Heavy metals in drinking water and periodontitis: evidence from the national oral health survey from China

The National Oral Health Survey of China (CNOHS) is a nationally representative oral health survey and oral clinical examination assessing the oral health status of Chinese children and adults. The fourth CNOHS used a multistage cluster sampling method, and a more detailed description of the study design and sampling procedure was published elsewhere [11, 25]. A total of 124 districts (91 cities) were selected from 31 provinces using the probability-proportional-to-size (PPS) method, considering the variation in population size. From these districts, participants between 35 and 74 years old were chosen from 372 communities (three communities from each urban and rural district) using quota sampling. In this study, due to the availability of water quality data, and the periodontal examinations only conducted in adults, we adopted the study participants to those 35–75 years of age from selected geographic and cultural representative cities. Therefore, our study population includes 1,348 participants, which represent over forty million adults from twelve counties/districts in six provinces. The Chinese Stomatological Association Ethics Committee approved this study (Approval no. 2014-003). Written informed consent from participants was obtained before they completed the questionnaires.

Two chlorinated water samples (dry [May] and wet [September] seasons) prior to the oral examination date were collected from the pipeline terminal close to the participants’ residential addresses. The local Center for Disease Control and Prevention analyzed the terminal tap water samples for physicochemical indices. The trace heavy metals, including iron, manganese, copper, and zinc were measured by atomic absorption spectrophotometry or atomic absorption method. During the analysis, replicates were introduced to ensure the accuracy of the analysis. The relative errors of the replicates were within ± 5%, indicating acceptable analytical accuracy. To aid in the comparison of exposure test results, we calculated standardized Z scores scaled to the regional measurements by subtracting the regional measurement mean from each measurement site’s test values and dividing by the regional measurement standard deviation (SD) for those tests with approximately Gaussian distributions (manganese, iron, copper, and zinc).

Well-trained dentists performed periodontal examinations using a community periodontal index probe (CPI). A calibration training program was conducted before the survey to ensure the reliability of the results. In the survey, 5% of the random participants were selected to test the inter-examiner reproducibility. The kappa values for the three age groups ranged from 0.76 to 0.8. Four periodontal parameters, including bleeding on probing (BOP), presence of calculus, periodontal probing depth (PPD), and clinical attachment loss (CAL) were assessed. The diagnosis of periodontitis is according to the classification scheme proposed at the 2018 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions [1]. Participants with < 10% BOP-positive teeth and PPD ≤ 3 mm were classified as periodontal healthy. Periodontitis was classified to four stages depending on the four periodontal parameters. A CAL of 1–2 mm was classified as a stage I periodontal patient, a CAL of 3–4 mm was classified as a stage II periodontal patient, CAL over 5 mm was classified as a stage III periodontal patient, CAL over and equal 5 mm and with over and equal 5 missing teeth was classified as stage IV periodontal patient. Stage II periodontal patients were reclassified as stage III when the maximum PPD was over and equaled 6 mm [11]. In this study, periodontitis patients were diagnosed with stage I to IV periodontitis.

We selected potential confounders based on a literature review of previous studies. All the variables were collected using a standard structured questionnaire through face-to-face interviews by trained investigators. Potential risk factors about sociodemographic status, health behavior, and health conditions were controlled in our study. The sociodemographic status included age (35–49, 50–64, or 65–74), sex (male versus female), ethnicity (Han versus others), educational attainment (illiteracy, middle school and below, high school, or college), and residence status (living in rural versus urban areas). Health behavior covariates were measured by smoking status (never, former, or current), alcohol drinker(never, seldom, frequent, or former), frequency of having dessert or beverage (seldom, sometimes, or frequent), and whether the participants did teeth cleaning by dentists in the past 12 months (yes versus no). Covariates for health status included a history of stroke, diabetes, hypertension, heart disease, and chronic obstructive pulmonary disease (COPD) (yes versus no).

Descriptive analyses were conducted for all variables. We used weighted multivariable logistic regression models to estimate essential trace heavy metals exposure in drinking water and the risk of periodontitis with national representativeness. This study specified two models to adjust for a set of variables considered as risk factors. In the basic model (model 1), we adjusted for age, sex, ethnicity, educational attainment, and residence status. In the fully adjusted model (model 2), we further adjusted for current smoking status, alcohol consumption status, frequency of desserts or beverages, underwent teeth cleaning performed by dentists within the past 12 months, and medical history of stroke, diabetes, hypertension, heart diseases, and COPD. We conducted stratification analyses to explore whether the associations could be modified by sex, age, living region, smoking status, alcohol consumption, dessert consumption, and beverage consumption by including interaction terms between exposures and potential modifiers. For each potential effect modifier, we evaluated effect modification by likelihood ratio tests comparing models that included an interaction (product) term between essential trace heavy metals exposure and the effect modifier versus models without the interaction term. Stratum-specific ORs were obtained from the same interaction model by using the appropriate coefficients and variance-covariance matrix.

In the spline analysis, we explore the dose-response associations between essential trace heavy metals and the risk of periodontitis. All associations are presented as ORs with corresponding 95% CIs. A 2-sided p-value < 0.05 was considered statistically significant. Stata version 16.0 for Mac (Stata Corp, College Station, TX, USA) and R Studio Version 1.2.5042 (The R Project for Statistical Computing, Vienna, Austria) were used for the statistical analyses.