Smoking, Smoking Cessation, and Risk of Tooth Loss
Abstract
The aim of this study was to investigate the association between cigarette smoking and smoking cessation and the prevalence and
incidence of tooth loss in a large cohort study in Germany. We analyzed data of participants of the European Prospective
Investigation into Cancer and Nutrition (EPIC)–Potsdam study recruited between 1994 and 1998 from the general population in
Potsdam and other parts of Brandenburg, Germany, who had complete data on cigarette smoking, tooth loss, and covariates. Negative
binomial regression and tooth-specific logistic regression models were fit to evaluate the association between smoking and the baseline
prevalence and incidence of tooth loss during follow-up, respectively. Cigarette smoking was associated with higher prevalence of tooth
loss at baseline as well as higher incidence of tooth loss during follow-up. The association between smoking and the incidence of tooth
loss was stronger in men than women and stronger in younger versus older individuals. Heavy smoking (≥15 cigarettes/d) was associated
with >3 times higher risk of tooth loss in men (odds ratio, 3.6; 95% confidence interval, and more than twice the risk of tooth
loss in women (odds ratio, 2.5; 95% confidence interval) younger than 50 y when compared with never smokers. Smoking
cessation was consistently associated with a reduction in tooth loss risk, with the risk of tooth loss approaching that of never smokers
after approximately 10 to 20 y of cessation.
Introduction
Tooth loss remains to be a major public health problem worldwide . With
about 30% of people aged 65 to 74 y being edentulous globally,
“sustainable improvements in oral health and a reduction in
inequalities may be achieved by controlling the risk factors for
oral diseases”
Cigarette smoking is a major established risk factor for periodontitis and has been linked with higher prevalence of edentulism and fewer remaining teeth in a number of cross-sectional
studiesand with increased rates of tooth loss in
longitudinal studies
However, evidence is scarce from large
epidemiologic studies on the dose-dependent effects of smoking
and smoking cessation on the incidence of tooth loss in the general population .
The purpose of the present study was to evaluate the association between cigarette smoking and smoking cessation and
the prevalence and incidence of tooth loss in a large cohort
study in Germany.
Materials
Study Population
The European Prospective Investigation into Cancer and
Nutrition (EPIC)–Potsdam Study is part of the large-scale
Europe-wide prospective cohort study EPIC and includes
individuals. Details of the
cohort design and recruitment can be found elsewhere. Briefly, participants were recruited between 1994
and 1998 from the general population with the preferred ages of
35 to 65 y in women and 40 to 65 y in men. Participants attended
the study center for a baseline examination, self-administered
questionnaires, and personal computer-guided interviews.
Information on incident diseases and changes in lifestyle or diet
is assessed by mailed follow-up questionnaires Response rates for all follow-up rounds thus far have
exceeded 90% on each occasion. All participants gave written
informed consent, and the Ethics Committee of the Federal State
Brandenburg approved all study procedures. This study conforms to the STROBE guidelines.
Methods
Assessment of Cigarette Smoking
The baseline questionnaire asked detailed questions on the history of cigarette smoking, including the number of cigarettes
smoked per day, smoking duration (in years), and time since
cessation for former smokers (in years). Pack-years of smoking was calculated from these data (smoking duration in
years × numbers of cigarettes smoked per day / 20).
Assessment of Risk Factors and Covariates
Educational attainment was expressed as “vocational school or
less” (equal to or less than a lower education of 10 y of school
education with 2 y of additional professional training), “technical school” (10 y of school followed by >2 y of professional
training), or “university.” Body mass index and waist-to-hip
ratio were calculated from measured body weight and height
and from waist and hip circumferences, respectively.
Hypertension was defined as systolic blood pressure >140 mm
Hg, diastolic blood pressure >90 mm Hg, self-reporting of a
diagnosis, or use of antihypertensive medication. Self-reported
diagnosis of diabetes mellitus at baseline was validated by a
study physician using medical record review. Dietary habits—
including alcohol consumption and intake of vitamins or mineral supplements during the preceding year—were assessed by
a validated self-administered food frequency questionnaire
Assessment of Tooth Loss
In the fourth follow-up questionnaire returned between 2004
and 2006, patients reported the number of natural teeth and the
number of teeth lost since study baseline. We calculated the
baseline number of teeth by adding the number of teeth lost
during the follow-up period to the number of teeth present at
the fourth follow-up.
Data Analysis
With the exception of data on tooth loss, data assessed at study
baseline (1994 to 1998) were used for all analyses. However,
because data on tooth loss were collected only at the fourth
follow-up between 2004 and 2006, we restricted the analyses
to participants who returned the fourth follow-up questionnaire. Of these(1.2%) had to be excluded, as they did
not respond to either of the tooth loss questions, and an additional (0.4%) were excluded because they gave inconsistent responses to the questions on tooth loss. We also excluded
(1.7%) participants with missing data on cigarette smoking and (0.8%) participants with missing data in any of the
covariates.
For the analysis of the association between smoking and
number of teeth at baseline, negative binomial regression models were fit to obtain relative risks and 95% confidence intervals (CIs). These relative risks should be interpreted as the
ratio of the mean number of teeth in different smoking habit
categories, adjusted for covariates. For the analysis of the association between smoking and incidence of tooth loss during
follow-up, tooth-specific logistic models were fit, with tooth
loss as the dependent variable to obtain odds ratios (ORs) and
95% CIs. To account for clustering of teeth within subjects,
generalized estimating equations with an exchangeable correlation matrix were used. Here, ORs should be interpreted as the
relative risk of a tooth being lost as a function of smoking
exposure, adjusted for covariates.
Smoking exposure was modeled via mutually exclusive categories: never smokers, former smokers (<10, 10 to <20, ≥20 y
since cessation), and current smokers (<15 and ≥15 cigarettes/d).
Basic models adjusted for age (baseline prevalence) and age
and follow-up time (incidence during follow-up) and fully
adjusted models adjusted for age, sex, education, diabetes,
body mass index, waist-to-hip ratio, hormone replacement
therapy, contraception, intake of vitamins and mineral supplements, physical activity, alcohol intake, hypertension, and cardiovascular disease at baseline. These variables were included
in the fully adjusted models, as they are possible risk markers
of tooth loss or other causes of tooth loss, including dental disease and health behaviors.
Interaction terms were used to explore prespecified interactions of smoking with sex and 3 age groups (<50, 50 to 59,
≥60 y). For all subsequent analyses, separate models were fit
for men and women.
All statistical analyses were performed with STATA 11
(STATA Corp., College Station, TX, USA).
This report conforms to the STROBE statement.
Results
Characteristics of the Population
The final analytic sample consisted of participants
men and women; mean age at baseline, 50 y;
range, 20 to 70 y), who at baseline had a median of 25 natural
teeth present, with 1,566 (6.7%) being edentulous. The number
of teeth at baseline decreased with increasing age and was
slightly lower for women than men. Men <50 y had a median
of 28 teeth present, compared with 25 and 18 teeth for men
aged 50 to 59 and ≥60 y, respectively; the corresponding figures for women were 27, 23, and 17 teeth. A total of 4,394
(19%) participants were current cigarette smokers, and 7,268
(31%) were former cigarette smokers. Compared with never
smokers, current smokers were more likely to be male, less
educated, more likely to be hypertensive, and less likely to take
vitamins/mineral supplements, and they had higher alcohol
consumption
Cigarette Smoking and Baseline Prevalence
of Tooth Loss
We found a negative dose-dependent association between cigarette smoking and smoking cessation and number of natural
teeth at baseline. However, the association was
stronger among men than women (for interaction, P = 0.0004)
and stronger in older men than in men <50 y (for interaction,
P < 0.0001) as well as in older women than younger women
(for interaction, P = 0.0001).
Men <50 y who had never smoked had a median number of
28 teeth remaining, compared with 25 teeth for current smokers smoking >15 cigarettes/d (Table 2). After multivariate
adjustment, current heavy smokers had on average 13% (95%
CI: 7 to 18) fewer teeth than those who never smoked (Table
2). Men ≥60 y who had never smoked had a median number of
22 teeth remaining, compared with 7 teeth for current heavy
smokers. After multivariable adjustment, current heavy smokers had 48% (95% CI: 42 to 54), and current light smokers had
35% (95% CI: 27 to 42) fewer teeth on average than never
smokers (Table 2).
Female never smokers <50 y had a median number of 27
natural teeth, compared with 25 in those smoking ≥15
cigarettes/d. After multivariate adjustment, when compared
with those who never smoked, current heavy smokers had on
average 12% (95% CI: 7 to 16) fewer teeth and current light
smokers, 7% (95% CI: 4 to 10; Table 3). Women ≥60 y who
had never smoked had a median number of 18 teeth remaining,
compared with 11 teeth for current heavy smokers. After multivariable adjustment, when compared with never smokers,
current heavy smokers had 29% (95% CI: 17 to 39) fewer teeth
on average and current light smokers, 23% (95% CI: 16 to 30;
Among men and women, participants who had quit smoking
prior to baseline had lost more teeth than never smokers but had
on average more teeth remaining than current smokers. There
was a positive association between time since cessation and
number of remaining teeth across all subgroups; however, in
particular among older participants, former smokers had fewer
remaining teeth even after 10 y of abstinence
Cigarette Smoking and Incidence of Tooth Loss
The 21,810 participants who had at least 1 natural tooth
remaining at baseline were followed for a mean of 8.6 y (range,
6.8 to 13.1 y). Over the course of follow-up, 32,478 (6.4%) of
the 509,700 remaining teeth at baseline were lost, and 3,843
men (46%) and 5,707 women (42%) lost at least 1 tooth.
We found a positive dose- and time-dependent association
between baseline cigarette smoking and incidence of tooth loss
during follow-up. The association was stronger
among men than women (for interaction, P = 0.02), and significant effect measure modification by age was found among
men (for interaction, P = 0.02) and women (for interaction, P =
0.002). For men <50 y, each tooth in current heavy smokers
had 3 to 4 times higher odds of being lost during follow-up as
compared with never smokers (OR = 3.6; 95% CI: 3.0 to 4.4).
For men ≥60 y, teeth of current heavy smokers had >2 times
higher odds (OR = 2.5; 95% CI: 1.8 to 3.3) of being lost when
compared with teeth of never smokers . For women
<50 y, teeth of current heavy smokers (OR = 2.5; 95% CI: 2.1
to 2.9) and current light smokers (OR = 1.7; 95% CI: 1.5 to 2.0)
were more likely to be lost than those of never smokers. Similar
but somewhat weaker associations were observed among
women ≥60 y
Among men and women, teeth of individuals who had quit
smoking prior to baseline had a higher risk of being lost than
that of the teeth of never smokers. There was a negative association between time since cessation and risk of tooth loss
across all subgroups, with the risk of tooth loss approaching
that in never smokers after about 10 to 20 y of abstinence
.
Discussion
In this large population-based prospective cohort study of men
and women in the state of Brandenburg, Germany, we found a
strong dose-dependent association between cigarette smoking
and the prevalence and incidence of tooth loss, independent of
other risk factors and potential confounders. The risk of tooth
loss declined after smoking cessation as a function of time
since cessation; however, compared with that of never smokers, the risk of tooth loss remained elevated until approximately 20 y after smoking cessation. The association between
smoking and tooth loss was stronger in men than in women and
was stronger in younger compared with older men and women.
Periodontitis is likely to be an important intermediate factor
in the association between smoking and tooth loss observed in
this study. However, there is emerging evidence that cigarette
smoking may also be a risk factor for dental caries
and periapical pathosis
Following smoking cessation, there is a time-dependent
reduction of the smoking-associated excess risk of tooth loss,
with marked benefits of cessation evident within the first 10 y
after smoking cessation. However, it may take up to 20 y for
the risk to decline to that of never smokers. These results are
largely consistent with the findings from previous studies in
the United States and Australia
The observed sex differences are intriguing. In the present
study, stronger associations were consistently observed among
men compared with women. Women had more missing teeth at
baseline and a higher incidence of tooth loss during follow-up compared with men These latter findings are consistent with data from another large German
population based cohort study, where more tooth loss was
observed in women compared with men, even though, paradoxically, the prevalence of periodontitis was lower in women
than men . This illustrates that incident
tooth loss is not simply the end point of dental disease but is
also affected by factors related to dental care, including treatment preferences and ability/willingness to pay for dental
treatment
Although this is somewhat speculative, the higher relative
risks associated with smoking among men compared with
women may thus be merely the reflection of the lower background incidence of tooth loss as well as the higher risk of
periodontitis in men, rendering men—in relative terms—more
susceptible to the effects of smoking. However, this finding
may not generalize to other populations. For example, a study
of young adults in Finland reported a lower prevalence of tooth
loss and a stronger association between smoking and prevalent
tooth loss in women compared with men
Important strengths of this study are its population-based
prospective design, large sample size, and detailed data on
smoking history, allowing for fine exposure categories and precise estimates.
Considering that tooth loss is the outcome of a
complex process that may involve numerous factors that may
also be related to cigarette smoking
another strength of this study is our ability to control for a variety of potential confounders. Furthermore, the population is
comparatively homogeneous in terms of race and access to
dental care, as the study population is almost exclusively
Caucasian and dental care, including restorative dental care, is
included in the statutory general health insurance. A limitation
of this study is the lack of detailed data on oral health status,
oral hygiene habits, and dental care utilization, limiting our
ability to explore the mechanistic link between smoking and
tooth loss. Furthermore, there is the possibility that the effect
estimates in this study are confounded by differences in oral
hygiene status or oral hygiene habits between smokers and
nonsmokers. However, previous studies have failed to consistently demonstrate marked differences in oral hygiene between
smokers and nonsmokers . As
with any observational study, there is the possibility of residual
confounding and confounding by other unknown or unmeasured confounders, which could result in over- or underestimation of the causal association of cigarette smoking and tooth
loss. Assessment of tooth loss was based on individuals’ selfreport, which has been demonstrated to have high validity in
several studies in various populations ;
in terms of number of
teeth present as well as incident tooth loss. However, some
measurement error is likely to have occurred, given that in our
study respondents had to recall tooth loss over approximately a
10-y period. While it may be reasonable to assume that such
measurement error would have been independent of a subject’s
smoking status, this may have resulted in an attenuation of
relative risk estimates. Last, the results may not be generalizable to other populations, in particular to populations with a
different racial/ethnic composition or markedly different reasons for tooth loss or access to and/or utilization of dental care.
In conclusion, there is a strong dose-dependent association
between cigarette smoking and the risk of tooth loss. The risk
declines after cessation of cigarette smoking; however, the risk
may remain elevated for up to 20 y compared with never smokers. Efforts to improve the oral health of the population should
include the prevention of smoking as well the promotion of
smoking cessation.