|Home | About | Journals | Submit | Contact Us | Français|
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
How smokeless tobacco contributes to non-neoplastic oral diseases is unclear. It certainly increases risk of oral mucosal lesions, but reviewers disagree as to other conditions. In some areas, especially South-East Asia, risk is difficult to quantify due to the many products, compositions (including non-tobacco ingredients), and usage practices involved. This review considers studies from Europe (in practice mainly Scandinavia) and from the USA.
Experimental and epidemiological studies published in 1963–2007 were identified that related risk of oral lesions to smokeless tobacco use. Data were assessed separately for oral mucosal lesions, periodontal and gingival diseases, dental caries and tooth loss, and oral pain.
Oral mucosal lesions: Thirty-three epidemiological studies consistently show a strong dose-related effect of current snuff on oral mucosal lesion prevalence. In Scandinavia, users have a near 100% prevalence of a characteristic "snuff-induced lesion", but prevalence of the varied lesions reported in the USA is lower. Associations with chewing tobacco are weaker. The lack of clear association with former use suggests reversibility following cessation, consistent with experimental studies showing rapid lesion regression on quitting.
Periodontal and gingival diseases: Two of four studies report a significant association of snuff with attachment loss and four out of eight with gingival recession. Snuff is not clearly related to gingivitis or periodontal diseases. Limited evidence suggests chewing tobacco is unrelated to periodontal or gingival diseases.
Tooth loss: Swedish studies show no association with snuff, but one US study reported an association with snuff, and another with chewing tobacco.
Dental caries: Evidence from nine studies suggests a possible relationship with use of smokeless tobacco, particularly chewing tobacco, and the risk of dental caries.
Oral pain: Limited evidence precludes any clear conclusion.
This review confirms the strong association of current use of smokeless tobacco, particularly snuff, with prevalence of oral mucosal lesions. It provides suggestive evidence of an association of snuff use with gingival recession and attachment loss, and of chewing tobacco with dental caries. While smokeless tobacco clearly increases risk of oral mucosal lesions, interpretation for other endpoints is limited by study weaknesses, including poor confounding control.
This review concerns epidemiological studies that relate smokeless tobacco [ST] use to non-neoplastic oral diseases and is based primarily on 50 publications, 20 in Scandinavia [1-20], 29 in the USA [21-49] and one in England . In these countries, usage is mainly oral; nasal use of finely ground "dry snuff" has become rare . ST products are traditionally classified as snuff or chewing tobacco (CT) . In the USA, CT is typically mixed with saliva and then placed in the buccal pouch, while finely cut "moist snuff" is frequently held in the lower labial fold. In Scandinavia, snuff (or snus in Sweden) is generally placed under the upper or lower lip. As in some other reviews of ST effects [53-57] results from areas (such as India, South Asia, Africa and Saudi Arabia) where the tobacco chewed is often mixed with other substances, such as betel quid and areca nut are not considered . Nor is oral cancer considered, as its relationship with ST has been discussed elsewhere .
An earlier review of non-cancerous and precancerous oral health effects associated with ST use was published by the US Surgeon General in 1986 . This report concluded that "smokeless tobacco is responsible for the development of a portion of oral leukoplakias in both teenage and adult users" and that "dose response effects have been noted by a number of investigators". However, it also noted that "studies of the effects of smokeless tobacco use on gingival and periodontal tissues have resulted in equivocal findings" and that "negative health effects on the teeth from smokeless tobacco use are suspected but unconfirmed". Since that time there have been many reviews of the evidence relating ST use to oral changes [54,55,59-71].
While these reviews are consistent in concluding that ST use is associated with an increased risk of oral mucosal lesions, they are far less consistent with regard to other endpoints. Though various reviews [54,60,61,67] comment on a possible relationship with gingival recession, they disagree on whether an association is clearly demonstrated or not. Similarly, views vary over whether associations with periodontal diseases and dental caries have not been demonstrated , are possibly related, based on limited evidence , or have been shown to exist [54,67]. A number of these reviews [63,64,68,70,71] note that "ST-related" oral mucosal lesions reverse on quitting and go on to discuss the extent to which the presence of the lesion modifies the risk of oral cancer. A recent review by the US Surgeon General  notes that leukoplakia is a classic precursor lesion of oral cancer, but notes that "the leukoplakia that occurs in cigarette smokers differs morphologically from the keratoses caused by smokeless tobacco; although less common, the leukoplakia induced by cigarettes is more susceptible to malignant transformations," citing a review by Bouquot . The importance of distinguishing the oral mucosal lesions induced by ST from oral leukoplakias has also been emphasised by other reviewers [74,75].
Our objective, therefore, was to try to clarify the overall picture by conducting a detailed review of the available evidence.
Results were identified from a systematic search of MEDLINE, EMBASE, and SCOPUS through November 2007. The search was not limited by period or language. The main searches were based on combinations of the terms "smokeless tobacco", "chewing tobacco" and "snuff" for exposure, and "oral diseases", "mouth diseases", "periodontal diseases", "leukoplakia", "mucosal lesions", "gingivitis", "gingival recession", "teeth problems", and "dental caries" for outcome. Further articles were identified by inspection of reference lists in individual papers and reviews.
All reports had to satisfy the following Inclusion criteria: published in a peer reviewed journal or publicly available; based on research in humans; of experimental, cohort, case-control design or cross-sectional design; study location specified; conducted in North America or Europe; non-neoplastic form of oral disease (ICD 9 520–529) as the outcome; and CT, orally used moist snuff, or unspecified ST as the exposure. They also had to satisfy the Exclusion criteria: oral cancer, sample included in a more complete or recent study; inappropriate design (case report, qualitative study or review); and appropriate prevalence estimates or odds ratios (ORs) not reported and cannot be computed from the available data.
From each report, details were abstracted relating to the study (design, period, region, population, sample selection, and size), the exposure (method of assessment, type of ST investigated, exposure doses and durations considered), the outcome (type of oral disease and method of diagnosis), and issues relating to analysis (type of effect measure, analysis methods, stratification variables, and adjustment factors).
Data were considered separately under four disease groups: oral mucosal lesions, periodontal and gingival diseases, dental caries and tooth loss, and oral pain. Additionally, a summary description of each study was prepared, sorted by study design.
In some studies, the required estimates of prevalence, or of effect size and precision, were not given by the authors and had to be estimated from data presented. Crude ORs and 95% confidence intervals (CIs) were derived from the relevant 2 × 2 table using standard methods, with, where necessary, numbers estimated from proportions given numerically.
50 principal studies were identified, 20 of snuff, conducted in Scandinavia (two in Denmark, one in Finland and 17 in Sweden), 29 of CT, snuff or overall ST, conducted in the USA, and one of CT, conducted in England. No relevant studies were identified from other parts of North America or Europe. Table Table11 (Europe) and Table Table22 (USA) give some details of the 50 studies considered. Both tables are sorted by type of study and then by year of publication within type. Six are experimental studies investigating the short-term effect on oral mucosal lesions of quitting ST use, switching to a different brand of snuff, or changing the site of placement of the snuff. Of the 44 epidemiological studies, three are prospective studies and two are case-control studies, though one  is a case-control study of oral and oropharyngeal cancer, with only the control group providing data relevant to this report. There are 39 cross-sectional studies, 26 of populations unselected by ST use and 13 of populations selected by ST use and/or presence of oral lesions. Some of the 50 studies are not independent. Thus, two studies [1,19] involve a subset of subjects selected from those in a third , while one study  gives combined results from a number of groups of baseball players, the results for two of these being previously reported [31,34]. Also, three studies [40,42,43] give results for different endpoints using differently selected samples from the third National Health and Nutrition Examination Survey (NHANES III).
The earliest publication  was in 1963, with five studies published in the 1970s, 13 in the 1980s, 18 in the 1990s and 13 since 2000. Many of the studies did not give details on when they were conducted. Of the 50 studies, 24 were of males and females, and 24 of males. Two studies [2,3] did not give details on the sex of their subjects. Most of the studies were mainly or wholly of adults, but there were 10 cross-sectional studies of children, mainly of school students, but including one of adolescent football players aged 11 to 18  and one of teenagers attending dental health check-up . Some of the studies of adults involved quite young populations, including air force trainees , military recruits , college athletes , 19 year olds  and baseball players [31,34-36,38,44]. Oral mucosal lesions were considered in 40 of the studies, with 19 considering indices of periodontal or gingival diseases, 10 dental caries, and three oral pain.
Fifteen of the studies in Scandinavia provided information relating ST use to oral mucosal lesions. As shown in Table Table3,3, 11 of these used the endpoint "snuff-dipper's lesion" as defined by Axéll et al  and a further three used an endpoint which appeared to be similar. For convenience we will refer to these 14 studies as being of snuff-induced lesions (SIL). Only in one study  was a clearly different endpoint used, of "oral leukoplakia", which could occur both in users and non-users of snuff.
Table Table44 summarizes the main results from the Scandinavian studies relating oral mucosal lesions to the use of moist snuff (snus). The single study of oral leukoplakia , in Denmark, had no control group so could not compare lesion prevalence in snuff users and non-users, though it reported that only 32 of the 450 oral leukoplakia cases studied used snuff and that the presence of symptoms (such as pain or rough feel to the tongue) was much less prevalent in those who used snuff (3.1%) than in those who did not (23.1%, p < 0.01).
The remaining 14 studies generally reported a 100% prevalence of SIL in snuff users. Partly this was because in seven studies [1,3,4,13,15-17] subjects had been, or appeared to have been, selected on the basis of SIL being present, but this was not always the case [2,5,18,19]. The only exceptions were a study in Finland  in which all the users with no lesion had either quit or started snuff in the previous three weeks, and a recent study of ice-hockey players in Sweden  which reported no lesions in five users of portion-bag snuff. One study in Sweden  reported total numbers of lesions, but prevalence seems likely to have been very high in snuff users.
As is evident from the results of the cross-sectional and case-control studies summarized in Table Table4,4, severity of SIL was clearly associated with the length of time per day snuff was used for and with the amount of snuff consumed per day [5,15,16,18,20], though the statistical significance of these relationships could not always be determined. The relationship of severity with duration of use was less marked [5,16,18]. Severity was also lower in users of portion-bag snuff than in users of loose snuff [19,20].
A number of the experimental and epidemiological studies investigated the effect of changing snuff habits on the presence and severity of SIL. The data in Table Table44 are all consistent with quitting in the short-term reducing the severity of the lesion  and in the longer term eliminating the lesion [1,4,7], and switching to lower nicotine, lower pH or portion-bag snuff also reducing severity [1-4].
While it is clear from the above that SIL occurs in virtually all current users of moist snuff, with severity related to extent of use, the data in Table Table44 tell us little about the long-term health implications of the lesion. The study by Roosaar et al  provides some additional information. Here 1,115 individuals with SIL were followed-up for 27–29 years, with a total of three incident cases of oral cancer seen, two in concomitant daily smokers. Though this was somewhat higher than expected (standardized incidence rate 2.3, 95% CI 0.5–6.7), none of the three occurred at the site of the original SIL.
Of the 15 Scandinavian studies considered, 14 consider use of moist snuff only and not other types of ST. The study of Andersson et al , however, considered nine users of CT as well as 45 snuff users. They noted that seven of the CT users showed leukoedema in the buccal mucosa, and that six who placed the tobacco on a permanent site had degree 1 or 2 (Axéll) lesions at the site of placement.
The single study in England , of coal miners, reported that 10 out of 280 CT users, 3.6%, had leukoplakia, as compared to none out of 122 non-chewers (p = 0.05). Here leukoplakia was defined as "a white patch that could not be rubbed off and which could not be diagnosed as being due to any other disease".
Twenty-four US studies provided evidence relating ST use to oral mucosal lesions. As shown in Table Table5,5, the definitions of the endpoint used varied widely. Eight studies [22,27,28,30,33,36,44,48] used a definition modified by Greer and Poulson  from that used by Axéll , and at least three studies [34,35,38,39] used a similar definition. However, in some studies the endpoints were clearly not comparable, including one  where the endpoint was any type of oral lesion and one  where the definition specifically excluded alterations that resolved rapidly on discontinuation of ST use. Unlike the studies in Scandinavia, the conditions were usually referred to as oral leukoplakia or as oral lesions, with no specific reference to ST or snuff in the name of the condition, as in SIL. An exception was the Tomar et al study  which referred to the endpoint as "smokeless tobacco lesions". Indeed it is clear from Table Table6,6, which presents the prevalence of these lesions in relation to ST use, that the endpoint definitions did not exclude the possibility of the lesion being present in non-users of ST. In fact, in all but one of the 15 studies which presented findings separately for users and non-users, including the Tomar et al study, prevalence is greater than zero in the non-users.
Exposure is classified in Table Table66 by type (ST, snuff or CT) and by time (current, former, ever or unspecified use which may be equivalent to current). Fourteen of the US studies provided evidence on snuff use. Ignoring one study where the subjects were selected as having a lesion , lesion prevalence in current snuff users was typically in the range 30–70%, though lower in two studies [45,47] where the definition implied a more severe lesion, and in one study  where the endpoint of "soft tissue lesion" was not further defined. Though lower than the near 100% prevalence reported in Scandinavia, it was clearly much higher than in non-users, with ORs ranging from 8.2 to 97.1 in six studies [23,26,34-36,39]. In contrast the evidence of any increased prevalence in former snuff users is much weaker, with one case-control study  reporting no increase, one cross-sectional study  an increase of borderline significance, another cross-sectional study  an increase, but based on very few subjects, and one experimental study  reporting resolution of the great majority of lesions in US Air Force trainees after 6 weeks of a mandatory tobacco-free period.
Nine of the US studies provided evidence on use of CT, and in all of these lesion prevalence was lower in current users of CT than in current users of snuff. Indeed in the five studies where RRs and CIs could be estimated, risk was not elevated (OR 0.97) in one study , was elevated but not significantly (ORs 3.39 and 3.16) in two studies [23,36] and was only significantly elevated (ORs 15.0 and 2.5) in two [34,39]. Oral mucosal lesion prevalence was not significantly associated with former CT use [26,39].
Nineteen of the studies provided evidence on overall ST use; in ten of these data by specific type of ST were not available. Given the variation that could have existed in the type of ST used in the different studies (and in the endpoints considered), it is not surprising that there is also marked variation in the prevalence of oral mucosal lesions reported in the different studies. However, it is abundantly clear that there was a marked effect of use on lesion prevalence. Thus, all of the studies for which ORs (with CIs) could be estimated for current or ever ST use showed a significant increase, with ORs over 25 in three studies [21,23,34,48] and the lowest OR, 3.90, being in a study  where the endpoint was wide ranging and prevalence relatively high in non-users. Again, the evidence of an association was much weaker for former than for current users. Of the six studies reporting relevant data, five [26,34,36,38,42] found no relationship, with a significant increase in former ST users reported only in one .
Aspects of the dose-response relationship have been examined in many of the US studies. In one detailed analysis , odds ratios compared to non-users of ST increased significantly with increasing amount of ST used, hours in the month, and recency of last use, with estimates of 354 (129–970) for users of four or more cans of snuff a week, 34.7 (9.4–128) for users of four or more pouches of CT a week, 361 (107–1215) for more than four hours ST use in the day and 201 (84-9-475) for used ST in the last hour. These associations were all found to be independently significant in a multifactorial analysis, but weaker associations with age of initiation and duration of ST use were not independent risk factors. The findings were unaffected by adjustment for age, race, education, cigarette smoking, alcohol consumption, and dental hygiene.
Though not usually analyzed in such detail, similar findings have been reported by other authors. In users of ST or of snuff, prevalence of oral lesions has consistently been found to increase with increasing amount used and/or frequency of use [23,27,28,30,32,35,39,48] and with recency of use [23,35]. These studies also found that prevalence increased with duration of use, though they generally did not test whether the effect of duration remained after adjustment for extent of current usage. Relationships of severity of oral lesions to extent and duration of usage have also been found [23,48], though not always .
The epidemiological data from Europe and the USA are consistent with demonstrating a striking and dose-related effect of current snuff use on the prevalence of oral mucosal lesions. In Scandinavia, snuff users have a near 100% prevalence of SIL, but the prevalence of the more varied lesions studied in the USA is typically lower. The association between current CT use and lesion prevalence, mainly studied in the USA, is clearly very much weaker. The lack of any very clear association with former use is consistent with reversibility of the lesion following cessation, and fits in with findings from the experimental studies, which typically show regression of the lesion on quitting after a relatively short period.
A variety of indices relevant to periodontal and gingival diseases are considered here, as illustrated in the endpoint column for Tables Tables77 and and8,8, where the results relating to ST use are given. Table Table77 summarizes findings from seven studies in Sweden, no other Scandinavian studies reporting results, while Table Table88 gives results from 13 studies in the USA. Changing and inconsistent diagnostic criteria, as well as a lack of standardized use of methods and terminology [76-81] complicate the summarizing of evidence in this area. The endpoints in Tables Tables77 and and88 are as named by the authors of the original publications cited, and are defined more precisely there. The results are sorted so that similar endpoints are classified together, in order to facilitate the interpretation of the data. Some of the endpoints considered, such as plaque and calculus, do not of themselves indicate disease, but are indirectly related and the results are shown for additional information.
The results from Sweden show no significant relationship, in any study, between snuff use and the presence of plaque or calculus, pocket depth, attachment loss, alveolar bone level, bone height or periodontal disease (defined as three or more teeth with pocket depth ≥ 5 mm). One study  reported that snuff users had a significantly (p < 0.001) increased gingival index, but other studies [10-12,20] showed no relationship with gingivitis, gingival index or gingival bleeding. Gingival recession was considered by five Swedish studies. Two of these [17,20] provided no data for non-users of snuff, but of the others, one  reported the highest rates of recession, but no significant increase in relation to ever snuff use, one  reported a significant increase (OR 3.72, 95% CI 1.40–9.99), and one  reported a significantly (p < 0.05) higher prevalence in users of loose snuff than portion-bag snuff.
The two US studies to provide some data relating to use of CT [31,34] found no evidence of a relationship with plaque, gingival bleeding or attachment loss, though one of these  reported a non-significant increase for gingival recession. However, some of the studies found a relationship between snuff or unspecified ST with some indices of periodontal or gingival diseases. While no significant increases were seen for plaque, calculus, gingivitis, gingival bleeding or pocket depth, the data, though somewhat inconsistent, suggested a relationship with gingival recession. Here, six studies provide relevant data, three in children or adolescents [29,30,32] and three in baseball players [31,34,38]. The two which reported very low rates of gingival recession [30,32] found no association, but four [29,31,34,38] did, significant except in the case of one small study . One of the studies  reported high rates of recession and a strong relationship (OR 9.15, 95% CI 5.40–15.5), even stronger in patients with gingivitis. Two studies [24,34] also reported a significant association of snuff or ST with attachment loss, though another study  reported no significant relationship. Of two studies of periodontal diseases (see footnotes to Table Table88 for definitions), one  found no relationship with ST use, but the other  reported a significant increase in current users.
Based on the overall data from the US and Sweden, the evidence of a relationship with snuff or unspecified ST use is weak for gingivitis (or gingival bleeding), where a significant association is reported in only one out of eight studies, and for the more general term periodontal disease, where one of the three studies found a relationship. It is, however, rather stronger for attachment loss, where significant associations were reported in two out of four studies, and particularly for gingival recession, where four out of eight studies reported a significant relationship, and an additional study  found that prevalence was almost nine times higher in loose snuff users than in portion-bag snuff users. Very limited evidence for CT did not demonstrate any relationship with periodontal or gingival diseases.
Table Table99 (Sweden) and Table Table1010 (USA) summarize results relating ST use to presence of teeth and indices of dental caries. They are laid out in a style similar to Tables Tables77 and and8,8, with results grouped according to the endpoints considered. Endpoints in Tables Tables99 and and1010 are named as in the original publications cited here, with the reader referred to these publications for further definition.
Two studies from Sweden [11,20] provided no indication of an effect of snuff use on the number of teeth present. One of these studies  also reported no relationship with having filled teeth. The third study , of 14–19 year olds attending dental check-up, reported significant (p < 0.001) increases in snuff users in the mean number of decayed, missing and filled teeth, of decayed and filled proximal surfaces and of initially decayed proximal surfaces, though not in the mean number of decayed proximal surfaces. It should be noted, however, that these analyses were not adjusted for age, the snuff dippers being markedly older (71% aged 17–19) than the non-tobacco users (52% aged 17–19, p < 0.001). Within specific years of age, significant differences were generally not seen (except for decayed, missing and filled teeth in 17 year olds), but an overall age-adjusted test was unfortunately not reported. The analyses were also not adjusted for education.
Tooth loss in relation to ST use was investigated in three US studies. While one  found no association with ST use, one  found a significant relationship with current snuff but not current CT use, and the third  reported a significant 14% increase in the rate of tooth loss in a large prospective study in which adjustment was made for a range of potential confounding variables, including age and social status.
Dental caries was studied in seven US studies. In six of these [27-29,34,38,44] there was no significant evidence of a relationship with unspecified ST use. The seventh and by far the most comprehensive investigation  was based on the NHANES III study. Results for a range of the endpoints studied are shown in Table Table10.10. No significant difference was seen for snuff use for any index of dental caries (though, as noted above, it was for number of teeth present). In contrast, CT use was significantly associated with decayed or filled permanent teeth, decayed or filled root surfaces, and decayed root surfaces. In an analysis adjusted for age, race/ethnicity, education, and past-year dental visits, the OR for decayed or filled root surfaces was 4.18 (1.96–9.92) for current users of CT only and 0.67 (0.26–1.74) for current users of snuff only, compared to never users of tobacco. In current users of CT, the OR increased with the number of packs per week (trend p = 0.0002) and with years of use (trend p = 0.002). None of the other studies reported dose-response data for dental caries.
The overall evidence is suggestive of a possible relationship of ST use, particularly CT, with the risk of dental caries.
Information relating ST use to oral pain is extremely limited. In the Florida Dental Care Study , 33 adults aged 45+ were current users of CT or snuff, 69 were former users, and 604 had never been users. Based on reports over a 48 month period, analyses were carried out relating current and former ST use to five pain variables (tooth pain, painful gums, temperature sensitivity, activity reduction, and trouble sleeping because of oral pain), with adjustment for race, sex, age, oral hygiene, dental care, and education. Former use was unrelated to any of the five pain variables. For current use, a significant increase was noted for painful gums (OR 1.7, 95% CI 1.2–2.1), but not for the other four variables, with the odds ratio for tooth pain being 1.0 (95% CI 0.4–2.3). Separate results for CT and for snuff were not presented.
No other study directly related ST to the prevalence of oral pain, but a study in Denmark  found that among oral leukoplakia patients, snuff users were less likely to experience pain from the lesion than were non-snuff users (1/32 = 3.1% vs 83/399 = 20.8%, p < 0.01). The North Carolina study  reported that ST use and history of pain both predicted new cases of attachment loss over a three year period but did not relate ST use to pain.
It is not possible to draw any clear conclusion on the relationship between ST use and oral pain from these data.
This report is based on data from 50 studies published between 1963 and 2007 relating the risk of non-neoplastic oral diseases to the consumption of CT and snuff as used in Europe (mainly Scandinavia) and the USA. Six were experimental studies (or had an experimental component); the epidemiological studies were mainly of cross-sectional design. The number of studies considered is substantially larger than in other recent reviews (e.g. [66-71]). Many of the 50 reports have limitations and present less information than is ideal. Problems encountered include small numbers of subjects or exposed cases, unrepresentativeness of the studied populations, inconsistently defined outcomes, and heterogeneous methods of exposure assessment. Exposure details such as type of ST and duration and frequency of use were often not reported. A number of the studies only presented data on populations selected by ST use and/or presence of oral lesions, and therefore, did not allow estimation of prevalences and odds ratios. At the analytical level, one major weakness was incomplete presentation of findings. Another was failure to adjust for important potential confounders, with only a handful of studies adjusting for variables such as age, smoking, education or frequency of dental visits. A third was the frequent failure to present results separately for major subgroups, particularly regarding alternative tobacco consumption, either smoked or smokeless. Thus comparisons of ever snuff users with never snuff users may also compare groups differing on the use of cigarettes and other ST. These shortcomings inevitably limit the inferences that can be drawn as well as the possibility to conduct meaningful meta-analyses. Nevertheless, we feel that it is possible to draw some conclusions from the present results.
It is abundantly clear, as numerous reviewers agree [53-55,59-71], that snuff use markedly increases the risk of oral mucosal lesions. In Scandinavia, users have a near 100% prevalence of the characteristic SIL, a lesion which appears not to occur in non-users, though direct evidence from the available publications is limited to one study . In the USA, the types of lesion studied are much more varied. There many snuff users do not have a lesion and some non-users do, but prevalence is much higher in users than in non-users, with reported odds ratios of up to almost 100 , and frequency dose-related to daily usage. Though strong dose-related relationships are also clearly seen in relation to unspecified ST use, the association between current CT use and lesion prevalence, mainly studied in the USA, is clearly very much weaker.
In both the USA and Scandinavia prevalence of oral mucosal lesions in former users of snuff, CT or unspecified ST is typically low and not clearly higher than in never users, suggesting the reversibility of the effect. This is consistent with the findings from experimental studies where rapid regression upon quitting was observed as well as rapid onset in subjects starting to use ST in the first place or switching to a new habitual oral site. The experimental studies also suggest that the severity (or even the appearance) of the lesion can be affected by switching the type of product used.
The rapid reversibility on quitting ST of SIL, and many of the oral mucosal lesions studied in the USA, suggests that presence of these lesions may not be of great clinical significance. Additional evidence that there is a low probability of SIL transforming into oral cancer comes from the recent follow-up study of Roosaar et al  which found no cases of oral cancers at the site of snuff placement in over 1000 individuals with SIL followed for over 25 years, the low incidence of oral cancer in Sweden, where use of oral snuff is very common , and evidence that the lesions induced may be less susceptible to malignant transformation than those induced by cigarette smoke . While it is well documented that cigarette smoking increases the risk of oral cancer [72,82], a recent review of ST as currently used in Western populations suggests that it carries little or no increased risk of oral cancer .
The evidence available for oral pain is too limited to draw a clear conclusion regarding an association. Somewhat more evidence is available (see Tables Tables55 to to8)8) regarding periodontal diseases, caries and related conditions. Although one study  reported a significantly higher prevalence of gingivitis in snuff users in Sweden, a number of other studies found no relationship of gingivitis or gingival bleeding with the use of ST. Perhaps more indicative of an effect is the evidence relating ST use to gingival recession, attachment loss, dental caries and tooth loss. For all four endpoints, however, the evidence is somewhat inconsistent. From the data available, it is difficult to judge whether this variation is due to differences in type of product used, different definitions of endpoints, or failure to take into account other factors associated with poor dental health, including age, smoking, socio-economic status (education) and the number of visits to the dentist. At present the evidence on attachment loss, dental caries, and tooth loss must be regarded as no more than suggestive of an association with the use of ST.
The evidence relating to gingival recession seems rather stronger with four out of eight studies of snuff or unspecified ST reporting a significant association. There is huge variation in the reported prevalence between studies, marked heterogeneity in the strength of the reported associations in most of the studies reporting a relationship (the extremely high odds ratio in one study  being clearly inconsistent with the other findings), and failure to adjust for relevant potential confounding variables. However, given also that two of the three studies not finding an association were in children or adolescents where reported prevalence was very low indeed, and that the results are supported by an experimental study  showing a clearly lower risk in users of portion-bag than loose snuff, a true effect seems quite probable.
Detailed assessment of the overall risks and benefits of ST use to the public health would require consideration of the whole spectrum of its possible health effects and is beyond the scope of this review. However, we do note that there are numerous reports, including our own publications on oral cancer  and on circulatory disease , which support the risks of smoking-related diseases from ST as being generally much less than those from smoking. This review confirms the strong relationship of oral mucosal lesions to ST use, shows that prevalence and severity is related to the type and amount of the product used, and that the lesion is reversible on quitting. The evidence relating other oral lesions to ST use is less clear. A causal relationship of snuff use with gingival recession seems probable, but not certain. The relationships between CT use and dental caries and between ST use and attachment loss are less clear, and the evidence here may be regarded only as suggestive of a causal relationship. There seems no real indication that ST use affects gingivitis (or gingival bleeding). Data are too limited to draw reliable conclusions for other endpoints, including oral pain.
GK and RW work for Philip Morris International (PMI), R&D. PNL, founder of P.N. Lee Statistics and Computing Ltd., is an independent consultant in statistics and an advisor in the fields of epidemiology and toxicology to a number of tobacco, pharmaceutical, and chemical companies.
GK and RW completed an unpublished review on this subject in 2006. PNL re-examined the literature and extensively revised the review into a form suitable for publication. All authors read and approved the final manuscript.
The pre-publication history for this paper can be accessed here:
All costs of this project were covered by Philip Morris International. This is an independent scientific assessment, and the views expressed are those of the authors.
We thank the referees for comments that have considerably improved this paper, Pauline Wassell and Diane Morris for typing the various drafts of it and Veronique Riccitelli, Sandra Vaucher, and Yvonne Cooper for assistance in obtaining the relevant literature.