With the notable exception of Axéll,⁴ who reported the results of a random sample of 20,333 drawn from approximately 250,000 residents older than the age of 15 years from Uppsala County, Sweden, the literature is devoid of credible prevalence data for the adult general population. Axéll⁴ found that among the most prevalent lesions were recurrent aphthae (two-year prevalence rate: 17.70 percent), recurrent herpes labialis (two-year prevalence rate: 17.38 percent), denture stomatitis (two-year prevalence rate: 16.02 percent) and geographic tongue (two-year prevalence rate: 8.57 percent).

With the notable exception of one article, the literature is devoid of credible prevalence data for the adult general population.

Because of the paucity of information about the prevalence of oral mucosal lesions, we report the results of the oral mucosal examination from the Third National Health and Nutrition Examination Survey, or NHANES III, a probability sample of civilian, noninstitutionalized people in the United States conducted from 1988 through 1994.⁵

	METHODS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Oral mucosal examinations were performed as part of NHANES III, a periodic survey conducted by the National Center for Health Statistics on the basis of a complex, multistage sample plan. The survey was designed to provide national estimates of the health and nutritional status of the civilian, noninstitutionalized U.S. population two months of age and older. From 19,528 randomly selected representative households, 33,994 subjects were interviewed by NHANES III staff members; 30,818 were examined in mobile examination centers and 493 were examined at home. Calibrated physicians and dentists performed all examinations and extensive health, social and nutritional medical histories were obtained via interviews with the subjects or their parents. A detailed discussion of the survey methods is presented in an article by Drury and colleagues.⁶

Dentist examiners were trained to recognize, classify and record in a standardized manner the clinical characteristics of each of the 48 conditions of interest, to include diagnosis, size, location, surface morphology, color consistency, pain, duration and history using procedures based on the World Health Organization’s Guide to Epidemiology and Diagnosis of Oral Mucosal Diseases and Conditions.⁷ While smears for yeast were taken on lesions suspected of being angular cheilitis, candidiasis, denture stomatitis, erythroplakias, leukoplakias, lichen planus and median rhomboid glossitis, all diagnoses were clinical.⁸

We analyzed point prevalence—the proportion of subjects who had a lesion—in the aggregate and by sex, age, race/ethnicity (non-Hispanic white, non-Hispanic black and Mexican-American), use of removable dentures and tobacco-use history. Tobacco-use history was determined during the health interview. We created six mutually exclusive categories of tobacco-use history:

– smokers who also use smokeless tobacco;

– mokeless-tobacco users;

– smokers;

– former smokers;

– former smokeless-tobacco users;

– those who never smoked more than 100 cigarettes, 20 cigars or 20 pipefuls or used smokeless tobacco or snuff in their lifetime.

We excluded from analyses using the race/ethnicity variable any subject who did not fall into one of the three race/ethnicity categories. This resulted in the removal of 702 subjects, categorized as "other," from the analysis.

Since the survey used complex, multistage sampling, we used a specialized software package (SAS-callable SUDAAN 8.0, Research Triangle Institute, Research Triangle Park, N.C.) to compute standard errors for all variables, adjusting for the effects of the survey design (design effect) as well providing the (weighted) population size to which the prevalence data can be projected. For example, the 17,245 subjects aged 17 years and older who received an oral mucosal examination are projected to represent 182,292,666 people in the U.S. population.

Results are presented in tables showing the actual number of lesions found, weighted count and standard error adjusted for the design effect, 95 percent confidence intervals, or CIs. In addition, odds ratios, or ORs, from bivariate and multivariate logistic regressions are presented with 95 percent CIs for the adjusted ORs. The multivariate model adjusts the ORs for the effect of the other variables in the model. If unity does not fall within the confidence limits of an OR, the OR is significantly different from unity at (at least) the .05 level.

	RESULTS

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Oral mucosal examinations were performed on 17,235 adults aged 17 years and older, of whom 4,801 (27.9 percent) had a total of 6,003 clinically detectable lesions. After adjusting for the design effect, we can project that 28.24 percent of the U.S. population has at least one clinically recognized lesion. Figure 1 shows the locations of the lesions. The hard palate had the most lesions (25.9 percent), followed by the gin-giva (20.4 percent), lip (15.4 percent) and dorsum of the tongue (14.2 percent). Figure 2 compares lesion prevalence by site for four categories of tobaccouse history. Smokeless-tobacco users had significantly higher prevalences of lesions on the lips, mucosa, tongue, commissures and vestibule than did nonusers. Cigarette smokers who used smokeless tobacco had a severalfold higher prevalence of lesions on the floor of the mouth than did users of either tobacco product alone and nonusers of tobacco. Smokers had a higher prevalence of palatal lesions than did nonsmokers.

View larger version (40K): [in this window] [in a new window]   Figure 1. Site distribution of mucosal lesions.

View larger version (33K): [in this window] [in a new window]   Figure 2. Prevalence of lesion site, by tobacco-use category.

Smokeless-tobacco use was the strongest risk factor, followed by use of removable dentures. The odds of having a lesion increased with age.

Table 2 (page 1284) shows the prevalence of mucosal lesions by age group, sex, race/ethnicity, tobacco use and use of removable dentures; ORs from bivariate and multivariate logistic regressions; and 95 percent CIs for the adjusted ORs. The odds of having a mucosal lesion increase steadily with age; people aged 70 years and older have almost twice the odds of having a lesion as do those aged from 17 through 29 years. Note that the adjusted ORs are appreciably lower than the bivariate ORs for age and race/ethnicity suggesting that a substantial amount of the bivariate effect of age and race/ethnicity was due to other variables in the multivariate model. Whites had higher odds of having a lesion (OR = 1.25) than did blacks (OR = 1.03) or Mexican-Americans (the reference level). Male subjects (OR = 1.28) had significantly greater odds of having a lesion than did female subjects. Smokeless-tobacco users had almost four times the odds of having a lesion as did those who never used tobacco (OR = 3.90), smokeless-tobacco users who also smoked had more than two times the odds of having a lesion (OR = 2.29), and smokers’ odds of having a lesion were 25 percent greater (OR = 1.25) than those of subjects who had never used tobacco. Removable-denture wearers had more than three times the odds of having a lesion (OR = 3.57) than did those who did not wear removable dentures.

	DISCUSSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

The 48 mucosal lesions reported were diverse with respect to etiology and morbidity. For lesions associated with specific risk factors such as nicotinic and denture stomatitis, the overall prevalence depends on the underlying distribution of the risk factor (that is, smoking and denture-wearing, respectively). Consequently, comparing population prevalences for such lesions without knowing the underlying distribution of these risk factors is not advisable.

Smokeless-tobacco use was the strongest risk factor, followed by use of removable dentures. The odds of having a lesion increased with age, even after we adjusted for tobacco use, the presence of removable dentures, race/ethnicity and sex. While tobacco use was a substantial risk factor, it likely would have been greater if the dose (number of cigarettes smoked and amount of smokeless tobacco used per day) were included in the model.

Since NHANES III used similar methods as used in the Swedish study,⁴ it is instructive to compare the results of the two studies. Table 3 (page 1285) compares selected prevalence estimates for different types of lesions (shown in Table 1) with those from the Swedish study.⁴ Axéll⁴ found higher prevalence rates of most common mucosal lesions than were found in NHANES III. Keeping in mind that both studies represent large probability samples from the general population and used World Health Organization diagnostic categories,⁷ it is surprising that the prevalences are so different to the point that for every lesion, the prevalence estimates from one study fall outside the 95 percent CI of the other study. Moreover, the standard errors from the Swedish study generally were greater than those from NHANES III, perhaps reflecting the greater racial/ethnic homogeneity of the Swedish population.

View this table: [in this window] [in a new window]   TABLE 3 COMPARISON OF SELECTED STATISTICAL VARIABLES REGARDING LESION TYPE IN NHANES III* AND IN THE SWEDISH STUDY.

While NHANES III is a robust, well-respected population-based data source that used trained examiners and a sophisticated sampling methodology, it nonetheless has limitations. One limitation is inherent in its design. Since it is cross-sectional, it can be used to explore only associations, not causal relationships. Second, the lesions were identified without benefit of laboratory or histologic tests. While trained examiners can be expected to identify many entities (such as aphthous ulcers, nicotinic stomatitis and recurrent herpes labialis) without additional diagnostic aids, others (such as nevus) may have other differential diagnoses and can be diagnosed clinically with less certainty. Furthermore, some lesions (tumors, nonspecific lesions and lesions of unknown etiology) could not be identified at all. Finally, the study may understate the prevalence of mucosal lesions that present as acute problems. For example, people may be more inclined to seek treatment for mucoceles, ranulas (even buccal mucosa fibromas that suffer biting trauma at every meal) and other acute conditions, so these are less likely to be seen in prevalence studies.

	CONCLUSION

TOP ABSTRACT METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
We found that the prevalence of many mucosal lesions depends greatly on individual characteristics. Because of this, prevalence rates for oral mucosal lesions cited in texts should be based on studies of general populations and be stratified by risk factors for the specific lesion. These may include sex, age, race/ethnicity, tobacco use and use of removable dentures. Also, in comparing prevalences from different studies, it is important to know whether the subjects selected for examination were selected using a probability sample of a representative population. Finally, even well-executed studies in one country may not produce valid estimators of prevalences in other countries. Future studies should focus on multivariate models to explore the risk factors for lesions that are of clinical significance.