Participants of the study
The purpose and outline of the Japan Multi-Institutional Collaborative Cohort (J-MICC) Study was to prospectively investigate gene-environment interactions for the risk of NCDs, as described elsewhere [20]. In brief, approximately 100,000 study subjects aged 35–69 years participated in the baseline survey, and the J-MICC Study was executed using a systematic research protocol for all 13 research sites. Each collaborative research site was designed using a common protocol reflective of the J-MICC Study and additional original protocols were used. The study was performed over 20 years of follow-up to examine health outcomes. As part of the Shizuoka Sakuragaoka J-MICC Study [17], which was added to our regional research protocol at this site, the current cross-sectional study was conducted among 6395 men and women aged 35–79 years. Written informed consent was obtained from each participant after explaining the outline and purpose of the two study designs. All the research procedures were conducted in accordance with the Declaration of Helsinki and were approved by the Ethics Committee of Nagoya University Graduate School of Medicine (No.2010–0939-8), Aichi Cancer Center Research Institute (No.H2210001A), and University of Shizuoka (No.22–39). The participation rate was 27.6% for all participants. In the current study, 3494 men and 2552 women were finally selected as eligible participants after excluding the following criteria: 1) missing data on OHIs and measured body weight and height (n = 52), and 2) missing data on lifestyle information used as potential confounding factors (n = 231).
Definitions of obesity
During the health checkups, body weight and height were measured by trained nurses at five health checkup centers. The individual body mass index (BMI) was calculated by dividing the participants’ weight (kg) by the square of their height (m). Obesity is defined as a BMI ≥25 kg/m2 according to the Japan Society for the Study of Obesity (JASSO) [21], in consideration of WHO expert comments [22]. In this study, subjects with a BMI ≥25 and < 25 kg/m2 were defined as obese and non-obese group, respectively. The rationale for this study is that, compared to European populations, Asian populations have been shown to be at a higher risk for NCDs, even when BMI is lower than the existing WHO cutoff points for overweight and obesity. According to the guidelines for the management of obesity disease 2016 by JASSO [23], however, obesity should be diagnosed only after a physician examines the patient and comprehensively determines whether the patient has any health conditions or excessive accumulation of visceral fat.
Data collection of lifestyle information
The self-administered questionnaire consisted of both common (e.g., conventional lifestyle, medical history, and general tooth/gum conditions including tooth number) and original items (e.g., OHIs and dietary behaviors such as snacking) from all 13 research sites. Briefly, we collected the following data from each participant: physical activity, smoking status, habitual drinking, and snaking as a conventional lifestyle; medications for diabetes, dyslipidemia, hypertension, angina myocardial infarction, and cerebral stroke as medical history; tooth pain, gum bleeding, and gum swelling as general tooth/gum conditions; and age, education level, and tooth number as others. A scientifically validated short food frequency questionnaire (FFQ) was used to assess habitual diet, including total energy intake, and we have also reported success with the use of FFQ at our site using data from our site [24,25,26,27]. The validity and reproducibility tests for the selected foods have already been reported as follows: almost 0.50, as the median de-attenuated Spearman’s rank correlation coefficient between dietary records (for a total of 12 days over four seasons) and the first FFQ, and > 0.60 as the median energy-adjusted Spearman’s rank correlation coefficients between the first FFQ and the second FFQ. Physical activity and exercise were calculated as “Mets‧h/day” [28]. Trained scientific nurses reviewed the questionnaires to reduce missing data whenever possible. Two or three additional checks were performed by other trained nurses, each using a blue and red ink pen.
Definition of oral health indices
We used an oral health questionnaire with 15 questions administered to several municipalities [29]. In user’s guideline of the questionnaire, four OHIs of oral self-care habits, oral hygiene, oral function, and mastication ability were defined as follows: “tooth brushing frequency”, and “use of inter-dental care device” as oral self-care habits; “worried about dry mouth”, “sometimes food may be left in your mouth”, “a white moss-like substance is coating on your tongue”, and “worried about your own bad breath” as oral hygiene; “it is getting harder to rinse out your mouth”, “your voice has become hoarse”, “you can talk well”, “sometimes choking on tea or soup”, “you may choke during meals”, and “you may find it difficult to swallow food” as oral function; and “compared to six months ago, it is difficult for you to eat tough food”, “mealtimes are now longer than before”, and “sometimes you cannot taste it” as mastication ability, respectively. Fifteen items were answered on three levels (good/normal/poor), with the exception of two levels (good/poor) on items second, third, tenth, and thirteenth items. The four OHIs were scored on each composed question (good as “reference”), and also defined as “comprehensive OHI”, that is, the fifth OHI, according to their total score.
Statistical analyses
Data were analysed using R version 3.6.1. All reported probability values were two-tailed and the level of significance was set at p < 0.05. All analyses were performed separately for men and women. As appropriate, t-tests and chi-squared tests were used for continuous and categorical nominal variables, respectively. The three and two levels of the 15 questions according to the oral health questionnaire were scored with ordinal numbers (i.e., 1, 2, and 3 points) representing “≥3 times/day, twice/day, and ≤once/day”, and “good, normal, and poor”, and those (i.e., 1 and 3 points) for yes and no. Poor oral health was defined as higher scores on each of the five OHIs. Each poor level of OHIs and their composed questions is postulated to be related to higher risks of NCDs, acceleration of aging and deduction of social participation, and ultimately to inhibition of prolonging healthy life expectancy [30].
Multivariable adjusted logistic regression analysis was used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for obesity, compared with each 15 question on oral health (good as “reference”) and each five OHI (the lowest tertile as “reference”). With reference to previous studies [6, 14, 31], the potential confounding factors were applied as follows: age (years), physical activity (METs‧h/day), total energy intake (kcal/day), and tooth number (n) as continuous variables, and current smoker (no/yes = 0 / 1), current habitual drinker (no/yes = 0 / 1), snacking (none/sometimes/everyday = 1 / 2 / 3), education level (< 12 / 12 / > 12 years = 1 / 2 / 3), medications [no/yes = 0 / 1 for diabetes, dyslipidemia, hypertension, angina myocardial infarction, and cerebral stroke, respectively] and tooth/gum conditions [no/sometimes/yes = 1 / 2 / 3 for tooth pain, gum bleeding, and gum swelling, respectively] as categorical variables.
Using the scores described above as Model 1, trend associations were assessed with the median of each tertile score for the five OHIs. In Model 2, the OHI scores were recalculated by applying “1, 3, 5” and “ 1 and 5” points for the three and two levels of the questions comprised the OHI. Trends were calculated in the same way and compared with the results of Model 1 to check 1) whether the positive or negative β values were the same in Model 2, and 2) whether their statistical significances were also the same.