocols were approved by the ethics committees of Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences and Okayama University Hospital (no. 1512-018). All targeted participants gave their informed written consent for study participation. STROBE guidelines were followed (Table A1).

Self-Questionnaires
At baseline (2013), students answered questions concerning their name, age, sex, general condition, experience of orthodontic treatment, awareness of bruxism, and oral habits at baseline [22]. According to a previous study [22], we asked students to identify their awareness of bruxism as follows: during the past 3 months, (i) "Has anyone heard you grinding your teeth at night?"; (ii) "Are you ever aware of grinding your teeth during the daytime?"; and (iii) "Are you ever aware of clenching your teeth during the daytime?" [8,22,[25][26][27]. Each question was answered by selecting a frequency (frequently, sometimes, rarely, or never). We combined "rarely" and "never" responses into a single category of negative awareness, and the remaining two responses into a single category of positive awareness [22]. The validity and reliability of the questionnaire have been confirmed as useful for evaluating bruxism [8,[25][26][27]. For oral habits, "yes/no" answers were given by participants as follows: biting fingernails/pens/pencils, biting mucosa of the cheeks/lips, and gum chewing [22,[28][29][30]. At follow-up in 2016, students provided their history of orthodontic treatment during the 3 year study period, non-nutritive sucking [22], habitual mouth breathing [31], early loss of primary teeth [32], and parents' history of malocclusion in a "yes/no" format.

Assessment of Malocclusion
Five dentists (D.E., K.K., M.Y-T., S.M., and T.A.) examined malocclusion in the participants during the oral examinations. We used a modified version of the Index of Orthodontic Treatment Need (IOTN) that does not define a definite aesthetic need for treatment (Aesthetic Component grades 8, 9 and 10) to assess malocclusion (Table 1) [23]. Our modified version of the IOTN and the original modified version [33] are useful for screening malocclusion by non-orthodontists in oral health surveys [22]. The dental health component of the modified IOTN consists of a two-grade scale (0 = no definite need for orthodontic treatment; i.e. we defined it as a normal occlusion [normal occlusion group]; 1 = definite need for orthodontic treatment; i.e. we defined it as a malocclusion [malocclusion group]). The type of malocclusion (missing teeth, overjet, crossbite, crowding, or overbite) was recorded using the community periodontal index (CPI) probe (YDM, Tokyo, Japan) in accordance with a previous study [22]. When one of them was positive at least, we defined that the participant had malocclusion. All dentists were trained and calibrated to use the modified IOTN. For this, an orthodontist acted as the gold standard. The kappa value was > 0.8. Table 1. Modified version of the Index of Orthodontic Treatment Need [22,33].

Missing teeth
Hypodontia requiring pre-restorative orthodontics or orthodontic space closure to obviate the need of a prosthesis. Impeded eruption of teeth, presence of supernumerary teeth, and retained deciduous teeth.

Overjet
Increased overjet greater than 6 mm. Reverse overjet greater than 3.5 mm with no masticatory or speech difficulties. Reverse overjet greater than 1 mm but less than 3.5 mm with recorded masticatory and speech difficulties.

Crossbite
Anterior or posterior crossbites with greater than 2 mm discrepancy between retruded contact position and intercuspal position.
Displacement of contact points (crowding) Contact point displacements greater than 4 mm.

Overbite
Lateral or anterior open bites greater than 4 mm. Deep overbite with gingival or palatal trauma.

Assessment of Body Mass Index (BMI)
In the general health examination at baseline and follow-up, public health nurses at the university measured the participants' height and body weight using a Tanita body fat analyser (BF-220; Tanita Co., Tokyo, Japan). Since BMI may be related to jaw growth, BMI was computed as weight in kilograms divided by height in meters squared. For this analysis, categories of BMI were calculated based on the accepted cut-off values for underweight (BMI < 18.5 kg/m 2 ), normal weight (BMI 18.5-24.9 kg/m 2 ), and overweight (BMI ≥ 25.0 kg/m 2 ) [34].

Statistical Analyses
We used SPSS version 20 (IBM, Tokyo, Japan) for statistical analyses. P values < 0.05 were considered to indicate significant associations. The McNemar-Bowker or paired t-test was used to investigate significant differences between baseline and follow-up. The chi-square test was used to determine significant differences between the normal occlusion and malocclusion groups, whereas in cases of awareness of bruxism at baseline and oral habits at baseline, the chi-square test with Bonferroni correction to control the false discovery rate (P < 0.05/3) was used [35]. Since the majority (approximately 83%) of malocclusions involved crowding, we also investigated associations between crowding and other parameters.
Odds ratios (ORs) and 95% Confidence Intervals (CIs) were calculated using a series of logistic regression models. The onset of malocclusion or crowding was used as the dependent variable. Based on binary analyses and previous studies [22], BMI category, clenching during the daytime, and sex were added as independent variables in multiple logistic regression models as items associated with outcome. We assessed model fit using the Hosmer-Lemeshow goodness-of-fit test for logistic regression. Figure 1 shows a flow chart of participants in this three-year cohort study from baseline to follow-up. We selected 1,092 students who matched the study criteria at baseline. At the follow-up, 838 students had not undergone an oral examination and 16 met the exclusion criteria (five students had received orthodontic treatment and 11 had provided incomplete data). Finally, 238 students were analysed (normal occlusion vs. malocclusion). The follow-up rate was 21.8% (238/1092). Furthermore, we performed subgroup analysis to focus on crowding. We excluded 22 students who had other types of malocclusion (overjet, overbite, crossbite, and missing teeth). Finally, 216 students were analysed (normal occlusion vs. crowding).

Statistical analyses
We used SPSS version 20 (IBM, Tokyo, Japan) for statistical analyses. P values < 0.05 were considered to indicate significant associations. The McNemar-Bowker or paired t-test was used to investigate significant differences between baseline and follow-up. The chi-square test was used to determine significant differences between the normal occlusion and malocclusion groups, whereas in cases of awareness of bruxism at baseline and oral habits at baseline, the chi-square test with Bonferroni correction to control the false discovery rate (P < 0.05/3) was used [35]. Since the majority (approximately 83%) of malocclusions involved crowding, we also investigated associations between crowding and other parameters.
Odds ratios (ORs) and 95% Confidence Intervals (CIs) were calculated using a series of logistic regression models. The onset of malocclusion or crowding was used as the dependent variable. Based on binary analyses and previous studies [22], BMI category, clenching during the daytime, and sex were added as independent variables in multiple logistic regression models as items associated with outcome. We assessed model fit using the Hosmer-Lemeshow goodness-of-fit test for logistic regression. Figure 1 shows a flow chart of participants in this three-year cohort study from baseline to follow-up. We selected 1,092 students who matched the study criteria at baseline. At the follow-up, 838 students had not undergone an oral examination and 16 met the exclusion criteria (five students had received orthodontic treatment and 11 had provided incomplete data). Finally, 238 students were analysed (normal occlusion vs. malocclusion). The follow-up rate was 21.8% (238/1092). Furthermore, we performed subgroup analysis to focus on crowding. We excluded 22 students who had other types of malocclusion (overjet, overbite, crossbite, and missing teeth). Finally, 216 students were analysed (normal occlusion vs. crowding).

Changes in Parameters from Baseline to Follow-Up
The incidences of malocclusion and crowding were 53.8% and 44.5%, respectively (Table 2). No significant difference in BMI distribution was seen between baseline and follow-up (McNemar-Bowker tests; P > 0.05). On the other hand, mean height and weight differed significantly between stages (paired t-test; P < 0.05; 95%CI of height, 0.15-0.33; 95%CI of weight, 0.13-1.15).

Association between Malocclusion/Crowding and Other Parameters
Among the parameters examined, a significant difference in BMI distribution was seen between the normal occlusion and malocclusion groups (chi-square test; P = 0.04) ( Table 3). Significant difference in BMI distribution were seen during the daytime between the normal occlusion and crowding groups (chi-square test; P = 0.02) ( Table 4).  Logistic regression analysis showed the risk of malocclusion was significantly related to underweight (BMI < 18.5 kg/m 2 ; multiple logistic regression analysis; P = 0.03; 95%CI, 1.11-4.92) ( Table 5). However, no significant association was identified between malocclusion and awareness of clenching during the daytime (multiple logistic regression analysis; P > 0.05; 95%CI, 0.91-9.88). The Hosmer-Lemeshow test found acceptable model fit, with a chi-square statistic of 0.78 (P = 0.68). On logistic regression analysis, risk of crowding correlated significantly with underweight (BMI < 18.5 kg/m 2 ) and awareness of clenching during the daytime (multiple logistic regression analysis; P < 0.05) ( Table 5). The Hosmer-Lemeshow test found acceptable model fit, with a chi-square statistic of 0.88 (P = 0.83).

Discussion
A previous cross-sectional study showed a significant association between prevalence of malocclusion (crowding) and awareness of clenching in Japanese university students [22]. To the best of our knowledge, the present study is the first prospective cohort study to investigate whether awareness of clenching during the daytime is a risk factor for malocclusion in young adults. In this study, the results showed that awareness of clenching during the daytime was associated with the onset of crowding (adjusted OR, 3.63; 95% CI, 1.08-12.17). These findings may support our hypothesis that awareness of clenching during the daytime is a risk factor for crowding.
Previous studies have suggested that clenching forces contribute to tooth movement [21,36,37]. This force can induce tooth movement and contribute to malocclusion [21]. The mean clenching force is 720 N (162 lb) with a range of 244-1243 N (55-280 lb) [36]. The bite force needed to contribute to displacement is approximately 100 N [18,38]. Because the force of clenching is higher than the threshold of tooth displacement, clenching during the daytime may represent a risk factor for crowding through tooth movement.
Underweight (BMI < 18.5 kg/m 2 ) was associated with the onset of crowding. Underweight is related to skeletal maturation [39], and delayed maturation might affect the onset of malocclusion [40]. Kataoka et al. [22] showed a significant association between underweight and the prevalence of malocclusion in a cross-sectional study. In the present study, mean height in students with normal weight (18.5  Oral habits, including biting fingernails/pens/pencils, biting the mucosa of the cheeks/lips, and gum chewing [28][29][30], were not significantly associated with the onset of malocclusion (chi-square test; P >0.05). Previous studies have shown that bad oral habits are related to the prevalence of malocclusion [41,42]. Some inconsistencies are apparent between our study and previous studies. The reasons for this are unclear, but may involve differences in age (young adults vs. children) and dentition (permanent vs. primary).
The results of the present study might be clinically relevant. Treatment for awake bruxism is based on behaviour modification and habit reversal. Behaviour modification has the potential to stop or reduce awake bruxism [2,43]. Treatment for awake bruxism could therefore prevent the onset of malocclusion. When clinicians encounter younger patients who are aware of clenching during the daytime, increased efforts may be needed to prevent malocclusion. The present work was only an observational study. Further studies are therefore needed to clarify whether clinical interventions can help prevent the onset of malocclusion.
The prevalence of malocclusion was higher than that of the previous studies. In the present study, the rate of onset of malocclusion was 53.8% (Table 2). In previous research using IOTN, the prevalence of malocclusion varied widely among subjects (21-44.9%) [23,[44][45][46]. As participant age, country, sample size, and study design differed between the present and other investigations, caution is warranted in regard to the generalizability of the results.
The participants were not considered an unusual sample based on the two aspects as below. The prevalence of awake bruxism was 8.0% in the present work, which is within the 8-34% range reported in previous studies [9][10][11][12][13][14][15]. Moreover, the distributions of underweight, normal weight, and overweight based on BMI classifications established by the World Health Organization were 17.2%, 75.6% and 7.1%, respectively. As a reference, the 2013 Japan National Health and Nutrition Survey showed distributions of underweight, normal weight, and overweight of 22.3%, 70.2%, and 7.4%, respectively (age range, 15-19 years) [47]. The distribution of BMI categories in the present study did not differ significantly from that in the Japanese national survey (chi-square test; P > 0.05).
No relationship was observed between sex and the onset of malocclusion. Previous studies have also reported that sex was not significantly associated with the prevalence of malocclusion based on the IOTN [48][49][50]. Those studies support our results.
Several limitations of the present study must be considered when interpreting the results. First, these findings to young adults in general. Second, we did not investigate bite force or clenching force; these are difficult to measure in general oral examinations because they require special instruments [51]. Third, there may have been a selection bias, given the low follow-up rate. In the present study, analysed students (n = 238) comprised 21.8% of all eligible students (n = 1,092). However, no significant differences were seen in the ratios of bruxism, oral habits, or BMI between the analysed and non-analysed students (238 vs. 854 students, chi-square test; P > 0.05), with the exception of sex (chi-square test; P < 0.05). Any effects of a selection bias would have therefore been negligible. Forth, we could not investigate malocclusion in the participants' parents. We have to pay attention to deal with the data of parents' malocclusion based on the questionnaire because of bias. Finally, we did not investigate tongue thrust, which might affect malocclusion, based on a recent case report [52].

Conclusions
In conclusion, in this prospective cohort study, awareness of clenching during the daytime and underweight were found to be related to the onset of malocclusion (crowding) among university students. These findings suggest that clinicians may need to apply increased efforts to prevent malocclusion in younger patients who are aware of clenching during the daytime.

Funding 22
Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based 9 * Give information separately for exposed and unexposed groups.