Oral Microbiome Profile of the US Population

Abstract

Importance: The oral microbiome likely plays key roles in human health. Yet, population-representative characterizations are lacking.

Objective: To characterize the composition, diversity, and correlates of the oral microbiome in US adults.

Design, setting, and participants: This cross-sectional study analyzed data from the population-representative National Health and Nutrition Examination Survey (NHANES) from 2009 to 2012. Microbiome data were made publicly available in 2024. NHANES participants were aged 18 to 69 years and provided oral rinse samples in 1 of 2 consecutive NHANES cycles (2009-2010 and 2011-2012).

Exposures: Demographic, socioeconomic, behavioral, anthropometric, metabolic, and clinical characteristics.

Main outcomes and measures: Oral microbiome measures, characterized through 16S ribosomal RNA gene sequencing, included α diversity (observed amplicon sequence variants [ASVs], Faith phylogenetic diversity, Shannon-Weiner Index, and Simpson Index); β diversity (unweighted UniFrac, weighted UniFrac, and Bray-Curtis dissimilarity); and prevalence and relative abundance at phylum level through genus level. Analyses accounted for the NHANES complex sample design.

Results: This study included 8237 US adults aged 18 to 69 years, representing 202 314 000 individuals (102 813 000 men [50.8%]; mean [SD] age, 42.3 [14.4] years; 9.3% self-reported as Mexican American, 12.1% as non-Hispanic Black, 64.7% as non-Hispanic White, 5.9% as other Hispanic, and 8.1% as other non-Hispanic individuals). The oral microbiome encompassed 37 bacterial phyla, 99 classes, 212 orders, 446 families, and 1219 genera. Five phyla (Firmicutes, Actinobacteria, Bacteroidetes, Proteobacteria, and Fusobacteria) and 6 genera (Veillonella, Streptococcus, Prevotella 7, Rothia, Actinomyces, and Gemella) were present in nearly all US adults (weighted prevalence, >99%). These genera were the most abundant, accounting for 65.7% of total abundance. Observed ASVs showed a quadratic pattern with age (peak at 30 years), were similar by sex, significantly lower among non-Hispanic White individuals, and increased with greater body mass index (BMI), alcohol use, and periodontal disease severity. All covariates together accounted for a modest proportion of oral microbiome variability as measured by β diversity: R2 = 8.7% (95% CI, 8.4%-9.1%) for unweighted UniFrac, R2 = 7.2% (95% CI, 6.6%-7.7%) for weighted UniFrac, and R2 = 6.3% (95% CI, 3.1%-6.7%) for Bray-Curtis matrices. By contrast, relative abundance of a few genera explained a high percentage of variability in β diversity for weighted UniFrac: Aggregatibacter (R2 = 22.4%; 95% CI, 22.1%-22.8%), Lactococcus (R2 = 21.6%; 95% CI, 20.9%-22.3%), and Haemophilus (R2 = 18.4%; 95% CI, 18.1%-18.8%). Prevalence and relative abundance of numerous genera were associated with age, race and ethnicity, smoking, BMI categories, alcohol use, and periodontal disease severity.

Conclusions and relevance: This cross-sectional study of the oral microbiome in US adults showed that a few genera were universally present and a different set of genera explained a high percentage of oral microbiome diversity across the population. This comprehensive characterization provides a contemporary reference standard for future studies.

Figure 1.. Prevalence of Phyla and Genera in the Oral Microbiome of Adults in the 2009 to 2012 National Health and Nutrition Examination Surveys

Circles represent weighted mean relative abundance. The inset in panel B shows the proportionate weighted mean relative abundance for genera with more than 1% mean relative abundance and for the remaining genera. Prevalence estimates of taxa at the phylum and genus levels are shown in eTables 2 and 6, respectively, in Supplement 2. NA indicates not applicable.

Figure 2.. Adjusted Predictive Margins for the Number of Observed Amplicon Sequence Variants (ASVs) by Age in the Oral Microbiome of Adults in the 2009 to 2012 National Health and Nutrition Examination Surveys

Age was modeled through 5-knot restricted cubic splines. Predictive margins (solid line) were estimated in survey design–adjusted linear regression models, with concomitant adjustment for sex, self-reported race and ethnicity, educational level, marital status, income to poverty ratio, body mass index, risk behaviors (smoking and alcohol use), medical conditions (diabetes and hypertension), oral health (periodontal disease, tooth count, and edentulism), and use of prescription medications within the past 30 days (antibiotics, antilipidemics, respiratory inhalants, and for gastroesophageal reflux disease [GERD]). eAppendix 2 in Supplement 1 provides variable definitions. The shaded area represents the 95% CIs.

Figure 3.. Overall and Covariate-Specific Proportion of Variability in β Diversity Matrices in the Oral Microbiome of Adults in the 2009 to 2012 National Health and Nutrition Examination Surveys

The estimates were based on models that incorporated concomitant adjustment for age (modeled as 5-knot restricted cubic splines), sex, self-reported race and ethnicity, educational level, marital status, income to poverty ratio, body mass index (BMI), risk behaviors (smoking and alcohol use), medical conditions (diabetes and hypertension), oral health (periodontal disease, tooth count, and edentulism), and use of prescription medications within the past 30 days (antibiotics, antilipidemics, respiratory inhalants, and for gastroesophageal reflux disease). eAppendix 2 in Supplement 1 provides variable definitions. The Fast Adonis algorithm was used for estimation. Error bars represent 95% CIs.

Figure 4.. Covariates Associated With the Number of Genera and Adjusted Odds Ratio by Age for Prevalence of Genera in Adults in the 2009 to 2012 National Health and Nutrition Examination Surveys

Associations with relative abundance were estimated in Poisson regression models. Associations with prevalence were estimated in binary logistic regression models. All models included concomitant adjustment for age (modeled as 5-knot restricted cubic splines), sex, self-reported race and ethnicity, educational level, marital status, income to poverty ratio, body mass index (BMI), risk behaviors (smoking and alcohol use), medical conditions (diabetes and hypertension), oral health (periodontal disease, tooth count, and edentulism), and use of prescription medications within the past 30 days (antibiotics, antilipidemics, and for gastroesophageal reflux disease). eAppendix 2 in Supplement 1 provides variable definitions.