Abstract
Background
Tooth looseness is a clinically relevant self-reported oral health outcome that may reflect compromised tooth support. Sleep disorders have been associated with adverse oral health, but whether metabolic dysfunction contributes to the association between sleep disorder status and tooth looseness remains unclear. This study examined the association between self-reported sleep disorder status and tooth looseness and evaluated whether metabolic dysfunction, reflected by continuous metabolic score for insulin resistance (METS-IR), statistically accounted for part of this association.
Materials and methods
This study was a secondary data analysis of pooled data from the National Health and Nutrition Examination Survey (NHANES) 2011–2020. Tooth looseness without injury was the outcome, self-reported physician-diagnosed sleep disorder status was the exposure, and continuous METS-IR was specified as the primary intermediate metabolic variable. Elevated METS-IR, defined as METS-IR ≥ 45, was retained for sensitivity analyses. Survey weights and complex design variables were applied. Survey-weighted logistic regression was used to examine the association between sleep disorder status and tooth looseness and the association between continuous METS-IR and tooth looseness. Survey-weighted linear regression was used to examine the association between sleep disorder status and continuous METS-IR. Model 1 was unadjusted. Model 2, the primary adjusted model, adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. Model 3 was further adjusted for body mass index as a conservative sensitivity analysis. A regression-based indirect pathway framework was used to evaluate the model-based indirect pathway through continuous METS-IR.
Results
A total of 4112 participants were included in the final analytic sample, of whom 711 reported tooth looseness without injury. In the primary adjusted model, sleep disorder status was associated with higher odds of tooth looseness (OR = 1.46, 95% CI: 1.04–2.06, p = 0.038). Sleep disorder status was also associated with higher continuous METS-IR levels (β = 9.07, 95% CI: 7.30–10.85, p < 0.001), and continuous METS-IR was associated with tooth looseness (OR = 1.01 per 1-unit increase, 95% CI: 1.01–1.02, p < 0.001). After additional adjustment for BMI, the associations involving METS-IR were attenuated. In the model-based indirect pathway analysis, continuous METS-IR accounted for 23.72% of the total association between sleep disorder status and tooth looseness. Given the secondary analysis of cross-sectional survey data, this estimate was interpreted as a statistical decomposition of observed associations rather than evidence of causal mediation.
Conclusions
In this secondary data analysis of NHANES 2011–2020, self-reported sleep disorder status was associated with higher odds of tooth looseness, and continuous METS-IR statistically accounted for a modest proportion of this association. These findings suggest that metabolic dysfunction may be relevant to the association between sleep-related health and oral health. However, the results should be interpreted as cross-sectional, model-based associations rather than causal evidence. Longitudinal studies with objective sleep assessment, clinical periodontal evaluation, assessment of sleep bruxism, dietary assessment, and repeated metabolic measurements are needed to clarify temporality and potential causal relationships.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12903-026-09383-3.
Keywords: Sleep disorders, Tooth looseness, METS-IR, Insulin resistance, Metabolic dysfunction, NHANES, Secondary data analysis
Introduction
Tooth looseness in adults is a clinically relevant self-reported oral health outcome that may reflect compromised tooth support. Tooth looseness can arise from multiple periodontal, mechanical, dental, and systemic conditions. Among these, periodontal tissue breakdown is one of the major clinical contexts in which loosened teeth may occur. Clinically assessed tooth mobility has also been associated with a higher long-term risk of tooth extraction or tooth loss among patients with periodontitis, supporting the relevance of compromised tooth support as an upstream marker of adverse oral outcomes [1, 2]. Therefore, self-reported tooth looseness may provide useful population-level information on oral health, although it should not be interpreted as equivalent to a clinical periodontal diagnosis.
Sleep disorders are common and have been increasingly linked to systemic dysregulation and adverse oral health profiles. Previous epidemiological studies have reported associations between sleep-related problems and periodontitis, tooth loss, and poorer self-reported oral health [3]. However, most existing studies have focused on periodontitis or tooth loss, whereas less attention has been paid to tooth looseness as a self-reported indicator of compromised tooth support. Furthermore, whether metabolic dysfunction statistically accounts for part of the association between sleep disorder status and tooth looseness remains insufficiently characterized.
Metabolic dysfunction is a plausible factor connecting sleep disorders with compromised oral health. Sleep disturbance has been associated with impaired glycaemic regulation, insulin resistance, inflammatory activation, and neuroendocrine dysregulation [4, 5]. These systemic alterations may be related to periodontal inflammation, impaired tissue repair, and compromised tooth support. The metabolic score for insulin resistance (METS-IR) is a non-insulin-based index derived from fasting plasma glucose, triglycerides, high-density lipoprotein cholesterol, and body mass index, and has been used as a practical surrogate marker of insulin resistance and cardiometabolic risk in epidemiological studies [6]. In the present study, continuous METS-IR was used as the primary indicator of metabolic dysfunction.
Despite growing evidence linking sleep disorders with oral health, few population-based studies have specifically examined tooth looseness as the oral health outcome, and the role of METS-IR in this association remains unclear. To address this gap, we conducted a secondary data analysis of pooled NHANES 2011–2020 data to examine the association between self-reported sleep disorder status and tooth looseness and to evaluate whether continuous METS-IR represented a model-based indirect pathway in this association. We hypothesized that sleep disorder status would be associated with higher odds of tooth looseness and that continuous METS-IR would statistically account for a modest component of this association.
Methods
Study design and data source
This study was designed as a secondary data analysis of publicly available data from the National Health and Nutrition Examination Survey (NHANES), conducted by the National Center for Health Statistics (NCHS). NHANES uses a complex, multistage probability sampling design to generate nationally representative estimates of the non-institutionalized U.S. population. Data from the 2011–2012, 2013–2014, 2015–2016, and 2017–March 2020 pre-pandemic NHANES cycles were pooled for the present analysis. Because sleep disorder status, METS-IR, and tooth looseness were assessed within the same survey framework, the findings should be interpreted as cross-sectional associations. Ethics approval and consent procedures for NHANES are described in the Declarations section.
Participants
Participants were eligible for inclusion if they had valid data on tooth looseness, sleep disorder status, and METS-IR components. Tooth looseness was assessed using the NHANES oral health questionnaire item, “Any teeth became loose without an injury?” METS-IR required fasting plasma glucose, triglycerides, high-density lipoprotein cholesterol, and body mass index. Participants with unavailable, inapplicable, missing, or invalid tooth-looseness data were first excluded. Participants with missing or non-informative data on sleep disorder status or METS-IR components were then excluded. Missing covariate data were handled by available-case analysis in descriptive analyses and by complete-case analysis within each corresponding adjusted regression model. Non-informative responses, including “Refused,” “Don’t know,” and “Not ascertained,” were treated as missing values.
To assess potential selection bias related to analytic sample selection, baseline characteristics were compared between included participants and excluded participants among those with valid tooth-looseness data. These comparisons are presented in Supplementary Table S1.
Measures
Outcome: tooth looseness
Tooth looseness was assessed using the NHANES oral health questionnaire item, “Any teeth became loose without an injury?” Responses were coded as a binary outcome: yes or no. In this study, tooth looseness was used consistently as the outcome term throughout the manuscript. In the model-based indirect pathway framework, tooth looseness was specified as the outcome.
Exposure: sleep disorder status
Sleep disorder status was defined according to self-report of having been told by a doctor or other health professional that the participant had a sleep disorder. Responses were coded as a binary variable: yes or no. In the model-based indirect pathway framework, sleep disorder status was specified as the exposure.
METS-IR and elevated METS-IR
The metabolic score for insulin resistance (METS-IR) was calculated using the following formula proposed by Bello-Chavolla et al. [6]: METS-IR = ln [(2 × FPG) + TG] × BMI / ln (HDL-C), where FPG denotes fasting plasma glucose, TG denotes triglycerides, BMI denotes body mass index, and HDL-C denotes high-density lipoprotein cholesterol. FPG, TG, and HDL-C were expressed in mg/dL, and BMI was expressed in kg/m².
Continuous METS-IR was used as the primary metabolic variable in the main regression analyses and as the primary intermediate variable in the model-based indirect pathway analysis. Elevated METS-IR was defined as METS-IR ≥ 45 and was retained for sensitivity analyses. METS-IR < 45 was classified as non-elevated METS-IR. This threshold was selected a priori as an operational high-risk cut-off rather than as a universally validated diagnostic threshold. The rationale was based on the original validation of METS-IR as a non-insulin-based surrogate marker of insulin sensitivity and cardiometabolic risk [6], together with subsequent epidemiological evidence reporting METS-IR cut-off values in a similar range for metabolic risk classification, including sex-specific cut-offs for incident type 2 diabetes that were approximately 40–43 in women and 44–45 in men [7].
Covariates
Covariates were selected based on epidemiologic relevance and availability in NHANES. They included age, sex, race and ethnicity, education level, poverty-to-income ratio (PIR), and ever-smoking status. Race and ethnicity were categorized according to NHANES classifications. Education level was categorized as less than 9th grade, 9–11th grade, high school graduate, some college or AA degree, and college graduate or above. Ever-smoking status was defined according to self-report of having smoked at least 100 cigarettes in one’s lifetime. Participants who reported having smoked at least 100 cigarettes were classified as ever smokers, whereas those who had not were classified as never smokers.
Because BMI is incorporated in the METS-IR formula, BMI was not included in the primary adjusted models involving METS-IR. Instead, BMI was added in conservative sensitivity models to assess robustness to potential index overlap and overadjustment.
Statistical analysis
All analyses were conducted using R version 4.3.0. NHANES sampling weights, strata, and primary sampling units were incorporated to account for the complex survey design. When pooling multiple NHANES cycles, sampling weights were rescaled in accordance with NCHS analytic guidance. Because METS-IR required fasting laboratory variables, fasting subsample weights were used for analyses involving METS-IR. Continuous variables were summarized as survey-weighted means with standard errors, and categorical variables were summarized as unweighted counts with survey-weighted percentages. Group differences were assessed using design-based tests appropriate for complex survey data.
To evaluate potential selection bias related to analytic sample selection, baseline characteristics were compared between included participants and excluded participants with available baseline-comparison data among those with valid tooth-looseness data. Continuous variables were compared using design-based ANOVA, and categorical variables were compared using design-based χ² tests.
Survey-weighted regression models were used to examine the main associations. Survey-weighted logistic regression was used to examine the association between sleep disorder status and tooth looseness. Survey-weighted linear regression was used to examine the association between sleep disorder status and continuous METS-IR. Survey-weighted logistic regression was used to examine the association between continuous METS-IR and tooth looseness, with continuous METS-IR modeled per 1-unit increase. Sequential models were fitted. Model 1 was unadjusted. Model 2, the primary adjusted model, included age, sex, race and ethnicity, education level, PIR, and ever-smoking status. Model 3 was further adjusted for BMI as a conservative sensitivity analysis. Because BMI is incorporated in the METS-IR formula, Model 3 was interpreted as a sensitivity model for potential index overlap and overadjustment rather than as the primary inferential model. For adjusted regression models, missing covariate data were handled by complete-case analysis within each corresponding model.
A regression-based indirect pathway framework was used to evaluate whether continuous METS-IR statistically accounted for part of the association between sleep disorder status and tooth looseness. Sleep disorder status was specified as the exposure, continuous METS-IR as the intermediate variable, and tooth looseness as the outcome. The primary mediator and outcome models were adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. BMI was not included in the primary indirect pathway model because BMI is incorporated in the METS-IR calculation.
Estimation was performed using the R package mediation with 1000 quasi-Bayesian Monte Carlo simulations. The package provides indirect and direct effect estimates. Because the present study was a secondary analysis of cross-sectional survey data, these estimates were interpreted as model-based statistical decomposition measures rather than as evidence of causal mediation. Results were therefore described as the statistical indirect pathway estimate, statistical direct pathway estimate, total association, and proportion accounted for by the indirect pathway.
Sensitivity analyses were conducted using elevated METS-IR, defined as METS-IR ≥ 45, to preserve comparability with previous threshold-based epidemiological studies and to provide clinically interpretable high-risk classification. These analyses included the association between sleep disorder status and elevated METS-IR, the association between elevated METS-IR and tooth looseness, and the corresponding threshold-based indirect pathway analysis with elevated METS-IR specified as the intermediate variable. These threshold-based sensitivity analyses are presented in Supplementary Tables S2–S4.
Because sleep disorder status and tooth looseness were assessed as binary self-reported variables, dose-response analyses based on sleep disorder severity, frequency, duration, or graded tooth-looseness severity could not be performed. As an additional analysis, restricted cubic spline analysis was conducted using continuous METS-IR to evaluate the dose-response association between METS-IR and tooth looseness. The model was adjusted for age, sex, race and ethnicity, education level, PIR, ever-smoking status, and sleep disorder status. BMI was not included in this spline model because BMI is a component of METS-IR. The results are presented in Supplementary Figure S1.
All statistical tests were two-sided, and p < 0.05 was considered statistically significant.
Results
Participant selection and baseline characteristics
A total of 45,462 participants from pooled NHANES 2011–2020 cycles were initially considered. After excluding participants with unavailable, inapplicable, missing, or invalid tooth-looseness data, 9359 participants had valid tooth-looseness data. Among these, participants with missing or non-informative data on sleep disorder status or METS-IR components were excluded, resulting in a final analytic sample of 4112 participants. Of these, 711 reported tooth looseness without injury, and 3401 did not (Fig. 1).
Fig. 1.
Flow diagram of participant selection
Weighted baseline characteristics of the analytic sample are shown in Table 1. Compared with participants without tooth looseness, those with tooth looseness were older (56.16 vs. 52.41 years, p < 0.001), had a lower poverty-to-income ratio (2.26 vs. 3.13, p < 0.001), and had a higher BMI (30.35 vs. 29.17 kg/m², p = 0.011). Group distributions also differed by race and ethnicity (p < 0.001), education level (p < 0.001), sex (p = 0.023), and ever-smoking status (p < 0.001). Sleep disorder status was more prevalent among participants with tooth looseness than among those without tooth looseness (16.31% vs. 10.51%, p = 0.002). Participants with tooth looseness also had higher continuous METS-IR levels than those without tooth looseness (46.21 vs. 43.41, p = 0.002), and elevated METS-IR was more prevalent in the tooth-looseness group (48.21% vs. 37.30%, p < 0.001).
Table 1.
Weighted baseline characteristics of the study population by tooth looseness status
| Variable | Tooth looseness | ||||
|---|---|---|---|---|---|
| Total (n = 4112) | Yes (n = 711) | No (n = 3401) | Statistic | p | |
| Age (year) | 52.94 (0.36) | 56.16 (0.66) | 52.41 (0.41) | F = 21.44 | < 0.001 |
| Race and ethnicity, n (%) | χ²=30.63 | < 0.001 | |||
| Mexican American | 467 (7.43) | 80 (7.18) | 387 (7.47) | ||
| Other Hispanic | 395 (5.36) | 56 (4.64) | 339 (5.48) | ||
| Non-Hispanic White | 1786 (70.52) | 288 (66.83) | 1498 (71.13) | ||
| Non-Hispanic Black | 860 (9.68) | 212 (15.61) | 648 (8.70) | ||
| Non-Hispanic Asian | 516 (5.02) | 60 (3.81) | 456 (5.22) | ||
| Other Race | 88 (2.00) | 15 (1.93) | 73 (2.01) | ||
| Education level, n (%) | χ²=106.08 | < 0.001 | |||
| Less than 9th grade | 415 (6.23) | 69 (7.41) | 346 (6.03) | ||
| 9-11th grade | 595 (11.53) | 138 (17.37) | 457 (10.57) | ||
| High school graduate | 889 (20.30) | 190 (25.59) | 699 (19.44) | ||
| Some college or AA degree | 1123 (29.92) | 223 (35.51) | 900 (29.00) | ||
| College graduate or above | 1088 (32.00) | 90 (14.09) | 998 (34.95) | ||
| Ratio of family income to poverty | 3.01 (0.10) | 2.26 (0.11) | 3.13 (0.10) | F = 67.24 | < 0.001 |
| BMI, kg/m² | 29.34 (0.19) | 30.35 (0.46) | 29.17 (0.18) | F = 7.34 | 0.011 |
| Sex, n (%) | χ²=6.52 | 0.023 | |||
| Male | 2006 (48.58) | 384 (53.49) | 1622 (47.77) | ||
| Female | 2106 (51.42) | 327 (46.51) | 1779 (52.23) | ||
| Sleep disorder status, n (%) | χ²=16.67 | 0.002 | |||
| Yes | 430 (11.33) | 101 (16.31) | 329 (10.51) | ||
| No | 3682 (88.67) | 610 (83.69) | 3072 (89.49) | ||
| Ever-smoking status, n (%) | χ²=105.51 | < 0.001 | |||
| Yes | 1875 (45.85) | 436 (65.53) | 1439 (42.62) | ||
| No | 2233 (54.11) | 275 (34.47) | 1958 (57.34) | ||
| METS-IR, mean (SE) | 43.81 (0.41) | 46.21 (0.91) | 43.41 (0.38) | F = 12.04 | 0.002 |
| Elevated METS-IR, n (%) | χ²=24.97 | < 0.001 | |||
| < 45 | 2516 (61.16) | 390 (51.79) | 2126 (62.70) | ||
| ≥ 45 | 1596 (38.84) | 321 (48.21) | 1275 (37.30) | ||
Values are presented as survey-weighted means with standard errors for continuous variables and unweighted counts with survey-weighted percentages for categorical variables. Group comparisons were performed using design-based ANOVA for continuous variables and design-based χ² tests for categorical variables, accounting for the complex NHANES survey design. Ever-smoking status was defined as having smoked at least 100 cigarettes in one’s lifetime
BMI Body mass index, NHANES National Health and Nutrition Examination Survey, SE Standard error. For variables with missing responses, available-case counts are shown
To assess potential selection bias related to analytic sample selection, baseline characteristics were compared between included participants and excluded participants with available baseline-comparison data among those with valid tooth-looseness data. Included and excluded participants were generally similar in age, education level, poverty-to-income ratio, BMI, sex, self-reported tooth looseness, sleep disorder status, and ever-smoking status. However, race and ethnicity distribution differed significantly between included and excluded participants (p = 0.002). These findings indicate that potential selection bias related to missing data cannot be fully excluded (Supplementary Table S1).
Association between sleep disorder status and tooth looseness
Survey-weighted logistic regression showed that sleep disorder status was associated with tooth looseness (Table 2). In the unadjusted model, participants reporting sleep disorders had higher odds of tooth looseness than those without sleep disorders (OR = 1.66, 95% CI: 1.19–2.31, p = 0.005). After adjustment for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status, this association remained statistically significant but was attenuated (OR = 1.46, 95% CI: 1.04–2.06, p = 0.038). After additional adjustment for BMI, the association was further attenuated and was no longer statistically significant (OR = 1.34, 95% CI: 0.94–1.91, p = 0.124).
Table 2.
Association between sleep disorder status and tooth looseness
| Variables | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
| Sleep disorder status | ||||||
| No | 1.00 (Reference) | 1.00 (Reference) | 1.00 (Reference) | |||
| Yes | 1.66 (1.19–2.31) | 0.005 | 1.46 (1.04–2.06) | 0.038 | 1.34 (0.94–1.91) | 0.124 |
Outcome: tooth looseness. Exposure: sleep disorder status, with no sleep disorder as the reference group. ORs and 95% CIs were estimated using survey-weighted logistic regression accounting for the complex NHANES sampling design. Model 1 was unadjusted. Model 2, the primary adjusted model, adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. Model 3 was further adjusted for BMI as a sensitivity analysis
BMI Body mass index, CI Confidence interval, NHANES National Health and Nutrition Examination Survey, OR Odds ratio
Association between sleep disorder status and continuous METS-IR
Sleep disorder status was associated with higher continuous METS-IR levels in survey-weighted linear regression models (Table 3). In the unadjusted model, participants reporting sleep disorders had higher METS-IR levels than those without sleep disorders (β = 8.91, 95% CI: 7.15–10.66, p < 0.001). After adjustment for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status, this association remained statistically significant (β = 9.07, 95% CI: 7.30–10.85, p < 0.001). After additional adjustment for BMI, the association was markedly attenuated and was no longer statistically significant (β = 0.72, 95% CI: -0.30 to 1.73, p = 0.180). Because BMI is incorporated in the METS-IR formula, the BMI-adjusted model was interpreted as a conservative sensitivity analysis rather than as the primary inferential model.
Table 3.
Association between sleep disorder status and continuous METS-IR
| Variables | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| β (95% CI) | p | β (95% CI) | p | β (95% CI) | p | |
| Sleep disorder status | ||||||
| No | 0.00 (Reference) | 0.00 (Reference) | 0.00 (Reference) | |||
| Yes | 8.91 (7.15–10.66) | < 0.001 | 9.07 (7.30–10.85) | < 0.001 | 0.72 (-0.30–1.73) | 0.180 |
Outcome: continuous METS-IR. Exposure: sleep disorder status, with no sleep disorder as the reference group. β coefficients and 95% CIs were estimated using survey-weighted linear regression accounting for the complex NHANES sampling design. Model 1 was unadjusted. Model 2, the primary adjusted model, was adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. Model 3 was further adjusted for BMI as a conservative sensitivity analysis. Because BMI is incorporated in the METS-IR formula, Model 3 was interpreted as a sensitivity analysis for potential index overlap and overadjustment rather than as the primary inferential model
BMI Body mass index, CI Confidence interval, METS-IR Metabolic score for insulin resistance, NHANES National Health and Nutrition Examination Survey
Association between continuous METS-IR and tooth looseness
Continuous METS-IR was associated with tooth looseness in survey-weighted logistic regression models (Table 4). In the unadjusted model, each 1-unit increase in METS-IR was associated with higher odds of tooth looseness (OR = 1.02, 95% CI: 1.01–1.02, p < 0.001). After adjustment for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status, the association remained statistically significant (OR = 1.01 per 1-unit increase, 95% CI: 1.01–1.02, p < 0.001). After additional adjustment for BMI, the association was attenuated and was no longer statistically significant (OR = 1.02, 95% CI: 1.00–1.03, p = 0.093). Because BMI is incorporated in the METS-IR formula, the BMI-adjusted model was interpreted as a conservative sensitivity analysis rather than as the primary inferential model.
Table 4.
Association between continuous METS-IR and tooth looseness
| Variables | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | OR (95% CI) | p | |
| Continuous METS-IR, per 1-unit increase | 1.02 (1.01–1.02) | < 0.001 | 1.01 (1.01–1.02) | < 0.001 | 1.02 (1.00–1.03) | 0.093 |
Outcome: self-reported tooth looseness. Exposure: continuous METS-IR, modeled per 1-unit increase. ORs and 95% CIs were estimated using survey-weighted logistic regression accounting for the complex NHANES sampling design. Model 1 was unadjusted. Model 2, the primary adjusted model, was adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. Model 3 was further adjusted for BMI as a conservative sensitivity analysis. Because BMI is incorporated in the METS-IR formula, Model 3 was interpreted as a sensitivity analysis for potential index overlap and overadjustment rather than as the primary inferential model
BMI Body mass index, CI Confidence interval, METS-IR Metabolic score for insulin resistance, NHANES National Health and Nutrition Examination Survey, OR Odds ratio
Model-based indirect pathway analysis
A model-based indirect pathway analysis was conducted to assess whether continuous METS-IR statistically accounted for part of the association between sleep disorder status and tooth looseness in this cross-sectional setting. Sleep disorder status was specified as the exposure, continuous METS-IR as the intermediate variable, and tooth looseness as the outcome. In the primary adjusted model, the estimated indirect pathway through continuous METS-IR was statistically significant (estimate = 0.010, p < 0.001). By contrast, the remaining direct pathway between sleep disorder status and tooth looseness after accounting for continuous METS-IR was not statistically significant (estimate = 0.040, p = 0.102). The total association was statistically significant (estimate = 0.050, p = 0.032). Overall, continuous METS-IR accounted for 23.72% of the total association between sleep disorder status and tooth looseness (Fig. 2).
Fig. 2.
Statistical indirect pathway analysis of the association between sleep disorder status and tooth looseness through continuous METS-IR
Given the secondary analysis of cross-sectional survey data, these estimates were interpreted as a model-based statistical decomposition of observed associations rather than evidence of causal mediation. Threshold-based sensitivity analyses using elevated METS-IR are presented in Supplementary Tables S2–S4.
Sleep disorder status was specified as the exposure (X), continuous METS-IR as the intermediate variable (M), and tooth looseness as the outcome (Y). Path a represents the association between sleep disorder status and continuous METS-IR. Path b represents the association between continuous METS-IR and tooth looseness after adjustment for sleep disorder status. Path c′ represents the remaining association between sleep disorder status and tooth looseness after accounting for continuous METS-IR. The statistical indirect pathway was calculated as a × b. The statistical indirect pathway estimate was 0.010 (p < 0.001), the statistical direct pathway estimate was 0.040 (p = 0.102), and the total association was 0.050 (p = 0.032). Continuous METS-IR accounted for 23.72% of the total association. Models were adjusted for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. BMI was not included in the primary indirect pathway model because BMI is incorporated into the METS-IR calculation.
Dose-response analysis of METS-IR and tooth looseness
Because sleep disorder status and tooth looseness were assessed as binary self-reported variables, dose-response analyses based on sleep disorder severity, frequency, duration, or graded tooth-looseness severity could not be performed. Therefore, restricted cubic spline analysis was conducted using continuous METS-IR to further evaluate the dose-response association between metabolic dysfunction and tooth looseness.
The restricted cubic spline analysis showed a significant overall association between METS-IR and tooth looseness (p for overall < 0.001), with no evidence of significant nonlinearity (p for nonlinearity = 0.358). The curve suggested an approximately linear positive association, with higher METS-IR levels corresponding to higher odds of tooth looseness (Supplementary Figure S1).
Discussion
Overview and principal interpretation
In this secondary data analysis of NHANES 2011–2020, self-reported sleep disorder status was associated with higher odds of tooth looseness in the primary adjusted model. Sleep disorder status was also associated with higher continuous METS-IR levels, and continuous METS-IR was associated with tooth looseness. In the model-based indirect pathway analysis, continuous METS-IR accounted for a modest proportion of the total association between sleep disorder status and tooth looseness. Threshold-based sensitivity analyses using elevated METS-IR yielded supportive findings and are presented in supplementary materials. Together, these findings suggest that metabolic dysfunction, as reflected by METS-IR, may be relevant to the observed association between sleep-related health and tooth looseness.
The results should be interpreted cautiously. The study was based on a secondary analysis of cross-sectional survey data, and sleep disorder status, METS-IR, and tooth looseness were assessed within the same survey framework. Therefore, the observed indirect pathway represents a statistical decomposition of associations rather than evidence of causal mediation or temporal ordering. The modest proportion accounted for by continuous METS-IR also suggests that metabolic dysfunction is unlikely to be the only factor involved. Other pathways, including inflammation, oral health behaviors, dental care utilization, medication use, comorbid systemic conditions, and sleep bruxism-related mechanical loading, may also contribute to the association between sleep disorder status and tooth looseness.
Sleep disorder status and tooth looseness
Sleep disorder status was associated with tooth looseness after adjustment for age, sex, race and ethnicity, education level, poverty-to-income ratio, and ever-smoking status. This finding is consistent with previous evidence linking sleep-related problems with adverse oral health outcomes, including periodontitis, tooth loss, and poorer self-reported oral health [3, 8–11]. However, after additional adjustment for BMI, the association between sleep disorder status and tooth looseness was attenuated and was no longer statistically significant. This attenuation suggests that body size or related metabolic factors may partly account for the observed association, although BMI adjustment should be interpreted cautiously in this study because BMI is also a component of the METS-IR construct.
Tooth looseness should not be interpreted as a clinical diagnosis of periodontitis. Nevertheless, it is a clinically meaningful self-reported indicator because it may reflect compromised tooth support. Clinically assessed tooth mobility has been associated with a higher long-term risk of tooth extraction or tooth loss among patients with periodontitis [1, 2]. Therefore, the present findings extend prior sleep–oral health research by focusing on tooth looseness as a population-level oral health outcome that may capture compromised tooth support before tooth loss occurs.
In addition to metabolic or periodontal pathways, sleep disorders may also be linked to tooth looseness through mechanical pathways. Sleep bruxism, tooth grinding, and occlusal overload may increase mechanical stress on periodontal support and contribute to tooth mobility [12, 13]. Because NHANES did not provide detailed information on sleep bruxism, occlusal trauma, parafunctional habits, or clinical mobility grading in the analytic framework used, this alternative non-metabolic pathway could not be directly evaluated.
Sleep disorder status and continuous METS-IR
Sleep disorder status was strongly associated with higher continuous METS-IR levels in the primary adjusted model. This finding is consistent with the broader sleep-metabolic literature. Experimental sleep restriction studies have shown impaired insulin sensitivity and metabolic endocrine alterations in healthy adults [14, 15]. Epidemiological and meta-analytic evidence also links abnormal sleep duration, insomnia symptoms, and poor sleep quality with glycaemic dysregulation and type 2 diabetes risk [4, 16, 17].
After additional adjustment for BMI, the association between sleep disorder status and continuous METS-IR was markedly attenuated and was no longer statistically significant. This finding is expected because BMI is directly incorporated in the METS-IR formula. Therefore, the BMI-adjusted model should not be interpreted as replacing the primary model. Instead, it serves as a conservative sensitivity analysis evaluating the extent to which the association involving METS-IR overlaps with BMI-related information.
Continuous METS-IR and tooth looseness
Continuous METS-IR was associated with higher odds of tooth looseness in the primary adjusted model. This association is biologically plausible given the established links between metabolic dysfunction, insulin resistance, dysglycaemia, and periodontal tissue breakdown. Diabetes and periodontitis are widely recognized as bidirectionally related chronic conditions [18, 19], and metabolic syndrome has been associated with periodontitis in systematic reviews and meta-analyses [20]. These findings support the possibility that metabolic vulnerability may be related to compromised tooth support.
The restricted cubic spline analysis using continuous METS-IR further supported this pattern. METS-IR showed a significant overall association with tooth looseness, without evidence of significant nonlinearity, suggesting an approximately linear positive association. This finding supports the use of continuous METS-IR in the primary analyses and reduces reliance on a threshold-based definition. However, this dose-response analysis was limited to METS-IR because sleep disorder status and tooth looseness were assessed as binary self-reported variables in NHANES.
Model-based indirect pathway through continuous METS-IR
The model-based indirect pathway analysis suggested that continuous METS-IR accounted for 23.72% of the total association between sleep disorder status and tooth looseness. The indirect pathway was statistically significant, whereas the remaining direct pathway was not statistically significant after accounting for continuous METS-IR. This pattern is statistically consistent with a modest indirect pathway involving metabolic dysfunction.
This finding is plausible in the context of prior evidence linking sleep disturbance with inflammatory activation, neuroendocrine dysregulation, and metabolic impairment [5, 14–17]. In parallel, periodontal tissue breakdown and compromised tooth support are closely related to host–microbe interactions and dysregulated inflammatory responses [21, 22]. These lines of evidence support the biological plausibility of a sleep–metabolic–oral health pathway.
Nevertheless, the present analysis should not be interpreted as demonstrating a causal mechanism. Estimates from regression-based indirect-pathway models in observational and cross-sectional data rely on assumptions that are difficult to verify, including the absence of unmeasured confounding and the correct temporal ordering of exposure, intermediate variable, and outcome [23, 24]. In this study, sleep disorder status, METS-IR, and tooth looseness were measured within the same survey setting. Therefore, the indirect pathway should be understood as a model-based statistical decomposition of observed associations. Longitudinal studies are needed to determine whether sleep disorders precede metabolic dysfunction and whether metabolic dysfunction subsequently contributes to tooth looseness.
Interpretation of BMI sensitivity models
BMI sensitivity models should be interpreted with particular caution. BMI is incorporated in the METS-IR formula, and additional adjustment for BMI may introduce index overlap and overadjustment when METS-IR is treated as an exposure or intermediate variable [25, 26]. In the present study, associations involving METS-IR were attenuated after BMI adjustment. This attenuation does not necessarily contradict the primary findings. Rather, it indicates that part of the information captured by METS-IR overlaps with BMI-related variation.
METS-IR should not be interpreted as merely equivalent to BMI. BMI mainly reflects body size or adiposity, whereas METS-IR integrates BMI with fasting plasma glucose, triglycerides, and high-density lipoprotein cholesterol. Therefore, METS-IR may capture insulin-resistance-related metabolic dysfunction more comprehensively than BMI alone. For this reason, the non-BMI-adjusted model was specified as the primary adjusted model for analyses involving METS-IR, whereas the BMI-adjusted model was presented as a conservative sensitivity analysis. This approach allows the primary inference to reflect the METS-IR construct while still showing how results change when BMI is additionally controlled.
Strengths and limitations
This study has several strengths. It used pooled NHANES data across multiple survey cycles, incorporated sampling weights and complex survey design variables, and applied a consistent analytic strategy to examine the associations among sleep disorder status, continuous METS-IR, and tooth looseness. The analysis focused on tooth looseness, a self-reported oral health outcome that has received less attention than periodontitis or tooth loss in previous sleep-related oral health research. In response to concerns regarding dichotomization of METS-IR, continuous METS-IR was used as the primary metabolic variable, while elevated METS-IR was retained only for threshold-based sensitivity analyses. In addition, we incorporated ever-smoking status, included a participant flow diagram, compared included and excluded participants, and supplemented the main analysis with a restricted cubic spline analysis of continuous METS-IR.
Several limitations should be considered. First, because this was a secondary analysis of cross-sectional survey data, temporality and causality cannot be inferred. The model-based indirect pathway involving continuous METS-IR should be interpreted as a statistical decomposition of observed associations rather than evidence of causal mediation. Longitudinal studies are required to clarify whether sleep disorder status precedes metabolic dysfunction and subsequent tooth looseness. Reverse causality is also possible. Tooth looseness or compromised dentition may impair mastication, reduce intake of fibrous foods, shift dietary choices toward softer or more processed foods, and thereby worsen metabolic profiles, including METS-IR.
Second, sleep disorder status was based on a broad self-reported NHANES item asking whether a doctor or other health professional had ever told the participant that they had a sleep disorder. This measure does not distinguish between insomnia, obstructive sleep apnea, circadian rhythm disorders, or other sleep-related conditions. These subtypes may have different metabolic, inflammatory, behavioral, and oral health profiles. In particular, obstructive sleep apnea may be more closely linked to obesity and metabolic dysfunction, whereas insomnia may involve different behavioral, neuroendocrine, or inflammatory pathways. Therefore, the findings cannot be attributed to any specific sleep disorder subtype.
Third, tooth looseness was assessed using a self-reported binary item and should be interpreted as a proxy for clinical tooth mobility rather than as an objective periodontal measurement. The item did not capture the number of affected teeth, looseness severity, periodontal pocket depth, clinical attachment loss, radiographic bone loss, alveolar bone status, or clinical mobility grade. Previous studies have shown that self-reported periodontal measures may be useful for surveillance but cannot replace standardized clinical periodontal examinations [27, 28]. In addition, NHANES did not provide sufficient information in the analytic framework used to evaluate sleep bruxism, occlusal trauma, or tooth grinding-related mechanical loading, which may represent alternative pathways linking sleep disorders to tooth looseness.
Fourth, residual confounding may remain. Although the models adjusted for major sociodemographic variables and ever-smoking status, several potentially important oral health-related and systemic confounders were not included in the primary models. These include oral hygiene practices, dental visit frequency, periodontal treatment history, medication use, oral health status, systemic inflammatory conditions, hormonal changes, osteoporosis, other bone-related diseases, diet quality, and masticatory function. These factors may influence tooth looseness and may also affect the observed indirect pathway through metabolic dysfunction.
Fifth, potential selection bias cannot be fully excluded. The final analytic sample was substantially smaller than the initial pooled NHANES sample, largely because tooth-looseness data were unavailable, inapplicable, missing, or invalid for many participants, and because METS-IR required fasting laboratory data. To evaluate this issue, we presented a participant flow diagram and compared included and excluded participants among those with valid tooth-looseness data. Although included and excluded participants were generally similar in several characteristics, race and ethnicity distribution differed significantly between groups. Therefore, missing-data-related selection bias may still have influenced the findings.
Finally, dose-response analyses based on sleep disorder severity, frequency, duration, or graded tooth-looseness severity could not be performed because the NHANES sleep disorder and tooth-looseness items were binary. The supplementary restricted cubic spline analysis was therefore limited to continuous METS-IR. Future studies using graded sleep assessments, objective periodontal examinations, assessment of sleep bruxism, dietary assessment, masticatory function measures, and repeated metabolic measurements are needed to evaluate more detailed dose-response patterns.
Implications
The present findings suggest that metabolic dysfunction may be relevant to the association between sleep-related health and tooth looseness. From a population-health perspective, individuals with sleep disorders may represent a group in whom oral health and metabolic risk deserve integrated attention. However, the results should not be interpreted as supporting clinical screening or intervention decisions based solely on the present analysis. Future longitudinal and mechanistic studies are needed to clarify whether improving sleep health or metabolic status is associated with a lower risk of compromised tooth support, and whether non-metabolic pathways such as sleep bruxism and occlusal loading also contribute to this association.
Conclusions
In this secondary data analysis of NHANES 2011–2020, self-reported sleep disorder status was associated with higher odds of tooth looseness in the primary adjusted model. Continuous METS-IR was also associated with tooth looseness and statistically accounted for a modest proportion of the association between sleep disorder status and tooth looseness. Additional restricted cubic spline analysis suggested an approximately linear positive association between continuous METS-IR and tooth looseness. These findings indicate that metabolic dysfunction may be relevant to the association between sleep-related health and oral health. However, the results should be interpreted as cross-sectional, model-based associations rather than evidence of causality. Longitudinal studies with objective sleep assessment, clinical periodontal evaluation, assessment of sleep bruxism, dietary assessment, masticatory function measures, and repeated metabolic measurements are needed to clarify temporality and to determine whether graded sleep or metabolic abnormalities are associated with tooth looseness.
Supplementary Information
Acknowledgements
The authors would like to thank The Seventh Clinical Medical School, Guangzhou University of Chinese Medicine for their resources and support. The authors also thank SciBio Innovation Technology Limited for administrative and logistical support.
Abbreviations
- BMI
Body mass index
- CI
Confidence interval
- FPG
Fasting plasma glucose
- HDL-C
High-density lipoprotein cholesterol
- METS-IR
Metabolic score for insulin resistance
- NCHS
National Center for Health Statistics
- NHANES
National Health and Nutrition Examination Survey
- OR
Odds ratio
- PIR
Poverty-to-income ratio
- PSU
Primary sampling unit
- SE
Standard error
- TG
Triglycerides
Authors’ contributions
Xiarui Zhang and Guohai Li contributed equally to the manuscript.Conceptualization: Xiarui Zhang, Lingdu Wen, Junquan Liang.Methodology: Guohai Li, Junquan Liang.Formal analysis: Guohai Li.Data curation: Guohai Li.Visualization: Junquan Liang, Wei Zhang, Tongzhou Liang.Writing-original draft: Junquan Liang.Writing-review & editing: Junquan Liang, Rundong Tang.Supervision: Xiarui Zhang, Lingdu Wen, Junquan Liang, Tongzhou Liang.
Funding
This work was supported by the National Natural Science Foundation of China (No. 82305372); the Shenzhen Natural Science Foundation General Program for Basic Research (No. JCYJ20230807115516032); the 2024 High-Quality Development Research Project of Public Hospitals in Bao’an District, Shenzhen (No. BAGZL2024108).
Data availability
NHANES data are publicly available from the CDC/NCHS website. This study used pooled data from the 2011–2012, 2013–2014, 2015–2016, and 2017–March 2020 pre-pandemic NHANES cycles.
Declarations
Ethics approval and consent to participate
The NHANES survey protocol was approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board, and all procedures were conducted in accordance with the ethical principles of the Declaration of Helsinki. Written informed consent was obtained from all participants at the time of NHANES data collection. For minors, written parental permission was obtained, and children/adolescents provided assent as appropriate. The present study was a secondary analysis of publicly available, de-identified data and therefore did not require additional institutional review board approval.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Junquan Liang, Email: gliangzixuan@163.com.
Lingdu Wen, Email: wenlingdu@126.com.
References
- 1.Tonetti MS, Greenwell H, Kornman KS. Staging and grading of periodontitis: Framework and proposal of a new classification and case definition. J Periodontol. 2018;89(Suppl 1):S159–172. [DOI] [PubMed] [Google Scholar]
- 2.Peditto M, Rupe C, Gambino G, Di Martino M, Barbato L, Cairo F, Oteri G, Cavalcanti R. Influence of mobility on the long-term risk of tooth extraction/loss in periodontitis patients. A systematic review and meta-analysis. J Periodontal Res. 2024;59(6):1047–61. [DOI] [PubMed] [Google Scholar]
- 3.Jin H, Zhang Y, Ren R, Shi Y, Feng X, Vitiello MV, Tang X. Associations between insomnia symptoms and self-reported oral health in 59370 adults. Sleep Breath. 2024;29(1):16. [DOI] [PubMed] [Google Scholar]
- 4.Chen Y, Tong E, Rao Y, Yu EY, Zeegers M, Wesselius A. The association between insomnia (related symptoms) and glycaemic control: a systematic review and meta-analysis. J Glob Health. 2025;15:04016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Irwin MR. Sleep and inflammation: partners in sickness and in health. Nat Rev Immunol. 2019;19(11):702–15. [DOI] [PubMed] [Google Scholar]
- 6.Bello-Chavolla OY, Almeda-Valdes P, Gomez-Velasco D, Viveros-Ruiz T, Cruz-Bautista I, Romo-Romo A, Sánchez-Lázaro D, Meza-Oviedo D, Vargas-Vázquez A, Campos OA, et al. METS-IR, a novel score to evaluate insulin sensitivity, is predictive of visceral adiposity and incident type 2 diabetes. Eur J Endocrinol. 2018;178(5):533–44. [DOI] [PubMed] [Google Scholar]
- 7.Xing Z, Xiao M, Schocken DD, Zgibor JC, Alman AC. Sex-specific optimal cut-off points for metabolic health indicators to predict incident type 2 diabetes mellitus. Nutr metabolism Cardiovasc diseases: NMCD. 2025;35(8):103963. [DOI] [PubMed] [Google Scholar]
- 8.Alhassani AA, Al-Zahrani MS. Is inadequate sleep a potential risk factor for periodontitis? PLoS ONE. 2020;15(6):e0234487. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zhou Q, Ye F, Zhang S. Sleep Duration and Risk of Periodontitis-A Systematic Review and Meta-Analysis. Int Dent J. 2024;74(3):418–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lei F, Hu X. Oral health and sleep disorders: A systematic review and meta-analysis. Biomed Rep. 2025;22(3):37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Carra MC, Balagny P, Bouchard P. Sleep and periodontal health. Periodontol 2000. 2024;96(1):42–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Fan J, Caton JG. Occlusal trauma and excessive occlusal forces: Narrative review, case definitions, and diagnostic considerations. J Periodontol. 2018;89(Suppl 1):S214–222. [DOI] [PubMed] [Google Scholar]
- 13.Pauletto P, Polmann H, Conti Réus J, Massignan C, de Souza BDM, Gozal D, Lavigne G, Flores-Mir C, De Canto L. Sleep bruxism and obstructive sleep apnea: association, causality or spurious finding? A scoping review. Sleep. 2022;45(7):zsac073. [DOI] [PubMed] [Google Scholar]
- 14.Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt on metabolic and endocrine function. Lancet. 1999;354(9188):1435–9. [DOI] [PubMed] [Google Scholar]
- 15.Buxton OM, Pavlova M, Reid EW, Wang W, Simonson DC, Adler GK. Sleep restriction for 1 week reduces insulin sensitivity in healthy men. Diabetes. 2010;59(9):2126–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Shan Z, Ma H, Xie M, Yan P, Guo Y, Bao W, Rong Y, Jackson CL, Hu FB, Liu L. Sleep duration and risk of type 2 diabetes: a meta-analysis of prospective studies. Diabetes Care. 2015;38(3):529–37. [DOI] [PubMed] [Google Scholar]
- 17.Cappuccio FP, D’Elia L, Strazzullo P, Miller MA. Quantity and quality of sleep and incidence of type 2 diabetes: a systematic review and meta-analysis. Diabetes Care. 2010;33(2):414–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sanz M, Ceriello A, Buysschaert M, Chapple I, Demmer RT, Graziani F, Herrera D, Jepsen S, Lione L, Madianos P, et al. Scientific evidence on the links between periodontal diseases and diabetes: Consensus report and guidelines of the joint workshop on periodontal diseases and diabetes by the International diabetes Federation and the European Federation of Periodontology. Diabetes Res Clin Pract. 2018;137:231–41. [DOI] [PubMed] [Google Scholar]
- 19.Preshaw PM, Alba AL, Herrera D, Jepsen S, Konstantinidis A, Makrilakis K, Taylor R. Periodontitis and diabetes: a two-way relationship. Diabetologia. 2012;55(1):21–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nibali L, Tatarakis N, Needleman I, Tu YK, D’Aiuto F, Rizzo M, Donos N. Clinical review: Association between metabolic syndrome and periodontitis: a systematic review and meta-analysis. J Clin Endocrinol Metab. 2013;98(3):913–20. [DOI] [PubMed] [Google Scholar]
- 21.Hajishengallis G. Periodontitis: from microbial immune subversion to systemic inflammation. Nat Rev Immunol. 2015;15(1):30–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Papapanou PN, Sanz M, Buduneli N, Dietrich T, Feres M, Fine DH, Flemmig TF, Garcia R, Giannobile WV, Graziani F, et al. Periodontitis: Consensus report of workgroup 2 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. J Periodontol. 2018;89(Suppl 1):S173–182. [DOI] [PubMed] [Google Scholar]
- 23.Imai K, Keele L, Tingley D. A general approach to causal mediation analysis. Psychol Methods. 2010;15(4):309–34. [DOI] [PubMed] [Google Scholar]
- 24.VanderWeele TJ. Mediation Analysis: A Practitioner’s Guide. Annu Rev Public Health. 2016;37:17–32. [DOI] [PubMed] [Google Scholar]
- 25.Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology. 2009;20(4):488–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Cole SR, Hernán MA. Fallibility in estimating direct effects. Int J Epidemiol. 2002;31(1):163–5. [DOI] [PubMed] [Google Scholar]
- 27.Eke PI, Dye B. Assessment of self-report measures for predicting population prevalence of periodontitis. J Periodontol. 2009;80(9):1371–9. [DOI] [PubMed] [Google Scholar]
- 28.Eke PI, Dye BA, Wei L, Slade GD, Thornton-Evans GO, Beck JD, Taylor GW, Borgnakke WS, Page RC, Genco RJ. Self-reported measures for surveillance of periodontitis. J Dent Res. 2013;92(11):1041–7. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
NHANES data are publicly available from the CDC/NCHS website. This study used pooled data from the 2011–2012, 2013–2014, 2015–2016, and 2017–March 2020 pre-pandemic NHANES cycles.