Association Between Oral Health and Metabolic Dysfunction–Associated Steatotic Liver Disease Among US Adults

Abstract

INTRODUCTION:

Metabolic dysfunction–associated steatotic liver disease (MASLD) is a leading cause of chronic liver disease, yet extrametabolic contributors such as oral health remain underexplored. While periodontitis has been linked to nonalcoholic fatty liver disease, untreated caries and unmet dental care needs have been less examined under new MASLD criteria. We evaluated associations between examiner-assessed oral health indicators and MASLD in a nationally representative sample of US adults.

METHODS:

We analyzed 2,528 adults aged 18 years or older from the 2017–2020 National Health and Nutrition Examination Survey with valid liver transient elastography. MASLD was defined as steatosis (controlled attenuation parameter ≥285 dB/m) plus ≥1 metabolic risk factor. Examined oral health indicators included examiner-assessed need for dental care, decayed teeth, gum disease, and a composite of decayed teeth or gum disease. Survey-weighted logistic regression estimated odds ratios (ORs) adjusted for sociodemographic and behavioral factors.

RESULTS:

MASLD prevalence was 38.9%. In fully adjusted models, needing dental care (OR = 1.42, 95% CI: 1.02–1.95) and having decayed teeth (OR = 1.52, 95% CI: 1.05–2.20) were associated with higher MASLD odds. After false discovery rate correction, only dental care need remained significant (q = 0.043). Sex-stratified analyses revealed pronounced associations in women, who had 91% higher MASLD odds if dental care was needed (OR = 1.91, 95% CI: 1.18–3.10) and 156% higher odds with decayed teeth (OR = 2.56, 95% CI: 1.35–4.84). Significant associations were also observed in adults aged 45–59 and 60 years or older.

DISCUSSION:

Unmet dental needs and caries are associated with MASLD, with particularly strong associations observed in women. These findings highlight oral health as a potential marker for MASLD risk and underscore the value of integrated oral-systemic assessments in preventive care.

INTRODUCTION

Metabolic dysfunction–associated steatotic liver disease (MASLD), previously known as non-alcoholic fatty liver disease (NAFLD), is a leading cause of chronic liver disease worldwide and affects nearly one-third of the adult population (1,2). Its rising prevalence has been driven by the worldwide increase in metabolic conditions such as obesity, insulin resistance, and dyslipidemia (3). By 2040, the burden of MASLD is projected to increase by more than 40% and further support the need to identify modifiable risk factors beyond those traditionally considered metabolic (4). Given this rising burden, clarifying nonmetabolic, potentially modifiable correlates that refine risk stratification is a timely public health priority. In particular, markers that are easy to observe in routine clinical care and that reflect both biological and social vulnerability could be especially valuable for case-finding and prevention.

The recent transition in nomenclature from NAFLD to MASLD reflects an effort to center metabolic dysfunction as a unifying driver of disease. Under the updated consensus, MASLD is defined by the presence of hepatic steatosis together with at least one cardiometabolic risk factor, while individuals with steatosis and higher levels of alcohol intake are now reclassified as having MetALD or alcoholic liver disease rather than MASLD (5). Studies have shown that there is a high degree of diagnostic concordance between the 2 terms, particularly in real-world clinical populations, indicating that prior findings on NAFLD remain applicable under the MASLD framework (5). In practice, the MASLD criteria refine the clinical context in which steatosis is interpreted rather than replacing the underlying disease biology. At the same time, the new definition emphasizes the need to reassess non-metabolic contributors and risk markers within a framework that explicitly includes cardiometabolic risk.

Beyond metabolic factors, oral health is increasingly recognized as a potential contributor to MASLD pathogenesis (6–9). The oral cavity hosts a dense microbiome and may fuel systemic inflammation through the oral–gut–liver axis (10). Periodontitis, in particular, has been linked to cardiovascular disease, diabetes, and fatty liver disease (8,11). Proposed mechanisms include gut barrier disruption, endotoxemia, and inflammatory cascades that promote insulin resistance and hepatic steatosis (12,13). However, oral health encompasses more than periodontitis. Dental caries and unmet dental care needs represent distinct aspects of oral disease burden and healthcare access, which may also influence systemic inflammation and MASLD risk but remain understudied.

Experimental studies show that oral administration of Porphyromonas gingivalis can induce hepatic steatosis and fibrosis through endotoxemia and gut dysbiosis (14–16). Apart from the microbiome, periodontal infections may exacerbate vascular injury by promoting endothelial dysfunction which is a hallmark of cardiometabolic disorders that frequently co-occur with MASLD (17). Studies have shown that periodontal treatment can improve glycemic control and reduce cardiovascular risk, particularly among individuals with diabetes or prediabetes (18–20). Population-based studies suggest that genetic markers of inflammation may modify the association between periodontal disease and liver steatosis and highlight a potential gene–environment interaction (21). These findings support the biological plausibility that oral disease may serve both as a marker of systemic inflammation and as a modifiable contributor to MASLD pathogenesis. However, much of this experimental and genetic evidence has been generated in models or cohorts defined using pre-MASLD definition, underscoring the need to test whether these pathways translate to the new MASLD definition. In addition, most work has focused on periodontal disease, leaving uncertainty about the relative importance of untreated caries and unmet dental care needs, which are highly prevalent and clinically visible but less often studied.

Despite growing biological plausibility, few epidemiologic studies have evaluated oral health and MASLD under the updated diagnostic criteria. Given shared risk factors such as poor diet, low socioeconomic status, and chronic inflammation, oral health may serve not only as a surrogate marker but also as a modifiable risk factor for MASLD (12–14). Moreover, previous research has focused primarily on periodontal disease, while indicators such as dental caries remain underexplored despite their shared inflammatory basis. Examiner-recommended dental care is even less studied, yet it may capture a broader spectrum of disease severity and unmet need than periodontal probing alone. Prior National Health and Nutrition Examination Survey (NHANES) studies largely examined NAFLD or earlier criteria. Our study is the first to apply the MASLD consensus definition using transient elastography. Most population-based studies of oral health and fatty liver disease have relied on biochemical indices rather than transient elastography and MASLD criteria, so it is uncertain whether their findings fully generalize to contemporary MASLD definitions. Furthermore, it is unclear whether any associations between oral health and MASLD are uniform across patient subgroups or are concentrated among specific populations, such as women, younger or older adults, or individuals without overt metabolic syndrome. To our knowledge, no nationally representative analysis has simultaneously evaluated dental caries, periodontal signs, and examiner-recommended dental care in relation to this new updated MASLD definition.

We therefore analyzed nationally representative data from the 2017–2020 NHANES to investigate whether clinically assessed oral health indicators are associated with MASLD. We hypothesized that poorer oral health, particularly untreated caries and unmet dental needs, would be independently associated with higher odds of MASLD after adjustment for sociodemographic and behavioral factors, and we also explored whether these associations varied by sex, age, and metabolic syndrome status.

METHODS

Study population and data source

We used data from the 2017–2020 NHANES, an ongoing, nationally representative survey of the civilian, noninstitutionalized US population conducted by the National Center for Health Statistics (22). We restricted the analysis to adults aged 18 years or older who attended the mobile examination center (MEC) and underwent liver ultrasound transient elastography (FibroScan). We excluded participants who tested positive for hepatitis B or C markers (hepatitis B surface antigen or core antibody, hepatitis C antibody or RNA), those with self-reported physician diagnosis of autoimmune hepatitis, and those with excessive alcohol intake (≥1 alcoholic drinks per day for women or ≥2 for men). Pregnant women were also excluded. We further excluded anyone missing the controlled attenuation parameter (CAP) and participants with liver stiffness measurement (LSM) suggesting cirrhosis (LSM ≥13.6 kPa) (23). Among adults with valid elastography, additional exclusions for missing oral health or covariate data yielded a final analytic sample of 2,528 participants. A flow diagram summarizing these steps, including the number excluded for LSM ≥13.6 kPa and missing data, is presented in Figure 1.

Figure 1.

Flow diagram of analytic sample selection for metabolic dysfunction–associated steatotic liver disease analysis, National Health and Nutrition Examination Survey 2017–2020.

All analyses applied MEC examination weights, strata, and primary sampling units to account for the complex survey design, oversampling, and nonresponse, thereby providing estimates generalizable to the US adult population. NHANES protocols were approved by the National Center for Health Statistics Institutional Review Board, and all participants provided written informed consent. Detailed NHANES sampling and examination procedures are described elsewhere (22).

Oral health exposures

Oral health exposures were derived from the standardized NHANES oral health examination and included 3 clinically assessed indicators: examiner-recommended dental care, presence of untreated dental caries, and presence of gum disease. These were supplemented by a composite measure indicating the presence of either caries or gum disease. These variables represent clinically assessed indicators of oral disease burden.

Examiner-recommended dental care was assessed at the conclusion of the oral exam, when trained dental examiners classified participants into 1 of 4 categories: (i) see a dentist immediately, (ii) see a dentist within 2 weeks, (iii) see a dentist at their earliest convenience, or (iv) continue routine care. For this analysis, we dichotomized the variable into “needs dental care” (categories 1–3) vs “continue routine care” (category 4). NHANES dental examiners complete centralized training, calibration exercises, and ongoing quality assurance based on standardized Oral Health Examiners Manuals to promote consistent application of these criteria. However, this item has not been formally validated as a graded severity scale and reflects a combination of objective clinical findings and examiner judgment.

Presence of decayed teeth was defined as ≥1 untreated coronal carious lesion based on visual-tactile examination. This dichotomization was chosen to reflect clinically relevant caries requiring restorative care and to enable comparison with prior NHANES studies. Presence of gum disease was based on examiner-identified periodontal signs. Participants were classified as having gum disease or no gum disease based on this assessment. We acknowledge that NHANES uses a partial-mouth periodontal protocol, which has lower sensitivity compared with full-mouth exams. This likely underestimate true prevalence and may bias estimates toward the null. A composite variable was created to indicate whether participants had either decayed teeth or gum disease (yes/no).

MASLD outcome

The primary outcome was MASLD, defined according to recent consensus criteria (5). Hepatic steatosis was identified using FibroScan-CAP, with steatosis defined as CAP ≥285 dB/m, a threshold supported by validation studies in NHANES and other cohorts (24).

Participants were classified as having MASLD if they met both of the following: (i) hepatic steatosis (CAP ≥285 dB/m) and (ii) at least 1 cardiometabolic risk factor. Metabolic risk factors included (i) overweight/obesity (Body Mass Index [BMI] ≥25 kg/m² [≥23 for Asians]) or elevated waist circumference (>94 cm [men]/>80 cm [women]); (ii) dysglycemia (fasting glucose ≥100 mg/dL, HbA1c ≥5.7%) or current use of diabetes medication (insulin or oral hypoglycemic agents); (iii) hypertension (systolic ≥130 mm Hg, diastolic ≥85 mm Hg, or current use of antihypertensive medication); (iv) hypertriglyceridemia (fasting triglycerides ≥ 150 mg/dL) or current use of lipid-lowering medication (e.g., fibrates or other triglyceride-lowering agents); or (v) low high density lipoprotein (HDL) cholesterol (<40 mg/dL [men]/<50 mg/dL [women]).

For exploratory stratified analyses, metabolic syndrome was defined using standard criteria as the presence of ≥3 of the following: elevated waist circumference, elevated triglycerides or triglyceride-lowering medication use, reduced HDL cholesterol, elevated blood pressure or antihypertensive medication use, and dysglycemia (elevated fasting glucose or diabetes medication).

Covariates

Covariates were selected based on their established or biologically plausible associations with oral health and MASLD.

Demographic variables included age (categorized as 18–44, 45–59, or 60 years or older), sex (male or female), and self-reported race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Hispanic, Asian/Other/Mixed).

Socioeconomic indicators included educational attainment (less than high school, high school graduate/GED, some college, or college graduate) and family income-to-poverty ratio, categorized as <1.0 (poor), 1.0–2.99 (low), or ≥3.0 (mid/high income). Marital status (married/living with partner, never married, widowed/divorced/separated) and insurance status were summarized descriptively but were not included in primary regression models because of collinearity with income-to-poverty ratio.

Health behavior covariates included smoking and physical activity. Smoking exposure was primarily captured using log-transformed serum cotinine as an objective biomarker of recent tobacco exposure. Self-reported smoking status (never, former, current) and pack-years ([average cigarettes per day ÷ 20] × years smoked) were used in descriptive analyses but not included as additional adjustment variables to avoid redundancy with cotinine. Physical activity was summarized as total weekly metabolic equivalent (MET) minutes based on self-reported leisure-time, transportation, and occupational activity and treated as a continuous variable in regression models. Physical activities categories (low, moderate, high) were used descriptively based on World Health Organization (WHO) thresholds. Dietary covariates considered included total energy intake (kcal/d), total carbohydrate, total sugars, total fat, saturated fat, and fiber (all in g/d), derived from the 24-hour dietary recall.

Anthropometric and metabolic biomarkers, measured during clinical examinations, included BMI (categorized as <25, 25–29.9, or ≥30 kg/m² and calculated from standardized height and weight measurements) and systolic and diastolic blood pressure (averaged across 2 seated measurements and treated as continuous variables in mm Hg). Laboratory biomarkers include glycohemoglobin (%), fasting glucose (mg/dL), HDL, low density lipoprotein (LDL), total cholesterol, and triglycerides (all in mg/dL), along with high-sensitivity C-reactive protein (hs-CRP, mg/L). These parameters were in the descriptives to characterize cardiometabolic risk but were not included as adjustment covariates in primary models because they contribute directly to the MASLD definition and could introduce overadjustment. High-sensitivity C-reactive protein (hs-CRP) was measured in serum and was included only in the fully adjusted sensitivity model, given its potential role as an intermediate on the causal pathway between oral inflammation and MASLD.

Statistical analysis

We accounted for the NHANES complex sampling design to produce nationally representative estimates. All analyses used NHANES MEC exam weights, strata, and primary sampling units. We generated survey-weighted descriptive statistics for demographic, socioeconomic, behavioral, metabolic, and oral health characteristics overall and by MASLD status. Categorical variables were summarized as weighted percentages and continuous variables as weighted means with SDs. Group differences by MASLD status were evaluated using Rao–Scott adjusted χ² tests for categorical variables and survey-weighted linear regression for continuous variables.

Survey-weighted logistic regression models were used to examine the associations between 4 binary oral health exposures: dental care recommendation (need vs continue routine care), presence of decayed teeth, presence of gum disease, and a composite indicator of decayed teeth or gum disease and MASLD. Three sequential models were constructed: Model 1 was unadjusted; Model 2 is the main model and adjusted for age, sex, race/ethnicity, educational attainment, income-to-poverty ratio, log-transformed serum cotinine, total weekly MET minutes of physical activity, and total dietary carbohydrate intake; Model 3 further adjusted for high-sensitivity C-reactive protein (hs-CRP). Model 2 was prespecified as the primary model. Model 3 was treated as a conservative sensitivity analysis because hs-CRP may lie on the causal pathway linking oral inflammation and MASLD, and adjusting for it could attenuate true associations. For each exposure, odds ratios (ORs) and 95% CIs were reported. To address multiple comparisons across the 4 correlated oral health exposures in Model 3, we applied false discovery rate (FDR) correction using the Benjamini–Hochberg procedure and report both raw P-values and FDR q-values.

Only total sugar intake was retained out of the dietary factors in model 2 because of multicollinearity among dietary variables (variance inflation factor (VIF) > 5). Covariates with VIF > 5 were excluded from final models. Variance inflation factors for oral health exposures were <2, suggesting no problematic collinearity. The composite indicator was retained as a summary burden measure but interpreted with caution. We tested multiplicative interactions by sex, age group, and race/ethnicity, smoking; education, and income and they were not statistically significant. Diabetic mellitus (DM) was not tested in the interaction model to avoid potential collider bias because DM is a definitional component of MASLD. Nonetheless, we performed stratified analyses by age (younger than 45, 45–59, 60 or older) and sex to explore potential subgroup heterogeneity, using the same fully adjusted covariates. We also conducted stratified analyses by metabolic syndrome status as an exploratory analysis to evaluate potential collider bias. These subgroup results were considered hypothesis-generating.

Missing covariate data were addressed through complete-case analysis. We conducted multiple imputations using Multivariate Imputation by Chained Equations (MICE) package in R (20 datasets) and confirmed that imputed results were consistent with complete-case estimates (see Supplementary Table 1, Supplementary Digital Content, http://links.lww.com/CTG/B465). Overall missingness for covariates was <8% which was consistent with missing at random. For sensitivity analysis, we repeated analyses after excluding participants with borderline steatosis (CAP 270–300 dB/m). The results remained consistent across all models (see Supplementary Table 2, Supplementary Digital Content, http://links.lww.com/CTG/B465).

All statistical tests were 2-sided with a significant threshold of P < 0.05. Analyses were performed in R (version 4.2) using the survey package for complex design modeling and ggplot2 for visualization. Our study followed the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines for cross-sectional studies.

RESULTS

Study population characteristics

A total of 2,528 adults from the 2017–2020 NHANES cycles met inclusion criteria for this analysis, of whom 984 (38.9%) met criteria for MASLD (Table 1). Compared with participants without MASLD, those with MASLD were older (mean age 57.1 vs 53.0 years, P = 0.00025) and more likely to be married or living with a partner (68.3% vs 61.3%, P = 0.046). The distribution of sex, educational status, income–poverty ratio, smoking status, serum cotinine levels, and WHO physical activity were similar between groups (P > 0.05), although participants with MASLD had slightly lower total MET-minutes per week on average (5,221.7 vs 6,058.5, P = 0.079). Adults with MASLD had worse cardiometabolic profile. They had higher mean BMI (33.4 vs 26.7 kg/m², P < 0.00001), systolic and diastolic blood pressure, glycohemoglobin, fasting glucose, triglycerides, and hs-CRP, along with lower HDL cholesterol (P ≤ 0.01). LDL cholesterol and total cholesterol did not differ significantly by MASLD status. Advanced fibrosis was substantially more frequent in the MASLD group than in those without MASLD (28.6% vs 7.1%, P < 0.00001).

Table 1.

Descriptive characteristics of the study population stratified by MASLD status

Variable	Total (N = 2,528)a	Non-MASLD (N = 1,544)a	MASLD (N = 984)a	Test statistic	P-value
Age, mean (SD)	54.64 (18.27)	52.98 (19.24)	57.11 (16.43)	4.3	0.00025
Age group, n (%)b				6.71	0.00266
18–44	690 (28.8)	482 (32.9)	208 (22.7)
45–59	601 (25.4)	345 (24.5)	256 (26.7)
Older than 60	1,237 (45.8)	717 (42.6)	520 (50.6)
Sex, n (%)b				3.78	0.06309
Male	1,452 (60.6)	844 (58.3)	608 (64.1)
Female	1,076 (39.4)	700 (41.7)	376 (35.9)
Race/ethnicity, n (%)b				5.89	0.00389
Asian/other/mixed race	411 (10.5)	265 (11.6)	146 (9.0)
Hispanic	541 (15.4)	303 (15.0)	238 (16.2)
Non-Hispanic Black	664 (12.0)	462 (14.4)	202 (8.5)
Non-Hispanic White	912 (62.0)	514 (59.0)	398 (66.4)
Educational status, n (%)b				0.73	0.45815
Less than high school	585 (16.4)	351 (17.2)	234 (15.1)
Completed high school	604 (29.7)	366 (28.5)	238 (31.4)
Some college/college graduate	1,202 (54.0)	721 (54.3)	481 (53.5)
Marital status, n (%)b				3.54	0.04561
Married/living with partner	1,426 (64.2)	826 (61.3)	600 (68.3)
Never married	351 (14.8)	240 (17.0)	111 (11.6)
Widowed/divorced/separated	616 (21.0)	372 (21.7)	244 (20.1)
Income to poverty ratio, n (%)b				0.46	0.58648
Low income (1.0–2.99)	969 (39.9)	588 (40.2)	381 (39.6)
Mid/high-income (≥3.0)	716 (45.7)	431 (46.4)	285 (44.8)
Poor (<1.0)	435 (14.3)	257 (13.4)	178 (15.6)
Income–poverty ratio, mean (SD)	2.89 (1.59)	2.92 (1.60)	2.84 (1.58)	−0.74	0.46874
Smoking status, n (%)b				0.83	0.44368
Current smoker	315 (12.1)	209 (12.6)	106 (11.2)
Former smoker	621 (27.1)	335 (25.3)	286 (29.8)
Never smoker	1,592 (60.8)	1,000 (62.1)	592 (59.0)
Pack-year, mean (SD)	10.01 (42.75)	8.97 (51.26)	11.56 (25.13)	1.59	0.12462
Serum cotinine, mean (SD)	45.33 (122.05)	48.74 (124.41)	40.36 (118.44)	−0.82	0.41873
WHO activity level, n (%)b				2.4	0.10149
Low (<600 total met min/wk)	325 (16.9)	190 (14.9)	135 (20.1)
Moderate (600–2,999 total met min/wk)	655 (38.3)	413 (38.8)	242 (37.4)
High (≥3,000 total met min/wk)	772 (44.8)	480 (46.3)	292 (42.5)
Total MET-min/week, mean (SD)	5,736.99 (7,510.71)	6,058.54 (7,708.54)	5,221.70 (7,157.88)	−1.84	0.07879
Recommendation for dental care, n (%)b				2.43	0.12
Continue routine care	1,229 (63.4)	765 (66.1)	464 (59.4)
Needs dental care	888 (36.6)	523 (33.9)	365 (40.6)
Presence of decayed teeth, n (%)b				6.12	0.02049
No	1,234 (66.1)	767 (69.0)	467 (62.0)
Yes	777 (33.9)	457 (31.0)	320 (38.0)
Presence of gum disease, n (%)b				0.66	0.42375
No	1,236 (84.2)	771 (85.1)	465 (82.7)
Yes	314 (15.8)	190 (14.9)	124 (17.3)
Fibrosis stage, n (%)b				90.02	<0.00001
F0–F1: no/mild	2,129 (84.2)	1,404 (92.9)	725 (71.4)
≥F2: significant/advanced fibrosis	399 (15.8)	140 (7.1)	259 (28.6)
BMI, mean (SD)	29.36 (6.64)	26.67 (5.17)	33.37 (6.58)	15.44	<0.00001
Systolic blood pressure, mean (SD)	125.39 (18.67)	124.28 (19.49)	127.01 (17.30)	2.9	0.00787
Diastolic blood pressure, mean (SD)	73.85 (11.25)	72.84 (11.25)	75.32 (11.10)	3.85	0.00078
Glycohemoglobin (%), mean (SD)	5.89 (1.08)	5.64 (0.77)	6.27 (1.34)	13.72	<0.00001
Fasting glucose (mg/dL), mean (SD)	116.46 (40.02)	106.33 (26.58)	132.33 (50.95)	6.34	<0.00001
HDL (mg/dL), mean (SD)	50.08 (13.55)	53.65 (13.67)	44.90 (11.56)	−11.52	<0.00001
LDL (mg/dL), mean (SD)	111.13 (36.72)	111.95 (37.26)	109.84 (35.85)	−0.81	0.42477
Total cholesterol (mg/dL), mean (SD)	183.98 (41.33)	184.94 (41.62)	182.57 (40.89)	−0.93	0.35992
Triglyceride (mg/dL), mean (SD)	125.28 (124.69)	105.15 (123.00)	156.07 (121.04)	5.33	0.00002
Hs-CRP (mg/L), mean (SD)	4.42 (10.64)	3.68 (11.32)	5.51 (9.47)	2.85	0.00887
Energy (kcal), mean (SD)	2,101.59 (926.66)	2058.82 (911.37)	2,162.91 (945.28)	1.89	0.07158
Total protein (gm), mean (SD)	79.64 (40.88)	77.22 (40.07)	83.10 (41.79)	2.16	0.04068
Total carbohydrate (gm), mean (SD)	249.34 (117.42)	247.05 (115.85)	252.63 (119.61)	0.79	0.43632
Total sugars (gm), mean (SD)	108.78 (72.53)	107.40 (70.62)	110.75 (75.19)	0.74	0.46494
Total fiber (gm), mean (SD)	16.85 (10.44)	16.86 (10.76)	16.85 (9.96)	−0.01	0.98916
Total fat (gm), mean (SD)	87.80 (48.56)	84.75 (46.49)	92.18 (51.10)	2.43	0.02309
Vitamin D (D2 + D3) (mcg), mean (SD)	4.73 (6.74)	4.60 (6.58)	4.92 (6.95)	0.72	0.4786
Calcium (mg), mean (SD)	934.90 (545.42)	920.55 (542.88)	955.49 (548.69)	0.98	0.33461
Iron (mg), mean (SD)	14.41 (8.80)	13.98 (8.04)	15.03 (9.75)	1.71	0.09924

Test statistic values are from survey-weighted linear regression for continuous variables and Rao–Scott design-adjusted χ² tests for categorical variables, accounting for the NHANES complex sampling design.

BMI, body mass index; CRP, C-reactive protein; Hs-CRP, high-sensitivity C-reactive protein; MASLD, metabolic dysfunction–associated steatotic liver disease; MET, metabolic equivalent; NHANES, National Health and Nutrition Examination Survey.

^aUnweighted sample size; weighted percentages shown in parentheses.

^bValues are weighted row percentages within each MASLD category.

Dietary energy, total carbohydrate, total sugars, total fiber, vitamin D, calcium, and iron intake were similar between groups (P > 0.05), although participants with MASLD had modestly higher total protein (83.1 vs 77.2 g/d, P = 0.041) and total fat intake (92.2 vs 84.8 g/d, P = 0.023). With respect to oral health indicators, participants with MASLD were more likely to be classified as needing dental care by the examiner, but this difference did not reach statistical significance (P = 0.12). The presence of decayed teeth was significantly more common in the MASLD group than in those without MASLD (38.0% vs 31.0%, P = 0.020). Self-reported gum disease was also more frequent among participants with MASLD (17.3% vs 14.9%), although this difference was not statistically significant (P = 0.424).

Associations between oral health and MASLD

In unadjusted models (Model 1), needing dental care, presence of decayed teeth, and the composite oral disease indicator were each associated with increased odds of MASLD, while gum disease was not (Table 2). Participants who needed dental care had 30% higher odds of MASLD compared with those advised to continue routine care (OR 1.30, 95% CI: 1.03–1.65). The presence of decayed teeth was associated with 36% higher odds of MASLD (OR 1.36, 95% CI: 1.07–1.74), and the composite indicator of decayed teeth or gum disease was associated with 34% higher odds (OR 1.34, 95% CI: 1.06–1.69). Gum disease alone was not significantly associated with MASLD (OR 1.19, 95% CI: 0.78–1.81).

Table 2.

Multivariable logistic regression models examining the association between oral health indicators and MASLD, NHANES 2017–2020

Exposures	Model 1: unadjusted OR (95% CI)	Model 2: Adjusted OR (95% CI)a	Model 3: Adjusted OR (95% CI)b	P-value (model 2)	Bonferroni P-value (model 2)	FDR q-value (BH)
Dental care recommendation
Continue routine care	Ref	Ref	Ref
Need dental care	1.30 (1.03–1.65)	1.45 (1.04–2.02)	1.42 (1.02–1.95)	0.038	0.128	0.043
Presence of decayed teeth
No	Ref	Ref	Ref
Yes	1.36 (1.07–1.74)	1.50 (1.04–2.15)	1.52 (1.05–2.20)	0.027	0.108	0.063
Presence of gum disease
No	Ref	Ref	Ref
Yes	1.19 (0.78–1.81)	0.99 (0.54–1.81)	0.97 (0.51–1.83)	0.926	3.704	0.926
Presence of decayed teeth or gum disease
No	Ref	Ref	Ref
Yes	1.34 (1.06–1.69)	1.46 (1.04–2.06)	1.48 (1.04–2.10)	0.03	0.12	0.058

FDR correction applied across the 4 oral health exposures in Model 2. Only dental care need remained significant after correction. Bold indicates statistically significant predictors (p < 0.05).

FDR, false discovery rate; MASLD, metabolic dysfunction–associated steatotic liver disease; MET, metabolic equivalent; NHANES, National Health and Nutrition Examination Survey; OR, odds ratio.

^aModel 2 adjusted for age, sex, race/ethnicity, education, income–poverty ratio, log-transformed serum cotinine, total MET minutes per week, and total dietary carbohydrate intake.

^bModel 3 additionally adjusted for high-sensitivity C-reactive protein.

In the main model (model 2), which adjusts for age, sex, race/ethnicity, education, income–poverty ratio, smoking exposure, physical activity, and dietary carbohydrate intake (Table 2) and serves as our primary model, needing dental care (OR 1.45, 95% CI: 1.04–2.02), decayed teeth (OR 1.50, 95% CI: 1.04–2.15), and the composite indicator (OR 1.46, 95% CI: 1.04–2.06) remained associated with higher odds of MASLD, whereas gum disease continued to show no significant association (OR 0.99, 95% CI: 0.54–1.81). In Model 3, which additionally included high-sensitivity C-reactive protein and is interpreted as conservative, needing dental care was associated with 42% higher odds of MASLD (OR 1.42, 95% CI: 1.03–1.95), decayed teeth with 52% higher odds (OR 1.52, 95% CI: 1.05–2.20), and the composite indicator with 48% higher odds (OR 1.48, 95% CI: 1.04–2.10). Gum disease remained not significantly associated with MASLD (OR 0.97, 95% CI: 0.51–1.83).

After controlling for multiple comparisons using FDR correction, only dental care need remained statistically significant (q = 0.043), while associations for decayed teeth (q = 0.063) and the composite indicator (q = 0.058) did not meet the corrected threshold (Table 2). Overall, the estimates from Models 2 and 3 suggest modest but consistent positive associations between poorer oral health and MASLD, with some expected loss of statistical significance after further adjustment and FDR correction.

Stratified analyses by age, sex, and metabolic syndrome status

Sex-stratified analyses (Table 3) revealed that associations between oral health indicators and MASLD were observed exclusively among women. In the main adjusted model (Model 2), women needing dental care had 89% higher odds of MASLD (OR: 1.89; 95% CI: 1.18–3.03; P = 0.018), and women with decayed teeth had 154% higher odds (OR: 2.54; 95% CI: 1.34–4.80; P = 0.012). In the sensitivity analysis adjusting additionally for CRP (Model 3), these associations persisted with similar magnitudes. No significant associations were observed in men in either model.

Table 3.

Stratified analysis of association between oral health indicators and MASLD by sex

Sex	Exposures	Model	OR (95% CI)	P-value
Male (n = 1,452)	Dental care recommendation (needs care vs routine care)	Model 1	1.04 (0.83–1.31)	0.722
		Model 2	1.00 (0.69–1.45)	0.99
		Model 3	0.96 (0.66–1.40)	0.841
	Decayed teeth (yes vs no)	Model 1	1.14 (0.79–1.64)	0.489
		Model 2	1.15 (0.67–1.98)	0.617
		Model 3	1.09 (0.62–1.92)	0.77
	Gum disease (yes vs no)	Model 1	0.97 (0.62–1.50)	0.879
		Model 2	0.67 (0.35–1.28)	0.242
		Model 3	0.64 (0.33–1.27)	0.227
Female (n = 1,076)	Dental care recommendation (needs care vs routine care)	Model 1	1.51 (1.07–2.13)	0.028a
		Model 2	1.89 (1.18–3.03)	0.018a
		Model 3	1.91 (1.18–3.10)	0.020a
	Decayed teeth (yes vs no)	Model 1	1.82 (1.12–2.96)	0.023a
		Model 2	2.54 (1.34–4.80)	0.012a
		Model 3	2.56 (1.35–4.84)	0.012a
	Gum disease (yes vs no)	Model 1	1.69 (0.81–3.51)	0.173
		Model 2	2.23 (1.02–4.91)	0.064
		Model 3	2.26 (1.00–5.09)	0.07

Model 1: Unadjusted.

Model 2 adjusted for age, sex, race/ethnicity, education, income–poverty ratio, log-transformed serum cotinine, total MET minutes per week, and total dietary carbohydrate intake.

Model 3 additionally adjusted for high-sensitivity C-reactive protein. OR (95% CI): odds ratio with 95% CI.

MASLD, metabolic dysfunction–associated steatotic liver disease; MET, metabolic equivalent; OR, odds ratio.

^aP < 0.05.

Age-stratified analyses (Table 4) showed differing patterns across the lifespan. Among younger adults (younger than 45 years), initial unadjusted associations (Model 1: dental needs OR: 1.51, P = 0.012; composite indicator OR: 1.54, P = 0.029) attenuated substantially and were not significant in either the main model (Model 2) or the model with additional CRP adjustment (Model 3). In middle-aged adults (45–59 years), the need for dental care was significantly associated with MASLD in the fully adjusted model (Model 3 OR: 1.52; 95% CI: 1.02–2.26; P = 0.038), while decayed teeth showed a borderline association (Model 3 OR: 1.51; 95% CI: 0.97–2.39; P = 0.069). Among older adults (60 years or older), significant associations in Model 3 were observed for dental needs (OR: 1.39; 95% CI: 1.02–1.98; P = 0.037), the composite indicator of decayed teeth or gum disease (OR: 1.48; 95% CI: 1.02–2.15; P = 0.040) and decayed teeth alone (OR: 1.52; 95% CI: 1.01–2.29; P = 0.045).

Table 4.

Age-stratified associations between oral health exposures and MASLD

Age group (yr)	Exposures	Model	OR (95% CI)	P-value
<45	Dental needs (yes vs no)	Model 1	1.51 (1.12–2.04)	0.012a
		Model 2	1.29 (0.93–1.79)	0.147
		Model 3	1.28 (0.85–1.93)	0.235
	Presence of decayed teeth or gum disease (yes vs no)	Model 1	1.54 (1.06–2.17)	0.029a
		Model 2	1.22 (0.83–1.79)	0.334
		Model 3	1.25 (0.78–2.00)	0.358
	Decayed teeth (yes vs no)	Model 1	1.49 (0.99–2.25)	0.071
		Model 2	1.26 (0.81–1.95)	0.325
		Model 3	1.30 (0.75–2.25)	0.354
	Gum disease (yes vs no)	Model 1	1.66 (1.14–2.41)	0.014a
		Model 2	0.86 (0.51–1.45)	0.589
		Model 3	0.92 (0.48–1.75)	0.801
45–59	Dental needs (yes vs no)	Model 1	1.42 (1.05–1.92)	0.025a
		Model 2	1.35 (0.96–1.90)	0.089
		Model 3	1.52 (1.02–2.26)	0.038a
	Presence of decayed teeth or gum disease (yes vs no)	Model 1	1.40 (0.95–2.06)	0.092
		Model 2	1.38 (0.92–2.07)	0.124
		Model 3	1.48 (0.94–2.33)	0.091
	Decayed teeth (yes vs no)	Model 1	1.45 (0.98–2.15)	0.064
		Model 2	1.42 (0.94–2.14)	0.097
		Model 3	1.51 (0.97–2.39)	0.069
	Gum disease (yes vs no)	Model 1	1.14 (0.70–1.86)	0.599
		Model 2	1.08 (0.65–1.80)	0.766
		Model 3	1.12 (0.64–1.96)	0.689
60+	Dental needs (yes vs no)	Model 1	1.18 (0.92–1.53)	0.21
		Model 2	1.32 (1.00–1.75)	0.069
		Model 3	1.39 (1.02–1.98)	0.037a
	Presence of decayed teeth or gum disease (yes vs no)	Model 1	1.19 (0.84–1.67)	0.34
		Model 2	1.35 (0.95–1.92)	0.098
		Model 3	1.48 (1.02–2.15)	0.040a
	Decayed teeth (yes vs no)	Model 1	1.19 (0.82–1.74)	0.367
		Model 2	1.32 (0.90–1.94)	0.158
		Model 3	1.52 (1.01–2.29)	0.045a
	Gum disease (yes vs no)	Model 1	1.09 (0.70–1.69)	0.712
		Model 2	1.10 (0.70–1.73)	0.684
		Model 3	1.18 (0.74–1.88)	0.491

Model 1: Unadjusted.

Model 2: Adjusted for age (continuous), sex, race/ethnicity, educational attainment, income-to-poverty ratio, log-transformed serum cotinine, total metabolic equivalent minutes per week, and total dietary carbohydrate intake.

Model 3: Model 2 + high-sensitivity C-reactive protein.

MASLD, metabolic dysfunction–associated steatotic liver disease; OR, odds ratio.

^aStatistically significant at α = 0.05.

Stratification by metabolic syndrome status (Table 5) revealed no statistically significant positive associations for oral health indicators in the main model (Model 2) in either group. The sensitivity analysis (Model 3) showed a borderline inverse association between decayed teeth and MASLD in participants without metabolic syndrome (OR: 0.38; 95% CI: 0.14–0.99; P = 0.070).

Table 5.

Stratified analysis of the association between oral health indicators and MASLD by metabolic syndrome status

Metabolic syndrome status	Exposures	Model	OR (95% CI)	P-value
No metabolic syndrome (n = 418)	Dental care recommendation (needs care vs routine care)	Model 1	1.16 (0.66–2.06)	0.61
		Model 2	0.64 (0.24–1.69)	0.384
		Model 3	0.59 (0.25–1.40)	0.254
	Decayed teeth (yes vs no)	Model 1	1.42 (0.53–3.76)	0.491
		Model 2	0.41 (0.13–1.26)	0.143
		Model 3	0.38 (0.14–0.99)	0.07
	Gum disease (yes vs no)	Model 1	1.16 (0.43–3.18)	0.769
		Model 2	0.81 (0.11–5.79)	0.837
		Model 3	0.88 (0.13–6.06)	0.901
Metabolic syndrome (n = 1,050)	Dental care recommendation (needs care vs routine care)	Model 1	1.16 (0.93–1.44)	0.201
		Model 2	1.26 (0.76–2.07)	0.381
		Model 3	1.29 (0.79–2.10)	0.336
	Decayed teeth (yes vs no)	Model 1	1.38 (1.00–1.89)	0.062
		Model 2	1.54 (0.87–2.73)	0.158
		Model 3	1.56 (0.88–2.78)	0.152
	Gum disease (yes vs no)	Model 1	1.01 (0.59–1.73)	0.962
		Model 2	0.88 (0.36–2.18)	0.788
		Model 3	0.95 (0.40–2.22)	0.901

Model 1: Unadjusted.

Model 2: Adjusted for age (continuous), sex, race/ethnicity, educational attainment, income-to-poverty ratio, log-transformed serum cotinine, total metabolic equivalent minutes per week, and total dietary carbohydrate intake.

Model 3: Model 2 + high-sensitivity C-reactive protein. MetS defined as ≥3 of (i) waist circumference >102/88 cm (M/F), (ii) triglycerides ≥150 mg/dL or medication, (iii) HDL <40/50 mg/dL (M/F), (iv) Hypertension ≥130/85 mm Hg or medication, (v) fasting glucose ≥100 mg/dL, HbA1c ≥5.7%, or diabetes medication.

MASLD, metabolic dysfunction–associated steatotic liver disease; MetS, metabolic syndrome; OR, odds ratio.

Sensitivity analysis

Sensitivity analyses confirmed the robustness of primary findings. Multiple imputation for missing covariates (n = 20 datasets) yielded results consistent with complete-case analysis: dental care need (OR 1.25, 95% CI: 1.09–1.43), decayed teeth (OR 1.21, 95% CI: 1.01–1.46), and composite oral disease (OR 1.22, 95% CI: 1.03–1.45) remained significantly associated with MASLD in fully adjusted models (see Supplementary Table 1, Supplementary Digital Content, http://links.lww.com/CTG/B465). These imputed estimates were similar in magnitude to those from Model 3 in the complete-case analysis and did not change the pattern of associations. Analyses excluding borderline steatosis (CAP 270–300 dB/m) showed attenuated but directionally consistent associations for dental care need (OR 1.22, 95% CI: 0.92–1.61) and composite indicator (OR 1.37, 95% CI: 0.98–1.89) in fully adjusted models. When using a lower CAP threshold (≥248 dB/m), no significant associations were observed (all P >0.12), although effect estimates remained positive (see Supplementary Table 2, Supplementary Digital Content, http://links.lww.com/CTG/B465). Overall, sensitivity analyses support that the observed associations are modest, somewhat sensitive to modeling choices and steatosis thresholds, but generally consistent in direction across specifications.

DISCUSSION

This study investigated associations between oral health indicators and metabolic dysfunction–associated steatotic liver disease (MASLD) under updated diagnostic criteria. Three main observations emerged. First, examiner-recommended dental care was consistently associated with higher odds of MASLD after adjustment for sociodemographic and behavioral factors, and this association remained statistically significant after FDR correction. Untreated dental caries and a composite indicator of caries or gum disease were also positively associated with MASLD in fully adjusted models, but did not meet the FDR-adjusted significance threshold. Second, exploratory stratified analyses suggested that these relationships were more evident among women and among middle-aged and older adults, whereas associations in younger adults were attenuated after adjustment. Third, metabolic syndrome status did not show a clear pattern of effect modification. We did not observe strong evidence that oral health–MASLD associations were confined to individuals without metabolic syndrome, and an isolated inverse estimate for decayed teeth in the non–metabolic-syndrome group was imprecise and likely reflects chance or collider-related bias rather than a protective effect. Overall, effect sizes were modest, and confidence intervals were sometimes wide, so these findings should be viewed as hypothesis-generating and interpreted with appropriate caution.

The demographic profile of our MASLD group which comprised mostly of older age, a higher proportion of men, and a greater proportion of participants who were married or living with a partner, was broadly consistent with prior NHANES-based estimates of MASLD burden in US adults, which have shown higher prevalence in middle-aged and older adults and in men (2). At the same time, oral disease burden and unmet dental care needs are known to cluster among individuals with lower income, lower educational attainment, and certain racial and ethnic minority groups (7). This overlap in social patterning provides an important backdrop for our findings. Oral health may act as a visible marker of the same structural and behavioral disadvantages that drive MASLD risk, while also contributing biologically through chronic inflammation and microbial translocation.

Our findings agree with a growing body of literature that implicate periodontal disease in the pathogenesis or progression of NAFLD. Prior studies have demonstrated a consistent relationship between poor periodontal status and liver steatosis across diverse populations. An analysis of NHANES data found that adults with moderate-to-severe periodontitis or fewer than 20 teeth had higher odds of NAFLD after adjusting for metabolic risk factors (7). Similarly, a Japanese study reported that individuals with periodontal probing depths ≥4 mm had nearly twice the odds of NAFLD (8) while a large Chinese male cohort linked the loss of more than 6 teeth with increased NAFLD risk (9). Furthermore, periodontal disease has also been linked to more advanced liver outcomes. A recent NHANES-based study by Moon et al (2021) found that individuals with mild/moderate and severe periodontitis had significantly higher odds of elevated liver fibrosis based on fibrosis-4 scores, compared with those without periodontitis suggesting that periodontitis may contribute not only to steatosis but also to fibrosis (25). Our findings suggest that caries-related indicators which are often overlooked in systemic disease research may also carry independent associations with hepatic steatosis. Given that both dental caries and periodontitis share microbial and inflammatory mechanisms, this broader oral disease burden may better reflect the systemic inflammatory milieu relevant to MASLD pathogenesis (26). Unlike other studies, however, our study applies the MASLD consensus definition with transient elastography, providing updated and clinically relevant estimates while evaluating caries, periodontal signs, and examiner-recommended care side-by-side.

Our findings are supported by large-scale studies across diverse populations. For example, a nationwide cohort of over 130,000 Israeli adults, Ram et al (2022) found that periodontitis was independently associated with NAFLD after adjustment for metabolic risk factors (27). A meta-analysis by Chen et al reported 19% increased odds of NAFLD in individuals with periodontal disease, with even stronger associations observed for cirrhosis and significant tooth loss (28). However, a meta-analysis by Xu and Tang reported null or inconsistent associations, likely because of heterogeneity in diagnostic criteria or study design, including reliance on self-reported oral health or variable methods for diagnosing NAFLD (29). Despite the limitations, multiple lines of evidence suggest that periodontal disease may actively contribute to liver injury. For example, a Finnish population-based cohort found higher liver-related mortality among individuals with both NAFLD and periodontitis compared with those with NAFLD alone (30). However, not all population studies have replicated these associations. Interestingly, we did not see this association in a Latino population, but it may be because NAFLD was so highly prevalent that the presence of periodontal disease was moot (31). One Mendelian randomization study did not also find a causal genetic link (32). Together, these heterogeneous findings underscore why triangulation across designs (observational, genetic, interventional) and standardized definitions is important for inference. They also provide a context in which modest, imprecise associations such as those observed in our study should be interpreted cautiously but not dismissed outright.

The biological plausibility of the oral–liver axis has been supported by experimental studies showing that oral pathogens such as P. gingivalis induce gut barrier dysfunction, endotoxemia, and hepatic inflammation through toll-like receptor activation (14–16). There is also evidence that specific inflammatory mediators like tumor necrosis factor-α, IL-6, prostaglandin E2 known to impair insulin signaling and hepatic lipid metabolism which are key processes in MASLD development (12,13). One study has shown that other microbes such as Aggregatibacter actinomycetemcomitans may similarly disrupt metabolic regulation through immune modulation and gut–liver axis perturbation (33). These findings suggest that poor oral health may act not only as a marker of metabolic risk but as an active contributor to hepatic injury. In this context, our signal for examiner-recommended dental care and untreated caries is consistent with a proinflammatory burden that may amplify early steatogenic pathways. The oral–gut–liver axis therefore seems to be the most direct and well-characterized mechanism, with systemic inflammatory and endothelial pathways better viewed as overlapping extensions of this core route rather than independent, competing explanations.

Stratified analyses revealed stronger associations among female patients, with significant findings for all oral health exposures. Although multiplicative interactions were not statistically significant, we included these subgroup analyses to explore possible effect heterogeneity. Estrogen's dual role in modulating both gingival inflammation (34) and hepatic lipid metabolism provides a plausible mechanism. Experimental studies show estrogen enhances hepatic de novo lipogenesis while simultaneously increasing gingival vascular permeability to oral pathogens (12). Furthermore, women demonstrate amplified Th1-mediated immune responses to oral biofilms (34), potentially exacerbating systemic inflammation from focal infections. Behavioral factors including dietary patterns and healthcare utilization disparities (35) may further contribute to this sex-specific risk profile. To support this, mice models demonstrate that oral pathogens such as P. gingivalis induce insulin resistance and hepatic steatosis in metabolically healthy animals through Farnesoid X receptor signaling disruption (14,15).

We also stratified analyses by metabolic syndrome (MetS) status to address collider bias due to MASLD metabolic criteria. However, in contrast to our initial expectation, associations between oral health indicators and MASLD were not clearly confined to participants without MetS, and estimates were generally imprecise in both strata. A borderline inverse association for decayed teeth among those without MetS is likely explained by sampling variability or conditioning on metabolic traits rather than a biologically protective effect. These sex-, age-, and MetS-stratified results were not primary hypotheses, are based on smaller strata, and may still be influenced by residual collider bias and confounding. They should therefore be interpreted as exploratory and hypothesis-generating rather than definitive evidence of effect modification (36,37).

Our findings have several clinical and public-health implications. Examiner-recommended dental care is simple to ascertain in both dental and medical settings and does not require specialized scoring or imaging. If confirmed in longitudinal studies, such a marker could contribute to risk stratification in primary care by prompting closer evaluation for MASLD among adults with evident unmet dental needs or untreated caries, particularly women and older adults. Conversely, recognition of MASLD in hepatology or primary care clinics might serve as a cue to inquire about dental care access and facilitate referral to dental services. More broadly, our results support ongoing efforts to integrate oral health into chronic disease prevention strategies. Interventions that improve oral hygiene, reduce sugar intake, address tobacco use, and expand access to dental care may deliver benefits that extend beyond the mouth, aligning with WHO noncommunicable disease priorities and Healthy People 2030 objectives for both liver and oral health (36,38,39). Randomized trials of periodontal or caries-directed therapy with liver fat and inflammatory endpoints in MASLD would be a particularly informative next step.

One strength of our study is the use of a large, population-representative dataset with standardized clinical measures. Unlike prior studies relying on self-report, we used examiner-assessed oral health indicators, which enhances validity. Use of CAP to determine hepatic steatosis and therefore better classify MASLD compared with prior NHANES years where no VCTE measures are available and multivariable adjustment for behavioral and metabolic confounders improves internal validity.

However, several limitations should be acknowledged. First, the cross-sectional design prevents assessment of temporal or causal relationships. It is therefore possible that MASLD itself may predispose to poorer oral health and we cannot disentangle whether oral disease primarily precedes, coincides with, or follows hepatic steatosis. Second, gum disease was measured using a partial-mouth protocol, which underestimates true periodontal disease prevalence and may bias results toward the null. The absence of adjunctive radiographs further reduces sensitivity for interproximal and apical disease, and our dichotomous classification does not capture the extent or severity of periodontal involvement. This measurement limitation may partly explain the gum-disease null in the main analyses. Third, without liver biopsy, we were unable to distinguish between simple steatosis and more advanced stages such as steatohepatitis or fibrosis beyond what CAP and LSM measurements provide. Despite adjustment for total sugar intake, residual confounding by overall dietary patterns is possible, future work should incorporate dietary pattern scores or dimension-reduction methods. Fourth, some biomarkers, particularly hs-CRP, may lie on the causal pathway between oral inflammation and MASLD. We therefore treated models that include hs-CRP as conservative sensitivity analyses, and it is possible that true etiologic associations are underestimated. Finally, residual confounding is likely, particularly from unmeasured variables such as chronic stress, health-seeking behaviors, and access to care. Insurance status was excluded because of collinearity with income-to-poverty ratio, which performed better as a socioeconomic predictor in our models. Despite these limitations, our findings add to a growing body of evidence linking oral and liver health and highlight the need for more integrated preventive approaches across disciplines. Our choice to treat hs-CRP as part of a sensitivity model reflects its likely position on the pathway between oral inflammation and MASLD rather than as a pure confounder. This approach is conservative but may underestimate associations.

In conclusion, we found that examiner-recommended dental care and untreated dental caries were associated with MASLD in a nationally representative sample of US adults. Associations appeared more pronounced among women and older adults, although subgroup findings were exploratory and should be confirmed in other cohorts. The magnitude of these associations was modest, but given the high prevalence of both MASLD and poor oral health, even small relative effects may have meaningful implications for population level risk stratification and prevention efforts. Our findings suggest that clinical dental indicators could be informative in metabolic liver disease risk profiling. They also point to opportunities for integrating dental and medical care in the prevention and management of chronic metabolic conditions. From a public health standpoint, promoting oral health in at-risk populations aligns with Healthy People 2030 objectives and WHO goals to reduce the burden of chronic disease. Longitudinal and interventional studies are needed to clarify directionality, quantify causal effects, and determine whether improving oral health can favorably influence core MASLD outcomes.

CONFLICTS OF INTEREST

Guarantor of the article: Chukwuemeka E. Ogbu, MD had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Specific author contributions: C.E.O.: conceptualization, methodology, formal analysis, data curation, writing—original draft, writing—review & editing. A.G., A.G., and J.D.: software, formal analysis, writing—original draft, writing—review and editing. C.O. and M.E.: writing—review and editing. C.U.: writing—original draft, writing—review and editing. A.S.B.: validation, writing—review and editing, supervision. All authors have read and approved the final manuscript.

Financial support: None to report.

Potential competing interests: None to report.

Study Highlights.

WHAT IS KNOWN

• ✓ Metabolic dysfunction–associated steatotic liver disease (MASLD) is a major global cause of chronic liver disease.

• ✓ Periodontitis has been linked to NAFLD in prior studies.

WHAT IS NEW HERE

• ✓ First study to assess oral health and MASLD using new diagnostic criteria.

• ✓ Unmet dental care need is a significant correlate for MASLD.

• ✓ Women with dental care needs had 90% higher odds of MASLD.

• ✓ Significant associations were identified in adults aged 45 years and older.

Supplementary Material

ct9-17-e00975-s001.pdf ^{(156.4KB, pdf)}

ABBREVIATIONS:

%

percent

adjusted p-value q

q-value

BMI

body mass index

CAP

controlled attenuation parameter

CI

confidence interval

cm

centimeters

CRP

C-reactive protein

dB/m

decibels per meter

DM

diabetes mellitus

DOI

digital object identifier

FDR

false discovery rate

FDR

false discovery rate

FXR

Farnesoid X receptor

g/day

grams per day

HbA1c

hemoglobin A1c

HCV

hepatitis C virus

HDL

high-density lipoprotein

hs-CRP

high-sensitivity C-reactive protein

IL-6

interleukin 6

IRB

Institutional Review Board

kcal/day

kilocalories per day

kPa

kilopascal(s)

LDL

low-density lipoprotein

LSM

liver stiffness measurement

MAR

missing at random

MASLD

metabolic dysfunction–associated steatotic liver disease

MEC

mobile examination center

MET

metabolic equivalent (task)

MetALD

metabolic dysfunction–associated alcohol-related liver disease

MetS

metabolic syndrome

mg/dL

milligrams per deciliter

mg/L

milligrams per liter

MICE

multiple imputation by chained equations

mmHg

millimeters of mercury

NAFLD

nonalcoholic fatty liver disease

NHANES

National Health and Nutrition Examination Survey

OR

odds ratio

P. gingivalis

porphyromonas gingivalis

RNA

ribonucleic acid

SD

standard deviation

STROBE

strengthening the reporting of observational studies in epidemiology

Th1

T helper type 1

TNF-α

tumor necrosis factor alpha

U.S.

United States

VIF

variance inflation factor

WHO

World Health Organization