A study of dose-response relationship between tobacco habits and oral leukoplakia

In a house-to-house survey in Ernakulam district, Kerala, India, 12,213 tobacco users were interviewed about the details of their tobacco usage and examined for the presence of leukoplakia. The frequency of tobacco habit was associated with the prevalence of leukoplakia indicating a positive doseresponse relationship. The dose-response relationship remained significant, taking age, sex, and the type of tobacco habit into account. After adjusting for all these variables jointly the association still remained significant. The dose-response relationship was stronger for the smoking habit than for the chewing habit. A weaker relationship in the chewing habit was not due to the duration of chewing habit or the habit of retaining the betel quid in the mouth while sleeping. Thus the dose-response relationship, although significant,

significant. The dose-response relationship was stronger for the smoking habit than for the chewing habit. A weaker relationship in the chewing habit was not due to the duration of chewing habit or the habit of retaining the betel quid in the mouth while sleeping. Thus the dose-response relationship, although significant, was different for tobacco smoking and chewing habits.
The association between oral leukoplakia and Subjects and methods tobacco habits is well established in numerous epidemiologic studies. The association has generally The district of Ernakulam in Kerala State was been found to abits of chosen for this study as the habits of chewing betel tobaCco _ quid and smoking bidis were known to bẽ principal aetiologic factors for oral l op!akia_ widespread in this district. In a house-to-house (Pindborg. 1980). survey 12,213 tobacco users aged 15 years and If the association is examined as a causal one above were interviewed about their tobacco habits according to the established principles in and examined for the presence of oral leukoplakia. epidemiology the hypothesis is confirmed on almost Leukoplakia was defined as a raised white or all counts for which the data are available; for greyish white patch 5 mm or more in diameter example, the association is biologically plausible, which could not be rubbed off and could not be has been found to be quite consistent in different attributed to any other diagnosable disease. This population groups (Pindborg et al., 1968; definition did not carry any histological  al. , 1969Banoczy, 1980), as well as in connotation. The methodology of the survey was different studies in similar population groups the same as that given by Mehta et al. (1969). (Pindborg et al., 1967;Mehta et al., 1969; In the study sample bidi smoking and pan Petersen et al., 1972), is confirmed through chewing were the most common forms of tobacco prospective studies Gupta et usage. Bidi is a cheap smoking stick made by al., 1980) and when tobacco habits are discontinued rolling a dried piece of temburni leaf (Diospyrous a significant increase in the regression of melanoxylon) into a conical shape and securing the leukoplakia is observed . roll with thread. The length of a bidi varies from 4 An important criterion for examining the causal to 8 cm and it contains 0.15 to 0.25 g of coarse hypothesis is the relationship between the disease powdered tobacco. Pan is a quid consisting of betel and the degree of exposure to the risk factor or the leaf, arecanut, lime (calcium hydroxide) and dose-response relationship. Some studies have tobacco. The usage of cigarette and tobacco alone reported the dose-response relationship between or tobacco with lime was infrequent (<7%) and oral leukoplakia and tobacco habits, (Dayal et al., therefore in this paper these habits have not been 1978; Baric et al., 1982) however, they have not categorized separately. controlled for confounding variables in their The tobacco users were asked about the duration analysis. In this paper, the dose-response and the frequency of their tobacco habit. The relationship between oral leukoplakia and tobacco chewers were also questioned about the habit of habits is investigated controlling the effects of retaining the betel quid in the mouth while retiring confounding variables. for sleep. For these who smoked as well as chewed details about both h'abits were recorded. The duration of tobacco habit was defined as the using tobacco for smoking or chewing. The frequency of tobacco habit was defined as the number of bidis smoked per day for smokers and the number of betel toabcco quids chewed per day for chewers. For simplicity of presentation, at times frequency data for smokers and chewers were combined. This, however, did not imply any assumption of equivalence between the dose represented by bidi and betel tobacco quid.
Relative risks of dose-response were calculated by dividing the prevalence of leukoplakia in the higher frequency group by the prevalence of leukoplakia in the lower frequency group.

Results
Among 12,213 individuals examined, 10,490 practised a single habit of either chewing or smoking and the rest of 1,723 practised chewing as well as smoking habits. Table I shows the distribution of the frequency of tobacco habit for 10,490 individuals who practised a single smoking or chewing habit and the prevalence of leukoplakia per 1,000. There is a clear and significant increase in prevalence with increase in the frequency of tabacco habit. To simplify further analysis and to avoid the problem of small numbers only two frequency classes: 1-10 and 11 and above, are given in the subsequent tables. x2=45.9, df=3. Table II shows the age distribution of frequency of tobacco habits, prevalence of leukoplakia, and relative risks. Lower frequency was more common in older individuals and higher frequency was more common among younger individuals. For each age group, however, the prevalence of leukoplakia was significantly higher in the frequency group 11 and above compared to the frequency group 1 to 10 showing that the relative risks were significant. Table III shows the distribution of the frequency of tobacco habit, the prevalence of leukoplakia and known that the occurrence of leukoplakia is also strongly associated with these three variables (Pindborg, 1980) and among these three variables age and the type of tobacco habit and sex and the  (Mehta et al., 1969). To eliminate possible confounding effects of there relationships on the association of leukoplakia and frequency of tobacco habit, Table IV shows the age adjusted prevalence of leukoplakia and the relative risk according to sex and type of tobacco habit. For males who smoked the relative risk (5.0) was highly significant and those those who chewed the relative risk (1.8) was just significant. For females who smoked, sufficient observations were not available and for those who chewed the relative risk (1.7) was not significant. Thus the dose-response relationship appeared to be stronger for smoking habit than for chewing habit.
To probe this phenomenon further, prevalence of leukoplakia was analysed by another component of dose for chewers, the habit of retaining the betel quid in the mouth while sleeping (Table V). It is clear that for females as well as for males there was no significant difference in prevalences.   Table VI) did not reveal any pattern either. In the study sample most individuals tended to start their tobacco habits at similar age resulting in a high correlation between age and the duration of the habit (correlation coefficient 0.6). Adjusting for age, therefore, also adjusted for the duration of the tobacco habit to a considerable extent. Another possible reason for a higher relative risk of dose response among smokers could be a more accurate assessment of frequency by smokers compared to chewers. The relative risks of dose response was therefore computed for 1,723 individuals who smoked as well as chewed. Among smokers of 1 to 10 bidis per day the relative risk for chewing over 10 quids was 1.1 (P<0.05). and among chewers of 1-10 quids the relative risk for smoking of over 10 bidis was 2.0 (P<0.05).

Discussion
In general "dose" or a measure of the degree of exposure to a risk factor can consist of several components. In the context of the present study, apart from the frequency and the duration of tobacco habit, several other components, specific to the type of tobacco habit, can be considered as representing, or at least affecting the dose. For example, for chewing habit, the duration for which a quid is kept in the mouth, amount of tobacco in a quid, the type of tobacco used, and for smoking habits, the degree of inhalation, the frequency of puffs, the left-over length of the butt, etc. could be considered as important components of dose in different circumstances. In the present study although information was collected on some of these components either there was not enough variation to justify separate analysis (type of tobacco by local names) or the information was not considered reliable enough (duration of keeping the quid in the mouth). Information on other aspects could not be collected. It is unlikely that the duration of keeping the quid would show any significant difference for the risk of leukoplakia because the habit of retaining the quid in the mouth while sleeping which effectively categorises the duration of keeping quid to less than 8 hours and more than 8 hours, did not show any difference.
The study shows that the dose-response relationship between leukoplakia and tobacco habits is significant after adjusting for age, sex and the type of tobacco habit. The striking result from this study, however, is that the dose-response relationship is stronger for smokers compared to chewers, and the difference is not attributable to the duration of chewing habit, habit of retaining quid in the mouth while sleeping, or better recall of frequency by smokers compared to chewers. The difference in the strength of the dose-response relationship could not be attributed to the choice of the cut-off point either.
It is interesting that another study from India which reported on the association between prevalence of leukoplakia and the frequency and the duration of chewing habit (Dayal et al., 1978) showed an increasing trend in the prevalence with increase in the duration and increase in the frequency. No statistical tests of significance were reported.
The stronger dose-response relationship for smokers than for chewers although, inexplicable, may not be surprising. It has been reported before that leukoplakia associated with smoking habit and leukoplakia associated with chewing behave differently with regard to incidence, spontaneous regression and malignant transformation (Mehta et al., 1981). It is therefore feasible that the two types of tobacco habit should show different results for dose-response relationships.